CLREX. STMicroelectronics STM32H7S3L8, STM32H7R3L8, STM32H7S7A8, STM32H7R3V8, STM32H7R3R8, STM32H7S3Z8, STM32H7S3I8, STM32H7S3V8, STM32H7S7Z8, STM32H7R7I8
Oglas
Oglas
The Cortex-M7 instruction set PM0253
Restrictions
In these instructions:
• Do not use PC.
• Do not use SP for Rd and Rt .
• For STREX, Rd must be different from both Rt and Rn .
• The value of offset must be a multiple of four in the range 0-1020.
Condition flags
These instructions do not change the flags.
Examples
MOV R1, #0x1 ; Initialize the ‘lock taken’ value try
LDREX R0, [LockAddr] ; Load the lock value
CMP R0, #0 ; Is the lock free?
ITT EQ ; IT instruction for STREXEQ and CMPEQ
STREXEQ R0, R1, [LockAddr] ; Try and claim the lock
CMPEQ R0, #0 ; Did this succeed?
BNE try ; No – try again
.... ; Yes – we have the lock.
3.4.10 CLREX
Clear Exclusive.
Syntax
CLREX{ cond }
Where: cond
Is an optional condition code. See Conditional execution on page 68 .
Operation
Use CLREX to make the next STREX, STREXB, or STREXH instruction write 1 to its destination register and fail to perform the store. It is useful in exception handler code to force the failure of the store exclusive if the exception occurs between a load exclusive instruction and the matching store exclusive instruction in a synchronization operation.
See
Synchronization primitives on page 37
for more information.
Condition flags
This instruction does not change the flags.
Examples
CLREX
84/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
3.5 General processing instructions
Table 30 shows the data processing instructions:
Mnemonic
Table 30. Data processing instructions
Brief description
LSR
MOV
MOVT
MOVW
MVN
ORN
ORR
RBIT
REV
REV16
ADC
ADD
ADDW
AND
ASR
BIC
CLZ
CMN
CMP
EOR
LSL
REVSH
Add with Carry
Add
Add
Logical AND
Arithmetic Shift Right
Bit Clear
Count leading zeros
Compare Negative
Compare
Exclusive OR
Logical Shift Left
Logical Shift Right
Move
Move Top
Move 16-bit constant
Move NOT
Logical OR NOT
Logical OR
Reverse Bits
Reverse byte order in a word
Reverse byte order in each halfword
Reverse byte order in bottom halfword and sign extend
ROR
RRX
RSB
SADD16
SADD8
SASX
Rotate Right
Rotate Right with Extend
Reverse Subtract
Signed Add 16
Signed Add 8
Signed Add and Subtract with Exchange
SSAX
SBC
Signed Subtract and Add with Exchange
Subtract with Carry
SHADD16 Signed Halving Add 16
See
ADD, ADC, SUB, SBC, and RSB on page 87
ADD, ADC, SUB, SBC, and RSB on page 87
ADD, ADC, SUB, SBC, and RSB on page 87
AND, ORR, EOR, BIC, and ORN on page 89
ASR, LSL, LSR, ROR, and RRX on page 90
AND, ORR, EOR, BIC, and ORN on page 89
AND, ORR, EOR, BIC, and ORN on page 89
ASR, LSL, LSR, ROR, and RRX on page 90
ASR, LSL, LSR, ROR, and RRX on page 90
AND, ORR, EOR, BIC, and ORN on page 89
AND, ORR, EOR, BIC, and ORN on page 89
REV, REV16, REVSH, and RBIT on page 95
REV, REV16, REVSH, and RBIT on page 95
REV, REV16, REVSH, and RBIT on page 95
REV, REV16, REVSH, and RBIT on page 95
ASR, LSL, LSR, ROR, and RRX on page 90
ASR, LSL, LSR, ROR, and RRX on page 90
ADD, ADC, SUB, SBC, and RSB on page 87
ADD, ADC, SUB, SBC, and RSB on page 87
PM0253 Rev 5 85/254
182
The Cortex-M7 instruction set PM0253
Mnemonic
Table 30. Data processing instructions (continued)
Brief description See
SHADD8
SHASX
Signed Halving Add 8
Signed Halving Add and Subtract with Exchange
SHSAX Signed Halving Subtract and Add with Exchange
SHSUB16 Signed Halving Subtract 16
SHSUB8
SSUB16
SSUB8
SUB
Signed Halving Subtract 8
Signed Subtract 16
Signed Subtract 8
Subtract
ADD, ADC, SUB, SBC, and RSB on page 87
SUBW
TEQ
TST
UADD16
UADD8
UASX
USAX
Subtract
Test Equivalence
Test
Unsigned Add 16
Unsigned Add 8
Unsigned Add and Subtract with Exchange
Unsigned Subtract and Add with Exchange
ADD, ADC, SUB, SBC, and RSB on page 87
UHADD16 Unsigned Halving Add 16
UHADD8 Unsigned Halving Add 8
UHASX
UHADD16 and UHADD8 on page 105
UHADD16 and UHADD8 on page 105
Unsigned Halving Add and Subtract with Exchange
UHSAX Unsigned Halving Subtract and Add with Exchange
UHSUB16 Unsigned Halving Subtract 16
UHSUB16 and UHSUB8 on page 107
UHSUB8
USAD8
USADA8
Unsigned Halving Subtract 8
Unsigned Sum of Absolute Differences
Unsigned Sum of Absolute Differences and
Accumulate
UHSUB16 and UHSUB8 on page 107
USUB16
USUB8
Unsigned Subtract 16
Unsigned Subtract 8
86/254 PM0253 Rev 5
PM0253
3.5.1
The Cortex-M7 instruction set
ADD, ADC, SUB, SBC, and RSB
Add, Add with carry, Subtract, Subtract with carry, and Reverse Subtract.
Syntax op {S}{ cond } { Rd ,} Rn , Operand2 op { cond } { Rd ,} Rn , # imm12 ; ADD and SUB only
Where: op
S cond
Rd
Rn
Operand2 imm12
Is one of:
ADD
Add.
ADC
SUB
Add with Carry.
Subtract.
SBC
RSB
Subtract with Carry.
Reverse Subtract.
Is an optional suffix. If
S
is specified, the condition code flags are updated on the result of the operation, see
Conditional execution on page 68
.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register. If
Rd
is omitted, the destination register is
Rn
.
Is the register holding the first operand.
Is a flexible second operand. See
Flexible second operand on page 64
for details of the options.
Is any value in the range 0-4095.
Operation
The ADD instruction adds the value of Operand2 or imm12 to the value in Rn .
The ADC instruction adds the values in Rn and Operand2 , together with the carry flag.
The SUB instruction subtracts the value of Operand2 or imm12 from the value in Rn .
The SBC instruction subtracts the value of Operand2 from the value in Rn . If the carry flag is clear, the result is reduced by one.
The RSB instruction subtracts the value in Rn from the value of Operand2 . This is useful because of the wide range of options for Operand2 .
Use ADC and SBC to synthesize multiword arithmetic, see
Multiword arithmetic examples on page 88
.
.
ADDW is equivalent to the ADD syntax that uses the imm12 operand. SUBW is equivalent to the SUB syntax that uses the imm12 operand.
PM0253 Rev 5 87/254
182
The Cortex-M7 instruction set PM0253
Restrictions
In these instructions:
• Operand2 must not be SP and must not be PC
• Rd can be SP only in ADD and SUB, and only with the additional restrictions:
– Rn must also be SP.
– Any shift in Operand2 must be limited to a maximum of 3 bits using LSL.
• Rn can be SP only in ADD and SUB
• Rd can be PC only in the ADD{ cond } PC, PC, Rm instruction where:
– The S suffix must not be specified.
– Rm must not be PC and must not be SP.
– If the instruction is conditional, it must be the last instruction in the IT block.
• With the exception of the ADD{ cond } PC, PC, Rm instruction, Rn can be PC only in
ADD and SUB, and only with the additional restrictions:
– The S suffix must not be specified.
– The second operand must be a constant in the range 0 to 4095.
– When using the PC for an addition or a subtraction, bits[1:0] of the PC are rounded to 0b00 before performing the calculation, making the base address for the calculation word-aligned.
– If the user wants to generate the address of an instruction, the constant has to be adjusted based on the value of the PC. Arm recommends using the ADR instruction instead of ADD or SUB with Rn equal to the PC, because the assembler automatically calculates the correct constant for the ADR instruction.
When Rd is PC in the ADD{ cond } PC, PC, Rm instruction:
• Bit[0] of the value written to the PC is ignored.
• A branch occurs to the address created by forcing bit[0] of that value to 0.
Condition flags
If
S
is specified, these instructions update the N, Z, C and V flags according to the result.
Examples
ADD R2, R1, R3
SUBS R8, R6, #240 ; Sets the flags on the result.
RSB R4, R4, #1280 ; Subtracts contents of R4 from 1280.
ADCHI R11, R0, R3 ; Only executed if C flag set and Z.
; flag clear.
Multiword arithmetic examples
Example 3-4: 64-bit addition shows two instructions that add a 64-bit integer contained in R2
and R3 to another 64-bit integer contained in R0 and R1, and place the result in R4 and R5.
Example 3-4: 64-bit addition
ADDS R4, R0, R2 ; Add the least significant words.
ADC R5, R1, R3 ; Add the most significant words with carry
.
The multiword values do not have to use consecutive registers. Example 3-5: 96-bit
88/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
shows instructions that subtract a 96-bit integer contained in R9, R1, and R11 from another contained in R6, R2, and R8. The example stores the result in R6, R9, and R2.
Example 3-5: 96-bit subtraction
SUBS R6, R6, R9 ; Subtract the least significant words.
SBCS R9, R2, R1 ; Subtract the middle words with carry.
SBC R2, R8, R11 ; Subtract the most significant words with
; carry.
Logical AND, OR, Exclusive OR, Bit Clear, and OR NOT.
Syntax op {S}{ cond } { Rd ,} Rn , Operand2
Where: op
S cond
Rd
Rn
Operand2
Is one of:
AND
Logical AND.
ORR
EOR
Logical OR, or bit set.
Logical Exclusive OR.
BIC
ORN
Logical AND NOT, or bit clear.
Logical OR NOT.
Is an optional suffix. If
S
is specified, the condition code flags are updated on the result of the operation, see
Conditional execution on page 68
.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the register holding the first operand.
Is a flexible second operand. See
Flexible second operand on page 64
for details of the options.
Operation
The AND, EOR, and ORR instructions perform bitwise AND, Exclusive OR, and OR operations on the values in Rn and Operand2 .
The BIC instruction performs an AND operation on the bits in Rn with the complements of the corresponding bits in the value of Operand2 .
The ORN instruction performs an OR operation on the bits in Rn with the complements of the corresponding bits in the value of Operand2 .
Restrictions
Do not use SP and do not use PC.
PM0253 Rev 5 89/254
182
The Cortex-M7 instruction set
3.5.3
PM0253
Condition flags
If S is specified, these instructions:
• Update the N and Z flags according to the result.
• Can update the C flag during the calculation of Operand2 , see
Example 3-5: 96-bit subtraction on page 89
.
• Do not affect the V flag.
Examples
AND R9, R2, #0xFF00
ORREQ R2, R0, R5
ANDS R9, R8, #0x19
EORS R7, R11, #0x18181818
BIC R0, R1, #0xab
ORN R7, R11, R14, ROR #4
ORNS R7, R11, R14, ASR #32
ASR, LSL, LSR, ROR, and RRX
Arithmetic Shift Right, Logical Shift Left, Logical Shift Right, Rotate Right, and Rotate Right with Extend.
Syntax op {S}{ cond } Rd , Rm , Rs op {S}{ cond } Rd , Rm , # n
RRX{S}{ cond } Rd , Rm
Where: op
S
Rd
Rm
Rs n
Is one of:
ASR
Arithmetic Shift Right.
LSL
LSR
ROR
Logical Shift Left.
Logical Shift Right.
Rotate Right.
Is an optional suffix. If
S
is specified, the condition code flags are updated on the result of the operation, see
Conditional execution on page 68
.
Is the destination register.
Is the register holding the value to be shifted.
Is the register holding the shift length to apply to the value in
Rm
. Only the least significant byte is used and can be in the range 0 to 255.
Is the shift length. The range of shift length depends on the instruction:
ASR
Shift length from 1 to 32
LSL
LSR
ROR
Shift length from 0 to 31
Shift length from 1 to 32
Shift length from 1 to 31
MOVS Rd, Rm
is the preferred syntax for
LSLS Rd, Rm, #0
.
90/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Operation
ASR, LSL, LSR, and ROR move the bits in the register Rm to the left or right by the number of places specified by constant n or register Rs .
RRX moves the bits in register Rm to the right by 1.
In all these instructions, the result is written to Rd , but the value in register Rm remains
.
Restrictions
Do not use SP and do not use PC.
Condition flags
If S is specified:
• These instructions update the N and Z flags according to the result.
• The C flag is updated to the last bit shifted out, except when the shift length is 0, see
.
Examples
ASR R7, R8, #9 ; Arithmetic shift right by 9 bits.
LSLS R1, R2, #3 ; Logical shift left by 3 bits with flag update.
LSR R4, R5, #6 ; Logical shift right by 6 bits.
ROR R4, R5, R6 ; Rotate right by the value in the bottom byte of
; R6.
RRX R4, R5 ; Rotate right with extend.
3.5.4 CLZ
Count Leading Zeros.
Syntax
CLZ{ cond } Rd , Rm
Where: cond
Rd
Rm
Is an optional condition code. See
Is the destination register.
Is the operand register.
Conditional execution on page 68 .
Operation
The CLZ instruction counts the number of leading zeros in the value in Rm and returns the result in Rd . The result value is 32 if no bits are set and zero if bit[31] is set.
Restrictions
Do not use SP and do not use PC.
Condition flags
This instruction does not change the flags.
PM0253 Rev 5 91/254
182
The Cortex-M7 instruction set
3.5.5
PM0253
Examples
CLZ R4,R9
CLZNE R2,R3
CMP and CMN
Compare and Compare Negative.
Syntax
CMP{ cond } Rn , Operand2
CMN{ cond } Rn , Operand2
Where: cond
Rn
Operand2
Is an optional condition code. See Conditional execution on page 68 .
Is the register holding the first operand.
Is a flexible second operand. See
Flexible second operand on page 64
for details of the options.
Operation
These instructions compare the value in a register with Operand2 . They update the condition flags on the result, but do not write the result to a register.
The CMP instruction subtracts the value of Operand2 from the value in Rn . This is the same as a SUBS instruction, except that the result is discarded.
The CMN instruction adds the value of Operand2 to the value in Rn . This is the same as an
ADDS instruction, except that the result is discarded.
Restrictions
In these instructions:
• Do not use PC.
• Operand2 must not be SP.
Condition flags
These instructions update the N, Z, C and V flags according to the result.
Examples
CMP R2, R9
CMN R0, #6400
CMPGT SP, R7, LSL #2
92/254 PM0253 Rev 5
PM0253
3.5.6
The Cortex-M7 instruction set
MOV and MVN
Move and Move NOT.
Syntax
MOV{S}{ cond } Rd , Operand2
MOV{ cond } Rd , # imm16
MVN{S}{ cond } Rd , Operand2
Where:
S cond
Rd
Operand2 imm16
Is an optional suffix. If
S
is specified, the condition code flags are updated on the result of the operation, see
Conditional execution on page 68
.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is a flexible second operand. See
Flexible second operand on page 64
for details of the options.
Is any value in the range 0-65535.
Operation
The MOV instruction copies the value of Operand2 into Rd .
When Operand2 in a MOV instruction is a register with a shift other than LSL #0, the preferred syntax is the corresponding shift instruction:
•
ASR{S}{cond} Rd, Rm, #n
is the preferred syntax for
MOV{S}{cond} Rd, Rm, ASR
#n
.
•
LSL{S}{cond} Rd, Rm, #n
is the preferred syntax for
MOV{S}{cond} Rd, Rm, LSL
#n if n != 0
.
•
LSR{S}{cond} Rd, Rm, #n
is the preferred syntax for
MOV{S}{cond} Rd, Rm, LSR
#n
.
•
ROR{S}{cond} Rd, Rm, #n
is the preferred syntax for
MOV{S}{cond} Rd, Rm, ROR
#n
.
•
RRX{S}{cond} Rd, Rm
is the preferred syntax for
MOV{S}{cond} Rd, Rm, RRX
.
Also, the MOV instruction permits additional forms of Operand2 as synonyms for shift instructions:
•
MOV{S}{cond} Rd, Rm, ASR Rs
is a synonym for
ASR{S}{cond} Rd, Rm, Rs
.
•
MOV{S}{cond} Rd, Rm, LSL Rs
is a synonym for
LSL{S}{cond} Rd, Rm, Rs
.
•
MOV{S}{cond} Rd, Rm, LSR Rs
is a synonym for
LSR{S}{cond} Rd, Rm, Rs
.
•
MOV{S}{cond} Rd, Rm, ROR Rs
is a synonym for
ROR{S}{cond} Rd, Rm, Rs
See
ASR, LSL, LSR, ROR, and RRX on page 90 .
The MVN instruction takes the value of Operand2 , performs a bitwise logical NOT operation on the value, and places the result into Rd .
The MOVW instruction provides the same function as MOV, but is restricted to using the imm16 operand.
PM0253 Rev 5 93/254
182
The Cortex-M7 instruction set PM0253
Restrictions
The user can use SP and PC only in the MOV instruction, with the following restrictions:
• The second operand must be a register without shift.
• The S suffix must not be specified.
When Rd is PC in a MOV instruction:
• Bit[0] of the value written to the PC is ignored.
• A branch occurs to the address created by forcing bit[0] of that value to 0.
Though it is possible to use MOV as a branch instruction, Arm strongly recommends the use of a BX or BLX instruction to branch for software portability to the Arm instruction set.
Condition flags
If S is specified, these instructions:
• Update the N and Z flags according to the result.
• Can update the C flag during the calculation of Operand2
, see Flexible second operand on page 64 .
• Do not affect the V flag.
Example
MOVS R11, #0x000B ; Write value of 0x000B to R11, flags get
; updated.
MOV R1, #0xFA05 ; Write value of 0xFA05 to R1, flags are not
; updated.
MOVS R10, R12 ; Write value in R12 to R10, flags get updated.
MOV R3, #23 ; Write value of 23 to R3.
MOV R8, SP ; Write value of stack pointer to R8.
MVNS R2, #0xF ; Write value of 0xFFFFFFF0 (bitwise inverse of
; 0xF) to the R2 and update flags.
3.5.7 MOVT
Move Top.
Syntax
MOVT{ cond } Rd , # imm16
Where: cond
Rd imm16
Is an optional condition code. See
Is the destination register.
Is a 16-bit immediate constant.
Conditional execution on page 68 .
Operation
MOVT writes a 16-bit immediate value, imm16 , to the top halfword, Rd [31:16], of its destination register. The write does not affect Rd [15:0].
The MOV, MOVT instruction pair enables to generate any 32-bit constant.
Restrictions
Rd must not be SP and must not be PC.
94/254 PM0253 Rev 5
PM0253
3.5.8
The Cortex-M7 instruction set
Condition flags
This instruction does not change the flags.
Examples
MOVT R3, #0xF123 ; Write 0xF123 to upper halfword of R3, lower
; halfword and APSR are unchanged.
REV, REV16, REVSH, and RBIT
Reverse bytes and Reverse bits.
Syntax op { cond } Rd , Rn
Where: op cond
Rd
Rn
Is one of:
REV
Reverse byte order in a word.
REV16
Reverse byte order in each halfword independently.
REVSH
Reverse byte order in the bottom halfword, and sign extend to 32 bits.
RBIT
Reverse the bit order in a 32-bit word.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the register holding the operand.
Operation
Use these instructions to change endianness of data:
REV
•
• converts either:
32-bit big-endian data into little-endian data.
32-bit little-endian data into big-endian data.
REV16
REVSH
•
• converts either:
16-bit big-endian data into little-endian data.
16-bit little-endian data into big-endian data.
•
• converts either:
16-bit signed big-endian data into 32-bit signed little-endian data.
16-bit signed little-endian data into 32-bit signed big-endian data.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not change the flags.
PM0253 Rev 5 95/254
182
The Cortex-M7 instruction set
3.5.9
PM0253
Examples
REV R3, R7 ; Reverse byte order of value in R7 and write it to R3.
REV16 R0, R0 ; Reverse byte order of each 16-bit halfword in R0.
REVSH R0, R5 ; Reverse Signed Halfword.
REVHS R3, R7 ; Reverse with Higher or Same condition.
RBIT R7, R8 ; Reverse bit order of value in R8 and write the result
; to R7.
SADD16 and SADD8
Signed Add 16 and Signed Add 8.
Syntax op { cond }{ Rd ,} Rn , Rm
Where: op cond
Rd
Rn
Rm
Is one of:
SADD16
Performs two 16-bit signed integer additions.
SADD8
Performs four 8-bit signed integer additions.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the first register holding the operand.
Is the second register holding the operand.
Operation
Use these instructions to perform a halfword or byte add in parallel.
The SADD16 instruction:
1.
Adds each halfword from the first operand to the corresponding halfword of the second operand.
2. Writes the result in the corresponding halfwords of the destination register.
The SADD8 instruction:
1.
Adds each byte of the first operand to the corresponding byte of the second operand.
2. Writes the result in the corresponding bytes of the destination register.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not change the flags.
Examples
SADD16 R1, R0 ; Adds the halfwords in R0 to the corresponding
; halfwords of R1 and writes to corresponding halfword
; of R1.
SADD8 R4, R0, R5 ; Adds bytes of R0 to the corresponding byte in R5 and
; writes to the corresponding byte in R4.
96/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
3.5.10 SHADD16 and SHADD8
Signed Halving Add 16 and Signed Halving Add 8.
Syntax op { cond }{ Rd ,} Rn , Rm
Where: op cond
Rd
Rn
Rm
Is one of:
SHADD16
Signed Halving Add 16
SHADD8
Signed Halving Add 8
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the first operand register.
Is the second operand register.
Operation
Use these instructions to add 16-bit and 8-bit data and then to halve the result before writing the result to the destination register.
The SHADD16 instruction:
1.
Adds each halfword from the first operand to the corresponding halfword of the second operand.
2. Shuffles the result by one bit to the right, halving the data.
3. Writes the halfword results in the destination register.
The SHADDB8 instruction:
1.
Adds each byte of the first operand to the corresponding byte of the second operand.
2. Shuffles the result by one bit to the right, halving the data.
3. Writes the byte results in the destination register.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not change the flags.
Examples
SHADD16 R1, R0 ; Adds halfwords in R0 to corresponding halfword of R1
; and writes halved result to corresponding halfword in
; R1.
SHADD8 R4, R0, R5 ; Adds bytes of R0 to corresponding byte in R5 and
; writes halved result to corresponding byte in R4.
PM0253 Rev 5 97/254
182
The Cortex-M7 instruction set PM0253
3.5.11 SHASX and SHSAX
Signed Halving Add and Subtract with Exchange and Signed Halving Subtract and Add with
Exchange.
Syntax op{ cond } { Rd }, Rn , Rm
Where: op cond
Rd
Rn , Rm
Is one of:
SHASX
Add and Subtract with Exchange and Halving.
SHSAX
Subtract and Add with Exchange and Halving.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Are registers holding the first and second operands.
Operation
The SHASX instruction:
1.
Adds the top halfword of the first operand with the bottom halfword of the second operand.
2. Writes the halfword result of the addition to the top halfword of the destination register, shifted by one bit to the right causing a divide by two, or halving.
3. Subtracts the top halfword of the second operand from the bottom highword of the first operand.
4. Writes the halfword result of the division in the bottom halfword of the destination register, shifted by one bit to the right causing a divide by two, or halving.
The SHSAX instruction:
1.
Subtracts the bottom halfword of the second operand from the top highword of the first operand.
2. Writes the halfword result of the addition to the bottom halfword of the destination register, shifted by one bit to the right causing a divide by two, or halving.
3. Adds the bottom halfword of the first operand with the top halfword of the second operand.
4. Writes the halfword result of the division in the top halfword of the destination register, shifted by one bit to the right causing a divide by two, or halving.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not affect the condition code flags.
Examples
SHASX R7, R4, R2 ; Adds top halfword of R4 to bottom halfword of
; R2 and writes halved result to top halfword of
; R7. Subtracts top halfword of R2 from bottom
98/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
; halfword of R4 and writes halved result to
; bottom halfword of R7.
SHSAX R0, R3, R5 ; Subtracts bottom halfword of R5 from top
; halfword of R3 and writes halved result to top
; halfword of R0.
; Adds top halfword of R5 to bottom halfword of
; R3 and writes halved result to bottom halfword
; of R0.
3.5.12 SHSUB16 and SHSUB8
Signed Halving Subtract 16 and Signed Halving Subtract 8.
Syntax op { cond }{ Rd ,} Rn , Rm
Where: op cond
Rd
Rn
Rm
Is one of:
SHSUB16
Signed Halving Subtract 16.
SHSUB8
Signed Halving Subtract 8.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the first operand register.
Is the second operand register
Operation
Use these instructions to add 16-bit and 8-bit data and then to halve the result before writing the result to the destination register.
The SHSUB16 instruction:
1.
Subtracts each halfword of the second operand from the corresponding halfwords of the first operand.
2. Shuffles the result by one bit to the right, halving the data.
3. Writes the halved halfword results in the destination register.
The SHSUBB8 instruction:
1.
Subtracts each byte of the second operand from the corresponding byte of the first operand.
2. Shuffles the result by one bit to the right, halving the data.
3. Writes the corresponding signed byte results in the destination register.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not change the flags.
PM0253 Rev 5 99/254
182
The Cortex-M7 instruction set PM0253
Examples
SHSUB16 R1, R0 ; Subtracts halfwords in R0 from corresponding
; halfword of R1 and writes to corresponding halfword
; of R1.
SHSUB8 R4, R0, R5 ; Subtracts bytes of R0 from corresponding byte in R5,
; and writes to corresponding byte in R4.
3.5.13 SSUB16 and SSUB8
Signed Subtract 16 and Signed Subtract 8.
Syntax op { cond }{ Rd ,} Rn , Rm
Where: op cond
Rd
Rn
Rm
Is one of:
SSUB16
Performs two 16-bit signed integer subtractions.
SSUB8
Performs four 8-bit signed integer subtractions.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the first operand register.
Is the second operand register.
Operation
Use these instructions to change endianness of data.
The SSUB16 instruction:
1.
Subtracts each halfword from the second operand from the corresponding halfword of the first operand.
2. Writes the difference result of two signed halfwords in the corresponding halfword of the destination register.
The SSUB8 instruction:
1.
Subtracts each byte of the second operand from the corresponding byte of the first operand.
2. Writes the difference result of four signed bytes in the corresponding byte of the destination register.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not change the flags.
Examples
SSUB16 R1, R0 ; Subtracts halfwords in R0 from corresponding halfword of R1
100/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
; halfword of R1 and writes to corresponding halfword
; of R1.
SSUB8 R4, R0, R5 ; Subtracts bytes of R5 from corresponding byte in
; R0, and writes to corresponding byte of R4.
3.5.14 SASX and SSAX
Signed Add and Subtract with Exchange and Signed Subtract and Add with Exchange.
Syntax op{ cond } { Rd }, Rm , Rn
Where: op cond
Rd
Rn , Rm
Is one of:
SASX
Signed Add and Subtract with Exchange.
SSAX
Signed Subtract and Add with Exchange.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Are registers holding the first and second operands.
Operation
The SASX instruction:
1.
Adds the signed top halfword of the first operand with the signed bottom halfword of the second operand.
2. Writes the signed result of the addition to the top halfword of the destination register.
3. Subtracts the signed bottom halfword of the second operand from the top signed highword of the first operand.
4. Writes the signed result of the subtraction to the bottom halfword of the destination register.
The SSAX instruction:
1.
Subtracts the signed bottom halfword of the second operand from the top signed highword of the first operand.
2. Writes the signed result of the addition to the bottom halfword of the destination register.
3. Adds the signed top halfword of the first operand with the signed bottom halfword of the second operand.
4. Writes the signed result of the subtraction to the top halfword of the destination register.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not affect the condition code flags.
Examples
SASX R0, R4, R5 ; Adds top halfword of R4 to bottom halfword of R5
PM0253 Rev 5 101/254
182
The Cortex-M7 instruction set PM0253
; and writes to top halfword of R0.
; Subtracts bottom halfword of R5 from top halfword
; of R4 and writes to bottom halfword of R0.
SSAX R7, R3, R2 ; Subtracts top halfword of R2 from bottom halfword
; of R3 and writes to bottom halfword of R7.
; Adds top halfword of R3 with bottom halfword of R
; R2 and writes to top halfword of R7.
3.5.15 TST and TEQ
Test bits and Test Equivalence.
Syntax
TST{ cond } Rn , Operand2
TEQ{ cond } Rn , Operand2
Where: cond
Rn
Operand2
Is an optional condition code. See Conditional execution on page 68 .
Is the register holding the first operand.
Is a flexible second operand. See
Flexible second operand on page 64
for details of the options.
Operation
These instructions test the value in a register against Operand2 . They update the condition flags based on the result, but do not write the result to a register.
The TST instruction performs a bitwise AND operation on the value in Rn and the value of
Operand2 . This is the same as the ANDS instruction, except that it discards the result.
To test whether a bit of Rn is 0 or 1, use the TST instruction with an Operand2 constant that has that bit set to 1 and all other bits cleared to 0.
The TEQ instruction performs a bitwise Exclusive OR operation on the value in Rn and the value of Operand2 . This is the same as the EORS instruction, except that it discards the result.
Use the TEQ instruction to test if two values are equal without affecting the V or C flags.
TEQ is also useful for testing the sign of a value. After the comparison, the N flag is the logical Exclusive OR of the sign bits of the two operands.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions:
• Update the N and Z flags according to the result.
• Can update the C flag during the calculation of Operand2
, see Flexible second operand on page 64 .
• Do not affect the V flag.
102/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Examples
TST R0, #0x3F8 ; Perform bitwise AND of R0 value to 0x3F8,
; APSR is updated but result is discarded
TEQEQ R10, R9 ; Conditionally test if value in R10 is equal to
; value in R9, APSR is updated but result is
; discarded.
3.5.16 UADD16 and UADD8
Unsigned Add 16 and Unsigned Add 8.
Syntax op { cond }{ Rd ,} Rn , Rm
Where: op cond
Rd
Rn
Rm
Is one of:
UADD16
Performs two 16-bit unsigned integer additions.
UADD8
Performs four 8-bit unsigned integer additions.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the first register holding the operand.
Is the second register holding the operand.
Operation
Use these instructions to add 16- and 8-bit unsigned data.
The UADD16 instruction:
1.
Adds each halfword from the first operand to the corresponding halfword of the second operand.
2. Writes the unsigned result in the corresponding halfwords of the destination register.
The UADD16 instruction:
1.
Adds each byte of the first operand to the corresponding byte of the second operand.
2. Writes the unsigned result in the corresponding byte of the destination register.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not change the flags.
Examples
UADD16 R1, R0 ; Adds halfwords in R0 to corresponding halfword of R1,
; writes to corresponding halfword of R1.
UADD8 R4, R0, R5 ; Adds bytes of R0 to corresponding byte in R5 and
; writes to corresponding byte in R4.
PM0253 Rev 5 103/254
182
The Cortex-M7 instruction set PM0253
3.5.17 UASX and USAX
Add and Subtract with Exchange and Subtract and Add with Exchange.
Syntax op{ cond } { Rd }, Rn , Rm
Where: op cond
Rd
Rn , Rm
Is one of:
UASX
Add and Subtract with Exchange.
USAX
Subtract and Add with Exchange.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Are registers holding the first and second operands.
Operation
The UASX instruction:
1.
Subtracts the top halfword of the second operand from the bottom halfword of the first operand.
2. Writes the unsigned result from the subtraction to the bottom halfword of the destination register.
3. Adds the top halfword of the first operand with the bottom halfword of the second operand.
4. Writes the unsigned result of the addition to the top halfword of the destination register.
The USAX instruction:
1.
Adds the bottom halfword of the first operand with the top halfword of the second operand.
2. Writes the unsigned result of the addition to the bottom halfword of the destination register.
3. Subtracts the bottom halfword of the second operand from the top halfword of the first operand.
4. Writes the unsigned result from the subtraction to the top halfword of the destination register.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not affect the condition code flags.
Examples
UASX R0, R4, R5 ; Adds top halfword of R4 to bottom halfword of R5
; and writes to top halfword of R0.
; Subtracts bottom halfword of R5 from top halfword
; of R0 and writes to bottom halfword of R0.
USAX R7, R3, R2 ; Subtracts top halfword of R2 from bottom halfword
; of R3 and writes to bottom halfword of R7.
; Adds top halfword of R3 to bottom halfword of R2
104/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
; and writes to top halfword of R7.
3.5.18 UHADD16 and UHADD8
Unsigned Halving Add 16 and Unsigned Halving Add 8.
Syntax op { cond }{ Rd ,} Rn , Rm
Where: op cond
Rd
Rn
Rm
Is one of:
UHADD16
Unsigned Halving Add 16.
UHADD8
Unsigned Halving Add 8.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the register holding the first operand.
Is the register holding the second operand.
Operation
Use these instructions to add 16- and 8-bit data and then to halve the result before writing the result to the destination register.
The UHADD16 instruction:
1.
Adds each halfword from the first operand to the corresponding halfword of the second operand.
2. Shuffles the halfword result by one bit to the right, halving the data.
3. Writes the unsigned results to the corresponding halfword in the destination register.
The UHADD8 instruction:
1.
Adds each byte of the first operand to the corresponding byte of the second operand.
2. Shuffles the byte result by one bit to the right, halving the data.
3. Writes the unsigned results in the corresponding byte in the destination register.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not change the flags.
Examples
UHADD16 R7, R3 ; Adds halfwords in R7 to corresponding halfword of R3
; and writes halved result to corresponding halfword
; in R7.
UHADD8 R4, R0, R5 ; Adds bytes of R0 to corresponding byte in R5 and
; writes halved result to corresponding byte in R4.
PM0253 Rev 5 105/254
182
The Cortex-M7 instruction set PM0253
3.5.19 UHASX and UHSAX
Unsigned Halving Add and Subtract with Exchange and Unsigned Halving Subtract and Add with Exchange.
Syntax op{ cond } { Rd }, Rn , Rm
Where: op cond
Rd
Rn , Rm
Is one of:
UHASX
Add and Subtract with Exchange and Halving.
UHSAX
Subtract and Add with Exchange and Halving.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Are registers holding the first and second operands.
Operation
The UHASX instruction:
1.
Adds the top halfword of the first operand with the bottom halfword of the second operand.
2. Shifts the result by one bit to the right causing a divide by two, or halving.
3. Writes the halfword result of the addition to the top halfword of the destination register.
4. Subtracts the top halfword of the second operand from the bottom highword of the first operand.
5. Shifts the result by one bit to the right causing a divide by two, or halving.
6. Writes the halfword result of the division in the bottom halfword of the destination register.
The UHSAX instruction:
1.
Subtracts the bottom halfword of the second operand from the top highword of the first operand.
2. Shifts the result by one bit to the right causing a divide by two, or halving.
3. Writes the halfword result of the subtraction in the top halfword of the destination register.
4. Adds the bottom halfword of the first operand with the top halfword of the second operand.
5. Shifts the result by one bit to the right causing a divide by two, or halving.
6. Writes the halfword result of the addition to the bottom halfword of the destination register.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not affect the condition code flags.
Examples
UHASX R7, R4, R2 ; Adds top halfword of R4 with bottom halfword of
106/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
; R2 and writes halved result to top halfword of
; R7.
; Subtracts top halfword of R2 from bottom
; halfword of R7 and writes halved result to
; bottom halfword of R7.
UHSAX R0, R3, R5 ; Subtracts bottom halfword of R5 from top
; halfword of R3 and writes halved result to top
; halfword of R0.
; Adds top halfword of R5 to bottom halfword of R3
; and writes halved result to bottom halfword of
; R0.
3.5.20 UHSUB16 and UHSUB8
Unsigned Halving Subtract 16 and Unsigned Halving Subtract 8.
Syntax op { cond }{ Rd ,} Rn , Rm
Where: op cond
Rd
Rn
Rm
Is one of:
UHSUB16
Performs two unsigned 16-bit integer additions, halves the results, and writes the results to the destination register.
UHSUB8
Performs four unsigned 8-bit integer additions, halves the results, and writes the results to the destination register.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the first register holding the operand.
Is the second register holding the operand.
Operation
Use these instructions to add 16-bit and 8-bit data and then to halve the result before writing the result to the destination register.
The UHSUB16 instruction:
1.
Subtracts each halfword of the second operand from the corresponding halfword of the first operand.
2. Shuffles each halfword result to the right by one bit, halving the data.
3. Writes each unsigned halfword result to the corresponding halfwords in the destination register.
The UHSUB8 instruction:
1.
Subtracts each byte of second operand from the corresponding byte of the first operand.
2. Shuffles each byte result by one bit to the right, halving the data.
3. Writes the unsigned byte results to the corresponding byte of the destination register.
Restrictions
Do not use SP and do not use PC .
PM0253 Rev 5 107/254
182
The Cortex-M7 instruction set PM0253
Condition flags
These instructions do not change the flags.
Examples
UHSUB16 R1, R0 ; Subtracts halfwords in R0 from corresponding
; halfword of R1 and writes halved result to
; corresponding halfword in R1.
UHSUB8 R4, R0, R5 ; Subtracts bytes of R5 from corresponding byte in R0
; and writes halved result to corresponding byte in
; R4.
3.5.21 SEL
Select Bytes. Selects each byte of its result from either its first operand or its second operand, according to the values of the GE flags.
Syntax
SEL{<c>}{<q>} {<Rd>,} <Rn>, <Rm>
Where:
<c>, <q>
<Rd>
<Rn>
<Rm>
Is a standard assembler syntax field.
Is the destination register.
Is the first operand register.
Is the second operand register.
Operation
The SEL instruction:
1.
Reads the value of each bit of APSR.GE.
either the first or second operand register.
Restrictions
None.
Condition flags
These instructions do not change the flags.
Examples
SADD16 R0, R1, R2 ; Set GE bits based on result.
SEL R0, R0, R3 ; Select bytes from R0 or R3, based on GE.
3.5.22 USAD8
Unsigned Sum of Absolute Differences.
Syntax
USAD8{ cond }{ Rd ,} Rn , Rm
108/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Where: cond
Rd
Rn
Rm
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the first operand register.
Is the second operand register.
Operation
The USAD8 instruction:
1.
Subtracts each byte of the second operand register from the corresponding byte of the first operand register.
2. Adds the absolute values of the differences together.
1.
Writes the result to the destination register.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not change the flags.
Examples
USAD8 R1, R4, R0 ; Subtracts each byte in R0 from corresponding byte
; of R4 adds the differences and writes to R1.
USAD8 R0, R5 ; Subtracts bytes of R5 from corresponding byte in
; R0, adds the differences and writes to R0.
3.5.23 USADA8
Unsigned Sum of Absolute Differences and Accumulate.
Syntax
USADA8{ cond }{ Rd ,} Rn , Rm , Ra
Where: cond
Rd
Rn
Rm
Ra
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the first operand register.
Is the second operand register.
Is the register that contains the accumulation value.
Operation
The USADA8 instruction:
PM0253 Rev 5 109/254
182
The Cortex-M7 instruction set PM0253
1.
Subtracts each byte of the second operand register from the corresponding byte of the first operand register.
2. Adds the unsigned absolute differences together.
3. Adds the accumulation value to the sum of the absolute differences.
4. Writes the result to the destination register.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not change the flags.
Examples
USADA8 R1, R0, R6 ; Subtracts bytes in R0 from corresponding halfword
; of R1 adds differences, adds value of R6, writes
; to R1.
USADA8 R4, R0, R5, R2 ; Subtracts bytes of R5 from corresponding byte in
; R0 adds differences, adds value of R2 writes to
; R4.
3.5.24 USUB16 and USUB8
Unsigned Subtract 16 and Unsigned Subtract 8.
Syntax op { cond }{ Rd ,} Rn , Rm
Where: op cond
Rd
Rn
Rm
Is one of:
USUB16
Unsigned Subtract 16.
USUB8
Unsigned Subtract 8.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the first operand register.
Is the second operand register.
Operation
Use these instructions to subtract 16-bit and 8-bit data before writing the result to the destination register.
The USUB16 instruction:
1.
Subtracts each halfword from the second operand register from the corresponding halfword of the first operand register.
2. Writes the unsigned result in the corresponding halfwords of the destination register.
The USUB8 instruction:
110/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
1.
Subtracts each byte of the second operand register from the corresponding byte of the first operand register.
2. Writes the unsigned byte result in the corresponding byte of the destination register.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not change the flags.
Examples
USUB16 R1, R0 ; Subtracts halfwords in R0 from corresponding halfword
; of R1 and writes to corresponding halfword in R1.
USUB8 R4, R0, R5 ; Subtracts bytes of R5 from corresponding byte in R0
; and writes to the corresponding byte in R4.
3.6
Mnemonic
MLA
MLS
MUL
SDIV
SMLA[B,T]
SMLAD,
SMLADX
SMLAL
SMLALD,
SMLALDX
SMLAW[B|T]
SMLSD
SMLSLD
SMMLA
SMMLS,
SMMLSR
Multiply and divide instructions
Table 31 shows the multiply and divide instructions:
Table 31. Multiply and divide instructions
Brief description
Multiply with Accumulate, 32-bit result
Multiply and Subtract, 32-bit result
Multiply, 32-bit result
Signed Divide
Signed Multiply Accumulate (halfwords)
See
Signed Multiply Accumulate Dual
Signed Multiply with Accumulate (32x32+64),
64-bit result
UMULL, UMLAL, SMULL, and SMLAL on page 126
Multiply
Signed Multiply Accumulate Long Dual
Signed Multiply Accumulate (word by halfword)
Signed Multiply Subtract Dual
Signed Multiply Subtract Long Dual
Signed Most Significant Word Multiply
Accumulate
Signed Most Significant Word Multiply
Subtract
PM0253 Rev 5 111/254
182
The Cortex-M7 instruction set PM0253
Mnemonic
SMUAD,
SMUADX
SMUL[B,T]
SMMUL,
SMMULR
SMULL
SMULWB,
SMULWT
SMUSD,
SMUSDX
UDIV
UMAAL
UMLAL
UMULL
Table 31. Multiply and divide instructions (continued)
Brief description
Signed Dual Multiply Add
Signed Multiply (word by halfword)
Signed Most Significant Word Multiply
Signed Multiply (32x32), 64-bit result
Signed Multiply (word by halfword)
See
UMULL, UMLAL, SMULL, and SMLAL on page 126
Signed Dual Multiply Subtract
Unsigned Divide
Unsigned Multiply Accumulate Accumulate
Long (32x32+32+32), 64-bit result
Unsigned Multiply with Accumulate
(32x32+64), 64-bit result
Unsigned Multiply (32x32), 64-bit result
UMULL, UMAAL, UMLAL on page 113
UMULL, UMAAL, UMLAL on page 113
UMULL, UMAAL, UMLAL on page 113
3.6.1 MUL, MLA, and MLS
Multiply, Multiply with Accumulate, and Multiply with Subtract, using 32-bit operands, and producing a 32-bit result.
Syntax
MUL{S}{ cond } { Rd ,} Rn , Rm ; Multiply
MLA{ cond } Rd , Rn , Rm , Ra ; Multiply with accumulate
MLS{ cond } Rd , Rn , Rm , Ra ; Multiply with subtract
Where: cond
S
Rd
Rn, Rm
Ra
Is an optional condition code. See Conditional execution on page 68 .
Is an optional suffix. If
S
is specified, the condition code flags are updated on the result of the operation, see
Conditional execution on page 68
.
Is the destination register. If
Rd
is omitted, the destination register is
Rn
.
Are registers holding the values to be multiplied.
Is a register holding the value to be added or subtracted from.
Operation
The MUL instruction multiplies the values from Rn and Rm , and places the least significant
32 bits of the result in Rd .
The MLA instruction multiplies the values from Rn and Rm , adds the value from Ra , and places the least significant 32 bits of the result in Rd .
112/254 PM0253 Rev 5
PM0253
3.6.2
The Cortex-M7 instruction set
The MLS instruction multiplies the values from Rn and Rm , subtracts the product from the value from Ra , and places the least significant 32 bits of the result in Rd .
The results of these instructions do not depend on whether the operands are signed or unsigned.
Restrictions
In these instructions, do not use SP and do not use PC.
If the S suffix is used with the MUL instruction:
• Rd , Rn , and Rm must all be in the range R0 to R7.
• Rd must be the same as Rm .
• The cond suffix must not be used.
Condition flags
If S is specified, the MUL instruction:
• Updates the N and Z flags according to the result.
• Does not affect the C and V flags.
Examples
MUL R10, R2, R5 ; Multiply, R10 = R2 x R5
MLA R10, R2, R1, R5 ; Multiply with accumulate, R10 = (R2 x R1) +
; R5
MULS R0, R2, R2 ; Multiply with flag update, R0 = R2 x R2
MULLT R2, R3, R2 ; Conditionally multiply, R2 = R3 x R2
MLS R4, R5, R6, R7 ; Multiply with subtract, R4 = R7 - (R5 x R6)
UMULL, UMAAL, UMLAL
Unsigned Long Multiply, with optional Accumulate, using 32-bit operands and producing a
64-bit result.
Syntax op { cond } RdLo , RdHi , Rn , Rm
Where: op
Is one of:
UMULL
Unsigned Long Multiply.
UMAAL
Unsigned Long Multiply with Accumulate Accumulate.
UMLAL
Unsigned Long Multiply, with Accumulate.
cond
Rn, Rm
Is an optional condition code. See Conditional execution on page 68 .
RdHi , RdLo
Are the destination registers. For
UMAAL
,
UMLAL
and
UMLAL
they also hold the accumulating value.
Are registers holding the first and second operands.
PM0253 Rev 5 113/254
182
The Cortex-M7 instruction set PM0253
Operation
These instructions interpret the values from Rn and Rm as unsigned 32-bit integers.
The UMULL instruction:
• Multiplies the two unsigned integers in the first and second operands.
• Writes the least significant 32 bits of the result in RdLo.
• Writes the most significant 32 bits of the result in RdHi .
The UMAAL instruction:
• Multiplies the two unsigned 32-bit integers in the first and second operands.
• Adds the unsigned 32-bit integer in RdHi to the 64-bit result of the multiplication.
• Adds the unsigned 32-bit integer in RdLo to the 64-bit result of the addition.
• Writes the top 32-bits of the result to RdHi.
• Writes the lower 32-bits of the result to RdLo .
The UMLAL instruction:
• Multiplies the two unsigned integers in the first and second operands.
• Adds the 64-bit result to the 64-bit unsigned integer contained in RdHi and RdLo .
• Writes the result back to RdHi and RdLo .
Restrictions
In these instructions:
• Do not use SP and do not use PC.
• RdHi and RdLo must be different registers.
Condition flags
These instructions do not affect the condition code flags.
Examples
UMULL R0, R4, R5, R6 ; Multiplies R5 and R6, writes the top 32 bits to
; R4 and the bottom 32 bits to R0.
UMAAL R3, R6, R2, R7 ; Multiplies R2 and R7, adds R6, adds R3, writes
; the top 32 bits to R6, and the bottom 32 bits
; to R3.
UMLAL R2, R1, R3, R5 ; Multiplies R5 and R3, adds R1:R2, writes to
; R1:R2.
114/254 PM0253 Rev 5
PM0253
3.6.3
The Cortex-M7 instruction set
SMLA and SMLAW
Signed Multiply Accumulate (halfwords).
Syntax op{ XY }{ cond } Rd , Rn , Rm op{ Y }{ cond } Rd , Rn , Rm , Ra
Where: op cond
Rd
Rn, Rm
Ra
Is one of:
SMLA
Signed Multiply Accumulate Long (halfwords)
X
and
Y
specifies which half of the source registers
Rn
and
Rm
are used as the first and second multiply operand.
If
X
is
B
, then the bottom halfword, bits [15:0], of
Rn
is used.
If
X
is
T
, then the top halfword, bits [31:16], of
Rn
is used.
If
Y
is
B
, then the bottom halfword, bits [15:0], of
Rm
is used.
If
Y
is
T
, then the top halfword, bits [31:16], of
Rm
is used.
SMLAW
Signed Multiply Accumulate (word by halfword)
Y
specifies which half of the source register
Rm
is used as the second multiply operand.
If
Y
is
T
, then the top halfword, bits [31:16] of
Rm
is used.
If
Y
is
B
, then the bottom halfword, bits [15:0] of
Rm
is used.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register. If
Rd
is omitted, the destination register is
Rn
.
Are registers holding the values to be multiplied.
Is a register holding the value to be added or subtracted from.
Operation
The SMALBB, SMLABT, SMLATB, SMLATT instructions:
• Multiplies the specified signed halfword, top or bottom, values from Rn and Rm .
• Adds the value in Ra to the resulting 32-bit product.
• Writes the result of the multiplication and addition in Rd .
The non-specified halfwords of the source registers are ignored.
The SMLAWB and SMLAWT instructions:
• Multiply the 32-bit signed values in Rn with:
– The top signed halfword of Rm , T instruction suffix.
– The bottom signed halfword of Rm , B instruction suffix.
• Add the 32-bit signed value in Ra to the top 32 bits of the 48-bit product
• Writes the result of the multiplication and addition in Rd .
The bottom 16 bits of the 48-bit product are ignored.
If overflow occurs during the addition of the accumulate value, the instruction sets the Q flag in the APSR. No overflow can occur during the multiplication.
PM0253 Rev 5 115/254
182
The Cortex-M7 instruction set PM0253
Restrictions
In these instructions, do not use SP and do not use PC.
Condition flags
If an overflow is detected, the Q flag is set.
Examples
SMLABB R5, R6, R4, R1 ; Multiplies bottom halfwords of R6 and R4, adds
; R1 and writes to R5.
SMLATB R5, R6, R4, R1 ; Multiplies top halfword of R6 with bottom
; halfword of R4, adds R1 and writes to R5.
SMLATT R5, R6, R4, R1 ; Multiplies top halfwords of R6 and R4, adds
; R1 and writes the sum to R5.
SMLABT R5, R6, R4, R1 ; Multiplies bottom halfword of R6 with top
; halfword of R4, adds R1 and writes to R5.
SMLABT R4, R3, R2 ; Multiplies bottom halfword of R4 with top
; halfword of R3, adds R2 and writes to R4.
SMLAWB R10, R2, R5, R3 ; Multiplies R2 with bottom halfword of R5, adds
; R3 to the result and writes top 32-bits to R10.
SMLAWT R10, R2, R1, R5 ; Multiplies R2 with top halfword of R1, adds R5
; and writes top 32-bits to R10.
3.6.4 SMLAD
Signed Multiply Accumulate Long Dual.
Syntax op{ X }{ cond } Rd , Rn , Rm , Ra
Where: op cond
Rd
Rn
Rm
Ra
Is one of:
SMLAD
Signed Multiply Accumulate Dual.
SMLADX
Signed Multiply Accumulate Dual Reverse.
X
specifies which halfword of the source register
Rn
is used as the multiply operand.
If
X
is omitted, the multiplications are bottom × bottom and top × top.
If
X
is present, the multiplications are bottom × top and top × bottom.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the first operand register holding the values to be multiplied.
Is the second operand register.
Is the accumulate value.
116/254 PM0253 Rev 5
PM0253
3.6.5
The Cortex-M7 instruction set
Operation
The SMLAD and SMLADX instructions regard the two operands as four halfword 16-bit values. The SMLAD and SMLADX instructions:
• If X is not present, multiply the top signed halfword value in Rn with the top signed halfword of Rm and the bottom signed halfword values in Rn with the bottom signed halfword of Rm .
• Or if X is present, multiply the top signed halfword value in Rn with the bottom signed halfword of Rm and the bottom signed halfword values in Rn with the top signed halfword of Rm .
• Add both multiplication results to the signed 32-bit value in Ra .
• Writes the 32-bit signed result of the multiplication and addition to Rd .
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not change the flags.
Examples
SMLAD R10, R2, R1, R5 ; Multiplies two halfword values in R2 with
; corresponding halfwords in R1, adds R5 and
; writes to R10.
SMLALDX R0, R2, R4, R6 ; Multiplies top halfword of R2 with bottom
; halfword of R4, multiplies bottom halfword of R2
; with top halfword of R4, adds R6 and writes to
; R0.
SMLAL and SMLALD
Signed Multiply Accumulate Long, Signed Multiply Accumulate Long (halfwords) and Signed
Multiply Accumulate Long Dual.
Syntax op { cond } RdLo , RdHi , Rn , Rm op {XY}{ cond } RdLo , RdHi , Rn , Rm op { X }{ cond } RdLo , RdHi , Rn , Rm
Where: op
Is one of:
SMLAL
Signed Multiply Accumulate Long.
SMLALXY
Signed Multiply Accumulate Long (halfwords, X and Y).
X and Y specify which halfword of the source registers
Rn
and
Rm
are used as the first and second multiply operand:
If
X
is
B
, then the bottom halfword, bits [15:0], of
Rn
is used.
If
X
is
T
, then the top halfword, bits [31:16], of
Rn
is used.
If
Y
is
B
, then the bottom halfword, bits [15:0], of
Rm
is used.
If
Y
is
T
, then the top halfword, bits [31:16], of
Rm
is used.
PM0253 Rev 5 117/254
182
The Cortex-M7 instruction set PM0253
Rn, Rm
SMLALD
Signed Multiply Accumulate Long Dual.
SMLALDX
Signed Multiply Accumulate Long Dual Reversed.
If the
X
is omitted, the multiplications are bottom × bottom and top × top.
If
X
is present, the multiplications are bottom × top and top × bottom.
cond
Is an optional condition code. See Conditional execution on page 68 .
RdHi , RdLo
Are the destination registers.
RdLo
is the lower 32 bits and
RdHi
is the upper 32 bits of the 64-bit integer.
For
SMLAL
,
SMLALBB
,
SMLALBT
,
SMLALTB
,
SMLALTT
,
SMLALD
and
SMLALDX,
they also hold the accumulating value.
Are registers holding the first and second operands.
Operation
The SMLAL instruction:
• Multiplies the two’s complement signed word values from Rn and Rm .
• Adds the 64-bit value in RdLo and RdHi to the resulting 64-bit product.
• Writes the 64-bit result of the multiplication and addition in RdLo and RdHi .
The SMLALBB, SMLALBT, SMLALTB and SMLALTT instructions:
• Multiplies the specified signed halfword, Top or Bottom, values from Rn and Rm .
• Adds the resulting sign-extended 32-bit product to the 64-bit value in RdLo and RdHi .
• Writes the 64-bit result of the multiplication and addition in RdLo and RdHi .
The non-specified halfwords of the source registers are ignored.
The SMLALD and SMLALDX instructions interpret the values from Rn and Rm as four halfword two’s complement signed 16-bit integers. These instructions:
• If X is not present, multiply the top signed halfword value of Rn with the top signed halfword of Rm and the bottom signed halfword values of Rn with the bottom signed halfword of Rm .
• Or if X is present, multiply the top signed halfword value of Rn with the bottom signed halfword of Rm and the bottom signed halfword values of Rn with the top signed halfword of Rm .
• Add the two multiplication results to the signed 64-bit value in RdLo and RdHi to create the resulting 64-bit product.
• Write the 64-bit product in RdLo and RdHi .
Restrictions
In these instructions:
• Do not use SP and do not use PC.
• RdHi and RdLo must be different registers.
Condition flags
These instructions do not affect the condition code flags.
Examples
SMLAL R4, R5, R3, R8 ; Multiplies R3 and R8, adds R5:R4 and writes
118/254 PM0253 Rev 5
PM0253
3.6.6
The Cortex-M7 instruction set
; to R5:R4.
SMLALBT R2, R1, R6, R7 ; Multiplies bottom halfword of R6 with top
; halfword of R7, sign extends to 32-bit,
; adds R1:R2 and writes to R1:R2.
SMLALTB R2, R1, R6, R7 ; Multiplies top halfword of R6 with bottom
; halfword of R7,sign extends to 32-bit, adds
; R1:R2 and writes to R1:R2.
SMLALD R6, R8, R5, R1 ; Multiplies top halfwords in R5 and R1 and b
; bottom halfwords of R5 and R1, adds R8:R6
; and writes to R8:R6.
SMLALDX R6, R8, R5, R1 ; Multiplies top halfword in R5 with bottom
; halfword of R1, and bottom halfword of R5
; with top halfword of R1, adds R8:R6 and
; writes to R8:R6.
SMLSD and SMLSLD
Signed Multiply Subtract Dual and Signed Multiply Subtract Long Dual.
Syntax op { X }{ cond } Rd , Rn , Rm , Ra
Where: op cond
Rd
Rn, Rm
Ra
Is one of:
SMLSD
Signed Multiply Subtract Dual.
SMLSDX
Signed Multiply Subtract Dual Reversed.
SMLSLD
Signed Multiply Subtract Long Dual.
SMLSLDX
Signed Multiply Subtract Long Dual Reversed.
If
X
is present, the multiplications are bottom × top and top × bottom.
If the
X
is omitted, the multiplications are bottom × bottom and top × top.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Are registers holding the first and second operands.
Is the register holding the accumulate value.
Operation
The SMLSD instruction interprets the values from the first and second operands as four signed halfwords. This instruction:
• Optionally rotates the halfwords of the second operand.
• Performs two signed 16 × 16-bit halfword multiplications.
• Subtracts the result of the upper halfword multiplication from the result of the lower halfword multiplication.
• Adds the signed accumulate value to the result of the subtraction.
• Writes the result of the addition to the destination register.
PM0253 Rev 5 119/254
182
The Cortex-M7 instruction set PM0253
The SMLSLD instruction interprets the values from Rn and Rm as four signed halfwords.
This instruction:
• Optionally rotates the halfwords of the second operand.
• Performs two signed 16 × 16-bit halfword multiplications.
• Subtracts the result of the upper halfword multiplication from the result of the lower halfword multiplication.
• Adds the 64-bit value in RdHi and RdLo to the result of the subtraction.
• Writes the 64-bit result of the addition to the RdHi and RdLo .
Restrictions
In these instructions:
• Do not use SP and do not use PC.
Condition flags
This instruction sets the Q flag if the accumulate operation overflows. Overflow cannot occur during the multiplications or subtraction.
For the Thumb instruction set, these instructions do not affect the condition code flags.
Examples
SMLSD R0, R4, R5, R6 ; Multiplies bottom halfword of R4 with
; bottom halfword of R5, multiplies top
; halfword of R4 with top halfword of R5, sub
; substracts second from first, adds R6,
; writes to R0.
SMLSDX R1, R3, R2, R0 ; Multiplies bottom halfword of R3 with top
; halfword of R2, multiplies top halfword of
; R3 with bottom halfword of R2, subtracts se
; second from first, adds R0, writes to R1.
SMLSLD R3, R6, R2, R7 ; Multiplies bottom halfword of R6 with
; bottom halfword of R2, multiplies top
; halfword of R6 with top halfword of R2, sub
; substracts second first, adds R6:R3, writes
; to R6:R3.
SMLSLDX R3, R6, R2, R7 ; Multiplies bottom halfword of R6 with top
; halfword of R2, multiplies top halfword of
; R6 with bottom halfword of R2, subtracts
; second from first, adds R6:R3, writes to
; R6:R3.
120/254 PM0253 Rev 5
PM0253
3.6.7
The Cortex-M7 instruction set
SMMLA and SMMLS
Signed Most Significant Word Multiply Accumulate and Signed Most Significant Word
Multiply Subtract.
Syntax op{ R }{ cond } Rd , Rn , Rm , Ra
Where: op
R cond
Rd
Rn, Rm
Ra
Is one of:
SMMLA
Signed Most Significant Word Multiply Accumulate.
SMMLS
Signed Most Significant Word Multiply Subtract.
Is a rounding error flag. If
R
is specified, the result is rounded instead of being truncated. In this case the constant
0x80000000
is added to the product before the high word is extracted.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Are registers holding the first and second multiply operands.
Is the register holding the accumulate value.
Operation
The SMMLA instruction interprets the values from Rn and Rm as signed 32-bit words.
The SMMLA instruction:
• Multiplies the values in Rn and Rm .
• Optionally rounds the result by adding 0x80000000.
• Extracts the most significant 32 bits of the result.
• Adds the value of Ra to the signed extracted value.
• Writes the result of the addition in Rd .
The SMMLS instruction interprets the values from Rn and Rm as signed 32-bit words.
The SMMLS instruction:
• Multiplies the values in Rn and Rm .
• Optionally rounds the result by adding 0x80000000.
• Extracts the most significant 32 bits of the result.
• Subtracts the extracted value of the result from the value in Ra .
• Writes the result of the subtraction in Rd .
Restrictions
In these instructions:
• Do not use SP and do not use PC.
Condition flags
These instructions do not affect the condition code flags.
PM0253 Rev 5 121/254
182
The Cortex-M7 instruction set PM0253
Examples
SMMLA R0, R4, R5, R6 ; Multiplies R4 and R5, extracts top 32 bits,
; adds R6, truncates and writes to R0.
SMMLAR R6, R2, R1, R4 ; Multiplies R2 and R1, extracts top 32 bits,
; adds R4, rounds and writes to R6.
SMMLSR R3, R6, R2, R7 ; Multiplies R6 and R2, extracts top 32 bits,
; subtracts R7, rounds and writes to R3.
SMMLS R4, R5, R3, R8 ; Multiplies R5 and R3, extracts top 32 bits,
; subtracts R8, truncates and writes to R4.
3.6.8 SMMUL
Signed Most Significant Word Multiply.
Syntax op{ R }{ cond } Rd , Rn , Rm
Where: op
R cond
Rd
Rn, Rm
Is one of:
SMMUL
Signed Most Significant Word Multiply
Is a rounding error flag. If
R
is specified, the result is rounded instead of being truncated. In this case the constant
0x80000000
is added to the product before the high word is extracted.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Are registers holding the first and second operands.
Operation
The SMMUL instruction interprets the values from Rn and Rm as two’s complement 32-bit signed integers. The SMMUL instruction:
• Multiplies the values from Rn and Rm .
• Optionally rounds the result, otherwise truncates the result.
• Writes the most significant signed 32 bits of the result in Rd .
Restrictions
In this instruction:
• Do not use SP and do not use PC.
Condition flags
This instruction does not affect the condition code flags.
Examples
SMULL R0, R4, R5 ; Multiplies R4 and R5, truncates top 32 bits
; and writes to R0.
SMULLR R6, R2 ; Multiplies R6 and R2, rounds the top 32
; bits and writes to R6.
122/254 PM0253 Rev 5
PM0253
3.6.9
The Cortex-M7 instruction set
SMUAD and SMUSD
Signed Dual Multiply Add and Signed Dual Multiply Subtract.
Syntax op{X}{ cond } Rd , Rn , Rm
Where: op cond
Rd
Rn, Rm
Is one of:
SMUAD
Signed Dual Multiply Add.
SMUADX
Signed Dual Multiply Add Reversed.
SMUSD
Signed Dual Multiply Subtract.
SMUSDX
Signed Dual Multiply Subtract Reversed.
If
X
is present, the multiplications are bottom × top and top × bottom.
If the
X
is omitted, the multiplications are bottom × bottom and top × top.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Are registers holding the first and the second operands.
Operation
The SMUAD instruction interprets the values from the first and second operands as two signed halfwords in each operand. This instruction:
• Optionally rotates the halfwords of the second operand.
• Performs two signed 16 × 16-bit multiplications.
• Adds the two multiplication results together.
• Writes the result of the addition to the destination register.
The SMUSD instruction interprets the values from the first and second operands as two’s complement signed integers. This instruction:
• Optionally rotates the halfwords of the second operand.
• Performs two signed 16 × 16-bit multiplications.
• Subtracts the result of the top halfword multiplication from the result of the bottom halfword multiplication.
• Writes the result of the subtraction to the destination register.
Restrictions
In these instructions:
• Do not use SP and do not use PC.
Condition flags
Sets the Q flag if the addition overflows. The multiplications cannot overflow.
Examples
SMUAD R0, R4, R5 ; Multiplies bottom halfword of R4 with the
; bottom halfword of R5, adds multiplication of
PM0253 Rev 5 123/254
182
The Cortex-M7 instruction set PM0253
; top halfword of R4 with top halfword of R5,
; writes to R0.
SMUADX R3, R7, R4 ; Multiplies bottom halfword of R7 with top
; halfword of R4, adds multiplication of top
; halfword of R7 with bottom halfword of R4,
; writes to R3.
SMUSD R3, R6, R2 ; Multiplies bottom halfword of R4 with bottom
; halfword of R6, subtracts multiplication of top
; halfword of R6 with top halfword of R3, writes
; to R3.
SMUSDX R4, R5, R3 ; Multiplies bottom halfword of R5 with top
; halfword of R3, subtracts multiplication of top
; halfword of R5 with bottom halfword of R3,
; writes to R4.
3.6.10 SMUL and SMULW
Signed Multiply (halfwords) and Signed Multiply (word by halfword).
Syntax op { XY }{ cond } Rd , Rn , Rm op { Y }{ cond } Rd . Rn , Rm
For
SMULXY
only: op
Is one of: cond
Rd
Rn, Rm
SMUL{ XY }
Signed Multiply (halfwords)
X
and
Y
specify which halfword of the source registers
Rn
and
Rm
is used as the first and second multiply operand.
If
X
is
B
, then the bottom halfword, bits [15:0] of
Rn
is used.
If
X
is
T
, then the top halfword, bits [31:16] of
Rn
is used.If
Y
is
B
, then the bottom halfword, bits [15:0], of
Rm
is used.
If
Y
is
T
, then the top halfword, bits [31:16], of
Rm
is used.
SMULW{ Y }
Signed Multiply (word by halfword)
Y
specifies which halfword of the source register
Rm
is used as the second multiply operand.
If
Y
is
B
, then the bottom halfword (bits [15:0]) of
Rm
is used.
If
Y
is
T
, then the top halfword (bits [31:16]) of
Rm
is used.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Are registers holding the first and second operands.
Operation
The SMULBB, SMULTB, SMULBT and SMULTT instructions interprets the values from Rn and Rm as four signed 16-bit integers.
These instructions:
• Multiply the specified signed halfword, Top or Bottom, values from Rn and Rm .
• Write the 32-bit result of the multiplication in Rd.
124/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
The SMULWT and SMULWB instructions interprets the values from Rn as a 32-bit signed integer and Rm as two halfword 16-bit signed integers. These instructions:
• Multiply the first operand and the top, T suffix, or the bottom, B suffix, halfword of the second operand.
• Write the signed most significant 32 bits of the 48-bit result in the destination register.
Restrictions
In these instructions:
• Do not use SP and do not use PC.
• RdHi and RdLo must be different registers.
Examples
SMULBT R0, R4, R5 ; Multiplies the bottom halfword of R4 with
; the top halfword of R5, multiplies results
; and writes to R0.
SMULBB R0, R4, R5 ; Multiplies the bottom halfword of R4 with
; the bottom halfword of R5, multiplies
; results and writes to R0.
SMULTT R0, R4, R5 ; Multiplies the top halfword of R4 with the
; top halfword of R5, multiplies results and
; writes to R0.
SMULTB R0, R4, R5 ; Multiplies the top halfword of R4 with the
; bottom halfword of R5, multiplies results
; and writes to R0.
SMULWT R4, R5, R3 ; Multiplies R5 with the top halfword of R3,
; extracts top 32 bits and writes to R4.
SMULWB R4, R5, R3 ; Multiplies R5 with the bottom halfword of
; R3, extracts top 32 bits and writes to R4.
PM0253 Rev 5 125/254
182
The Cortex-M7 instruction set PM0253
3.6.11 UMULL, UMLAL, SMULL, and SMLAL
Signed and Unsigned Long Multiply, with optional Accumulate, using 32-bit operands and producing a 64-bit result.
Syntax op { cond } RdLo , RdHi , Rn , Rm
Where: op
Is one of:
UMULL
Unsigned Long Multiply.
UMLAL
Unsigned Long Multiply, with Accumulate.
SMULL
Signed Long Multiply.
SMLAL
Signed Long Multiply, with Accumulate.
cond
Rn, Rm
Is an optional condition code. See Conditional execution on page 68 .
RdHi , RdLo
Are the destination registers. For
UMLAL
and
SMLAL
they also hold the accumulating value.
Are registers holding the operands.
Operation
The UMULL instruction interprets the values from Rn and Rm as unsigned integers. It multiplies these integers and places the least significant 32 bits of the result in RdLo , and the most significant 32 bits of the result in RdHi .
The UMLAL instruction interprets the values from Rn and Rm as unsigned integers. It multiplies these integers, adds the 64-bit result to the 64-bit unsigned integer contained in
RdHi and RdLo , and writes the result back to RdHi and RdLo .
The SMULL instruction interprets the values from Rn and Rm as two’s complement signed integers. It multiplies these integers and places the least significant 32 bits of the result in
RdLo , and the most significant 32 bits of the result in RdHi .
The SMLAL instruction interprets the values from Rn and Rm as two’s complement signed integers. It multiplies these integers, adds the 64-bit result to the 64-bit signed integer contained in RdHi and RdLo , and writes the result back to RdHi and RdLo .
Restrictions
In these instructions:
• Do not use SP and do not use PC.
• RdHi and RdLo must be different registers.
Condition flags
These instructions do not affect the condition code flags.
Examples
UMULL R0, R4, R5, R6 ; Unsigned (R4,R0) = R5 x R6
SMLAL R4, R5, R3, R8 ; Signed (R5,R4) = (R5,R4) + R3 x R8
126/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
3.6.12 SDIV and UDIV
Signed Divide and Unsigned Divide.
Syntax
SDIV{ cond } { Rd ,} Rn , Rm
UDIV{ cond } { Rd ,} Rn, Rm
Where: cond
Rd
Rn
Rm
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register. If
Rd
is omitted, the destination register is
Rn
.
Is the register holding the value to be divided.
Is a register holding the divisor.
Operation
The SDIV instruction performs a signed integer division of the value in Rn by the value in
Rm .
The UDIV instruction performs an unsigned integer division of the value in Rn by the value in Rm .
For both instructions, if the value in Rn is not divisible by the value in Rm , the result is rounded towards zero.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not change the flags.
Examples
SDIV R0, R2, R4 ; Signed divide, R0 = R2/R4
UDIV R8, R8, R1 ; Unsigned divide, R8 = R8/R1
PM0253 Rev 5 127/254
182
The Cortex-M7 instruction set PM0253
Table 32 shows the saturating instructions:
Mnemonic
Table 32. Saturating instructions
Brief description See
SSAT
SSAT16
USAT
USAT16
QADD
QSUB
QSUB16
QASX
Signed Saturate
Signed Saturate Halfword
Unsigned Saturate
Unsigned Saturate Halfword
Saturating Add
Saturating Subtract
Saturating Subtract 16
Saturating Add and Subtract with Exchange
QSAX
QDADD
QDSUB
Saturating Subtract and Add with Exchange
Saturating Double and Add
Saturating Double and Subtract
UQADD16 Unsigned Saturating Add 16
UQADD8
UQASX
UQSAX
Unsigned Saturating Add 8
Unsigned Saturating Add and Subtract with
Exchange
Unsigned Saturating Subtract and Add with
Exchange
UQSUB16 Unsigned Saturating Subtract 16
UQSUB8 Unsigned Saturating Subtract 8
For signed n -bit saturation, this means that:
• If the value to be saturated is less than -2 n 1
, the result returned is -2 n-1
• If the value to be saturated is greater than 2 n 1 -1, the result returned is 2 n-1 -1
• Otherwise, the result returned is the same as the value to be saturated.
For unsigned n -bit saturation, this means that:
• If the value to be saturated is less than 0, the result returned is 0
• If the value to be saturated is greater than 2 n
-1, the result returned is 2 n
-1
• Otherwise, the result returned is the same as the value to be saturated.
If the returned result is different from the value to be saturated, it is called saturation . If the saturation occurs, the instruction sets the Q flag to 1 in the APSR. Otherwise, it leaves the Q
flag unchanged. To clear the Q flag to 0, the MSR instruction must be used, see MSR on page 179
.
To read the state of the Q flag, use the MRS instruction, see
.
128/254 PM0253 Rev 5
PM0253
3.7.1
The Cortex-M7 instruction set
SSAT and USAT
Signed Saturate and Unsigned Saturate to any bit position, with optional shift before saturating.
Syntax op { cond } Rd , # n , Rm {, shift #s}
Where: op cond
Rd n
Rm shift #s
Is one of:
SSAT
Saturates a signed value to a signed range.
USAT
Saturates a signed value to an unsigned range.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
•
•
Specifies the bit position to saturate to: n
ranges from 1 to 32 for
SSAT
.
n
ranges from 0 to 31 for
USAT
.
Is the register containing the value to saturate.
Is an optional shift applied to
Rm
before saturating. It must be one of the following:
ASR # s where s
is in the range 1 to 31.
LSL # s where s
is in the range 0 to 31.
Operation
These instructions saturate to a signed or unsigned n -bit value.
The SSAT instruction applies the specified shift, then saturates to the signed range
− 2 n –1 ≤ x ≤ 2 n –1 − 1.
The USAT instruction applies the specified shift, then saturates to the unsigned range
0 ≤ x 2 n − 1.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not affect the condition code flags.
If saturation occurs, these instructions set the Q flag to 1.
Examples
SSAT R7, #16, R7, LSL #4 ; Logical shift left value in R7 by 4,
; then saturate it as a signed 16-bit
; value and write it back to R7.
USATNE R0, #7, R5 ; Conditionally saturate value in R5 as a
; an unsigned 7 bit value and write it to
; R0.
PM0253 Rev 5 129/254
182
The Cortex-M7 instruction set
3.7.2 SSAT16 and USAT16
Signed Saturate and Unsigned Saturate to any bit position for two halfwords.
Syntax
PM0253 op { cond } Rd , # n , Rm
Where: op cond
Rd n
Rm
Is one of:
SSAT16
Saturates a signed halfword value to a signed range.
USAT16
Saturates a signed halfword value to an unsigned range.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
•
•
Specifies the bit position to saturate to: n
ranges from 1 to 16 for
SSAT
.
n
ranges from 0 to 15 for
USAT
.
Is the register containing the value to saturate.
Operation
The SSAT16 instruction:
1.
Saturates two signed 16-bit halfword values of the register with the value to saturate from selected by the bit position in n .
2. Writes the results as two signed 16-bit halfwords to the destination register.
The USAT16 instruction:
1.
Saturates two unsigned 16-bit halfword values of the register with the value to saturate from selected by the bit position in n .
2. Writes the results as two unsigned halfwords in the destination register.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not affect the condition code flags.
If saturation occurs, these instructions set the Q flag to 1.
Examples
SSAT16 R7, #9, R2 ; Saturates the top and bottom highwords of R2
; as 9-bit values, writes to corresponding
; halfword of R7.
USAT16NE R0, #13, R5 ; Conditionally saturates the top and bottom
; halfwords of R5 as 13-bit values, writes to
; corresponding halfword of R0.
130/254 PM0253 Rev 5
PM0253
3.7.3
The Cortex-M7 instruction set
QADD and QSUB
Saturating Add and Saturating Subtract, signed.
Syntax op{ cond } { Rd }, Rn , Rm op{ cond } { Rd }, Rn , Rm
Where: op cond
Rd
Rn, Rm
Is one of:
QADD
Saturating 32-bit add.
QADD8
Saturating four 8-bit integer additions.
QADD16
Saturating two 16-bit integer additions.
QSUB
Saturating 32-bit subtraction.
QSUB8
Saturating four 8-bit integer subtraction.
QSUB16
Saturating two 16-bit integer subtraction.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Are registers holding the first and second operands.
Operation
These instructions add or subtract two, four or eight values from the first and second operands and then writes a signed saturated value in the destination register.
The QADD and QSUB instructions apply the specified add or subtract, and then saturate the result to the signed range − 2 n –1 the instruction, 32, 16 or 8.
≤ x ≤ 2 n –1 − 1, where x is given by the number of bits applied in
If the returned result is different from the value to be saturated, it is called saturation . If saturation occurs, the QADD and QSUB instructions set the Q flag to 1 in the APSR.
Otherwise, it leaves the Q flag unchanged. The 8-bit and 16-bit QADD and QSUB instructions always leave the Q flag unchanged.
To clear the Q flag to 0, the MSR instruction must be used, see
To read the state of the Q flag, use the MRS instruction, see
.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not affect the condition code flags.
If saturation occurs, these instructions set the Q flag to 1.
Examples
QADD16 R7, R4, R2 ; Adds halfwords of R4 with corresponding halfword of
; R2, saturates to 16 bits and writes to
; corresponding halfword of R7.
PM0253 Rev 5 131/254
182
The Cortex-M7 instruction set
3.7.4
132/254
PM0253
QADD8 R3, R1, R6 ; Adds bytes of R1 to the corresponding bytes of R6,
; saturates to 8 bits and writes to corresponding
; byte of R3.
QSUB16 R4, R2, R3 ; Subtracts halfwords of R3 from corresponding
; halfword of R2, saturates to 16 bits, writes to
; corresponding halfword of R4.
QSUB8 R4, R2, R5 ; Subtracts bytes of R5 from the corresponding byte
; in R2, saturates to 8 bits, writes to corresponding
; byte of R4.
QASX and QSAX
Saturating Add and Subtract with Exchange and Saturating Subtract and Add with
Exchange, signed.
Syntax op{ cond } { Rd }, Rm , Rn
Where: op cond
Rd
Rn , Rm
Is one of:
QASX
Add and Subtract with Exchange and Saturate.
QSAX
Subtract and Add with Exchange and Saturate.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Are registers holding the first and second operands.
Operation
The QASX instruction:
1.
Adds the top halfword of the source operand with the bottom halfword of the second operand.
2. Subtracts the top halfword of the second operand from the bottom highword of the first operand.
3. Saturates the result of the subtraction and writes a 16-bit signed integer in the range –
2 15 ≤ ≤ 2 15 – 1, where x equals 16, to the bottom halfword of the destination register.
4. Saturates the results of the sum and writes a 16-bit signed integer in the range
–2
15 ≤ x ≤ 2
15
– 1, where x equals 16, to the top halfword of the destination register.
The QSAX instruction:
1.
Subtracts the bottom halfword of the second operand from the top highword of the first operand.
2. Adds the bottom halfword of the source operand with the top halfword of the second operand.
3. Saturates the results of the sum and writes a 16-bit signed integer in the range
–2 15 ≤ x ≤ 2 15 – 1, where x equals 16, to the bottom halfword of the destination register.
4. Saturates the result of the subtraction and writes a 16-bit signed integer in the range –
2
15 ≤ ≤ 2
15
– 1, where x equals 16, to the top halfword of the destination register.
Restrictions
Do not use SP and do not use PC .
PM0253 Rev 5
PM0253
3.7.5
The Cortex-M7 instruction set
Condition flags
These instructions do not affect the condition code flags.
Examples
QASX R7, R4, R2 ; Adds top halfword of R4 to bottom halfword of R2,
; saturates to 16 bits, writes to top halfword of
; R7, Subtracts top highword of R2 from bottom
; halfword of R4, saturates to 16 bits and writes
; to bottom halfword of R7
QSAX R0, R3, R5 ; Subtracts bottom halfword of R5 from top halfword
; of R3, saturates to 16 bits, writes to top
; halfword of R0
; Adds bottom halfword of R3 to top halfword of R5,
; saturates to 16 bits, writes to bottom halfword
; of R0.
QDADD and QDSUB
Saturating Double and Add and Saturating Double and Subtract, signed.
Syntax op { cond } { Rd }, Rm , Rn
Where: op cond
Rd
Rm , Rn
Is one of:
QDADD
Saturating Double and Add.
QDSUB
Saturating Double and Subtract.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Are registers holding the first and second operands.
Operation
The QDADD instruction:
• Doubles the second operand value.
• Adds the result of the doubling to the signed saturated value in the first operand.
• Writes the result to the destination register.
The QDSUB instruction:
• Doubles the second operand value.
• Subtracts the doubled value from the signed saturated value in the first operand.
• Writes the result to the destination register.
Both the doubling and the addition or subtraction have their results saturated to the 32-bit signed integer range –2
31 ≤ x ≤ 2
31
– 1. If saturation occurs in either operation, it sets the Q flag in the APSR.
Restrictions
Do not use SP and do not use PC .
PM0253 Rev 5 133/254
182
The Cortex-M7 instruction set
3.7.6
PM0253
Condition flags
If saturation occurs, these instructions set the Q flag to 1.
Examples
QDADD R7, R4, R2 ; Doubles and saturates R4 to 32 bits, adds R2,
; Saturates to 32 bits, writes to R7
QDSUB R0, R3, R5 ; Subtracts R3 doubled and saturated to 32 bits
; from R5, saturates to 32 bits, writes to R0.
UQASX and UQSAX
Saturating Add and Subtract with Exchange and Saturating Subtract and Add with
Exchange, unsigned.
Syntax op{ cond } { Rd }, Rm , Rn
Where: type cond
Rd
Rn , Rm
Is one of:
UQASX
Add and Subtract with Exchange and Saturate.
UQSAX
Subtract and Add with Exchange and Saturate.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Are registers holding the first and second operands.
Operation
The UQASX instruction:
1.
Adds the bottom halfword of the source operand with the top halfword of the second operand.
2. Subtracts the bottom halfword of the second operand from the top highword of the first operand.
3. Saturates the results of the sum and writes a 16-bit unsigned integer in the range
0 ≤ x ≤ 16
– 1, where x
equals 16, to the top halfword of the destination register.
4. Saturates the result of the subtraction and writes a 16-bit unsigned integer in the range
0 ≤ x 2 16 – 1, where x
equals 16, to the bottom halfword of the destination register.
The UQSAX instruction:
1.
Subtracts the bottom halfword of the second operand from the top highword of the first operand.
2. Adds the bottom halfword of the first operand with the top halfword of the second operand.
3. Saturates the result of the subtraction and writes a 16-bit unsigned integer in the range
0 ≤ x ≤ 16
– 1, where x equals 16, to the top halfword of the destination register.
4. Saturates the results of the addition and writes a 16-bit unsigned integer in the range 0
≤ ≤ 2 16 – 1, where x equals 16, to the bottom halfword of the destination register.
134/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not affect the condition code flags.
Examples
UQASX R7, R4, R2 ; Adds top halfword of R4 with bottom halfword of R2,
; saturates to 16 bits, writes to top halfword of R7
; Subtracts top halfword of R2 from bottom halfword of
; R4, saturates to 16 bits, writes to bottom halfword
; of R7
UQSAX R0, R3, R5 ; Subtracts bottom halfword of R5 from top halfword of
; R3, saturates to 16 bits, writes to top halfword of
; R0
; Adds bottom halfword of R4 to top halfword of R5
; saturates to 16 bits, writes to bottom halfword of
; R0.
PM0253 Rev 5 135/254
182
The Cortex-M7 instruction set
3.7.7 UQADD and UQSUB
Saturating Add and Saturating Subtract Unsigned.
Syntax
PM0253 op{ cond } { Rd }, Rn , Rm op{ cond } { Rd }, Rn , Rm
Where: op cond
Rd
Rn, Rm
Is one of:
UQADD8
Saturating four unsigned 8-bit integer additions.
UQADD16
Saturating two unsigned 16-bit integer additions.
UDSUB8
Saturating four unsigned 8-bit integer subtractions.
UQSUB16
Saturating two unsigned 16-bit integer subtractions.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Are registers holding the first and second operands.
Operation
These instructions add or subtract two or four values and then writes an unsigned saturated value in the destination register.
The UQADD16 instruction:
• Adds the respective top and bottom halfwords of the first and second operands.
• Saturates the result of the additions for each halfword in the destination register to the unsigned range 0 ≤ x ≤ 2
16 − 1, where x is 16.
The UQADD8 instruction:
• Adds each respective byte of the first and second operands.
• Saturates the result of the addition for each byte in the destination register to the unsigned range 0 ≤ x ≤ 2
8 − 1, where x is 8.
The UQSUB16 instruction:
• Subtracts both halfwords of the second operand from the respective halfwords of the first operand.
• Saturates the result of the differences in the destination register to the unsigned range
0 ≤ x ≤ 2
16 − 1, where x is 16.
The UQSUB8 instructions:
• Subtracts the respective bytes of the second operand from the respective bytes of the first operand.
• Saturates the results of the differences for each byte in the destination register to the unsigned range 0 ≤ x ≤ 2
8 − 1, where x is 8.
Restrictions
Do not use SP and do not use PC .
136/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Condition flags
These instructions do not affect the condition code flags.
Examples
UQADD16 R7, R4, R2 ; Adds halfwords in R4 to corresponding halfword in
; R2, saturates to 16 bits, writes to corresponding
; halfword of R7
UQADD8 R4, R2, R5 ; Adds bytes of R2 to corresponding byte of R5,
; saturates to 8 bits, writes to corresponding bytes
; of R4
UQSUB16 R6, R3, R0 ; Subtracts halfwords in R0 from corresponding
; halfword in R3, saturates to 16 bits, writes to
; corresponding halfword in R6
UQSUB8 R1, R5, R6 ; Subtracts bytes in R6 from corresponding byte of
; R5, saturates to 8 bits, writes to corresponding
; byte of R1.
Mnemonic
PKH
SXTAB
SXTAB16
SXTAH
SXTB
SXTB16
SXTH
UXTAB
UXTAB16
UXTAH
UXTB
UXTB16
UXTH
Table 33 shows the instructions that operate on packing and unpacking data:
Table 33. Packing and unpacking instructions
Brief description See
Pack Halfword
Extend 8 bits to 32 and add
Dual extend 8 bits to 16 and add
Extend 16 bits to 32 and add
Sign extend a byte
Dual extend 8 bits to 16 and add
Sign extend a halfword
Extend 8 bits to 32 and add
Dual extend 8 bits to 16 and add
Extend 16 bits to 32 and add
Zero extend a byte
Dual zero extend 8 bits to 16 and add
Zero extend a halfword
PM0253 Rev 5 137/254
182
The Cortex-M7 instruction set
3.8.1 PKHBT and PKHTB
Pack Halfword.
Syntax
PM0253 op{ cond } { Rd }, Rn , Rm {, LSL # imm } op{ cond } { Rd }, Rn , Rm {, ASR # imm }
Where: op cond
Rd
Rn
Rm imm
Is one of:
PKHBT
Pack Halfword, bottom and top with shift.
PKHTB
Pack Halfword, top and bottom with shift.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the first operand register.
Is the second operand register holding the value to be optionally shifted.
Is the shift length. The type of shift length depends on the instruction:
For
PKHBT
LSL
A left shift with a shift length from 1 to 31, 0 means no shift.
For
PKHTB
:
ASR
An arithmetic shift right with a shift length from 1 to 32, a shift of 32-bits is encoded as
0b00000
.
Operation
The PKHBT instruction:
1.
Writes the value of the bottom halfword of the first operand to the bottom halfword of the destination register.
2. If shifted, the shifted value of the second operand is written to the top halfword of the destination register.
The PKHTB instruction:
1.
Writes the value of the top halfword of the first operand to the top halfword of the destination register.
2. If shifted, the shifted value of the second operand is written to the bottom halfword of the destination register.
Restrictions
Rd must not be SP and must not be PC.
Condition flags
This instruction does not change the flags.
138/254 PM0253 Rev 5
PM0253
3.8.2
The Cortex-M7 instruction set
Examples
PKHBT R3, R4, R5 LSL #0 ; Writes bottom halfword of R4 to bottom
; halfworfd of R3, writes top halfword of R5,
; unshifted, to top halfword of R3
PKHTB R4, R0, R2 ASR #1 ; Writes R2 shifted right by 1 bit to bottom
; halfword of R4, and writes top halfword of R0
; to top halfword of R4.
SXT and UXT
Sign extend and Zero extend.
Syntax op{ cond } { Rd ,} Rm {, ROR # n } op{ cond } { Rd }, Rm {, ROR # n }
Where: op cond
Rd
Rm
ROR # n
Is one of:
SXTB
Sign extends an 8-bit value to a 32-bit value.
SXTH
Sign extends a 16-bit value to a 32-bit value.
SXTB16
Sign extends two 8-bit values to two 16-bit values.
UXTB
UXTH
Zero extends an 8-bit value to a 32-bit value.
Zero extends a 16-bit value to a 32-bit value.
UXTB16
Zero extends two 8-bit values to two 16-bit values.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the register holding the value to extend.
Is one of:
ROR #8
Value from
Rm
is rotated right 8 bits.
ROR #16
Value from
Rm
is rotated right 16 bits.
ROR #24
Value from
Rm
is rotated right 24 bits.
If
ROR # n
is omitted, no rotation is performed.
Operation
These instructions do the following:
1.
Rotate the value from Rm right by 0, 8, 16 or 24 bits.
2. Extract bits from the resulting value:
• SXTB extracts bits[7:0] and sign extends to 32 bits.
•
• UXTB extracts bits[7:0] and zero extends to 32 bits.
SXTH extracts bits[15:0] and sign extends to 32 bits.
•
•
• UXTH extracts bits[15:0] and zero extends to 32 bits.
SXTB16 extracts bits[7:0] and sign extends to 16 bits, and extracts bits [23:16] and sign extends to 16 bits.
UXTB16 extracts bits[7:0] and zero extends to 16 bits, and extracts bits [23:16] and zero extends to 16 bits.
PM0253 Rev 5 139/254
182
The Cortex-M7 instruction set
3.8.3
PM0253
Restrictions
Do not use SP and do not use PC.
Condition flags
These instructions do not affect the flags.
Examples
SXTH R4, R6, ROR #16 ; Rotates R6 right by 16 bits, obtains bottom
; halfword of result, sign extends to 32 bits and
; writes to R4
UXTB R3, R10 ; Extracts lowest byte of value in R10, zero
; extends, and writes to R3.
SXTA and UXTA
Signed and Unsigned Extend and Add.
Syntax op { cond } { Rd ,} Rn , Rm {, ROR # n } op { cond } { Rd ,} Rn , Rm {, ROR # n }
Where: op cond
Rd
Rn
Rm
ROR # n
Is one of:
SXTAB
Sign extends an 8-bit value to a 32-bit value and add.
SXTAH
Sign extends a 16-bit value to a 32-bit value and add.
SXTAB16
Sign extends two 8-bit values to two 16-bit values and add.
UXTAB
Zero extends an 8-bit value to a 32-bit value and add.
UXTAH
Zero extends a 16-bit value to a 32-bit value and add.
UXTAB16
Zero extends two 8-bit values to two 16-bit values and add.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the first operand register.
Is the register holding the value to rotate and extend.
Is one of:
ROR #8
Value from
Rm
is rotated right 8 bits.
ROR #16
Value from
Rm
is rotated right 16 bits.
ROR #24
Value from
Rm
is rotated right 24 bits.
If
ROR # n
is omitted, no rotation is performed.
Operation
These instructions do the following:
1.
Rotate the value from Rm right by 0, 8, 16 or 24 bits.
2. Extract bits from the resulting value:
• SXTAB extracts bits[7:0] from Rm and sign extends to 32 bits.
• UXTAB extracts bits[7:0] from Rm and zero extends to 32 bits.
140/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
•
•
•
•
SXTAH extracts bits[15:0] from Rm and sign extends to 32 bits.
UXTAH extracts bits[15:0] from Rm and zero extends to 32 bits.
SXTAB16 extracts bits[7:0] from Rm and sign extends to 16 bits, and extracts bits [23:16] from Rm and sign extends to 16 bits.
UXTAB16 extracts bits[7:0] from Rm and zero extends to 16 bits, and extracts bits [23:16] from Rm and zero extends to 16 bits.
3. Adds the signed or zero extended value to the word or corresponding halfword of Rn and writes the result in Rd .
Restrictions
Do not use SP and do not use PC.
Condition flags
These instructions do not affect the flags.
Examples
SXTAH R4, R8, R6, ROR #16 ; Rotates R6 right by 16 bits, obtains bottom
; halfword, sign extends to 32 bits, adds
; R8, and writes to R4
UXTAB R3, R4, R10 ; Extracts bottom byte of R10 and zero extends
; to 32 bits, adds R4, and writes to R3.
3.9 Bit instructions
Table 34 shows the instructions that operate on adjacent sets of bits in registers or bit fields:
Mnemonic
BFC
BFI
SBFX
SXTB
SXTH
UBFX
UXTB
UXTH
Table 34. Packing and unpacking instructions
Brief description See
Bit Field Clear
Bit Field Insert
Signed Bit Field Extract
Sign extend a byte
Sign extend a halfword
Unsigned Bit Field Extract
Zero extend a byte
Zero extend a halfword
PM0253 Rev 5 141/254
182
The Cortex-M7 instruction set
3.9.1 BFC and BFI
Bit Field Clear and Bit Field Insert.
Syntax
PM0253
BFC{ cond } Rd , # lsb , # width
BFI{ cond } Rd , Rn , # lsb , # width
Where: cond
Rd
Rn lsb width
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the source register.
Is the position of the least significant bit of the bit field. lsb
must be in the range
0 to 31.
Is the width of the bit field and must be in the range 1 to 32 − lsb
.
Operation
BFC clears a bit field in a register. It clears width bits in Rd , starting at the low bit position lsb . Other bits in Rd are unchanged.
BFI copies a bit field into one register from another register. It replaces width bits in Rd starting at the low bit position lsb , with width bits from Rn starting at bit[0]. Other bits in Rd are unchanged.
Restrictions
Do not use SP and do not use PC.
Condition flags
These instructions do not affect the flags.
Examples
BFC R4, #8, #12 ; Clear bit 8 to bit 19 (12 bits) of R4 to 0
BFI R9, R2, #8, #12 ; Replace bit 8 to bit 19 (12 bits) of R9 with
; bit 0 to bit 11 from R2.
142/254 PM0253 Rev 5
PM0253
3.9.2
The Cortex-M7 instruction set
SBFX and UBFX
Signed Bit Field Extract and Unsigned Bit Field Extract.
Syntax
SBFX{ cond } Rd , Rn , # lsb , # width
UBFX{ cond } Rd , Rn , # lsb , # width
Where: cond
Rd
Rn lsb width
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the source register.
Is the position of the least significant bit of the bit field. lsb
must be in the range
0 to 31.
Is the width of the bit field and must be in the range 1 to 32 − lsb
.
Operation
SBFX extracts a bit field from one register, sign extends it to 32 bits, and writes the result to the destination register.
UBFX extracts a bit field from one register, zero extends it to 32 bits, and writes the result to the destination register.
Restrictions
Do not use SP and do not use PC .
Condition flags
These instructions do not affect the flags.
Examples
SBFX R0, R1, #20, #4 ; Extract bit 20 to bit 23 (4 bits) from R1 and
; sign extend to 32 bits and then write the
; result to R0.
UBFX R8, R11, #9, #10 ; Extract bit 9 to bit 18 (10 bits) from R11 and
; zero extend to 32 bits and then write the
; result to R8.
PM0253 Rev 5 143/254
182
The Cortex-M7 instruction set
3.9.3 SXT and UXT
Sign extend and Zero extend.
Syntax
PM0253
SXT extend { cond } { Rd ,} Rm {, ROR # n }
UXT extend { cond } { Rd }, Rm {, ROR # n }
Where: extend cond
Rd
Rm
ROR # n
Is one of:
B
Extends an 8-bit value to a 32-bit value.
H
Extends a 16-bit value to a 32-bit value.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the register holding the value to extend.
Is one of:
ROR #8
Value from
Rm
is rotated right 8 bits.
ROR #16
Value from
Rm
is rotated right 16 bits.
ROR #24
Value from
Rm
is rotated right 24 bits.
If
ROR # n
is omitted, no rotation is performed.
Operation
These instructions do the following:
1.
Rotate the value from Rm right by 0, 8, 16 or 24 bits.
2. Extract bits from the resulting value:
• SXTB extracts bits[7:0] and sign extends to 32 bits.
•
•
•
UXTB extracts bits[7:0] and zero extends to 32 bits.
SXTH extracts bits[15:0] and sign extends to 32 bits.
UXTH extracts bits[15:0] and zero extends to 32 bits.
Restrictions
Do not use SP and do not use PC.
Condition flags
These instructions do not affect the flags.
Examples
SXTH R4, R6, ROR #16 ; Rotate R6 right by 16 bits, then obtain the
; lower halfword of the result and then sign
; extend to 32 bits and write the result to R4.
UXTB R3, R10 ; Extract lowest byte of the value in R10 and
; zero extend it, and write the result to R3.
144/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
3.10 Branch and control instructions
Table 35 shows the branch and control instructions:
Mnemonic
B
BL
BLX
BX
CBNZ
CBZ
IT
TBB
TBH
Table 35. Branch and control instructions
Brief description See
Branch
Branch with Link
Branch indirect with Link
Branch indirect
Compare and Branch if Non Zero
Compare and Branch if Zero
If-Then
Table Branch Byte
Table Branch Halfword
B, BL, BX, and BLX on page 145
B, BL, BX, and BLX on page 145
B, BL, BX, and BLX on page 145
B, BL, BX, and BLX on page 145
3.10.1 B, BL, BX, and BLX
Branch instructions.
Syntax
B{ cond } label
BL{ cond } label
BX{ cond } Rm
BLX{ cond } Rm
Where:
B
BL
BX
BLX cond label
Rm
Is branch (immediate).
Is branch with link (immediate).
Is branch indirect (register).
Is branch indirect with link (register).
Is an optional condition code. See Conditional execution on page 68 .
Is a PC-relative expression. See PC-relative expressions on page 68 .
Is a register that indicates an address to branch to. Bit[0] of the value in
Rm must be 1, but the address to branch to is created by changing bit[0] to 0.
Operation
All these instructions cause a branch to label , or to the address indicated in Rm . In addition:
• The BL and BLX instructions write the address of the next instruction to LR (the link register, R14).
• The BX and BLX instructions result in a UsageFault exception if bit[0] of Rm is 0.
PM0253 Rev 5 145/254
182
The Cortex-M7 instruction set PM0253
B cond label is the only conditional instruction that can be either inside or outside an IT block. All other branch instructions must be conditional inside an IT block, and must be unconditional outside the IT block, see
Table 36 shows the ranges for the various branch instructions
.
Instruction
B label
B cond label (outside IT block)
B cond label (inside IT block)
BL{ cond } label
BX{ cond } Rm
BLX{ cond } Rm
Table 36. Branch ranges
Branch range
− 16 MB to +16 MB
− 1 MB to +1 MB
− 16 MB to +16 MB
− 16 MB to +16 MB
Any value in register
Any value in register
The user might have to use the .W suffix to get the maximum branch range. See
Instruction width selection on page 71 .
Restrictions
The restrictions are:
• Do not use PC in the BLX instruction.
• For BX and BLX, bit[0] of Rm must be 1 for correct execution but a branch occurs to the target address created by changing bit[0] to 0.
• When any of these instructions is inside an IT block, it must be the last instruction of the
IT block.
B cond is the only conditional instruction that is not required to be inside an IT block.
However, it has a longer branch range when it is inside an IT block.
Condition flags
These instructions do not change the flags.
Examples
B loopA ; Branch to loopA
BLE ng ; Conditionally branch to label ng
B.W target ; Branch to target within 16MB range
BEQ target ; Conditionally branch to target
BEQ.W target ; Conditionally branch to target within 1MB
BL funC ; Branch with link (Call) to function funC, return
; address stored in LR
BX LR ; Return from function call
BXNE R0 ; Conditionally branch to address stored in R0
BLX R0 ; Branch with link and exchange (Call) to a address
; stored in R0.
146/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
3.10.2 CBZ and CBNZ
Compare and Branch on Zero, Compare and Branch on Non-Zero.
Syntax
CBZ Rn , label
CBNZ Rn , label
Where:
Rn label
Is the register holding the operand.
Is the branch destination.
Operation
Use the CBZ or CBNZ instructions to avoid changing the condition code flags and to reduce the number of instructions.
CBZ Rn, label
does not change condition flags but is otherwise equivalent to:
CMP Rn, #0
BEQ label
CBNZ Rn, label
does not change condition flags but is otherwise equivalent to:
CMP Rn, #0
BNE label
Restrictions
The restrictions are:
• Rn must be in the range of R0 to R7.
• The branch destination must be within 4 to 130 bytes after the instruction.
• These instructions must not be used inside an IT block.
Condition flags
These instructions do not change the flags.
Examples
CBZ R5, target ; Forward branch if R5 is zero
CBNZ R0, target ; Forward branch if R0 is not zero
PM0253 Rev 5 147/254
182
The Cortex-M7 instruction set PM0253
3.10.3 IT
If-Then condition instruction.
Syntax
IT{ x { y { z }}} cond
Where: x y z cond specifies the condition switch for the second instruction in the IT block.
Specifies the condition switch for the third instruction in the IT block.
Specifies the condition switch for the fourth instruction in the IT block.
Specifies the condition for the first instruction in the IT block.
The condition switch for the second, third and fourth instruction in the IT block can be either:
T Then. Applies the condition cond to the instruction.
E Else. Applies the inverse condition of cond to the instruction.
It is possible to use AL (the always condition) for cond in an IT instruction. If this is done, all of the instructions in the IT block must be unconditional, and each of x , y , and z must be T or omitted but not E .
Operation
The IT instruction makes up to four following instructions conditional. The conditions can be all the same, or some of them can be the logical inverse of the others. The conditional instructions following the IT instruction form the IT block .
The instructions in the IT block, including any branches, must specify the condition in the
{ cond } part of their syntax.
The assembler might be able to generate the required IT instructions for conditional instructions automatically, so it is not needed to write them yourself. See the assembler documentation for details.
A BKPT instruction in an IT block is always executed, even if its condition fails.
Exceptions can be taken between an IT instruction and the corresponding IT block, or within an IT block. Such an exception results in entry to the appropriate exception handler, with suitable return information in LR and stacked PSR.
Instructions designed for use for exception returns can be used as normal to return from the exception, and execution of the IT block resumes correctly. This is the only way that a
PC-modifying instruction is permitted to branch to an instruction in an IT block.
Restrictions
The following instructions are not permitted in an IT block:
• IT.
• CBZ and CBNZ.
• CPSID and CPSIE.
148/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Other restrictions when using an IT block are:
• A branch or any instruction that modifies the PC must either be outside an IT block or must be the last instruction inside the IT block. These are:
– ADD PC, PC, Rm.
– MOV PC, Rm.
– B, BL, BX, BLX.
– Any LDM, LDR, or POP instruction that writes to the PC.
– TBB and TBH.
• Do not branch to any instruction inside an IT block, except when returning from an exception handler
• All conditional instructions except B cond must be inside an IT block. B cond can be either outside or inside an IT block but has a larger branch range if it is inside one
• Each instruction inside the IT block must specify a condition code suffix that is either the same or logical inverse as for the other instructions in the block.
The assembler might place extra restrictions on the use of IT blocks, such as prohibiting the use of assembler directives within them.
Condition flags
This instruction does not change the flags.
Example
ITTE NE ; Next 3 instructions are conditional
ANDNE R0, R0, R1 ; ANDNE does not update condition flags
ADDSNE R2, R2, #1 ; ADDSNE updates condition flags
MOVEQ R2, R3 ; Conditional move
CMP R0, #9 ; Convert R0 hex value (0 to 15) into ASCII
; ('0'-'9', 'A'-'F')
ITE GT ; Next 2 instructions are conditional
ADDGT R1, R0, #55 ; Convert 0xA -> 'A'
ADDLE R1, R0, #48 ; Convert 0x0 -> '0'
IT GT ; IT block with only one conditional instruction
ADDGT R1, R1, #1 ; Increment R1 conditionally
ITTEE EQ ; Next 4 instructions are conditional
MOVEQ R0, R1 ; Conditional move
ADDEQ R2, R2, #10 ; Conditional add
ANDNE R3, R3, #1 ; Conditional AND
BNE.W dloop ; Branch instruction can only be used in the last
; instruction of an IT block
IT NE ; Next instruction is conditional
ADD R0, R0, R1 ; Syntax error: no condition code used in IT block
PM0253 Rev 5 149/254
182
The Cortex-M7 instruction set PM0253
3.10.4 TBB and TBH
Table Branch Byte and Table Branch Halfword.
Syntax
TBB [ Rn , Rm ]
TBH [ Rn , Rm , LSL #1]
Where:
Rn
Rm
Is the register containing the address of the table of branch lengths.
If
Rn
is PC, then the address of the table is the address of the byte immediately following the
TBB
or
TBH
instruction.
Is the index register. This contains an index into the table. For halfword tables,
LSL #1
doubles the value in
Rm
to form the right offset into the table.
Operation
These instructions cause a PC-relative forward branch using a table of single byte offsets for
TBB, or halfword offsets for TBH. Rn provides a pointer to the table, and Rm supplies an index into the table. For TBB the branch offset is twice the unsigned value of the byte returned from the table, and for TBH the branch offset is twice the unsigned value of the halfword returned from the table. The branch occurs to the address at that offset from the address of the byte immediately after the TBB or TBH instruction.
Restrictions
The restrictions are:
• Rn must not be SP.
• Rm must not be SP and must not be PC.
• When any of these instructions is used inside an IT block, it must be the last instruction of the IT block.
Condition flags
These instructions do not change the flags.
Examples
ADR.W R0, BranchTable_Byte
TBB [R0, R1] ; R1 is the index, R0 is the base address of
; the branch table
Case1
; an instruction sequence follows
Case2
; an instruction sequence follows
Case3
; an instruction sequence follows
BranchTable_Byte
DCB 0 ; Case1 offset calculation
DCB ((Case2-Case1)/2) ; Case2 offset calculation
DCB ((Case3-Case1)/2) ; Case3 offset calculation
TBH [PC, R1, LSL #1] ; R1 is the index, PC is used as base of the
150/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
; branch table
BranchTable_H
DCW ((CaseA - BranchTable_H)/2) ; CaseA offset calculation
DCW ((CaseB - BranchTable_H)/2) ; CaseB offset calculation
DCW ((CaseC - BranchTable_H)/2) ; CaseC offset calculation
CaseA
; an instruction sequence follows
CaseB
; an instruction sequence follows
CaseC
; an instruction sequence follows
This section provides the instruction set that the single-precision and double-precision FPU uses.
Table 37 shows the floating-point instructions.
These instructions are only available if the FPU is included, and enabled, in the system.
See
Enabling the FPU on page 238 for information about enabling the floating-point unit.
Mnemonic
VABS
VADD
VCMP
VCMPE
VCVT
VCVT
VCVTR
VCVTB
VCVTT
VDIV
VFMA
VFNMA
VFMS
VFNMS
VLDM
Table 37. Floating-point instructions
Brief description See
Floating-point Absolute
Floating-point Add
Compares two floating-point registers, or one floating-point register and zero
Compares two floating-point registers, or one floating-point register and zero with Invalid
Operation check
Converts between floating-point and integer
Converts between floating-point and fixed point
VCVT between floating-point and fixed-point on page 156
VCVT between floating-point and fixed-point on page 156
Converts between floating-point and integer with rounding
VCVT, VCVTR between floating-point and integer on page 155
Converts half-precision value to single-precision
Converts single-precision register to half-precision
Floating-point Divide
Floating-point Fused Multiply Accumulate
Floating-point Fused Negate Multiply Accumulate
Floating-point Fused Multiply Subtract
Floating-point Fused Negate Multiply Subtract
Loads Multiple extension registers
PM0253 Rev 5 151/254
182
The Cortex-M7 instruction set PM0253
Mnemonic
VLDR
VMLA
VMLS
VMOV
VMOV
VMOV
VMOV
VMOV
VMOV
VMRS
VMSR
VMUL
VNEG
VNMLA
VNMLS
VNMUL
VPOP
VPUSH
VSQRT
VSTM
VSTR
VSUB
VSEL
VMAXNM,
VMINNM
VCVTA,
VCVTN,
VCVTP,
VCVTM
Table 37. Floating-point instructions (continued)
Brief description
Loads an extension register from memory
Floating-point Multiply Accumulate
Floating-point Multiply Subtract
Floating-point Move Immediate
Floating-point Move register
Copies Arm core register to single-precision
Copies 2 Arm core registers to 2 single-precision
VMOV Arm core register to single-precision on page 163
VMOV two Arm core registers to two singleprecision registers on page 164
VMOV Arm core register to scalar on page 165
VMOV scalar to Arm core register on page 163
Copies between Arm core register to scalar
Copies between Scalar to Arm core register
Moves to Arm core register from floating-point
System register
Moves to floating-point System register from Arm core register
Multiplies floating-point
Floating-point negate
Floating-point multiply and add
Floating-point multiply and subtract
Floating-point multiply
Pop extension registers
Pushes extension registers
Floating-point square root
Stores Multiple extension registers
VNMLA, VNMLS, VNMUL on page 167
VNMLA, VNMLS, VNMUL on page 167
VNMLA, VNMLS, VNMUL on page 167
Stores an extension register to memory
Floating-point Subtract
Selects register, alternative to a pair of conditional
VMOV
Maximum, Minimum with IEEE754-2008 NaN handling
Float to integer conversion with directed rounding
See
VCVTA, VCVTN, VCVTP, VCVTM on page 173
152/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Mnemonic
VRINTR,
VRINTX
VRINTA,
VRINTN,
VRINTP,
VRINTM
Table 37. Floating-point instructions (continued)
Brief description
Float to integer (in floating-point format) conversion
Float to integer (in floating-point format) conversion with directed rounding
See
VRINTA, VRINTN, VRINTP, VRINTM, VRINTZ on page 174
3.11.1 VABS
Floating-point Absolute.
Syntax
VABS{ cond }.F<32|64> <Sd|Dd> , <Sm|Dm>
Where: cond
Is an optional condition code. See Conditional execution on page 68 .
<Sd|Dd> , <Sm|Dm>
Are the destination floating-point value and the operand floating-point value.
Operation
This instruction:
1.
Takes the absolute value of the operand floating-point register.
2. Places the results in the destination floating-point register.
Restrictions
There are no restrictions.
Condition flags
This instruction does not change the flags.
Examples
VABS.F32 S4, S6
3.11.2 VADD
Floating-point Add.
Syntax
VADD{ cond }.F<32|64> { <Sd|Dd> ,} <Sn|Dn> , <Sm|Dm>
VADD{ cond }.F64 { Dd ,} Dn , Dm
Where: cond
<Sd|Dd>
Is an optional condition code. See Conditional execution on page 68
Is the destination floating-point value.
PM0253 Rev 5 153/254
182
The Cortex-M7 instruction set
<Sn|Dn> , <Sm|Dm>
Are the operand floating-point values.
Operation
This instruction:
1.
Adds the values in the two floating-point operand registers.
2. Places the results in the destination floating-point register.
Restrictions
There are no restrictions.
Condition flags
This instruction does not change the flags.
Examples
VADD.F32 S4, S6, S7
PM0253
154/254
Compares two floating-point registers, or one floating-point register and zero.
Syntax
VCMP{ E }{ cond }.F<32|64> <Sd|Dd> , <Sm|Dm>
VCMP{ E }{ cond }.F<32|64> <Sd|Dd> , #0.0
Where: cond
E
<Sd|Dd>
<Sm|Dm>
Is an optional condition code. See Conditional execution on page 68
Is the floating-point operand that is compared with.
If present, any
NaN
operand causes an
Invalid Operation
exception.
Otherwise, only a signaling
NaN
causes the exception.
Is the floating-point operand to compare.
Operation
This instruction:
1.
Compares either:
• Two floating-point registers.
• Or one floating-point register and zero.
2. Writes the result to the FPSCR flags.
Restrictions
This instruction can optionally raise an Invalid Operation exception if either operand is any type of NaN. It always raises an Invalid Operation exception if either operand is a signaling
NaN.
Condition flags
When this instruction writes the result to the FPSCR flags, the values are normally transferred to the Arm flags by a subsequent VMRS instruction, see
.
PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Examples
VCMP.F32 S4, #0.0VCMP.F32 S4, S2
3.11.4 VCVT, VCVTR between floating-point and integer
Converts a value in a register from floating-point to and from a 32-bit integer.
Syntax
VCVT{ R }{ cond }.
Tm .F<32|64> <Sd|Dd> , <Sm|Dm>
VCVT{ cond }.F<32|64>.
Tm <Sd|Dd> , <Sm|Dm>
Where:
R cond
Tm
•
•
If
R
is specified, the operation uses the rounding mode specified by the
FPSCR
.
If
R
is omitted. the operation uses the
Round towards Zero
rounding mode.
Is an optional condition code. See Conditional execution on page 68 .
Is the data type for the operand. It must be one of:
S32
U32
signed 32-bit value.
unsigned 32-bit value.
<Sd|Dd> , <Sm|Dm>
Are the destination register and the operand register.
Operation
These instructions:
1.
Either:
• Convert a value in a register from floating-point value to a 32-bit integer.
• Convert from a 32-bit integer to floating-point value.
2. Place the result in a second register.
The floating-point to integer operation normally uses the Round towards Zero rounding mode, but can optionally use the rounding mode specified by the FPSCR.
The integer to floating-point operation uses the rounding mode specified by the FPSCR.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
PM0253 Rev 5 155/254
182
The Cortex-M7 instruction set PM0253
Converts a value in a register from floating-point to and from fixed-point.
Syntax
VCVT{ cond }.
Td .F<32|64> <Sd|Dd> , <Sd|Dd> , # fbits
VCVT{ cond }.F<32|64>.
Td <Sd|Dd> , <Sd|Dd> , # fbits
Where: cond
Td
<Sd|Dd> fbits
•
•
Is an optional condition code. See
Is the data type for the fixed-point number. It must be one of:
S16
U16
signed 16-bit value.
unsigned 16-bit value.
Conditional execution on page 68
•
•
•
S32
signed 32-bit value.
•
U32
unsigned 32-bit value.
Is the destination register and the operand register.
Is the number of fraction bits in the fixed-point number:
If
Td
is
S16
or
U16
, fbits
must be in the range 0-16.
If
Td
is
S32
or
U32
, fbits
must be in the range 1-32.
Operation
This instruction:
1.
Either
• Converts a value in a register from floating-point to fixed-point.
• Converts a value in a register from fixed-point to floating-point.
2. Places the result in a second register.
The floating-point values are single-precision or double-precision.
The fixed-point value can be 16-bit or 32-bit. Conversions from fixed-point values take their operand from the low-order bits of the source register and ignore any remaining bits.
Signed conversions to fixed-point values sign-extend the result value to the destination register width.
Unsigned conversions to fixed-point values zero-extend the result value to the destination register width.
The floating-point to fixed-point operation uses the Round towards Zero rounding mode.
The fixed-point to floating-point operation uses the Round to Nearest rounding mode.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
156/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Converts between half-precision and single-precision or double-precision without intermediate rounding.
Syntax
VCVT{ y }{ cond }.F<32|64>.F16 <Sd|Dd> , Sm
VCVT{ y }{ cond }.F16.F<32|64> Sd , <Sm|Dm>
Where: y cond
<Sd|Dd>
<Sm|Dm>
•
Specifies which half of the operand register
Sm
or destination register
Sd
is used for the operand or destination:
If y is
B
, then the bottom half, bits [15:0], of
Sm
or
Sd
is used.
• If y
is
T
, then the top half, bits [31:16], of
Sm
or
Sd
is used.
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Is the operand register.
Operation
This instruction with the
.F16.F<32|64>
suffix:
1.
Converts the half-precision value in the top or bottom half of a single-precision register to single-precision or double-precision value.
2. Writes the result to a single-precision or double-precision register.
This instruction with the
.F<32|64>.F16
suffix:
1.
Converts the value in a double-precision or single-precision register to half-precision value.
2. Writes the result into the top or bottom half of a single-precision register, preserving the other half of the target register.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
3.11.7 VDIV
Divides floating-point values.
Syntax
VDIV{ cond }.F<32|64> { <Sd|Dd> ,} <Sn|Dn> , <Sm|Dm>
Where: cond
<Sd|Dd>
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
<Sn|Dn> , <Sm|Dm>
Are the operand registers.
PM0253 Rev 5 157/254
182
The Cortex-M7 instruction set
Operation
This instruction:
1.
Divides one floating-point value by another floating-point value.
2. Writes the result to the floating-point destination register.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
Floating-point Fused Multiply Accumulate and Subtract.
Syntax
VFMA{ cond }.F<32|64> { <Sd|Dd>, } <Sn|Dn> , <Sm|Dm>
VFMS{ cond }.F<32|64> { <Sd|Dd>, } <Sn|Dn> , <Sm|Dm>
Where: cond
<Sd|Dd>
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
<Sn|Dn> , <Sm|Dm>
Are the operand registers.
Operation
The VFMA instruction:
1.
Multiplies the floating-point values in the operand registers.
2. Accumulates the results into the destination register.
The result of the multiply is not rounded before the accumulation.
The VFMS instruction:
1.
Negates the first operand register.
2. Multiplies the floating-point values of the first and second operand registers.
3. Adds the products to the destination register.
4. Places the results in the destination register.
The result of the multiply is not rounded before the addition.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
PM0253
158/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Floating-point Fused Negate Multiply Accumulate and Subtract.
Syntax
VFNMA{ cond }.F<32|64> { <Sd|Dd>, } <Sn|Dn> , <Sm|Dm>
VFNMS{ cond }.F<32|64> { <Sd|Dd>, } <Sn|Dn> , <Sm|Dm>
Where: cond
<Sd|Dd>
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
<Sn|Dn> , <Sm|Dm>
Are the operand registers.
Operation
The VFNMA instruction:
1.
Negates the first floating-point operand register.
2. Multiplies the first floating-point operand with second floating-point operand.
3. Adds the negation of the floating -point destination register to the product
4. Places the result into the destination register.
The result of the multiply is not rounded before the addition.
The VFNMS instruction:
1.
Multiplies the first floating-point operand with second floating-point operand.
2. Adds the negation of the floating-point value in the destination register to the product.
3. Places the result in the destination register.
The result of the multiply is not rounded before the addition.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
3.11.10 VLDM
Floating-point Load Multiple.
Syntax
VLDM{ mode }{ cond }{.
size } Rn { !
}, list
Where: mode cond
Is the addressing mode:
IA
Increment after. The consecutive addresses start at the address specified in Rn .
DB
Decrement before. The consecutive addresses end just before the address specified in Rn .
Is an optional condition code. See Conditional execution on page 68 .
PM0253 Rev 5 159/254
182
The Cortex-M7 instruction set PM0253 size
Rn
!
list
Is an optional data size specifier.
Is the base register. The
SP
can be used.
Is the command to the instruction to write a modified value back to
Rn
. This is required if mode == DB
, and is optional if mode == IA
.
Is the list of extension registers to be loaded, as a list of consecutively numbered doubleword or singleword registers, separated by commas and surrounded by brackets.
Operation
This instruction loads multiple extension registers from consecutive memory locations using an address from an Arm core register as the base address.
Restrictions
The restrictions are:
• If size is present, it must be equal to the size in bits, 32 or 64, of the registers in list .
• For the base address, the SP can be used. In the Arm instruction set, if !
is not specified the PC can be used.
• list must contain at least one register. If it contains doubleword registers, it must not contain more than 16 registers.
• If using the Decrement Before addressing mode, the write back flag, !
, must be appended to the base register specification.
Condition flags
These instructions do not change the flags.
Example
VLDMIA.F64 r1, {d3,d4,d5}
3.11.11 VLDR
Loads a single extension register from memory.
Syntax
VLDR{ cond }{.
F<32 |64>} <sd|Dd >, [ Rn {# imm }]
VLDR{ cond }{.
F<32 |64>} <sd|Dd >, label
VLDR{ cond }{.
F<32 |64>} <sd|Dd >, [PC, # imm }]
Where: cond
32 , 64
Dd
Sd
Rn imm label
Is an optional condition code. See Conditional execution on page 68
Are the optional data size specifiers.
Is the destination register for a doubleword load.
Is the destination register for a singleword load.
Is the base register. The SP can be used.
values are multiples of 4 in the range 0 to 1020.
Is the label of the literal data item to be loaded.
Is the + or - immediate offset used to form the address. Permitted address
160/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Operation
This instruction loads a single extension register from memory, using a base address from an Arm core register, with an optional offset.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
Multiplies two floating-point values, and accumulates or subtracts the result.
Syntax
VMLA{ cond }.F<32|64> <Sd|Dd> , <Sn|Dn> , <Sm|Dm>
VMLS{ cond }.F<32|64> <Sd|Dd> , <Sn|Dn> , <Sm|Dm>
Where: cond
<Sd|Dd>
Is an optional condition code. See Conditional execution on page 68 .
Is the destination floating-point value.
<Sn|Dn> , <Sm|Dm>
Are the operand floating-point values.
Operation
The floating-point Multiply Accumulate instruction:
1.
Multiplies two floating-point values.
2. Adds the results to the destination floating-point value.
The floating-point Multiply Subtract instruction:
1.
Multiplies two floating-point values.
2. Subtracts the products from the destination floating-point value.
3. Places the results in the destination register.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
PM0253 Rev 5 161/254
182
The Cortex-M7 instruction set
Moves floating-point Immediate.
Syntax
VMOV{ cond }.F<32|64> <Sd|Dd> , # imm
Where: cond
<Sd|Dd> imm
Is an optional condition code. See
Is the destination register.
Is a floating-point constant.
Conditional execution on page 68 .
Operation
This instruction copies a constant value to a floating-point register.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
Copies the contents of one register to another.
Syntax
VMOV{ cond }.F<32|64> <Sd|Dd> , <Sm| Dm>
Where: cond
Dd
Dm
Sd
Sm
Is an optional condition code. See Conditional execution on page 68
Is the destination register, for a doubleword operation.
Is the source register, for a doubleword operation.
Is the destination register, for a singleword operation.
Is the source register, for a singleword operation.
Operation
This instruction copies the contents of one floating-point register to another.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
PM0253
162/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
3.11.15 VMOV scalar to Arm core register
Transfers one word of a doubleword floating-point register to an Arm core register.
Syntax
VMOV{ cond } Rt , Dn [ x ]
Where: cond
Rt
Dn x
•
•
Is an optional condition code. See Conditional execution on page 68 .
Is the destination Arm core register.
Is the 64-bit doubleword register.
Specifies which half of the doubleword register to use:
If
If x x
is 0, use lower half of doubleword register
is 1, use upper half of doubleword register.
Operation
This instruction transfers one word from the upper or lower half of a doubleword floatingpoint register to an Arm core register.
Restrictions
Rt cannot be PC or SP.
Condition flags
These instructions do not change the flags.
3.11.16 VMOV Arm core register to single-precision
Transfers a single-precision register to and from an Arm core register.
Syntax
VMOV{ cond } Sn , Rt
VMOV{ cond } Rt , Sn
Where: cond
<Sn>
Rt
Is an optional condition code. See
Is the Arm core register.
Conditional execution on page 68
Is the single-precision floating-point register.
Operation
This instruction transfers:
• The contents of a single-precision register to an Arm core register.
• The contents of an Arm core register to a single-precision register.
Restrictions
Rt cannot be PC or SP.
Condition flags
These instructions do not change the flags.
PM0253 Rev 5 163/254
182
The Cortex-M7 instruction set PM0253
3.11.17 VMOV two Arm core registers to two single-precision registers
Transfers two consecutively numbered single-precision registers to and from two Arm core registers.
Syntax
VMOV{ cond } Sm , Sm1 , Rt , Rt2
VMOV{ cond } Rt , Rt2 , Sm , Sm1
Where: cond
Sm
Sm1
Rt
Rt2
Is an optional condition code. See Conditional execution on page 68
Is the Arm core register that
<Sm>
is transferred to or from.
Is the The Arm core register that
<Sm1>
is transferred to or from.
Is the first single-precision register.
Is the second single-precision register. This is the next single-precision register after
<Sm>
.
Operation
This instruction transfers:
• The contents of two consecutively numbered single-precision registers to two Arm core registers.
• The contents of two Arm core registers to a pair of single-precision registers.
Restrictions
The restrictions are:
• The floating-point registers must be contiguous, one after the other.
• The Arm core registers do not have to be contiguous.
• Rt cannot be PC or SP.
Condition flags
These instructions do not change the flags.
3.11.18 VMOV two Arm core registers and a double-precision register
Transfers two words from two Arm core registers to a doubleword register, or from a doubleword register to two Arm core registers.
Syntax
VMOV{ cond } Dm , Rt , Rt2
VMOV{ cond } Rt , Rt2 , Dm
Where: cond
Dm
Rt, Rt2
Is an optional condition code. See
Is the double-precision register.
Are the two Arm core registers.
Conditional execution on page 68 .
164/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Operation
This instruction:
• Transfers two words from two Arm core registers to a doubleword register.
• Transfers a doubleword register to two Arm core registers.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
3.11.19 VMOV Arm core register to scalar
Transfers one word to a floating-point register from an Arm core register.
Syntax
VMOV{ cond }{.
32 } Dd[x] , Rt
Where: cond
32
Dd[x]
Rt
Is an optional condition code. See
Is an optional data size specifier.
as follows:
Conditional execution on page 68
• If x is 0, the lower half is extracted.
• If x is 1, the upper half is extracted.
Is the source Arm core register.
Is the destination, where [x] defines which half of the doubleword is transferred,
Operation
This instruction transfers one word to the upper or lower half of a doubleword floating-point register from an Arm core register.
Restrictions
Rt cannot be PC or SP.
Condition flags
These instructions do not change the flags.
3.11.20 VMRS
Moves to Arm core register from floating-point System register.
Syntax
VMRS{ cond } Rt , FPSCR
VMRS{ cond } APSR_nzcv , FPSCR
Where: cond
Rt
Is an optional condition code. See Conditional execution on page 68
Is the destination Arm core register. This register can be R0-R14.
PM0253 Rev 5 165/254
182
The Cortex-M7 instruction set PM0253
APSR_nzcv
Transfer floating-point flags to the APSR flags.
Operation
This instruction performs one of the following actions:
• Copies the value of the FPSCR to a general-purpose register.
• Copies the value of the FPSCR flag bits to the APSR N, Z, C, and V flags.
Restrictions
Rt cannot be PC or SP.
Condition flags
These instructions optionally change the N, Z, C, and V flags.
3.11.21 VMSR
Moves to floating-point System register from Arm core register.
Syntax
VMSR{ cond } FPSCR, Rt
Where: cond
Rt
Is an optional condition code. See Conditional execution on page 68
Is the general-purpose register to be transferred to the FPSCR.
Operation
This instruction moves the value of a general-purpose register to the FPSCR. See
Floatingpoint status control register on page 236 for more information.
Restrictions
Rt cannot be PC or SP.
Condition flags
This instruction updates the FPSCR.
3.11.22 VMUL
Floating-point Multiply.
Syntax
VMUL{ cond }.F<32|64> { <Sd|Dd> ,} <Sn|Dn> , <Sm|Dm>
Where: cond
<Sd|Dd>
Is an optional condition code. See Conditional execution on page 68 .
Is the destination floating-point value.
<Sn|Dn> , <Sm|Dm>
Are the operand floating-point values.
166/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Operation
This instruction:
1.
Multiplies two floating-point values.
2. Places the results in the destination register.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
3.11.23 VNEG
Floating-point Negate.
Syntax
VNEG{ cond }.F<32|64> <Sd|Dd> , <Sm|Dm>
Where: cond
<Sd|Dd>
<Sm|Dm>
Is an optional condition code. See
Is the operand floating-point value.
Conditional execution on page 68
Is the destination floating-point value.
Operation
This instruction:
1.
Negates a floating-point value.
2. Places the results in a second floating-point register.
The floating-point instruction inverts the sign bit.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
Floating-point multiply with negation followed by add or subtract.
Syntax
VNMLA{ cond }.F<32|64> <Sd|Dd> , <Sn|Dn> , <Sm|Dm>
VNMLS{ cond }.F<32|64> <Sd|Dd> , <Sn|Dn> , <Sm|Dm>
VNMUL{ cond }.F<32|64> { <Sd|Dd> ,} <Sn|Dn> , <Sm|Dm>
Where: cond
<Sd|Dd>
Is an optional condition code. See Conditional execution on page 68
Is the destination floating-point register.
PM0253 Rev 5 167/254
182
The Cortex-M7 instruction set PM0253
<Sn|Dn> , <Sm|Dm> Are the operand floating-point registers.
Operation
The VNMLA instruction:
1.
Multiplies two floating-point register values.
2. Adds the negation of the floating-point value in the destination register to the negation of the product.
3. Writes the result back to the destination register.
The VNMLS instruction:
1.
Multiplies two floating-point register values.
2. Adds the negation of the floating-point value in the destination register to the product.
3. Writes the result back to the destination register.
The VNMUL instruction:
1.
Multiplies together two floating-point register values.
2. Writes the negation of the result to the destination register.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
3.11.25 VPOP
Floating-point extension register Pop.
Syntax
VPOP{ cond }{.
size } list
Where: cond size list
Is an optional condition code. See
Is an optional data size specifier. If present, it must be equal to the size in bits,
32 or 64, of the registers in list
.
Is a list of extension registers to be loaded, as a list of consecutively numbered doubleword or singleword registers, separated by commas and surrounded by brackets.
Conditional execution on page 68 .
Operation
This instruction loads multiple consecutive extension registers from the stack.
Restrictions
The list must contain at least one register, and not more than sixteen registers.
Condition flags
These instructions do not change the flags.
168/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
3.11.26 VPUSH
Floating-point extension register Push.
Syntax
VPUSH{ cond }{.
size } list
Where: cond size list
Is an optional condition code. See
Is an optional data size specifier. If present, it must be equal to the size in bits,
32 or 64, of the registers in list
.
Conditional execution on page 68 .
Is a list of the extension registers to be stored, as a list of consecutively numbered doubleword or singleword registers, separated by commas and surrounded by brackets.
Operation
This instruction stores multiple consecutive extension registers to the stack.
Restrictions list must contain at least one register, and not more than sixteen.
Condition flags
These instructions do not change the flags.
3.11.27 VSQRT
Floating-point Square Root.
Syntax
VSQRT{ cond }.F<32|64> <Sd|Dd> , <Sm|Dm>
Where: cond
<Sd|Dd>
<Sm|Dm>
Is an optional condition code. See
Is the operand floating-point value.
Conditional execution on page 68
Is the destination floating-point value.
Operation
This instruction:
• Calculates the square root of the value in a floating-point register.
• Writes the result to another floating-point register.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
PM0253 Rev 5 169/254
182
The Cortex-M7 instruction set PM0253
3.11.28 VSTM
Floating-point Store Multiple.
Syntax
VSTM{ mode }{ cond }{.
size } Rn { !
}, list
Where: mode cond size
Rn
!
list
•
Is the addressing mode:
IA Increment After . The consecutive addresses start at the address specified in
Rn
. This is the default and can be omitted.
• DB Decrement Before . The consecutive addresses end just before the address specified in
Rn
.
Is an optional condition code. See Conditional execution on page 68 .
Is an optional data size specifier. If present, it must be equal to the size in bits,
32 or 64, of the registers in list
.
Is the base register. The SP can be used.
Is the function that causes the instruction to write a modified value back to
Rn
.
Required if mode == DB
.
Is a list of the extension registers to be stored, as a list of consecutively numbered doubleword or singleword registers, separated by commas and surrounded by brackets.
Operation
This instruction stores multiple extension registers to consecutive memory locations using a base address from an Arm core register.
Restrictions
The restrictions are:
• list must contain at least one register. If it contains doubleword registers it must not contain more than 16 registers.
• Use of the PC as Rn is deprecated.
Condition flags
These instructions do not change the flags.
3.11.29 VSTR
Floating-point Store.
Syntax
VSTR{ cond }{.
32 } Sd , [ Rn {, # imm }]
VSTR{ cond }{.
64 } Dd , [ Rn {, # imm }]
Where: cond
32 , 64
Sd
Dd
Is an optional condition code. See Conditional execution on page 68
Are the optional data size specifiers.
Is the source register for a singleword store.
Is the source register for a doubleword store.
170/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Rn imm
Is the base register. The SP can be used.
Is the + or - immediate offset used to form the address. Values are multiples of
4 in the range 0-1020. imm
can be omitted, meaning an offset of
+0
.
Operation
This instruction stores a single extension register to memory, using an address from an Arm core register, with an optional offset, defined in imm :
Restrictions
The use of PC for Rn is deprecated.
Condition flags
These instructions do not change the flags.
3.11.30 VSUB
Floating-point Subtract.
Syntax
VSUB{ cond }.F<32|64> { <Sd|Dd> ,} <Sn|Dn> , <Sm|Dm>
Where: cond
<Sd|Dd>
Is an optional condition code. See Conditional execution on page 68 .
Is the destination floating-point value.
<Sn|Dn> , <Sm|Dm>
Are the operand floating-point values.
Operation
This instruction:
1.
Subtracts one floating-point value from another floating-point value.
2. Places the results in the destination floating-point register.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
PM0253 Rev 5 171/254
182
The Cortex-M7 instruction set PM0253
3.11.31 VSEL
Provides an alternative to a pair of conditional
VMOV
instructions.
Encoding
VSEL{ cond }.F<32|64> <Sd|Dd> , <Sn|Dn> , <Sm|Dm>
Where: c ond
<Sd|Dd>
Is an optional condition code. See Conditional execution on page 68 .
VSEL
has a subset of the condition codes. The condition codes for
VSEL
are limited to
GE
,
GT
,
EQ
and
VS
, with the effect that
LT
,
LE
,
NE
and
VC
is achievable by exchanging the source operands.
Is the destination single-precision or double-precision floating-point value.
<Sn|Dn> , <Sm|Dm>
Are the operand single-precision or double-precision floating-point values.
Operation
Depending on the result of the condition code, this instruction moves either:
• <Sn|Dn> source register to the destination register.
• <Sm|Dm> source register to the destination register.
Restrictions
The VSEL instruction must not occur inside an IT block.
Condition flags
These instructions do not change the flags.
Return the minimum or the maximum of two floating-point numbers with NaN handling as specified by IEEE754-2008.
Encoding
VMAXNM.F<32|64> <Sd|Dd> , <Sn|Dn> , <Sm|Dm>
VMINNM.F<32|64> <Sd|Dd> , <Sn|Dn> , <Sm|Dm>
Where:
<Sd|Dd>
Is the destination single-precision or double-precision floating-point value.
<Sn|Dn> , <Sm|Dm>
Are the operand single-precision or double-precision floating-point values.
Operation
The VMAXNM instruction compares two source registers, and moves the largest to the destination register.
The VMINNM instruction compares two source registers, and moves the lowest to the destination register.
172/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
3.11.33 VCVTA, VCVTN, VCVTP, VCVTM
Floating-point to integer conversion with directed rounding.
Syntax
VCVT <rmode> .S32.F<32|64> <Sd> , <Sm|Dm>
VCVT <rmode> .U32.F<32|64> <Sd> , <Sm|Dm>
Where:
<Sd|Dd>
Is the destination single-precision or double-precision floating-point value.
<Sn|Dn> , <Sm|Dm>
Are the operand single-precision or double-precision floating-point values.
<rmode>
Is one of:
A
Round to nearest ties away.
M
N
P
Round to nearest even.
Round towards plus infinity.
Round towards minus infinity.
Operation
These instructions:
1.
Read the source register.
2. Convert to integer with directed rounding.
3. Write to the destination register.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
Round a floating-point value to an integer in floating-point format.
Encoding
VRINT{R,X}{ cond }.F<32|64> <Sd|Dd> , <Sm|Dm>
Where: cond
<Sd|Dd>
<Sm|Dm>
Is an optional condition code. See Conditional execution on page 68
Is the destination floating-point value.
Are the operand floating-point values.
PM0253 Rev 5 173/254
182
The Cortex-M7 instruction set PM0253
Operation
These instructions:
1.
Read the source register.
2. Round to the nearest integer value in floating-point format using the rounding mode specified by the FPSCR.
3. Write the result to the destination register.
exact.
Restrictions
There are no restrictions.
Condition flags
These instructions do not change the flags.
Round a floating-point value to an integer in floating-point format using directed rounding.
Encoding
VRINT <rmode> .F<32|64> <Sd|Dd> , <Sm|Dm>
Where:
<Sd|Dd>
Is the destination single-precision or double-precision floating-point value.
<Sn|Dn> , <Sm|Dm>
Are the operand single-precision or double-precision floating-point values.
<rmode>
P
Z
M
N
Is one of:
A
Round to nearest ties away.
Round to Nearest Even.
Round towards Plus Infinity.
Round towards Minus Infinity.
Round towards Zero.
Operation
These instructions:
1.
Read the source register.
2. Round to the nearest integer value with a directed rounding mode specified by the instruction.
3. Write the result to the destination register.
Restrictions
These instructions cannot be conditional. These instructions cannot generate an inexact exception even if the result is not exact.
174/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Condition flags
These instructions do not change the flags.
3.12 Miscellaneous instructions
Table 38 shows the remaining Cortex
®
-M7 instructions:
Mnemonic
MSR
NOP
SEV
SVC
WFE
WFI
BKPT
CPSID
CPSIE
DMB
DSB
ISB
MRS
Table 38. Miscellaneous instructions
Brief description
Breakpoint
Change Processor State, Disable Interrupts
Change Processor State, Enable Interrupts
Data Memory Barrier
Data Synchronization Barrier
Instruction Synchronization Barrier
Move from special register to register
Move from register to special register
No Operation
Send Event
Supervisor Call
Wait For Event
Wait For Interrupt
See
3.12.1 BKPT
Breakpoint.
Syntax
BKPT # imm
Where: imm
Is an expression evaluating to an integer in the range 0-255 (8-bit value).
Operation
The BKPT instruction causes the processor to enter Debug state. Debug tools can use this to investigate system state when the instruction at a particular address is reached.
imm is ignored by the processor. If required, a debugger can use it to store additional information about the breakpoint.
The BKPT instruction can be placed inside an IT block, but it executes unconditionally, unaffected by the condition specified by the IT instruction.
PM0253 Rev 5 175/254
182
The Cortex-M7 instruction set PM0253
Condition flags
This instruction does not change the flags.
Examples
BKPT #0x3 ; Breakpoint with immediate value set to 0x3 (debugger can
; extract the immediate value by locating it using the PC)
Arm does not recommend the use of the BKPT instruction with an immediate value set to 0xAB for any purpose other than Semi-hosting.
3.12.2 CPS
Change Processor State.
Syntax
CPS effect iflags
Where: effect iflags
Is one of:
IE
Clears the special purpose register.
ID
Sets the special purpose register.
Is a sequence of one or more flags: i
Set or clear PRIMASK.
f
Set or clear FAULTMASK.
Operation
CPS changes the PRIMASK and FAULTMASK special register values. See
Exception mask registers on page 25 for more information about these registers.
Restrictions
The restrictions are:
• Use CPS only from privileged software. It has no effect if used in unprivileged software.
• CPS cannot be conditional and so must not be used inside an IT block.
Condition flags
This instruction does not change the condition flags.
Examples
CPSID i ; Disable interrupts and configurable fault handlers (set
; PRIMASK)
CPSID f ; Disable interrupts and all fault handlers (set FAULTMASK)
CPSIE i ; Enable interrupts and configurable fault handlers (clear
; PRIMASK)
CPSIE f ; Enable interrupts and fault handlers (clear FAULTMASK)
176/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
3.12.3 DMB
Data Memory Barrier.
Syntax
DMB{ cond }
Where: cond
Is an optional condition code. See Conditional execution on page 68 .
Operation
DMB acts as a data memory barrier. It ensures that all explicit memory accesses that appear, in program order, before the DMB instruction are completed before any explicit memory accesses that appear, in program order, after the DMB instruction. DMB does not affect the ordering or execution of instructions that do not access memory.
Condition flags
This instruction does not change the flags.
Examples
DMB ; Data Memory Barrier
3.12.4 DSB
Data Synchronization Barrier.
Syntax
DSB{ cond }
Where: cond
Is an optional condition code. See Conditional execution on page 68 .
Operation
DSB acts as a special data synchronization memory barrier. Instructions that come after the
DSB, in program order, do not execute until the DSB instruction completes. The DSB instruction completes when all explicit memory accesses before it complete.
Condition flags
This instruction does not change the flags.
Examples
DSB ; Data Synchronisation Barrier
PM0253 Rev 5 177/254
182
The Cortex-M7 instruction set PM0253
3.12.5 ISB
Instruction Synchronization Barrier.
Syntax
ISB{ cond }
Where: cond
Is an optional condition code. See Conditional execution on page 68 .
Operation
ISB acts as an instruction synchronization barrier. It flushes the pipeline of the processor, so that all instructions following the ISB are fetched from cache or memory again, after the ISB instruction has been completed.
Condition flags
This instruction does not change the flags.
Examples
ISB ; Instruction Synchronisation Barrier
3.12.6 MRS
Move the contents of a special register to a general-purpose register.
Syntax
MRS{ cond } Rd , spec_reg
Where: cond
Rd spec_reg
Is an optional condition code. See Conditional execution on page 68 .
Is the destination register.
Can be any of:
APSR
,
IPSR
,
EPSR
,
IEPSR
,
IAPSR
,
EAPSR
,
PSR
,
MSP
,
PSP
,
PRIMASK
,
BASEPRI
,
BASEPRI_MAX
,
FAULTMASK
, or
CONTROL
.
All the
EPSR
and
IPSR
fields are zero when read by the
MRS
instruction.
Operation
Use MRS in combination with MSR as part of a read-modify-write sequence for updating a
PSR, for example to clear the Q flag.
In process swap code, the programmers model state of the process being swapped out must be saved, including relevant PSR contents. Similarly, the state of the process being swapped in must also be restored. These operations use MRS in the state-saving instruction sequence and MSR in the state-restoring instruction sequence.
BASEPRI_MAX is an alias of BASEPRI when used with the MRS instruction.
See
178/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
Restrictions
Rd must not be SP and must not be PC.
Condition flags
This instruction does not change the flags.
Examples
MRS R0, PRIMASK ; Read PRIMASK value and write it to R0
3.12.7 MSR
Move the contents of a general-purpose register into the specified special register.
Syntax
MSR{ cond } spec_reg , Rn
Where: cond
Rn spec_reg
Is an optional condition code. See Conditional execution on page 68 .
Is the source register.
Can be any of:
APSR_nzcvq
,
APSR_g
,
APSR_nzcvqg
,
MSP
,
PSP
,
PRIMASK
,
BASEPRI
,
BASEPRI_MAX
,
FAULTMASK
, or
CONTROL
.
APSR
can be used to refer to
APSR_nzcvq
.
Operation
The register access operation in MSR depends on the privilege level. Unprivileged software
can only access the APSR, see Table 4 on page 23
. Privileged software can access all special registers.
In unprivileged software writes to unallocated or execution state bits in the PSR are ignored.
When writing to BASEPRI_MAX, the instruction writes to BASEPRI only if either:
• Rn is non-zero and the current BASEPRI value is 0.
• Rn is non-zero and less than the current BASEPRI value.
See
Restrictions
Rn must not be SP and must not be PC.
Condition flags
This instruction updates the flags explicitly based on the value in Rn .
Examples
MSR CONTROL, R1 ; Read R1 value and write it to the CONTROL register
PM0253 Rev 5 179/254
182
The Cortex-M7 instruction set PM0253
3.12.8 NOP
No Operation.
Syntax
NOP{ cond }
Where: cond
Is an optional condition code. See Conditional execution on page 68 .
Operation
NOP does nothing. NOP is not necessarily a time-consuming NOP. The processor might remove it from the pipeline before it reaches the execution stage.
Use NOP for padding, for example to place the following instruction on a 64-bit boundary.
Condition flags
This instruction does not change the flags.
Examples
NOP ; No operation
3.12.9 SEV
Send Event.
Syntax
SEV{ cond }
Where: cond
Is an optional condition code. See Conditional execution on page 68 .
Operation
SEV is a hint instruction that causes an event to be signaled to all processors within a
multiprocessor system. It also sets the local event register to 1, see Power management on page 50
.
Condition flags
This instruction does not change the flags.
Examples
SEV ; Send Event
180/254 PM0253 Rev 5
PM0253 The Cortex-M7 instruction set
3.12.10 SVC
Supervisor Call.
Syntax
SVC{ cond } # imm
Where: cond imm
Is an optional condition code. See Conditional execution on page 68 .
Is an expression evaluating to an integer in the range 0-255 (8-bit value).
Operation
The SVC instruction causes the SVC exception.
imm is ignored by the processor. If required, it can be retrieved by the exception handler to determine what service is being requested.
Condition flags
This instruction does not change the flags.
Examples
SVC #0x32 ; Supervisor Call (SVCall handler can extract the immediate
; value by locating it through the stacked PC)
3.12.11 WFE
Wait For Event.
Syntax
WFE{ cond }
Where: cond
Is an optional condition code. See Conditional execution on page 68 .
Operation
WFE is a hint instruction.
If the event register is 0, WFE suspends execution until one of the following events occurs:
• An exception, unless masked by the exception mask registers or the current priority level.
• An exception enters the Pending state, if SEVONPEND in the System Control register is set.
• A Debug Entry request, if Debug is enabled.
• An event signaled by a peripheral or another processor in a multiprocessor system using the SEV instruction.
If the event register is 1, WFE clears it to 0 and returns immediately.
For more information see
PM0253 Rev 5 181/254
182
The Cortex-M7 instruction set PM0253
Condition flags
This instruction does not change the flags.
Examples
WFE ; Wait for event
3.12.12 WFI
Wait for Interrupt.
Syntax
WFI{ cond }
Where: cond
Is an optional condition code. See Conditional execution on page 68 .
Operation
WFI is a hint instruction that suspends execution until one of the following events occurs:
• A non-masked interrupt occurs and is taken.
• An interrupt masked by PRIMASK becomes pending.
• A Debug Entry request.
Condition flags
This instruction does not change the flags.
Examples
WFI ; Wait for interrupt
182/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
4.1 About the Cortex-M7 peripherals
The address map of the Private peripheral bus (PPB) is:
Address
Table 39. Core peripheral register regions
Core peripheral
0xE000E008-0xE000E00F
0xE000E010-0xE000E01F
0xE000E100-0xE000E4EF
System control block
System timer
Nested Vectored Interrupt Controller
0xE000ED00-0xE000ED3F System control block
0xE000ED78- 0xE000ED84 Processor features
0xE000ED90-0xE000EDB8 Memory Protection Unit
0xE000EF00-0xE000EF03
0xE000EF30-0xE000EF44
0xE000EF50-0xE000EF78
0xE000EF90-0xE000EFA8
Nested Vectored Interrupt Controller
Floating-Point Unit
Cache maintenance operations
Access control
Description
In the register descriptions:
• the register type is described as follows:
RW Read and write.
RO
WO
Read-only.
Write-only.
• the required privilege gives the privilege level required to access the register, as follows:
Privileged
Only privileged software can access the register.
Unprivileged
Both unprivileged and privileged software can access the register.
Attempting to access a privileged register from unprivileged software results in a BusFault.
PM0253 Rev 5 183/254
253
Cortex-M7 peripherals PM0253
4.2 Nested Vectored Interrupt Controller
This section describes the NVIC and the registers it uses. The NVIC supports:
• 1 to 240 interrupts.
• A programmable priority level of 0-255 for each interrupt. A higher level corresponds to a lower priority, so level 0 is the highest interrupt priority.
• Level and pulse detection of interrupt signals.
• Dynamic reprioritization of interrupts.
• Grouping of priority values into group priority and subpriority fields.
• Interrupt tail-chaining.
• An external Non Maskable Interrupt (NMI)
The processor automatically stacks its state on exception entry and unstacks this state on exception exit, with no instruction overhead. This provides low latency exception handling.
The hardware implementation of the NVIC registers is:
Address Name
0xE000E100-
0xE000E11C
0XE000E180-
0xE000E19C
0XE000E200-
0xE000E21C
0XE000E280-
0xE000E29C
0xE000E300-
0xE000E31C
0xE000E400-
0xE000E4EF
NVIC_ISER0-
NVIC_ISER7
NVIC_ICER0-
NVIC_ICER7
NVIC_ISPR0-
NVIC_ISPR7
NVIC_ICPR0-
NVIC_ICPR7
NVIC_IABR0-
NVIC_IABR7
NVIC_IPR0-
NVIC_IPR59
Table 40. NVIC register summary
Type
Required privilege
Reset value
RW
RW
RW
RW
RW
Privileged
Privileged
Privileged
Privileged
Privileged
0x00000000
0x00000000
0x00000000
0x00000000
0x00000000
Description
Interrupt set-enable registers on page 185
Interrupt clear-enable registers on page 186
Interrupt set-pending registers on page 186
Interrupt clear-pending registers on page 187
Interrupt active bit registers on page 188
RW Privileged 0x00000000
Interrupt priority registers on page 188
0xE000EF00 STIR WO Configurable
(1)
0x00000000
Software trigger interrupt register on page 189
1. See the register description for more information.
184/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
Cortex
®
-M7 NVIC registers using CMSIS
CMSIS functions enable the software portability between different Cortex ® -M profile processors. To access the NVIC registers when using CMSIS, use the following functions:
Table 41. CMSIS access NVIC functions
CMSIS function void NVIC_EnableIRQ(IRQn_Type IRQn)
(1) void NVIC_DisableIRQ(IRQn_Type IRQn)
Description
Enables an interrupt or exception.
void NVIC_SetPendingIRQ(IRQn_Type
IRQn)
Disables an interrupt or exception.
Sets the pending status of interrupt or exception to 1.
void NVIC_ClearPendingIRQ(IRQn_Type
IRQn)
Clears the pending status of interrupt or exception to 0.
uint32_t NVIC_GetPendingIRQ(IRQn_Type
IRQn)
Reads the pending status of interrupt or exception. This function returns non-zero value if the pending status is set to 1.
void NVIC_SetPriority(IRQn_Type IRQn,
Sets the priority of an interrupt or exception with configurable priority level to 1.
uint32_t NVIC_GetPriority(IRQn_Type
IRQn)
Reads the priority of an interrupt or exception with configurable priority level. This function return the current priority level.
1. The input parameter IRQn is the IRQ number, see
Table 19 on page 40 for more information.
The NVIC_ISER0-NVIC_ISER7 registers enable interrupts, and show which interrupts are
enabled. See the register summary in Table 40 on page 184 for the register attributes.
The bit assignments are:
Figure 17. ISER bit assignments
31 0
SETENA bits
MSv39675V1
[31:0]
Bits Name
Table 42. ISER bit assignments
Function
SETENA
Interrupt set-enable bits.
Write:
0: No effect.
1: Enable interrupt.
Read:
0: Interrupt disabled.
1: Interrupt enabled.
PM0253 Rev 5 185/254
253
Cortex-M7 peripherals PM0253
If a pending interrupt is enabled, the NVIC activates the interrupt based on its priority. If an interrupt is not enabled, asserting its interrupt signal changes the interrupt state to pending, but the NVIC never activates the interrupt, regardless of its priority.
The NVIC_ICER0-NVIC_ICER7 registers disable interrupts, and show which interrupts are
enabled. See the register summary in Table 40 on page 184 for the register attributes.
The bit assignments are:
Figure 18. ICER bit assignment
31 0
CLRENA bits
MSv39671V1
[31:0]
Bits Name
Table 43. ICER bit assignments
Function
CLRENA
Interrupt clear-enable bits.
Write:
0: No effect.
1: Disable interrupt.
Read:
0: Interrupt disabled.
1: Interrupt enabled.
The NVIC_ISPR0-NVIC_ISPR7 registers force interrupts into the pending state, and show
which interrupts are pending. See the register summary in Table 40 on page 184 for the
register attributes.
The bit assignments are:
Figure 19. ISPR bit assignments
31 0
SETPEND bits
MSv39676V1
186/254 PM0253 Rev 5
PM0253
Bits
[31:0]
Name
SETPEND
Table 44. ISPR bit assignments
Function
Interrupt set-pending bits.
Write:
0: No effect.
1: Changes interrupt state to pending.
Read:
0: Interrupt is not pending.
1: Interrupt is pending.
Writing 1 to the ISPR bit corresponding to:
• An interrupt that is pending has no effect.
• A disabled interrupt sets the state of that interrupt to pending.
Cortex-M7 peripherals
The NVIC_ICPR0-NCVIC_ICPR7 registers remove the pending state from interrupts, and
show which interrupts are pending. See the register summary in Table 40 on page 184 for
the register attributes.
The bit assignments are:
Figure 20. ICPR bit assignments
31 0
CLRPEND bits
MSv39672V1
Bits
[31:0]
Name
CLRPEND
Table 45. ICPR bit assignments
Function
Interrupt clear-pending bits.
Write:
0: No effect.
1: Removes pending state an interrupt.
Read:
0: Interrupt is not pending.
1: Interrupt is pending.
Writing 1 to an ICPR bit does not affect the active state of the corresponding interrupt.
PM0253 Rev 5 187/254
253
Cortex-M7 peripherals
4.2.6 Interrupt active bit registers
PM0253
The NVIC_IABR0-NVIC_IABR7 registers indicate which interrupts are active. See the register summary in
for the register attributes.
The bit assignments are:
31
Figure 21. IABR bit assignments
ACTIVE bits
0
MSv39670V1
[31:0]
Bits Name
ACTIVE
Table 46. IABR bit assignments
Function
Interrupt active flags:
0: Interrupt not active.
1: Interrupt active.
A bit is read as one if the status of the corresponding interrupt is active or active and pending.
The NVIC_IPR0-NVIC_IPR59 registers provide an 8-bit priority field for each interrupt.
These registers are byte-accessible. See the register summary in
for their attributes. Each register holds four priority fields as shown:
Figure 22. IPR bit assignments
31 24 23 16 15 8 7 0
IPR 59 PRI_N 239 PRI_N 238 PRI_N 237 PRI_N 236
IPR n PRI_N 4n+3
IPR0 PRI_N3
PRI_N 4n+2
PRI_N2
PRI_N 4n+1
PRI_N1
PRI_N 4n
PRI_N0
MSv39674V1
188/254 PM0253 Rev 5
PM0253
4.2.8
Cortex-M7 peripherals
Bits Name
[31:24] Priority, byte offset 3
[23:16] Priority, byte offset 2
[15:8] Priority, byte offset 1
[7:0] Priority, byte offset 0
Table 47. IPR bit assignments
Function
Each priority field holds a priority value, 0-255. The lower the value, the greater the priority of the corresponding interrupt. The processor implements only bits[7:n] of each field, bits[n-1:0] read as zero and ignore writes.
See
Accessing the Cortex ® -M7 NVIC registers using CMSIS on page 185 for more
information about the access to the interrupt priority array, which provides the software view of the interrupt priorities.
Find the IPR number and byte offset for interrupt m as follows:
• the corresponding IPR number, see
n is given by n = m DIV 4
• the byte offset of the required Priority field in this register is m MOD 4, where:
– Byte offset 0 refers to register bits[7:0].
– Byte offset 1 refers to register bits[15:8].
– Byte offset 2 refers to register bits[23:16].
– Byte offset 3 refers to register bits[31:24].
Software trigger interrupt register
Write to the STIR to generate an interrupt from software. See the register summary in
for the STIR attributes.
When the USERSETMPEND bit in the SCR is set to 1, unprivileged software can access the STIR, see
System control register on page 199 .
Only privileged software can enable unprivileged access to the STIR.
The bit assignments are:
Figure 23. STIR bit assignments
31 9 8 0
Reserved INTID
Bits
[31:9]
[8:0]
Field
-
INTID
MSv39692V1
Table 48. STIR bit assignments
Function
Reserved.
Interrupt ID of the interrupt to trigger, in the range 0-239. For example, a value of
0x03
specifies interrupt IRQ3.
PM0253 Rev 5 189/254
253
Cortex-M7 peripherals PM0253
The processor supports both level-sensitive and pulse interrupts. The pulse interrupts are also described as edge-triggered interrupts.
A level-sensitive interrupt is held asserted until the peripheral deasserts the interrupt signal.
Typically this happens because the ISR accesses the peripheral, causing it to clear the interrupt request. A pulse interrupt is an interrupt signal sampled synchronously on the rising edge of the processor clock. To ensure the NVIC detects the interrupt, the peripheral must assert the interrupt signal for at least one clock cycle, during which the NVIC detects the pulse and latches the interrupt.
When the processor enters the ISR, it automatically removes the pending state from the interrupt, see
Hardware and software control of interrupts on page 190
. For a level-sensitive interrupt, if the signal is not deasserted before the processor returns from the ISR, the interrupt becomes pending again, and the processor must execute its ISR again. This means that the peripheral can hold the interrupt signal asserted until it no longer requires servicing.
Hardware and software control of interrupts
• The Cortex
®
-M7 latches all interrupts. A peripheral interrupt becomes pending for one of the following reasons:
• The NVIC detects that the interrupt signal is HIGH and the interrupt is not active.
• The NVIC detects a rising edge on the interrupt signal.
• Software writes to the corresponding interrupt set-pending register bit, see
Interrupt set-pending registers on page 186
, or to the STIR to make an interrupt pending, see
Software trigger interrupt register on page 189 .
A pending interrupt remains pending until one of the following:
• The processor enters the ISR for the interrupt. This changes the state of the interrupt from pending to active. Then:
– For a level-sensitive interrupt, when the processor returns from the ISR, the NVIC samples the interrupt signal. If the signal is asserted, the state of the interrupt changes to pending, which might cause the processor to immediately re-enter the
ISR. Otherwise, the state of the interrupt changes to inactive.
– For a pulse interrupt, the NVIC continues to monitor the interrupt signal, and if this is pulsed the state of the interrupt changes to pending and active. In this case, when the processor returns from the ISR the state of the interrupt changes to pending, which might cause the processor to immediately re-enter the ISR.
– If the interrupt signal is not pulsed while the processor is in the ISR, when the processor returns from the ISR the state of the interrupt changes to inactive.
• The software writes to the corresponding interrupt clear-pending register bit.
For a level-sensitive interrupt, if the interrupt signal is still asserted, the state of the interrupt does not change. Otherwise, the state of the interrupt changes to inactive.
For a pulse interrupt, the state of the interrupt changes to:
– Inactive, if the state was pending.
– Active, if the state was active and pending.
190/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
4.2.10 NVIC design hints and tips
Ensure that the software uses correctly aligned register accesses. The processor does not support unaligned accesses to NVIC registers. See the individual register descriptions for the supported access sizes.
An interrupt can enter pending state even if it is disabled. Disabling an interrupt only prevents the processor from taking that interrupt.
Before programming VTOR to relocate the vector table, ensure the vector table entries of the new vector table are set up for fault handlers, NMI, and all enabled exception-like
interrupts. For more information see Vector table offset register on page 197
.
NVIC programming hints
The software uses the CPSIE I and CPSID I instructions to enable and disable interrupts.
The CMSIS provides the following intrinsic functions for these instructions: void __disable_irq(void) // Disable Interrupts void __enable_irq(void) // Enable Interrupts
In addition, the CMSIS provides a number of functions for NVIC control, including:
Table 49. CMSIS functions for NVIC control
CMSIS interrupt control function Description void NVIC_SetPriorityGrouping(uint32_t priority_grouping)
Set the priority grouping void NVIC_EnableIRQ(IRQn_t IRQn)
Enable IRQn
Disable IRQn void NVIC_DisableIRQ(IRQn_t IRQn) uint32_t NVIC_GetPendingIRQ (IRQn_t IRQn)
Return true (IRQ-Number) if IRQn is pending void NVIC_SetPendingIRQ (IRQn_t IRQn)
Set IRQn pending void NVIC_ClearPendingIRQ (IRQn_t IRQn) uint32_t NVIC_GetActive (IRQn_t IRQn)
Clear IRQn pending status
Return the IRQ number of the active interrupt void NVIC_SetPriority (IRQn_t IRQn, uint32_t priority) uint32_t NVIC_GetPriority (IRQn_t IRQn)
Set priority for IRQn
Read priority of IRQn
The input parameter IRQn is the IRQ number, see
Table 19 on page 40 . For more
information about these functions see the CMSIS documentation.
PM0253 Rev 5 191/254
253
Cortex-M7 peripherals PM0253
The System Control Block (SCB) provides system implementation information, and system control. This includes configuration, control, and reporting of the system exceptions. The system control block registers are:
Address
Table 50. Summary of the system control block registers
Name Type
Required privilege
Reset value
Description
0xE000E008
0xE000ED00
0xE000ED04
0xE000ED08
0xE000ED34
0xE000ED38
ACTLR
CPUID
ICSR
VTOR
0xE000ED0C AIRCR
0xE000ED10
0xE000ED20
0xE000ED24
SCR
0xE000ED14 CCR
SHPR3
SHCRS
0xE000ED28 CFSR
0xE000ED28 MMSR (2)
0xE000ED29 BFSR
0xE000ED2A UFSR
0xE000ED2C HFSR
MMAR
BFAR
0xE000ED3C AFSR
RW
RO
RW (1)
RW
RW
RW
RW
0xE000ED18 SHPR1 RW
0xE000ED1C SHPR2 RW
RW
RW
Privileged 0x00000000
Auxiliary control register on page 193
Privileged 0x410FC270
CPUID base register on page 194
Privileged 0x00000000
Interrupt control and state register on page 194
Privileged Unknown
Privileged 0xFA050000
Vector table offset register on page 197
Application interrupt and reset control register on page 197
Privileged 0x00000000
System control register on page 199
Privileged 0x00000200
Configuration and control register on page 200
Privileged 0x00000000
System handler priority register 1 on page 202
Privileged 0x00000000
System handler priority register 2 on page 203
Privileged 0x00000000
System handler priority register 3 on page 203
Privileged 0x00000000
System handler control and state register on page 204
RW
RW
RW
Privileged 0x00000000
Configurable fault status register on page 205
Privileged 0x00
MemManage fault address register on page 211
Privileged 0x00
BusFault status register on page 207
RW
RW
RW
Privileged 0x0000
Privileged 0x00000000
Privileged Unknown
RW Privileged Unknown
RAZ/WI Privileged -
Auxiliary control register on page 193
HardFault status register on page 210
MemManage fault address register on page 211
BusFault address register on page 212
Auxiliary Fault Status register not implemented
1. See the register description for more information.
2. A subregister of the CFSR.
192/254 PM0253 Rev 5
PM0253
4.3.1
Cortex-M7 peripherals
Auxiliary control register
The ACTLR provides disable bits for the following processor functions:
• FPU exception outputs.
• Dual-issue functionality.
• Flushing of the trace output from the ITM and DWT.
• Dynamic read allocate mode.
By default this register is set to provide optimum performance from the Cortex
®
-M7 processor, and does not normally require modification.
See the register summary in
for the ACTLR attributes. The bit assignments are:
Figure 24. ACTLR bit assignments
31 13 12 11 10 9 3 2 1 0
Reserved Reserved
DISITMATBFLUSH
DISRAMODE
FPEXCODIS
DISFOLD
Reserved
MSv39648V1
Bits
[31:13] -
[12]
[11]
[10]
[9:3]
[2]
[1:0] -
-
Name
DISITMATBFLUSH
DISRAMODE
FPEXCODIS
DISFOLD
Table 51. ACTLR bit assignments
Function
Reserved
Disables ITM and DWT ATB flush:
0: Normal operation.
1: ITM and DWT ATB flush disabled.
Disables dynamic read allocate mode for Write-Back Write-Allocate memory regions:
0: Normal operation.
1: Dynamic read allocate mode disabled.
Disables FPU exception outputs:
0: Normal operation.
1: FPU exception outputs are disabled.
Reserved.
Disables dual-issue functionality:
0: Normal operation.
1: Dual-issue functionality is disabled. Setting this bit decreases performance.
Reserved.
PM0253 Rev 5 193/254
253
Cortex-M7 peripherals PM0253
4.3.3
The CPUID register contains the processor part number, version, and implementation
information. See the register summary in Table 50 on page 192 for its attributes. The bit
assignments are:
Figure 25. CPUID bit assignments
31
Implementer
24 23
Variant
20 19
Constant
16 15
PartNo
4 3
Revision
0
Bits Name
[31:24] Implementer
[23:20] Variant
[19:16] Constant
[15:4] PartNo
[3:0] Revision
MSv39661V1
Table 52. CPUID bit assignments
Function
Implementer code:
0x41 Arm
Variant number, the r value in the r n p n product revision identifier:
0x0 Revision 0
Reads as 0xF
Part number of the processor:
0xC27: Cortex
®
-M7
Revision number, the p value in the r n p n product revision identifier:
0x0 : Patch 0
Interrupt control and state register
The ICSR:
• provides:
– A set-pending bit for the Non Maskable Interrupt (NMI) exception.
– Set-pending and clear-pending bits for the PendSV and SysTick exceptions.
• indicates:
– The exception number of the exception being processed.
– Whether there are preempted active exceptions.
– The exception number of the highest priority pending exception
– Whether any interrupts are pending.
See the register summary in Table 50 on page 192 , and the type descriptions in
, for the ICSR attributes. The bit assignments are:
194/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
Bits
[31]
[30:29] -
[28]
[27]
[26]
NMIPENDSET
Reserved
PENDSVSET
PENDSVCLR
PENDSTSET
Figure 26. ICSR bit assignments
31 30 29 28 27 26 25 24 23 22 21 20 12 11 10 9 8
VECTPENDING
Reserved
ISRPENDING
Reserved for Debug use
Reserved
PENDSTCLR
VECTACTIVE
Reserved
RETTOBASE
0
MSv39673V1
Name
NMIPENDSET
PENDSVSET
PENDSVCLR
PENDSTSET
-
Type
Table 53. ICSR bit assignments
Function
RW
RW
WO
RW
NMI set-pending bit.
Write:
0: No effect.
1: Changes NMI exception state to pending.
Read:
0: NMI exception is not pending.
1: NMI exception is pending.
Because NMI is the highest-priority exception, normally the processor enters the NMI exception handler as soon as it registers a write of 1 to this bit, and entering the handler clears this bit to 0. A read of this bit by the
NMI exception handler returns 1 only if the NMI signal is reasserted while the processor is executing that handler.
Reserved.
PendSV set-pending bit.
Write:
0: No effect.
1: Changes PendSV exception state to pending.
Read:
0: PendSV exception is not pending.
1: PendSV exception is pending.
Writing 1 to this bit is the only way to set the PendSV exception state to pending.
PendSV clear-pending bit.
Write:
0: No effect.
1: Removes the pending state from the PendSV exception.
SysTick exception set-pending bit.
Write:
0: No effect.
1: Changes SysTick exception state to pending.
Read:
0: SysTick exception is not pending.
1: SysTick exception is pending.
PM0253 Rev 5 195/254
253
Cortex-M7 peripherals PM0253
Bits Name
Table 53. ICSR bit assignments (continued)
Type Function
[25]
[24]
[23]
[22]
[21]
-
Reserved for Debug use
[20:12] VECTPENDING
[11]
[10:9] -
-
PENDSTCLR
ISRPENDING
-
-
RETTOBASE RO
-
WO
RO
RO
RO
SysTick exception clear-pending bit.
Write:
0: No effect.
1: Removes the pending state from the SysTick exception.
This bit is WO. On a register read its value is
Unknown
.
Reserved.
This bit is reserved for Debug use and reads-as-zero when the processor is not in Debug.
Interrupt pending flag, excluding NMI and Faults:
0: Interrupt not pending.
1: Interrupt pending.
Reserved.
Indicates the exception number of the highest priority pending enabled exception:
0: No pending exceptions.
Nonzero: The exception number of the highest priority pending enabled exception.
The value indicated by this field includes the effect of the BASEPRI and
FAULTMASK registers, but not any effect of the PRIMASK register.
Indicates whether there are preempted active exceptions:
0: There are preempted active exceptions to execute.
1: There are no active exceptions, or the currently-executing exception is the only active exception.
Reserved.
[8:0] VECTACTIVE
RO
Contains the active exception number:
0: Thread mode
1: The exception number
(1)
of the currently active exception.
Subtract 16 from this value to obtain the CMSIS IRQ number required to index into the Interrupt Clear-Enable, Set-Enable, Clear-Pending, Set-
Pending, or Priority Registers, see
.
1. This is the same value as IPSR bits[8:0], see
Interrupt program status register on page 23
.
When writing to the ICSR, the effect is unpredictable if the user:
• Writes 1 to the PENDSVSET bit and writes 1 to the PENDSVCLR bit.
• Writes 1 to the PENDSTSET bit and writes 1 to the PENDSTCLR bit.
196/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
The VTOR indicates the offset of the vector table base address from memory address
0x00000000. See the register summary in Table 50 on page 192 for its attributes. The bit
assignments are:
Figure 27. VTOR bit assignments
31 9 8 0
TBLOFF Reserved
MSv39699V1
Bits Name
[31:9] TBLOFF
[8:0] -
Table 54. VTOR bit assignments
Function
Vector table base offset field. It contains bits [29:7] of the offset of the table base from the bottom of the memory map.
Reserved.
When setting TBLOFF, the user must align the offset to the number of exception entries in the vector table.
The table alignment requirements mean that bits [8:0] of the table offset are always zero.
The AIRCR provides priority grouping control for the exception model, endian status for data accesses, and reset control of the system. See the register summary in
and
for its attributes.
To write to this register, the user must write 0x5FA to the VECTKEY field, otherwise the processor ignores the write.
The bit assignments are:
Figure 28. AIRCR bit assignments
31 16 15 14 11 10
On read: VECTKEYSTAT
On write: VECTKEY
Reserved
ENDIANNESS PRIGROUP
Reserved for Debug use
8 7
Reserved
3
SYSRESETREQ
VECTCLRACTIVE
VECTRESET
2 1 0
MSv39651V1
PM0253 Rev 5 197/254
253
Cortex-M7 peripherals PM0253
[31:16]
[15]
[14:11] -
[10:8]
[7:3]
[2]
[1]
[0]
Bits Name
Read:
VECTKEYSTAT
Write: VECTKEY
ENDIANNESS
PRIGROUP
- -
SYSRESETREQ
VECTCLRACTIVE WO
VECTRESET
Table 55. AIRCR bit assignments
Type Function
-
RW
RO
RW
WO
WO
Register key:
Reads as 0xFA05
On writes, write 0x5FA to VECTKEY, otherwise the write is ignored.
Data endianness bit:
0: Little-endian.
Reserved
Interrupt priority grouping field. This field determines the split of group priority from subpriority, see
.
Reserved.
System reset request:
0: No system reset request.
1: Asserts a signal to the outer system that requests a reset.
This is intended to force a large system reset of all major components except for debug.
This bit reads as 0.
Reserved for Debug use. This bit reads as 0. When writing to the register the user must write 0 to this bit, otherwise behavior is Unpredictable.
Reserved for Debug use. This bit reads as 0. When writing to the register the user must write 0 to this bit, otherwise behavior is Unpredictable.
Binary point
The PRIGROUP field indicates the position of the binary point that splits the PRI_ n fields in the Interrupt Priority Registers into separate group priority and subpriority fields.
shows how the PRIGROUP value controls this split.
If the user implements fewer than 8 priority bits he might require more explanation here, and want to remove invalid rows from the table, and modify the entries in the number of columns.
Table 56. Priority grouping
Interrupt priority level value, PRI_ N [7:0]
PRIGROUP Binary point
(1)
0b000
0b001
0b010
0b011
0b100 bxxxxxxx.y
bxxxxxx.yy
bxxxxx.yyy
bxxxx.yyyy
bxxx.yyyyy
Number of
Group priority bits
[7:1]
[7:2]
[7:3]
[7:4]
[7:5]
Subpriority
[0]
[1:0]
[2:0]
[3:0]
[4:0] bits
128
64
32
16
8
Group priorities
Subpriorities
2
4
8
16
32
198/254 PM0253 Rev 5
PM0253
Note:
4.3.6
Cortex-M7 peripherals
Table 56. Priority grouping (continued)
Interrupt priority level value, PRI_ N [7:0] Number of
PRIGROUP Binary point
(1)
Group priority bits
Subpriority bits
Group priorities
Subpriorities
0b101
0b110
0b111 bxx.yyyyyy
bx.yyyyyyy
b.yyyyyyyy
[7:6]
[7]
None
[5:0]
[6:0]
[7:0]
4
2
1
64
128
256
1. PRI_ n [7:0] field showing the binary point. x denotes a group priority field bit, and y denotes a subpriority field bit.
System control register
The SCR controls features of entry to and exit from Low-power state. See the register summary in
for its attributes. The bit assignments are
Figure 29. SCR bit assignments:
31 5 4 3 2 1 0
Reserved
Bits
[31:5] -
[4]
[3] -
Name
SEVONPEND
SEVONPEND
Reserved
SLEEPDEEP
SLEEPONEXIT
Reserved
MSv39687V1
Table 57. SCR bit assignments
Function
Reserved.
Send Event on Pending bit:
0: Only enabled interrupts or events can wakeup the processor, disabled interrupts are excluded.
1: Enabled events and all interrupts, including disabled interrupts, can wakeup the processor.
When an event or interrupt enters pending state, the event signal wakes up the processor from WFE. If the processor is not waiting for an event, the event is registered and affects the next WFE.
The processor also wakes up on execution of an SEV instruction or an external event.
Reserved.
PM0253 Rev 5 199/254
253
Cortex-M7 peripherals PM0253
Bits
[2]
[1]
[0]
Name
Table 57. SCR bit assignments (continued)
Function
-
SLEEPDEEP
SLEEPONEXIT
Controls whether the processor uses sleep or deep sleep as its Lowpower mode:
0: Sleep.
1: Deep sleep.
Indicates sleep-on-exit when returning from Handler mode to Thread mode:
0: Do not sleep when returning to Thread mode.
1: Enter sleep, or deep sleep, on return from an ISR.
Setting this bit to 1 enables an interrupt driven application to avoid returning to an empty main application.
Reserved.
The CCR controls entry to Thread mode and enables:
• The handlers for NMI, hard fault and faults escalated by FAULTMASK to ignore
BusFaults.
• Trapping of divide by zero and unaligned accesses.
•
Access to the STIR by unprivileged software, see Software trigger interrupt register on page 189
.
• Instruction and data cache enable control.
See the register summary in
for the CCR attributes.
The bit assignments are:
Figure 30. CCR bit assignments
31 19 18 17 16 15 10 9 8 7 5 4 3 2 1 0
Reserved Reserved
BP
IC
DC
STKALIGN
BFHFNMIGN
Reserved
DIV_0_TRP
UNALIGN_TRP
Reserved
USERSETMPEND
NONBASETHRDENA
MSv39656V1
200/254 PM0253 Rev 5
PM0253
Bits
[31:19] -
[18] BP
Name
[17] IC
[16]
[15:10] -
DC
[9]
[8]
[7:5] -
[4]
[3]
[2] -
STKALIGN
BFHFNMIGN
DIV_0_TRP
UNALIGN_TRP
Cortex-M7 peripherals
Table 58. CCR bit assignments
Type Function
-
-
-
Reserved.
RO Always reads-as-one. It indicates branch prediction is enabled.
RW
RW
Enables L1 instruction cache:
0: L1 instruction cache disabled.
1: L1 instruction cache enabled.
Enables L1data cache:
0: L1 data cache disabled.
1: L1 data cache enabled.
RO
RW
RW
RW
Reserved.
Always reads-as-one. It indicates stack alignment on exception entry is 8-byte aligned.
On exception entry, the processor uses bit[9] of the stacked
PSR to indicate the stack alignment. On return from the exception it uses this stacked bit to restore the correct stack alignment.
Enables handlers with priority -1 or -2 to ignore data BusFaults caused by load and store instructions. This applies to the hard fault, NMI, and FAULTMASK escalated handlers:
0:Data bus faults caused by load and store instructions cause a lock-up.
1: Handlers running at priority -1 and -2 ignore data bus faults caused by load and store instructions.
Set this bit to 1 only when the handler and its data are in absolutely safe memory. The normal use of this bit is to probe system devices and bridges to detect control path problems and fix them.
Reserved.
Enables faulting or halting when the processor executes an
SDIV or UDIV instruction with a divisor of 0:
0: Do not trap divide by 0.
1: Trap divide by 0.
When this bit is set to 0, a divide by zero returns a quotient of
0.
Enables unaligned access traps:
0: Do not trap unaligned halfword and word accesses.
1: Trap unaligned halfword and word accesses.
If this bit is set to 1, an unaligned access generates a
UsageFault.
Unaligned LDM, STM, LDRD, and STRD instructions always fault irrespective of whether UNALIGN_TRP is set to 1.
Reserved.
PM0253 Rev 5 201/254
253
Cortex-M7 peripherals PM0253
Bits
[1]
[0]
Name
Table 58. CCR bit assignments (continued)
Type Function
USERSETMPEND RW
NONBASETHRDE
NA
RW
Enables unprivileged software access to the STIR, see
Software trigger interrupt register on page 189
:
0: Disable.
1: Enable.
Indicates how the processor enters Thread mode:
0: Processor can enter Thread mode only when no exception is active.
1: Processor can enter Thread mode from any level under
the control of an EXC_RETURN value, see Exception return on page 46 .
The SHPR1-SHPR3 registers set the priority level, 0 to 255 of the exception handlers that have configurable priority.
SHPR1-SHPR3 are byte accessible. See the register summary in Table 50 on page 192
for their attributes.
The system fault handlers and the priority field and register for each handler are:
Handler
MemManage
BusFault
UsageFault
SVCall
PendSV
SysTick
Table 59. System fault handler priority fields
Field Register description
PRI_4
PRI_5
PRI_6
PRI_11
PRI_14
PRI_15
System handler priority register 1
System handler priority register 2
System handler priority register 3
Each PRI_ n field is 8 bits wide, but the processor implements only bits[7:M] of each field, and bits[M-1:0] read as zero and ignore writes.
System handler priority register 1
The bit assignments are:
Figure 31. SHPR1 bit assignements
31 24 23 16 15 8 7 0
Reserved PRI_6 PRI_5 PRI_4
MSv39689V1
202/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
Bits
[31:24]
[23:16]
[15:8]
[7:0]
Name
PRI_7
PRI_6
PRI_5
PRI_4
Table 60. SHPR1 register bit assignments
Function
Reserved
Priority of system handler 6, UsageFault
Priority of system handler 5, BusFault
Priority of system handler 4, MemManage
System handler priority register 2
The bit assignments are:
Figure 32. SHPR2 bit assignments
31 24 23
PRI_11 Reserved
0
MSv39690V1
Bits
[31:24]
[23:0]
Table 61. SHPR2 register bit assignments
Name Function
-
PRI_11 Priority of system handler 11, SVCall
Reserved
System handler priority register 3
The bit assignments are:
Figure 33. SHPR3 bit assignments
31 24 23 16 15
PRI_15 PRI_14 Reserved
0
MSv39691V1
Bits
[31:24]
[23:16]
[15:0]
Table 62. SHPR3 register bit assignments
Name Function
-
PRI_15
PRI_14
Priority of system handler 15, SysTick exception
Priority of system handler 14, PendSV
Reserved
PM0253 Rev 5 203/254
253
Cortex-M7 peripherals
4.3.9 System handler control and state register
PM0253
The SHCSR enables the system handlers, and indicates:
• The pending status of the BusFault, MemManage fault, and SVC exceptions.
• The active status of the system handlers.
See the register summary in
for the SHCSR attributes. The bit assignments are:
31
Reserved
Figure 34. SHCSR bit assignments
19 18 17 16 15 14 13 12 11 10 9 8 7 6 4 3 2 1 0
USGFAULTENA
BUSFAULTENA
MEMFAULTENA
SVCALLPENDED
BUSFAULTPENDED
MEMFAULTPENDED
USGFAULTPENDED
SYSTICKACT
PENDSVACT
Reserved
MONITORACT
SVCALLACT
MEMFAULTACT
BUSFAULTACT
Reserved
USGFAULTACT
Reserved
MSv39688V1
[11]
[10]
[9]
[8]
[7]
[6:4]
[3]
Bits Name
[31:19] -
[18]
[17]
USGFAULTENA
BUSFAULTENA
[16]
[15]
MEMFAULTENA
SVCALLPENDED
[14]
[13]
[12]
BUSFAULTPENDED
USGFAULTPENDED
SYSTICKACT
PENDSVACT
-
MONITORACT
-
SVCALLACT
USGFAULTACT
Table 63. SHCSR bit assignments
Function
MEMFAULTPENDED
Reserved
UsageFault enable bit, set to 1 to enable
(1)
BusFault enable bit, set to 1 to enable
MemManage enable bit, set to 1 to enable
SVCall pending bit, reads as 1 if exception is pending
(2)
BusFault exception pending bit, reads as 1 if exception is
MemManage exception pending bit, reads as 1 if exception is
UsageFault exception pending bit, reads as 1 if exception is
SysTick exception active bit, reads as 1 if exception is active
(3)
PendSV exception active bit, reads as 1 if exception is active
Reserved
Debug monitor active bit, reads as 1 if Debug monitor is active
SVCall active bit, reads as 1 if SVC call is active
Reserved
UsageFault exception active bit, reads as 1 if exception is active
204/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
Bits
Table 63. SHCSR bit assignments (continued)
Name Function
[2]
[1]
[0]
-
BUSFAULTACT
MEMFAULTACT
Reserved
BusFault exception active bit, reads as 1 if exception is active
MemManage exception active bit, reads as 1 if exception is active
1. Enable bits, set to 1 to enable the exception, or set to 0 to disable the exception.
2. Pending bits, read as 1 if the exception is pending, or as 0 if it is not pending. The user can write to these bits to change the pending status of the exceptions.
3. Active bits, read as 1 if the exception is active, or as 0 if it is not active. The user can write to these bits to change the active status of the exceptions, but see the Caution in this section.
If the user disables a system handler and the corresponding fault occurs, the processor treats the fault as a hard fault.
The user can write to this register to change the pending or active status of system exceptions. An OS kernel can write to the active bits to perform a context switch that changes the current exception type.
• A software that changes the value of an active bit in this register without correct adjustment to the stacked content can cause the processor to generate a fault exception. Ensure a software that writes to this register retains and subsequently restores the current active status.
• After having enabled the system handlers, if the user has to change the value of a bit in this register he must use a read-modify-write procedure to ensure that only the required bit is changed.
The CFSR indicates the cause of a MemManage fault, BusFault, or UsageFault. See the register summary in
for its attributes. The bit assignments are:
Figure 35. CFSR bit assignments
31
Usage Fault Status Register
16 15
Bus Fault Status Register
8 7
Memory Management
Fault Status Register
0
UFSR BFSR MMFSR
MSv39658V1
The following subsections describe the subregisters that make up the CFSR:
•
MemManage fault status register on page 206 .
•
BusFault status register on page 207 .
•
UsageFault status register on page 209
.
PM0253 Rev 5 205/254
253
Cortex-M7 peripherals PM0253
The CFSR is byte accessible. The CFSR or its subregisters can be accessed as follows:
• Access the complete CFSR with a word access to 0xE000ED28.
• Access the MMFSR with a byte access to 0xE000ED28.
• Access the MMFSR and BFSR with a halfword access to 0xE000ED28.
• Access the BFSR with a byte access to 0xE000ED29.
• Access the UFSR with a halfword access to 0xE000ED2A.
MemManage fault status register
The flags in the MMFSR indicate the cause of memory access faults. The bit assignments are:
Figure 36. MMFSR bit assignments
7 6 5 4 3 2 1 0
MMARVALID
Reserved
MLSPERR
MSTKERR
IACCVIOL
DACCVIOL
Reserved
MUNSTKERR
MSv39680V1
Bits
[7]
[6]
[5]
[4]
-
Name
MMARVALID
MLSPERR
MSTKERR
Table 64. MMFSR bit assignments
Function
MemManage Fault Address register (MMFAR) valid flag:
0: Value in MMAR is not a valid fault address.
1: MMAR holds a valid fault address.
If a MemManage fault occurs and is escalated to a HardFault because of priority, the HardFault handler must set this bit to 0. This prevents problems on return to a stacked active MemManage fault handler whose MMAR value has been overwritten.
Reserved.
0: No MemManage fault occurred during floating-point lazy state preservation.
1: A MemManage fault occurred during floating-point lazy state preservation.
MemManage fault on stacking for exception entry:
0: No stacking fault.
1: Stacking for an exception entry has caused one or more access violations.
When this bit is 1, the SP is still adjusted but the values in the context area on the stack might be incorrect. The processor has not written a fault address to the MMAR.
206/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
Bits
[3]
[2]
[1]
[0]
Name
MUNSTKERR
-
DACCVIOL
IACCVIOL
Table 64. MMFSR bit assignments (continued)
Function
MemManage fault on unstacking for a return from exception:
0: No unstacking fault.
1: Unstack for an exception return has caused one or more access violations.
This fault is chained to the handler. This means that when this bit is 1, the original return stack is still present. The processor has not adjusted the SP from the failing return, and has not performed a new save. The processor has not written a fault address to the MMAR.
Reserved
Data access violation flag:
0: No data access violation fault.
1: The processor attempted a load or store at a location that does not permit the operation.
When this bit is 1, the PC value stacked for the exception return points to the faulting instruction. The processor has loaded the MMAR with the address of the attempted access.
Instruction access violation flag:
0: No instruction access violation fault.
1: The processor attempted an instruction fetch from a location that does not permit execution.
This fault occurs on any access to an XN region, even when the MPU is disabled or not present.
When this bit is 1, the PC value stacked for the exception return points to the faulting instruction. The processor has not written a fault address to the
MMAR.
The MMFSR bits are sticky. This means as one or more fault occurs, the associated bits are set to 1. A bit that is set to 1 is cleared to 0 only by writing 1 to that bit, or by a reset.
BusFault status register
The flags in the BFSR indicate the cause of a bus access fault. The bit assignments are:
Figure 37. BFSR bit assignments
7 6 5 4 3 2 1 0
BFARVALID
Reserved
LSPERR
STKERR
IBUSERR
PRECISERR
IMPRECISERR
UNSTKERR
MSv39653V1
PM0253 Rev 5 207/254
253
Cortex-M7 peripherals
Bits Name
[7] BFARVALID
[6]
[5]
-
LSPERR
[4] STKERR
[3] UNSTKERR
[2] IMPRECISERR
[1] PRECISERR
[0] IBUSERR
PM0253
Table 65. BFSR bit assignments
Function
BusFault Address register (BFAR) valid flag:
0: Value in BFAR is not a valid fault address.
1: BFAR holds a valid fault address.
The processor sets this bit to 1 after a BusFault where the address is known. Other faults can set this bit to 0, such as a MemManage fault occurring later.
If a BusFault occurs and is escalated to a hard fault because of priority, the hard fault handler must set this bit to 0. This prevents problems if returning to a stacked active
BusFault handler whose BFAR value has been overwritten.
Reserved.
0: No bus fault occurred during floating-point lazy state preservation.
1: A bus fault occurred during floating-point lazy state preservation.
BusFault on stacking for exception entry:
0: No stacking fault.
1: Stacking for an exception entry has caused one or more BusFaults.
When the processor sets this bit to 1, the SP is still adjusted but the values in the context area on the stack might be incorrect. The processor does not write a fault address to the BFAR.
BusFault on unstacking for a return from exception:
0: No unstacking fault.
1: Unstack for an exception return has caused one or more BusFaults.
This fault is chained to the handler. This means that when the processor sets this bit to 1, the original return stack is still present. The processor does not adjust the SP from the failing return, does not performed a new save, and does not write a fault address to the BFAR.
Imprecise data bus error:
0: No imprecise data bus error.
1: A data bus error has occurred, but the return address in the stack frame is not related to the instruction that caused the error.
When the processor sets this bit to 1, it does not write a fault address to the BFAR.
This is an asynchronous fault. Therefore, if it is detected when the priority of the current process is higher than the BusFault priority, the BusFault becomes pending and becomes active only when the processor returns from all higher priority processes. If a precise fault occurs before the processor enters the handler for the imprecise BusFault, the handler detects both IMPRECISERR set to 1 and one of the precise fault status bits set to 1.
Precise data bus error:
0: No precise data bus error.
1: A data bus error has occurred, and the PC value stacked for the exception return points to the instruction that caused the fault.
When the processor sets this bit to 1, it writes the faulting address to the BFAR.
Instruction bus error:
0: No instruction bus error.
1: Instruction bus error.
The processor detects the instruction bus error on prefetching an instruction, but it sets the IBUSERR flag to 1 only if it attempts to issue the faulting instruction.
When the processor sets this bit to 1, it does not write a fault address to the BFAR.
208/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
The BFSR bits are sticky. This means as one or more fault occurs, the associated bits are set to 1. A bit that is set to 1 is cleared to 0 only by writing 1 to that bit, or by a reset.
UsageFault status register
The UFSR indicates the cause of a UsageFault. The bit assignments are:
Figure 38. UFSR bit assignments
15
Reserved
10 9 8 7
Reserved
4 3 2 1 0
Bits
[15:10] -
[9]
[8]
[7:4]
[3]
[2]
DIVBYZERO
UNALIGNED
NOCP
INVPC
INVSTATE
UNDEFINSTR
-
Name
DIVBYZERO
UNALIGNED
NOCP
INVPC
MSv39698V1
Table 66. UFSR bit assignments
Function
Reserved.
Divide by zero UsageFault:
0: No divide by zero fault, or divide by zero trapping not enabled.
1: The processor has executed an SDIV or UDIV instruction with a divisor of 0.
When the processor sets this bit to 1, the PC value stacked for the exception return points to the instruction that performed the divide by zero.
Enable trapping of divide by zero by setting the DIV_0_TRP bit in the CCR
to 1, see Configuration and control register on page 200
.
Unaligned access UsageFault:
0: No unaligned access fault, or unaligned access trapping not enabled.
1: The processor has made an unaligned memory access.
Enable trapping of unaligned accesses by setting the UNALIGN_TRP bit
in the CCR to 1, see Configuration and control register on page 200 .
Unaligned LDM, STM, LDRD, and STRD instructions always fault irrespective of the setting of UNALIGN_TRP.
Reserved.
No coprocessor UsageFault:
0: No UsageFault caused by attempting to access a coprocessor.
1: The processor has attempted to access a coprocessor.
The processor does not support coprocessor instructions:
Invalid PC load UsageFault, caused by an invalid PC load by
EXC_RETURN:
0: No invalid PC load UsageFault.
1: The processor has attempted an illegal load of EXC_RETURN to the
PC, as a result of an invalid context, or an invalid EXC_RETURN value.
When this bit is set to 1, the PC value stacked for the exception return points to the instruction that tried to perform the illegal load of the PC.
PM0253 Rev 5 209/254
253
Cortex-M7 peripherals PM0253
[1]
[0]
Bits Name
Table 66. UFSR bit assignments (continued)
Function
INVSTATE
UNDEFINSTR
Invalid state UsageFault:
0: No invalid state UsageFault.
1: The processor has attempted to execute an instruction that makes illegal use of the EPSR.
When this bit is set to 1, the PC value stacked for the exception return points to the instruction that attempted the illegal use of the EPSR.
This bit is not set to 1 if an undefined instruction uses the EPSR.
Undefined instruction UsageFault:
0: No undefined instruction UsageFault.
1: The processor has attempted to execute an undefined instruction.
When this bit is set to 1, the PC value stacked for the exception return points to the undefined instruction.
An undefined instruction is an instruction that the processor cannot decode.
The UFSR bits are sticky. This means as one or more fault occurs, the associated bits are set to 1. A bit that is set to 1 is cleared to 0 only by writing 1 to that bit, or by a reset.
The HFSR gives information about events that activate the HardFault handler. See the register summary in
for its attributes.
This register is read, write to clear. This means that bits in the register read normally, but writing 1 to any bit clears that bit to 0. The bit assignments are:
Figure 39. HFSR bit assignments
31 30 29 2 1 0
Reserved
FORCED
DEBUGEVT
VECTTBL
Reserved
MSv39669V1
210/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
Bits Name
Table 67. HFSR bit assignments
Function
[31] DEBUGEVT
[30]
[29:2] -
[1]
[0] -
FORCED
VECTTBL
Reserved for Debug use. When writing to the register the user must write 1 to this bit, otherwise behavior is
UNPREDICTABLE
.
Indicates a forced hard fault, generated by escalation of a fault with configurable priority that cannot be handled, either because of priority or because it is disabled:
0: No forced HardFault.
1: Forced HardFault.
When this bit is set to 1, the HardFault handler must read the other fault status registers to find the cause of the fault.
Reserved.
Indicates a BusFault on a vector table read during exception processing:
0: No BusFault on vector table read.
1: BusFault on vector table read.
This error is always handled by the hard fault handler.
When this bit is set to 1, the PC value stacked for the exception return points to the instruction that was preempted by the exception.
Reserved.
The HFSR bits are sticky. This means as one or more fault occurs, the associated bits are set to 1. A bit that is set to 1 is cleared to 0 only by writing 1 to that bit, or by a reset.
The MMFAR contains the address of the location that generated a MemManage fault. See the register summary in
for its attributes. The bit assignments are:
Bits Name
Table 68. MMFAR bit assignments
Function
When an unaligned access faults, the address is the actual address that faulted. Because a single read or write instruction can be split into multiple aligned accesses, the fault address can be any address in the range of the requested access size.
Flags in the MMFSR indicate the cause of the fault, and whether the value in the MMFAR is valid. See
Configuration and control register on page 200
.
PM0253 Rev 5 211/254
253
Cortex-M7 peripherals PM0253
4.3.13 BusFault address register
The BFAR contains the address of the location that generated a BusFault. See the register summary in
for its attributes. The bit assignments are:
Bits Name
[31:0] ADDRESS
Table 69. BFAR bit assignments
Function
When the BFARVALID bit of the BFSR is set to 1, this field holds the address of the location that generated the BusFault
When an unaligned access faults the address in the BFAR is the one requested by the instruction, even if it is not the address of the fault.
Flags in the BFSR indicate the cause of the fault, and whether the value in the BFAR is valid. See
.
4.3.14 System control block design hints and tips
Ensure software uses aligned accesses of the correct size to access the system control block registers:
• Except for the CFSR and SHPR1-SHPR3, it must use aligned word accesses.
• For the CFSR and SHPR1-SHPR3 it can use byte or aligned halfword or word accesses.
The processor does not support unaligned accesses to system control block registers.
In a fault handler, to determine the true faulting address:
1.
Read and save the MMFAR or BFAR value.
2. Read the MMARVALID bit in the MMFSR, or the BFARVALID bit in the BFSR. The
MMFAR or BFAR address is valid only if this bit is 1.
The software must follow this sequence because another higher priority exception might change the MMFAR or BFAR value. For example, if a higher priority handler preempts the current fault handler, the other fault might change the MMFAR or BFAR value.
In addition, the CMSIS provides a number of functions for system control, including:
Table 70. CMSIS function for system control
CMSIS system control function Description void NVIC_SystemReset (void)
Reset the system
4.4 System timer, SysTick
The processor has a 24-bit system timer, SysTick, that counts down from the reload value to zero, reloads, that is wraps to, the value in the SYST_RVR register on the next clock edge, then counts down on subsequent clocks.
When the processor is halted for debugging the counter does not decrement.
The system timer registers are:
212/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
Address Name
Table 71. System timer registers summary
Type
Required privilege
Reset value
Description
0xE000E010 SYST_CSR RW Privileged 0x00000004
0xE000E014 SYST_RVR RW Privileged UNKNOWN
0xE000E018 SYST_CVR RW Privileged UNKNOWN
0xE000E01C SYST_CALIB RO Privileged 0xC0000000
SysTick control and status register
SysTick current value register
SysTick calibration value register
The SysTick SYST_CSR register enables the SysTick features. See the register summary in
Table 71 for its attributes. The bit assignments are:
Figure 40. SysTick SYST_CSR bit assignments
31 17 16 15
Reserved Reserved
3 2 1 0
0 0 0
COUNTFLAG
CLKSOURCE
TICKINT
ENABLE
MSv39696V1
Bits
[31:17] -
[16]
[15:3] -
COUNTFLAG
[2]
[1]
[0]
Name
CLKSOURCE
TICKINT
ENABLE
Table 72. SysTick SYST_CSR bit assignments
Function
Reserved.
Returns 1 if timer counted to 0 since last time this was read.
Reserved.
Indicates the clock source:
– 0: External clock.
– 1: Processor clock.
Enables SysTick exception request:
0: Counting down to zero does not assert the SysTick exception request.
1: Counting down to zero asserts the SysTick exception request.
Software can use COUNTFLAG to determine if SysTick has ever counted to zero.
Enables the counter:
0: Counter disabled.
1: Counter enabled.
PM0253 Rev 5 213/254
253
Cortex-M7 peripherals
4.4.2
PM0253
When ENABLE is set to 1, the counter loads the RELOAD value from the SYST_RVR register and then counts down. On reaching 0, it sets the COUNTFLAG to 1 and optionally asserts the SysTick depending on the value of TICKINT. It then loads the RELOAD value again, and begins counting.
SysTick reload value register
The SYST_RVR register specifies the start value to load into the SYST_CVR register. See the register summary in
for its attributes. The bit assignments are:
Figure 41. SYST_RVR bit assignments
31 24 23 0
Reserved RELOAD
4.4.3
MSv39695V1
Bits Name
Table 73. SYST_RVR bit assignments
Function
[31:24] -
[23:0] RELOAD
Reserved.
Value to load into the SYST_CVR register when the counter is enabled and when it reaches 0, see
Calculating the RELOAD value .
Calculating the RELOAD value
The RELOAD value can be any value in the range 0x00000001-0x00FFFFFF. A start value of 0 is possible, but has no effect because the SysTick exception request and COUNTFLAG are activated when counting from 1 to 0.
The RELOAD value is calculated according to its use. For example, to generate a multi-shot timer with a period of N processor clock cycles, use a RELOAD value of N-1. If the SysTick interrupt is required every 100 clock pulses, set RELOAD to 99.
SysTick current value register
The SYST_CVR register contains the current value of the SysTick counter. See the register summary in
for its attributes. The bit assignments are
31
Figure 42. SYST_CVR bit assignments:
24 23 0
Reserved CURRENT
MSv39694V1
214/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
Bits
[31:24] -
Name
[23:0] CURRENT
Table 74. SYST_CVR bit assignments
Function
Reserved.
Reads return the current value of the SysTick counter.
A write of any value clears the field to 0, and also clears the SYST_CSR
COUNTFLAG bit to 0.
The SYST_CALIB register indicates the SysTick calibration properties. See the register summary in
for its attributes. The bit assignments are:
31 30 29
Figure 43. SYST_CALIB bit assignments
24 23 0
Reserved TENMS
SKEW
NOREF
MSv39693V1
Bits Name
Table 75. SYST_CALIB bit assignments
Function
[31] NOREF
Indicates whether the device provides a reference clock to the processor:
0: Reference clock provided.
1: No reference clock provided.
If the device does not provide a reference clock, the SYST_CSR.CLKSOURCE bit reads-as-one and ignores writes.
[30]
[29:24] -
SKEW
[23:0] TENMS
Indicates whether the TENMS value is exact:
0: TENMS value is exact.
1: TENMS value is inexact, or not given.
An inexact TENMS value can affect the suitability of SysTick as a software real time clock.
Reserved.
Reload value for 10ms (100Hz) timing, subject to system clock skew errors. If the value reads as zero, the calibration value is not known.
If the calibration information is not known, calculate the calibration value required from the frequency of the processor clock or external clock.
PM0253 Rev 5 215/254
253
Cortex-M7 peripherals
4.4.5 SysTick design hints and tips
PM0253
The SysTick counter runs on the processor clock. If this clock signal is stopped for Low- power mode, the SysTick counter stops.
Ensure the software uses aligned word accesses to access the SysTick registers.
The SysTick counter reload and current value are undefined at reset, the correct initialization sequence for the SysTick counter is:
1.
Program reload value.
2. Clear current value.
3. Program Control and Status register.
In addition, the CMSIS provides a number of functions for SysTick control, including:
Table 76. CMSIS functions for SysTick control
CMSIS SysTick control function Description uint32_t
SysTick_Config(uint32_t ticks)
Creates a periodic SysTick interrupt using the SysTick timer, with a interval defined by the ticks parameter.
216/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
4.5.1
The processor features registers provide a software with cache configuration information.
The identification space registers are:
Address
Table 77. Identification space summary
Name Type
Required privilege
Reset value
Description
0xE000ED78 CLIDR
0xE000ED7C CTR
RO Privileged 0x09000003
RO Privileged 0x8303C003
0xE000ED80 CCSIDR RO Privileged Unknown
0xE000ED84 CSSELR RW Privileged Unknown
Cache type register on page 218
Cache size ID register on page 219
Cache size selection register on page 220
All the registers are only accessible by privileged loads and stores. Unprivileged accesses to these registers result in a BusFault.
Cache level ID register
The CLIDR identifies the type of cache, or caches, implemented at each level, and the level of coherency and unification. See the register summary in
for its attributes. The bit assignments are:
Figure 44. CLIDR bit assignments
31 30 29
LoU
27 26
LoC
24 23 21 20
LoUIS CL 7
18 17
CL 6
15 14
CL 5
12 11 10
CL 4
8
CL 3
6 5
CL 2
3 2
CL 1
0
Reserved
Bits Name
[31:30] -
[29:27]
[26:24]
[23:21]
[20:18]
[17:15]
LoU
LoC
LoUIS
CL 7
CL 6
SBZ.
Table 78. CLIDR bit assignments
Function
Level of Unification.
0b001 : Level 2, if either cache is implemented.
0b000:Level 1, if neither instruction nor data cache is implemented.
Level of Coherency.
0b001: Level 2, if either cache is implemented.
0b000: Level 1, if neither instruction nor data cache is implemented.
RAZ.
0b000 : No cache at CL 7.
0b000 : No cache at CL 6.
MSv39659V1
PM0253 Rev 5 217/254
253
Cortex-M7 peripherals
Table 78. CLIDR bit assignments (continued)
Bits Name Function
[14:12] CL 5
[11:9] CL 4
[8:6]
[5:3]
[2]
CL 3
CL 2
CL 1
[1]
[0]
CL 1
CL 1
0b000 : No cache at CL 5.
0b000 : No cache at CL 4.
0b000 : No cache at CL 3.
0b000 : No cache at CL 2.
RAZ: Indicates no unified cache at CL1.
1: Data cache is implemented.
0: No data cache is implemented.
1: An instruction cache is implemented.
0: No instruction cache is implemented.
PM0253
The CTR provides information about the cache architecture. See the register summary in
for its attributes. The bit assignments are:
31 29 28 27 24 23
Figure 45. CTR bit assignments
20 19 16 15 14 13 4 3 0
1 0 0 0 CWG ERG DMinLine 1 1 0 0 0 0 0 0 0 0 0 0 IminLine
Format
MSv39663V1
Bits Name
Table 79. CTR bit assignments
Description
[31:29] Format
[28] -
[27:24] CWG
Register format.
0b100 : Armv7 register format.
Reserved, RAZ.
Cache Writeback Granule.
0b0011 : 8 word granularity for the Cortex
®
-M7 processor.
[23:20] ERG
[19:16] DMinLine
[15:14] -
Exclusives Reservation Granule.
0b0000 : The local monitor within the processor does not hold any physical address. It treats any STREX instruction access as matching the address of the previous LDREX instruction. This means that the implemented exclusive reservation granule is the entire memory address range.
Smallest cache line of all the data and unified caches under the core control.
0b0011: 8 words for the Cortex
®
-M7 processor.
All bits RAO.
218/254 PM0253 Rev 5
PM0253
4.5.3
Cortex-M7 peripherals
Bits
[13:4] -
Name
[3:0] IminLine
Table 79. CTR bit assignments (continued)
Description
Reserved, RAZ.
Smallest cache line of all the instruction caches under the control of the processor.
0b0011 : 8 words for the Cortex ® -M7 processor.
Cache size ID register
The CCSIDR identifies the configuration of the cache currently selected by the CSSELR. If no instruction or data cache is configured, the corresponding CCSIDR is RAZ. See the register summary in
for its attributes. The bit assignments are:
Figure 46. CCSIDR bit assignments
31 30 29 28 27
NumSet
13 12
Assoc
3 2 0
LineSize
WA
RA
WB
WT
MSv39657V1
Table 80. CCSIDR bit assignments
Bits Name Function
(1)
[31]
[30]
[29]
[28]
[27:13]
[12:3]
[2:0]
WT
WB
RA
WA
NumSets
Associativity
LineSize
Indicates support available for Write-Through:
1: Write-Through support available.
Indicates support available for Write-Back:
1: Write-Back support available.
Indicates support available for read allocation:
1: Read allocation support available.
Indicates support available for write allocation:
1: Write allocation support available.
Indicates the number of sets as:
(number of sets) - 1.
Indicates the number of ways as:
(number of ways) - 1.
Indicates the number of words in each cache line.
1. See
for valid bit field encodings.
The LineSize field is encoded as 2 less than log(2) of the number of words in the cache line.
For example, a value of 0x0 indicates there are four words in a cache line, that is the minimum size for the cache. A value of 0x1 indicates there are eight words in a cache line.
PM0253 Rev 5 219/254
253
Cortex-M7 peripherals PM0253
Table 81 shows the individual bit field and complete register encodings for the CCSIDR. Use
this to determine the cache size for the L1 data or instruction cache selected by the Cache
Size Selection Register
(CSSELR). See Cache size selection register .
CSSELR Cache
0x0
0x1
Size
Table 81. CCSIDR encodings
Register bit field encoding
Complete register encoding WT WB RA WA NumSets
Assoc iativit y
LineSize
Data cache
4 Kbytes 0xF003E019
8 Kbytes 0xF007E019
16 Kbytes 0xF00FE019 1 1
32 Kbytes 0xF01FE019
64 Kbytes 0xF03FE019
4 Kbytes 0xF007E009
8 Kbytes 0xF00FE009
Instruction cache
16 Kbytes 0xF01FE009
32 Kbytes 0xF03FE009
1 1
64 Kbytes 0xF07FE009
1 1
1 1
0x001F
0x003F
0x007F
0x00FF
0x01FF
0x003F
0x007F
0x00FF
0x01FF
0x03FF
0x3
0x1
0x1
0x1
4.5.4 Cache size selection register
The CSSELR selects the cache whose configuration is currently visible in the CCSIDR. See the register summary in
for its attributes The bit assignments are:
Figure 47. CSSELR bit assignments
31
Reserved
4 3
Level
1 0
InD
MSv39662V1
Bit
[31:4] -
Name
[3:1]
[0]
Level
InD
Table 82. CSSELR bit assignments
Description
RESERVED
Identifies the cache level selected.
0b000: Level 1 cache.
This field is read only, writes are ignored.
Enables selection of instruction or data cache:
0: Data cache.
1: Instruction cache.
220/254 PM0253 Rev 5
PM0253
4.6
Cortex-M7 peripherals
Memory protection unit
The Memory Protection Unit (MPU) divides the memory map into a number of regions, and defines the location, size, access permissions, and memory attributes of each region. It supports:
• Independent attribute settings for each region.
• Overlapping regions.
• Export of memory attributes to the system.
The memory attributes affect the behavior of memory accesses to the region. The Cortex
M7 MPU defines:
® -
• 8 or 16 separate memory regions, 0-7 or 0-15.
• A background region.
When memory regions overlap, a memory access is affected by the attributes of the region with the highest number. For example, the attributes for region 7 take precedence over the attributes of any region that overlaps region 7.
The background region has the same memory access attributes as the default memory map, but is accessible from privileged software only.
The Cortex ® -M7 MPU memory map is unified. This means instruction accesses and data accesses have same region settings.
If a program accesses a memory location that is prohibited by the MPU, the processor generates a MemManage fault. This causes a fault exception, and might cause termination of the process in an OS environment. In an OS environment, the kernel can update the MPU region setting dynamically based on the process to be executed. Typically, an embedded
OS uses the MPU for memory protection.
The configuration of MPU regions is based on memory types, see
Memory regions, types and attributes on page 33 .
Table 83 shows the possible MPU region attributes. These include Shareability and cache
behavior attributes that are generally only relevant when the processor is configured with caches.
Memory type
Shareability
Table 83. Memory attributes summary
Other attributes Description
-
Strongly- ordered
-
Device Shared
Non-shared -
-
-
All accesses to Strongly-ordered memory occur in program order. All Strongly-ordered regions are assumed to be shared.
Memory-mapped peripherals that several processors share.
Memory-mapped peripherals that only a single processor uses.
PM0253 Rev 5 221/254
253
Cortex-M7 peripherals PM0253
-
Table 83. Memory attributes summary (continued)
Memory type
Shareability Other attributes Description
Normal Shared
Non-shared
Non-cacheable Writethrough Cacheable
Write-back Cacheable
Non-cacheable Writethrough Cacheable
Write-back Cacheable
Normal memory that is shared between several processors.
Normal memory that only a single processor uses.
Use the MPU registers to define the MPU regions and their attributes. The MPU registers are:
Address
Table 84. MPU registers summary
Name Type
Required privilege
Reset value
Description
0xE000ED90 MPU_TYPE
0xE000ED94 MPU_CTRL
0xE000ED98 MPU_RNR
0xE000ED9C MPU_RBAR
0xE000EDA0 MPU_RASR
RO Privileged 0x00000800
RW Privileged 0x00000000
MPU control register on page 223
RW Privileged Unknown
RW
RW
Privileged
Privileged
Unknown
-
(1)
MPU region number register on page 225
MPU region base address register on page 225
MPU region attribute and size register on page 226
0xE000EDA4 MPU_RBAR_A1 RW
0xE000EDA8 MPU_RASR_A1 RW
0xE000EDAC MPU_RBAR_A2 RW
0xE000EDB0 MPU_RASR_A2 RW
0xE000EDB4 MPU_RBAR_A3 RW
0xE000EDB8 MPU_RASR_A3 RW
Privileged
Privileged
Privileged
Privileged
Privileged
Privileged
Unknown
-
Unknown
-
Unknown
-
Alias of RBAR, see
MPU region base address register on page 225
Alias of RASR, see
MPU region attribute and size register on page 226
Alias of RBAR, see
MPU region base address register on page 225
Alias of RASR, see
MPU region attribute and size register on page 226
Alias of RBAR, see
MPU region base address register on page 225
Alias of RASR, see
MPU region attribute and size register on page 226
1. Unknown apart from the ENABLE field, which is reset to 0.
222/254 PM0253 Rev 5
PM0253
4.6.1
Cortex-M7 peripherals
MPU type register
The MPU_TYPE register indicates whether the MPU is present, and if so, how many regions it supports. If the MPU is not present the MPU_TYPE register is RAZ. See the register summary in
Table 84 for its attributes. The bit assignments are:
31 24 23
Figure 48. TYPE bit assignments
16 15 8 7 1 0
Reserved IREGION DREGION Reserved
4.6.2
SEPARATE
MSv39697V1
Bits Name
Table 85. TYPE bit assignments
Function
[31:24] -
[23:16] IREGION
Reserved.
Indicates the number of supported MPU instruction regions.
Always contains 0x00. The MPU memory map is unified and is described by the DREGION field.
[15:8]
[7:1]
[0]
DREGION
Indicates the number of supported MPU data regions:
0x08: 8 MPU regions.
0x10: 16 MPU regions.
- Reserved.
SEPARATE
Indicates support for unified or separate instruction and date memory maps:
0: Unified.
MPU control register
The MPU_CTRL register:
• Enables the MPU.
• Enables the default memory map background region.
• Enables use of the MPU when in the hard fault, Non Maskable Interrupt (NMI), and
FAULTMASK escalated handlers.
See the register summary in
for the MPU_CTRL attributes. The bit assignments are:
Figure 49. MPU_CTRL bit assignments
31 3 2 1 0
Reserved
PRIVDEFENA
HFNMIENA
ENABLE
MSv39681V1
PM0253 Rev 5 223/254
253
Cortex-M7 peripherals PM0253
Bits Name
Table 86. MPU_CTRL bit assignments
Function
[31:3] - Reserved.
[2]
[1]
PRIVDEFENA
HFNMIENA
Enables privileged software access to the default memory map:
0: If the MPU is enabled, disables use of the default memory map. Any memory access to a location not covered by any enabled region causes a fault.
1: If the MPU is enabled, enables use of the default memory map as a background region for privileged software accesses.
When enabled, the background region acts as if it is region number -1. Any region that is defined and enabled has priority over this default map.
If the MPU is disabled, the processor ignores this bit.
Enables the operation of MPU during hard fault, NMI, and FAULTMASK handlers.
When the MPU is enabled:
0: MPU is disabled during hard fault, NMI, and FAULTMASK handlers, regardless of the value of the ENABLE bit.
1: The MPU is enabled during hard fault, NMI, and FAULTMASK handlers.
When the MPU is disabled, if this bit is set to 1 the behavior is
Unpredictable.
[0] ENABLE
Enables the MPU:
0: MPU disabled.
1: MPU enabled.
When ENABLE and PRIVDEFENA are both set to 1:
• For privileged accesses, the default memory map
enabled memory region behaves as defined by the default memory map.
• Any access by unprivileged software that does not address an enabled memory region causes a MemManage fault.
XN and Strongly-ordered rules always apply to the System Control Space regardless of the value of the ENABLE bit.
When the ENABLE bit is set to 1, at least one region of the memory map must be enabled for the system to function unless the PRIVDEFENA bit is set to 1. If the PRIVDEFENA bit is set to 1 and no regions are enabled, then only privileged software can operate.
When the ENABLE bit is set to 0, the system uses the default memory map. This has the same memory attributes as if the MPU is not implemented, see
. The default memory map applies to accesses from both privileged and unprivileged software.
When the MPU is enabled, accesses to the System Control Space and vector table are always permitted. Other areas are accessible based on regions and whether PRIVDEFENA is set to 1.
Unless HFNMIENA is set to 1, the MPU is not enabled when the processor is executing the handler for an exception with priority –1 or –2. These priorities are only possible when handling a hard fault or NMI exception, or when FAULTMASK is enabled. Setting the
HFNMIENA bit to 1 enables the MPU when operating with these two priorities.
224/254 PM0253 Rev 5
PM0253
4.6.3
Cortex-M7 peripherals
MPU region number register
The MPU_RNR selects which memory region is referenced by the MPU_RBAR and
MPU_RASR registers. See the register summary in
for its attributes.
The bit assignments are:
Figure 50. MPU_RNR bit assignments
31 8 7 0
Reserved REGION
MSv39684V1
Bits Name
Table 87. MPU_RNR bit assignments
Function
[31:8] -
[7:0] REGION
Reserved.
Indicates the MPU region referenced by the MPU_RBAR and MPU_RASR registers.
The MPU supports 8 or 16 memory regions, so the permitted values of this field are 0-7 or 0-15.
Normally, the user writes the required region number to this register before accessing the
MPU_RBAR or MPU_RASR. However the region number can be changed by writing to the
MPU RBAR with the VALID bit set to 1, see MPU region base address register . This write
updates the value of the REGION field.
The MPU_RBAR defines the base address of the MPU region selected by the MPU_RNR,
and can update the value of the MPU_RNR. See the register summary in Table 84 on page 222
for its attributes.
Write MPU_RBAR with the VALID bit set to 1 to change the current region number and update the MPU_RNR. See the register summary in
Table 84 on page 222 for its attributes.
The bit assignments are
Figure 51. MPU_RBAR bit assignments:
31
ADDR
N N-1
Reserved
5 4 3
REGION
0
If the region size is 32B, the ADDR field is bits [31:5] and there is no Reserved field
VALID
MSv39683V1
PM0253 Rev 5 225/254
253
Cortex-M7 peripherals PM0253
Bits Name
Table 88. MPU_RBAR bit assignments
Function
[31:N]
[(N-1):5] -
[4]
[3:0]
ADDR
VALID
Region base address field. The value of N depends on the region size. For
more information see The ADDR field
.
Reserved.
MPU Region Number valid bit:
Write:
0: MPU_RNR not changed, and the processor:
Updates the base address for the region specified in the MPU_RNR
Ignores the value of the REGION field
1: The processor:
Updates the value of the MPU_RNR to the value of the REGION field
Updates the base address for the region specified in the REGION field.
Always reads as zero.
REGION
MPU region field:
For the behavior on writes, see the description of the VALID field.
On reads, returns the current region number, as specified by the RNR.
The ADDR field
The ADDR field is bits[31:N] of the MPU_RBAR. The region size, as specified by the SIZE field in the MPU_RASR, defines the value of N:
N = Log
2
(Region size in bytes),
If the region size is configured to 4 Gbytes, in the MPU_RASR, there is no valid ADDR field.
In this case, the region occupies the complete memory map, and the base address is
0x00000000.
The base address is aligned to the size of the region. For example, a 64KB region must be aligned on a multiple of 64KB, for example, at 0x00010000 or 0x00020000.
The MPU_RASR defines the region size and memory attributes of the MPU region specified by the MPU_RNR, and enables that region and any subregions. See the register summary in
for its attributes.
MPU_RASR is accessible using word accesses:
• The most significant halfword holds the region attributes.
• The least significant halfword holds the region size and the region and subregion enable bits.
The bit assignments are:
226/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
Figure 52. MPU_RASR bit assignments
31 29 28 27 26
AP
24 23 22 21 19 18 17 16 15
TEX S C B SRD
8 7 6 5
SIZE
1 0
Reserved
XN
Reserved
Reserved Reserved
ENABLE
MSv39682V1
Bits Name
Table 89. MPU_RASR bit assignments
Function
[31:29]
[28]
-
XN
-
-
AP
Reserved.
Instruction access disable bit:
0: Instruction fetches enabled.
1: Instruction fetches disabled.
Reserved.
[27]
[26:24]
[23:22]
Access permission field, see
Reserved.
[21:19, 17, 16]
TEX, C, B Memory access attributes, see Table 91 on page 228
.
[18] S
Shareable bit, see Table 91 on page 228
.
[15:8]
[7:6]
[5:1]
[0]
SRD
Subregion disable bits. For each bit in this field:
0: Corresponding sub-region is enabled.
1: Corresponding sub-region is disabled.
See
for more information.
Region sizes of 128 bytes and less do not support subregions. When writing the attributes for such a region, write the SRD field as 0x00.
Reserved.
-
SIZE
Specifies the size of the MPU protection region. The minimum permitted value is 4 (0b00100), see
for more information.
ENABLE Region enable bit.
For information about access permission, see
MPU access permission attributes on page 228
.
SIZE field values
The SIZE field defines the size of the MPU memory region specified by the RNR. as follows:
(Region size in bytes) = 2 (SIZE+1)
The smallest permitted region size is 32B, corresponding to a SIZE value of 4.
gives example SIZE values, with the corresponding region size and value of N in the
MPU_RBAR.
PM0253 Rev 5 227/254
253
Cortex-M7 peripherals
4.6.6
PM0253
SIZE value
Table 90. Example SIZE field values
Region size Value of N
(1)
Note
0b00100 (4)
0b01001 (9)
0b10011 (19)
0b11101 (29)
0b11111 (31)
32 Bytes
1 Kbyte
1 Mbyte
1 Gbyte
4 Gbytes
5
10
20
30
32
1. In the MPU_RBAR, see
MPU region base address register on page 225 .
-
-
-
Minimum permitted size
Maximum possible size
MPU access permission attributes
This section describes the MPU access permission attributes. The access permission bits,
TEX, C, B, S, AP, and XN, of the RASR, control access to the corresponding memory region. If an access is made to an area of memory without the required permissions, then
the MPU generates a permission fault. Table 91
shows encodings for the TEX, C, B, and S access permission bits.
TEX
0b000
0b001
0b010
Table 91. TEX, C, B, and S encoding
C B S Memory type Shareability Other attributes
0
1
0
1
0 x
(1)
0
1
0
1
0
1
0
0
1
1
0
1
0
1
Stronglyordered
Device
Shareable -
Normal
Normal
Shareable -
Not shareable Outer and inner write-through. No write allocate.
Shareable
Not shareable
Shareable
Outer and inner write-back. No write allocate.
Normal
Not shareable
Shareable
Reserved encoding
Implementation defined attributes.
-
-
Outer and inner noncacheable.
Normal
Not shareable
Shareable
Device
Not shareable
Reserved encoding
Reserved encoding
-
-
Outer and inner write-back. Write and read allocate.
Nonshared Device.
228/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
TEX
Table 91. TEX, C, B, and S encoding (continued)
C B S Memory type Shareability Other attributes
0b1BB A A
0
1
Normal
Not shareable
Shareable
Cached memory, BB = outer policy,
AA = inner policy. See Table 92 on page 229
for the encoding of the AA and
BB bits.
1. The MPU ignores the value of this bit.
00
01
10
11
Table 92 shows the cache policy for memory attribute encodings with a TEX value is in the
range 4-7.
Table 92. Cache policy for memory attribute encoding
Encoding, AA or BB Corresponding cache policy
Non-cacheable
Write back, write and read allocate
Write through, no write allocate
Write back, no write allocate
Table 93 shows the AP encodings that define the access permissions for privileged and
unprivileged software.
011
100
101
110
111
AP[2:0]
000
001
010
Privileged permissions
Table 93. AP encoding
Unprivileged permissions
Description
No access
RW
RW
No access
No access
RO
All accesses generate a permission fault
Access from privileged software only
Writes by unprivileged software generate a permission fault
Full access RW RW
Unpredictable Unpredictable Reserved
RO No access Reads by privileged software only
RO
RO
RO
RO
Read only, by privileged or unprivileged software
Read only, by privileged or unprivileged software
PM0253 Rev 5 229/254
253
Cortex-M7 peripherals PM0253
4.6.8
When an access violates the MPU permissions, the processor generates a MemManage fault, see
Exceptions and interrupts on page 28
. The MMFSR indicates the cause of the
fault. See MemManage fault status register on page 206
for more information.
Updating an MPU region
To update the attributes for an MPU region, update the MPU_RNR, MPU_RBAR and
MPU_RASR registers. It is possible to program each register separately, or use a multipleword write to program all of these registers. The MPU_RBAR and MPU_RASR aliases can be used to program up to four regions simultaneously using an STM instruction.
Updating an MPU region using separate words
Simple code to configure one region:
; R1 = region number
; R2 = size/enable
; R3 = attributes
; R4 = address
LDR R0,=MPU_RNR ; 0xE000ED98, MPU region number register
STR R1, [R0, #0x0] ; Region Number
STR R4, [R0, #0x4] ; Region Base Address
STRH R2, [R0, #0x8] ; Region Size and Enable
STRH R3, [R0, #0xA] ; Region Attribute
Disable a region before writing new region settings to the MPU if the region being changed, has been previously enabled. For example:
; R1 = region number
; R2 = size/enable
; R3 = attributes
; R4 = address
LDR R0,=MPU_RNR ; 0xE000ED98, MPU region number register
STR R1, [R0, #0x0] ; Region Number
BIC R2, R2, #1 ; Disable
STRH R2, [R0, #0x8] ; Region Size and Enable
STR R4, [R0, #0x4] ; Region Base Address
STRH R3, [R0, #0xA] ; Region Attribute
ORR R2, #1 ; Enable
STRH R2, [R0, #0x8] ; Region Size and Enable
The software must use memory barrier instructions:
• Before MPU setup if there might be outstanding memory transfers, such as buffered writes, that might be affected by the change in MPU settings.
• After MPU setup if it includes memory transfers that must use the new MPU settings.
The software does not require any memory barrier instructions during MPU setup, because it accesses the MPU through the PPB, which is a Strongly-ordered memory region.
For example, if it is required that all of the memory access behavior to take effect immediately after the programming sequence, use a DSB instruction and an ISB instruction.
A DSB is required after changing MPU settings, such as at the end of context switch. An ISB is required if the code that programs the MPU region or regions is entered using a branch or call. If the programming sequence is entered by taking an exception and the programming sequence is exited by using a return from exception then an ISB instruction is not required.
230/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
Updating an MPU region using multi-word writes
The user can program directly using multi-word writes, depending on how the information is divided. Consider the following reprogramming:
; R1 = region number
; R2 = address
; R3 = size, attributes in one
LDR R0, =MPU_RNR ; 0xE000ED98, MPU region number register
STR R1, [R0, #0x0] ; Region Number
STR R2, [R0, #0x4] ; Region Base Address
STR R3, [R0, #0x8] ; Region Attribute, Size and Enable
Use an STM instruction to optimize this:
; R1 = region number
; R2 = address
; R3 = size, attributes in one
LDR R0, =MPU_RNR ; 0xE000ED98, MPU region number register
STM R0, {R1-R3} ; Region Number, address, attribute, size and enable
The user can do this in two words for pre-packed information. This means that the
MPU_RBAR contains the required region number and had the VALID bit set to 1, see
MPU region base address register on page 225 . Use this when the data is statically packed, for
example in a boot loader:
; R1 = address and region number in one
; R2 = size and attributes in one
LDR R0, =MPU_RBAR ; 0xE000ED9C, MPU Region Base register.
STR R1, [R0, #0x0] ; Region base address and region number combined
; with VALID (bit 4) set to 1.
STR R2, [R0, #0x4] ; Region Attribute, Size and Enable.
Subregions
Regions of 256 bytes or more are divided into eight equal-sized subregions. Set the corresponding bit in the SRD field of the MPU_RASR to disable a subregion, see
MPU region attribute and size register on page 226 . The least significant bit of SRD controls the
first subregion, and the most significant bit controls the last subregion. Disabling a subregion means another region overlapping the disabled range matches instead. If no other enabled region overlaps the disabled subregion, and the access is unprivileged or the background region is disabled, the MPU issues a fault.
Regions of 32, 64, and 128 bytes do not support subregions, With regions of these sizes,
The user must set the SRD field to 0x00, otherwise the MPU behavior is Unpredictable.
Example of SRD use
Two regions with the same base address overlap. Region one is 128 KB, and region two is
512 KB. To ensure the attributes from region one apply to the first 128 KB region, set the
SRD field for region two to 0b00000011 to disable the first two subregions, as the figure shows.
PM0253 Rev 5 231/254
253
Cortex-M7 peripherals
Figure 53. Example of disabling subregion
Region 2, with subregions
Base address of both regions
Region 1
Disabled subregion
Disabled subregion
Offset from base address
512KB
448KB
384KB
320KB
256KB
192KB
128KB
64KB
0
PM0253
MSv39664V1
To avoid an unexpected behavior, disable the interrupts before updating the attributes of a region that the interrupt handlers might access.
The processor does not support unaligned accesses to MPU registers.
The MPU registers support aligned word accesses only. The byte and halfword accesses are unpredictable.
When setting up the MPU, and if the MPU has previously been programmed, disable the unused regions to prevent any previous region settings from affecting the new MPU setup.
232/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
4.7.1
The Cortex
®
-M7 Floating-Point Unit (FPU) implements the FPv5 floating-point extensions.
The FPU fully supports single-precision and double-precision add, subtract, multiply, divide, multiply and accumulate, and square root operations. It also provides conversions between fixed-point and floating-point data formats, and floating-point constant instructions.
The FPU provides floating-point computation functionality that is compliant with the
ANSI/IEEE Std 754-2008, IEEE Standard for Binary Floating-Point Arithmetic , referred to as the IEEE 754 standard.
The silicon vendor should also include the following text when implementations support single-precision FPU only. The FPU contains 32 single-precision extension registers, which can be also accessed as 16 doubleword registers for load, store, and move operations.
Table 94 shows the floating-point system registers in the Cortex
®
-M7 processor with FPU.
Address
Table 94. Cortex
®
-M7 floating-point system registers
Name Type
Required privilege
Reset Description
-
0xE000ED88 CPACR
0xE000EF34 FPCCR
0xE000EF38 FPCAR
0xE000EF3C
FPSCR
(1)
FPDSC
R
RW
RW
RW
RW
RW
Privileged
Privileged
Privileged
Unprivileged -
Privileged
-
0x00000000
0xC0000000
Coprocessor access control register on page 233
Floating-point context control register on page 234
0x00000000
Floating-point context address register on page 236
Floating-point status control register on page 236
Floating-point default status control register on page 237
1. The FPSCR register is not memory-mapped, it can bee accessed using the VMSR and VMRS instructions, see
and
VMSR on page 166 . the software can only access the FPSCR when the FPU
is enabled, see Enabling the FPU on page 238
.
The following sections describe the floating-point system registers whose implementation is specific to this processor.
Coprocessor access control register
The CPACR register specifies the access privileges for coprocessors. See the register summary in
for its attributes. The bit assignments are:
Figure 54. CPACR bit assignments
31
Reserved
24 23 22 21 20 19
CP11 CP10 Reserved
0
MSv39660V1
PM0253 Rev 5 233/254
253
Cortex-M7 peripherals PM0253
Bits
[31:24]
[19:0] -
-
Name
[2 n +1:2 n ] for n values10 and
11
CP n
Table 95. CPACR bit assignments
Function
Reserved. Read as Zero, Write Ignore.
Access privileges for coprocessor n . The possible values of each field are:
0b00 : Access denied. Any attempted access generates a NOCP
UsageFault.
0b01 : Privileged access only. An unprivileged access generates a
NOCP fault.
0b10 : Reserved. The result of any access is Unpredictable.
0b11 : Full access.
Reserved. Read as Zero, Write Ignore.
for its attributes. The bit assignments are:
Figure 55. FPCCR bit assignments
31 30 29 9 8 7 6 5 4 3 2 1 0
Reserved
Bits
[31]
LSPEN
ASPEN
Name
ASPEN
[30]
[29:9] -
LSPEN
MONRDY
Reserved
BFRDY
MMRDY
HFRDY
THREAD
Reserved
USER
LSPACT
MSv39666V1
Table 96. FPCCR bit assignments
Function
Enables CONTROL.FPCA setting on execution of a floating-point instruction.
This results in automatic hardware state preservation and restoration, for floating-point context, on exception entry and exit.
0: Disable CONTROL.FPCA setting on execution of a floating-point instruction.
1: Enable CONTROL.FPCA setting on execution of a floating-point instruction.
0: Disable automatic lazy state preservation for floating-point context.
1: Enable automatic lazy state preservation for floating-point context.
Reserved.
234/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
Bits
[8]
[7]
[6]
[5]
[4]
[3]
[2]
[1]
[0]
-
-
Name
MONRDY
BFRDY
MMRDY
HFRDY
THREAD
USER
LSPACT
Table 96. FPCCR bit assignments (continued)
Function
0: DebugMonitor is disabled or priority did not permit setting MON_PEND when the floating-point stack frame was allocated.
1: DebugMonitor is enabled and priority permits setting MON_PEND when the floating-point stack frame was allocated.
Reserved.
0: BusFault is disabled or priority did not permit setting the BusFault handler to the pending state when the floating-point stack frame was allocated.
1: BusFault is enabled and priority permitted setting the BusFault handler to the pending state when the floating-point stack frame was allocated.
0: MemManage is disabled or priority did not permit setting the MemManage handler to the pending state when the floating-point stack frame was allocated.
1: MemManage is enabled and priority permitted setting the MemManage handler to the pending state when the floating-point stack frame was allocated.
0: Priority did not permit setting the HardFault handler to the pending state when the floating-point stack frame was allocated.
1: Priority permitted setting the HardFault handler to the pending state when the floating-point stack frame was allocated.
0: Mode was not Thread Mode when the floating-point stack frame was allocated.
1: Mode was Thread Mode when the floating-point stack frame was allocated.
Reserved.
0: Privilege level was not user when the floating-point stack frame was allocated.
1: Privilege level was user when the floating-point stack frame was allocated.
0: Lazy state preservation is not active.
1: Lazy state preservation is active. Floating-point stack frame has been allocated but saving state to it has been deferred.
PM0253 Rev 5 235/254
253
Cortex-M7 peripherals PM0253
The FPCAR register holds the location of the unpopulated floating-point register space allocated on an exception stack frame. See the register summary in
for its attributes. The bit assignments are:
Figure 56. FPCAR bit assignments
31 3 2 0
ADDRESS
Bits
[31:3]
[2:0]
Name
-
ADDRESS
Reserved
MSv39665V1
Table 97. FPCAR bit assignments
Function
The location of the unpopulated floating-point register space allocated on an exception stack frame.
Reserved. Read as Zero, Writes Ignored.
236/254
The FPSCR register provides all necessary User level control of the floating-point system.
The bit assignments are:
Figure 57. FPSCR bit assignments
31 30 29 28 27 26 25 24 23 22 21 8 7 6 5 4 3 2 1 0
N Z C V Reserved
Reserved
AHP
RMode
FZ
DN
IDC
Reserved
IXC
IOC
DZC
OFC
UFC
MSv39668V1
Bits Name
[31]
[30]
[29]
[28]
[27] -
N
Z
C
V
[26] AHP
Table 98. FPSCR bit assignments
Function
Condition code flags. Floating-point comparison operations update these flags.
N: Negative condition code flag.
Z: Zero condition code flag.
C: Carry condition code flag.
V: Overflow condition code flag.
Reserved.
Alternative half-precision control bit:
0: IEEE half-precision format selected.
1: Alternative half-precision format selected.
PM0253 Rev 5
PM0253 Cortex-M7 peripherals
Bits Name
Table 98. FPSCR bit assignments (continued)
Function
[25]
[24]
DN
FZ
[23:22] RMode
[21:8] -
[7] IDC
[6:5]
[4]
[3]
[2]
[1]
[0]
-
IXC
UFC
OFC
DZC
IOC
Default NaN mode control bit:
0: NaN operands propagate through to the output of a floating-point operation.
1: Any operation involving one or more NaNs returns the Default NaN.
Flush-to-zero mode control bit:
0: Flush-to-zero mode disabled. Behavior of the floating-point system is fully compliant with the IEEE 754 standard.
1: Flush-to-zero mode enabled.
Rounding Mode control field. The encoding of this field is:
0b00 : Round to Nearest (RN) mode
0b01 : Round towards Plus Infinity (RP) mode
0b10 : Round towards Minus Infinity (RM) mode
0b11 : Round towards Zero (RZ) mode.
The specified rounding mode is used by almost all floating-point instructions.
Reserved.
Input Denormal cumulative exception bit, see bits [4:0].
Reserved.
Cumulative exception bits for floating-point exceptions, see also bit [7]. Each of these bits is set to 1 to indicate that the corresponding exception has occurred since 0 was last written to it.
IDC, bit[7]: Input Denormal cumulative exception bit.
IXC: Inexact cumulative exception bit.
UFC: Underflow cumulative exception bit.
OFC: Overflow cumulative exception bit.
DZC: Division by Zero cumulative exception bit.
IOC: Invalid Operation cumulative exception bit.
The FPDSCR register holds the default values for the floating-point status control data. See the register summary in
for its attributes. The bit assignments are:
Figure 58. FPDSCR bit assignments
0 31
Reserved
27 26 25 24 23 22 21
0 0 0 0 0
AHP
DN
[31:27]
[26]
Bits
-
AHP
Reserved
RMode
FZ
Table 99. FPDSCR bit assignments
Name Function
Reserved
Default value for FPSCR.AHP
MSv39667V1
PM0253 Rev 5 237/254
253
Cortex-M7 peripherals
4.7.6
PM0253
[25]
[24]
[23:22]
[21:0]
Bits
Table 99. FPDSCR bit assignments (continued)
Name Function
DN
FZ
-
RMode
Default value for FPSCR.DN
Default value for FPSCR.FZ
Default value for FPSCR.RMode
Reserved
Enabling the FPU
The FPU is disabled from reset. The user must enable it before using any floating-point
instructions. Example 4-1: Enabling the FPU shows an example code sequence for enabling
the FPU in privileged mode. The processor must be in privileged mode to read from and write to the CPACR.
Example 4-1: Enabling the FPU
CPACR EQU 0xE000ED88
LDR R0, =CPACR ; Read CPACR
LDR r1, [R0] ; Set bits 20-23 to enable CP10 and CP11
; coprocessors
ORR R1, R1, #(0xF << 20)
STR R1, [R0] ; Write back the modified value to the CPACR
DSB
ISB ; Reset pipeline now the FPU is enabled.
238/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
4.7.7 Enabling clearing FPU exception interrupts
The FPU exception flags are generating an interrupt through the interrupt controller. The
FPU interrupt is globally controlled through the interrupt controller.
There is no individual mask and the enable/disable of the FPU interrupts is done at interrupt controller level. As it occurs very frequently, the IXC exception flag is not connected to the interrupt controller, and cannot generate an interrupt. If needed, it must be managed by polling.
Clearing the FPU exception flags depends on the FPU context save/restore configuration:
• No floating-point register saving: when Floating-point context control register (FPCCR)
Bit 30 LSPEN=0 and Bit 31 ASPEN=0.
You must clear interrupt source in Floating-point Status and Control Register (FPSCR).
Example: register uint32_t fpscr_val = 0; fpscr_val = __get_FPSCR();
{ check exception flags } fpscr_val &= (uint32_t)~0x8F; // Clear all exception flags
__set_FPSCR(fpscr_val);
• Lazy save/restore: when Floating-point context control register (FPCCR)
Bit 30 LSPEN=1 and Bit 31 ASPEN=X.
In the case of lazy floating-point context save/restore, a dummy read access should be made to Floating-point Status and Control Register (FPSCR) to force state preservation and FPSCR clear.
Then handle FPSCR in the stack.
Example: register uint32_t fpscr_val = 0; register uint32_t reg_val = 0; reg_val = __get_FPSCR(); //dummy access fpscr_val=*(__IO uint32_t*)(FPU->FPCAR +0x40);
{ check exception flags } fpscr_val &= (uint32_t)~0x8F ; // Clear all exception flags
*(__IO uint32_t*)(FPU->FPCAR +0x40)=fpscr_val;
__DMB() ;
• Automatic floating-point registers save/restore: when Floating-point context control register (FPCCR)
Bit 30 LSPEN=0 and Bit 31 ASPEN=1.
In the case of automatic floating-point context save/restore, a read access should be made to Floating-point Status and Control Register (FPSCR) to force clear.
Then handle FPSCR in the stack.
Example:
// FPU Exception handler void FPU_ExceptionHandler(uint32_t lr, uint32_t sp)
{ register uint32_t fpscr_val; if(lr == 0xFFFFFFE9)
{
sp = sp + 0x60;
} else if(lr == 0xFFFFFFED)
PM0253 Rev 5 239/254
253
Cortex-M7 peripherals PM0253
{
sp = __get_PSP() + 0x60 ;
} fpscr_val = *(uint32_t*)sp;
{ check exception flags } fpscr_val &= (uint32_t)~0x8F ; // Clear all exception flags
*(uint32_t*)sp = fpscr_val;
__DMB() ;
}
// FPU IRQ Handler void __asm FPU_IRQHandler(void)
{
IMPORT FPU_ExceptionHandler
MOV R0, LR
MOV R1, SP
// move LR to R0
// Save SP to R1 to avoid any modification to
VMRS R2, FPSCR
B
BX
}
// the stack pointer from FPU_ExceptionHandler
// dummy read access, to force clear
FPU_ExceptionHandler
LR
The cache maintenance operations are only accessible by privileged loads and stores.
Unprivileged accesses to these registers always generate a BusFault.
Address
Table 100. Cache maintenance space register summary
Name Type
Required privilege
Reset value
Description
0xE000EF50 ICIALLU
0xE000EF54 -
0xE000EF58 ICIMVAU
0xE000EF5C DCIMVAC
0xE000EF60 DCISW
-
WO
WO
WO
WO
0xE000EF64 DCCMVAU WO
Privileged Unknown
Instruction cache invalidate all to the Point of
Unification (PoU) (1)
Reserved
Privileged Unknown Instruction cache invalidate by address to the PoU
Privileged Unknown
Data cache invalidate by address to the Point of
Coherency (PoC)
(2)
Privileged Unknown Data cache invalidate by set/way
Privileged Unknown Data cache by address to the PoU
Privileged Unknown Data cache clean by address to the PoC
0xE000EF68 DCCMVAC WO
0xE000EF6C DCCSW WO
0xE000EF70 DCCIMVAC WO
Privileged Unknown Data cache clean by set/way
Privileged Unknown
Data cache clean and invalidate by address to the
PoC
0xE000EF74 DCCISW
0xE000EF78 BPIALL
WO Privileged Unknown Data cache clean and invalidate by set/way
RAZ/WI Privileged The BPIALL register is not implemented
1. Cache maintenance operations by PoU can be used to synchronize data between the Cortex ®
Caches, for example when the software uses self-modifying code.
-M7 data and instruction
240/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
2. Cache maintenance operations by PoC can be used to synchronize data between the Cortex ® external agent such as a system DMA.
-M7 data cache and an
4.8.1
4.8.2
Full instruction cache operation
The ICIALLU is WO and write data is ignored and reads return 0. Writes to this register perform the requested cache maintenance operation. The BPIALL register is not implemented in the Cortex-M7 processor as branch predictor maintenance is not required.
The register is RAZ/WI.
Instruction and data cache operations by address
The cache maintenance operations registers are ICIMVAU, DCIMVAC, DCCMVAU,
DCCMVAC, and DCCIMVAC. These registers are WO, reads return 0. See the register summary in
Table 100 for their attributes. The bit assignments are:
[31:0]
Bit
Table 101. Cache operation registers bit assignments
Name Type Function
MVA WO MVA of requested operation
4.8.3 Data cache operations by set-way
The DCISW, DCCSW and DCCISW registers are WO. Reads return 0. See the register summary in
for their attributes. The bit assignments are:
31 30 29
Figure 59. Cache operation bit assignments
14 13 5 4 1 0
Way Reserved Set Reserved 0
MSv39654V1
Table 102. Cache operations by set-way bit assignments
Bit Name Type Function
[31:30] Way WO
[29:14] -
[13:5] Set
[4:1]
[0] -
-
-
-
-
WO
Way that operation applies to.
For the data cache, values 0, 1, 2 and 3 are supported.
Reserved
Set/index that operation applies to. The number of indices in a cache depends on the configured cache size. When this is less than the maximum, use the LSB of this field. The number of sets in the cache can be determined by reading the
Cache size ID register on page 219
.
Reserved
Always reads as zero.
PM0253 Rev 5 241/254
253
Cortex-M7 peripherals PM0253
4.8.4 Cortex
®
-M7 cache maintenance operations using CMSIS
CMSIS functions enable the software portability between different Cortex ® -M profile processors. To access cache maintenance operations when using CMSIS, use the following functions:
Table 103. CMSIS access cache maintenance operations
CMSIS function Descriptions void SCB_EnableICache(void) void SCB_DisableICache(void) void SCB_InvalidateICache(void) void SCB_EnableDCache(void) void SCB_DisableDCache(void)
Invalidate and then enable instruction cache
Disable instruction cache and invalidate its contents
Invalidate instruction cache
Invalidate and then enable data cache
Disable data cache and then clean and invalidate its contents void SCB_InvalidateDCache(void)
Invalidate data cache
Clean data cache void SCB_CleanDCache(void) void SCB_CleanInvlaidateDCache(void)
Clean and invalidate data cache
4.8.5 Initializing and enabling the L1-cache
The user can use cache maintenance operations for:
• Cache startup type operations.
• Manipulating the caches so that shared data is visible to other bus masters.
• Enabling data changed by an external DMA agent to be made visible to the Cortex ®
M7 processor.
-
After enabling or disabling the instruction cache, the user must issue an ISB instruction to flush the pipeline. This ensures that all subsequent instruction fetches see the effect of enabling or disabling the instruction cache.
After reset, the user must invalidate each cache before enabling it.
When disabling the data cache, the user must clean the entire cache to ensure that any dirty data is flushed to external memory.
Before enabling the data cache, the user must invalidate the entire data cache if external memory might have changed since the cache was disabled.
Before enabling the instruction cache, the user must invalidate the entire instruction cache if external memory might have changed since the cache was disabled.
L1 data and instruction cache must be invalidated before they are enabled in the software, otherwise unpredictable behavior can occur.
Invalidate the entire data cache
The software can use the following code example to invalidate the entire data cache, if it has been included in the processor. The operation is carried out by iterating over each line of the cache and using the DCISW register in the Private Peripheral Bus (PPB) memory region to
242/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals invalidate the line. The number of cache ways and sets is determined by reading the
CCSIDR register.
CCSIDR EQU 0xE000ED80
CSSELR EQU 0xE000ED84
DCISW EQU 0xE000EF60
MOV r0, #0x0
LDR r11, =CSSELR
STR r0, [r11] ; Select Data Cache size
DSB
LDR r11, =CCSIDR
LDR r2, [r11] ; Cache size identification
AND r1, r2, #0x7 ; Number of words in a cache line
ADD r7, r1, #0x4
MOV r1, #0x3ff
ANDS r4, r1, r2, LSR #3
MOV r1, #0x7fff
ANDS r2, r1, r2, LSR #13
CLZ r6, r4
LDR r11, =DCISW inv_loop1
MOV r1, r4 inv_loop2
LSL r3, r1, r6
LSL r8, r2, r7
ORRr 3, r3, r8
STR r3, [r11] ; Invalidate D-cache line
SUBS r1, r1, #0x1
BGE inv_loop2
SUBS r2, r2, #0x1
BGE inv_loop1
DSB
ISB
Invalidate instruction cache
The user can use the following code example to invalidate the entire instruction cache, if it has been included in the processor. The operation is carried out by writing to the ICIALLU register in the PPB memory region.
ICIALLU EQU 0xE000EF50
MOV r0, #0x0
LDR r11, =ICIALLU
STR r0, [r11]
DSB
ISB
PM0253 Rev 5 243/254
253
Cortex-M7 peripherals
4.8.6
4.8.7
PM0253
Enabling data and instruction caches
The user can use the following code example to enable the data and instruction cache after they have been initialized. The operation is carried out by modifying the CCR.IC and
CCR.DC fields in the PPB memory region.
CCR EQU 0xE000ED14
LDR r11, =CCR
LDR r0, [r11]
ORR r0, r0, #0x1:SHL:16 ; Set CCR.DC field
ORR r0, r0, #0x1:SHL:17 ; Set CCR.IC field
STR r0, [r11]
DSB
ISB
Faults handling considerations
Cache maintenance operations can result in a BusFault. Such fault events are asynchronous.
This type of BusFault:
• Does not cause escalation to HardFault where a BusFault handler is enabled.
• Never causes lockup.
Because the fault event is asynchronous, the software code for cache maintenance operations should use memory barrier instructions, such as DSB, on completion so that the fault event can be observed immediately.
Cache maintenance design hints and tips
The user must always place a DSB and ISB instruction sequence after a cache maintenance operation to ensure that the effect is observed by any following instructions in the software.
When using a cache maintenance operation by address or set/way a DSB instruction must be executed after any previous load or store, and before the maintenance operation, to guarantee that the effect of the load or store is observed by the operation. For example, if a store writes to the address accessed by a DCCMVAC the DSB instruction guarantees that the dirty data is correctly cleaned from the data cache.
When one or more maintenance operations have been executed, use of a DSB instruction guarantees that they have completed and that any following load or store operations executes in order after the maintenance operations.
Cache maintenance operations always complete in-order with respect to each other. This means only one DSB instruction is required to guarantee the completion of a set of maintenance operations.
The following code sequence shows how to use cache maintenance operations to synchronize the data and instruction caches for self-modifying code. The sequence is entered with <Rx> containing the new 32-bit instruction. Use STRH in the first line instead of
STR for a 16-bit instruction:
STR <Rx>, <inst_address1>
DSB ; Ensure the data has been written to the
; cache.
244/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
STR <inst_address1>, DCCMVAU ; Clean data cache by MVA to point of
; unification (PoU).
STR <inst_address1>, ICIMVAU ; Invalidate instruction cache by MVA to
; PoU.
DSB ; Ensure completion of the invalidations.
ISB ; Synchronize fetched instruction stream.
Control of the L1-cache ECC and attribute override, the priority of AHB slave traffic, and whether an access is mapped to TCM interfaces or AXI master interface, is defined by the access control registers. The access control registers are:
Address
Table 104. Access control register summary
Name Type
Required privilege
Reset value
Description
0xE000EF90 ITCMCR RW Privileged 0x00000000
0xE000EF94 DTCMCR RW Privileged 0x00000000
Instruction and data tightly-coupled memory control registers on page 246
0xE000EF98 AHBPCR RW Privileged 0x00000000
AHBP control register on page 248
0xE000EF9C CACR RW Privileged -
(1)
Auxiliary cache control register on page 249
0xE000EFA0 AHBSCR RW Privileged 0x00000800
AHB slave control register on page 250
0xE000EFA8 ABFSR RW Privileged 0x00000000
Auxiliary bus fault status register on page 251
1. The reset value is implementation and configuration dependent and the silicon vendor changes this. If cache ECC is configured the reset value is 0x00000000, if cache ECC is not configured the reset value is 0x00000002.
PM0253 Rev 5 245/254
253
Cortex-M7 peripherals
4.9.1 Instruction and data tightly-coupled memory control registers
PM0253
The ITCMCR and DTCMCR control whether access is mapped to the TCM interfaces or the
AXI master interface. The bit assignments are:
31
Figure 60. ITCMR and DTCMR bit assignments
7 6
Reserved SZ
3 2 1 0
RETEN
RMW
EN
MSv39677V1
Bits
Table 105. ITCMCR and DTCMCR bit assignments
Name Type Function
[31:7]
[6:3]
[2]
[1]
[0]
-
SZ
RETEN
RMW
EN
(2)
(1)
-
RO
RW
RW
RW
Reserved, RAZ/WI.
TCM size. Indicates the size of the relevant TCM:
0b0000: No TCM implemented.
0b0011: 4KB.
0b0100: 8KB.
0b0101: 16KB.
0b0110: 32KB.
0b0111: 64KB.
0b1000: 128KB.
0b1001: 256KB.
0b1010: 512KB.
0b1011: 1MB.
0b1100: 2MB.
0b1101: 4MB.
0b1110: 8MB.
0b1111: 16MB.
All other encodings are reserved.
Retry phase enable. When enabled the processor guarantees to honor the retry output on the corresponding TCM interface:
0: Retry phase disabled.
1: Retry phase enabled.
Read-Modify-Write (RMW) enable. Indicates that all sub-chunk writes to a given TCM use a RMW sequence:
0: RMW disabled.
1: RMW enabled.
TCM enable. When a TCM is disabled all accesses are made to the AXI master.
0: TCM disabled.
1: TCM enabled.
1. The RETEN field in the ITCMCR and DTCMCR is used to support error detection and correction in the
TCM.
2. The RMW field in the ITCMCR and DTCMCR is used to support error detection and correction in the TCM.
246/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
Enabling the TCM
The TCM interfaces can be enabled at reset in the system by an external signal on the processor. If they are disabled at reset then the following code example can be used to enable both the instruction and data TCM interfaces in software:
ITCMCR EQU 0xE000EF90
DTCMCR EQU 0xE000EF94
LDR r11, =ITCMCR
LDR r0, [r11]
ORR r0, r0, #0x1 ; Set ITCMCR.EN field
STR r0, [r11]
LDR r11, =DTCMCR
LDR r0, [r11]
ORR r0, r0, #0x1 ; Set DTCMCR.EN field
STR r0, [r11]
DSB
ISB
Enabling the TCM retry and read-modify-write
If the TCM connected to the processor supports error detection and correction, the TCM interface should be configured to support the retry and read-modify-write features. These can be enabled at reset in the system by external signals on the processor. If they are disabled at reset then the following code example can be used to enable them in software:
ITCMCR EQU 0xE000EF90
DTCMCR EQU 0xE000EF94
LDR r11, =ITCMCR
LDR r0, [r11]
ORR r0, r0, #0x1:SHL:1 ; Set ITCMCR.RMW field
ORR r0, r0, #0x1:SHL:2 ; Set ITCMCR.RETEN field
STR r0, [r11]
LDR r11, =DTCMCR
LDR r0, [r11]
ORR r0, r0, #0x1:SHL:1 ; Set DTCMCR.RMW field
ORR r0, r0, #0x1:SHL:2 ; Set DTCMCR.RETEN field
STR r0, [r11]
DSB
ISB
PM0253 Rev 5 247/254
253
Cortex-M7 peripherals PM0253
The AHBPCR controls accesses to the device on the AHBP or AXI master interface. The bit assignments are:
Figure 61. AHBPCR bit assignments
31 4 3 2 1 0
Reserved SZ
EN
MSv39649V1
Table 106. AHBPCR bit assignments
Bits Name Type Function
[31:4] -
[3:1] SZ
[0] EN
-
RO
RW
Reserved, RAZ/WI.
AHBP size:
0b001: 64 MBytes.
0b010: 128 MBytes.
0b011: 256 MBytes.
0b100: 512 MBytes.
AHBP enable:
0: AHBP disabled. When disabled all accesses are made to the AXI master.
1: AHBP enabled.
Enabling the AHBP interface
The AHBP interface can be enabled at reset in the system by an external signal on the processor. If it is disabled at reset then the following code example can be used to enable the AHPB interface from software:
AHBPCR EQU 0xE000EF98
LDR r11, =AHBPCR
LDR r0, [r11]
ORR r0, r0, #0x1 ; Set AHBPCR.EN field
STR r0, [r11]
DSB
ISB
248/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
4.9.3 Auxiliary cache control register
The CACR controls the L1-cache ECC and attribute override. The bit assignments are:
Figure 62. CACR bit assignments
31 3 2 1 0
Reserved
FORCEWT
ECCEN
SIWT
MSv39655V1
Bits
[31:3] -
Name
[2]
[1]
[0]
-
Table 107. CACR bit assignments
Type Function
FORCEWT RW
ECCEN
SIWT
RW
RW
Reserved, RAZ/WI.
Enables Force Write-through in the data cache:
0: Disables Force Write-Through.
1: Enables Force Write-Through. Cacheable Write-Back memory regions are treated as Write-Through.
This bit is RAZ if the data cache is excluded.
Enables ECC in the instruction and data cache:
0: Enables ECC in the instruction and data cache.
1: Disables ECC in the instruction and data cache.
This bit is WI if both data cache and instruction cache are excluded or if ECC is not configured. If ECC is included in the processor the reset value of ECCEN is 0. If ECC is excluded the reset value of ECCEN is 1
Enables cache coherency usage:
0: Normal Cacheable Shared locations are treated as being Non-cacheable.
Programmed inner cacheability attributes are ignored. This is the default mode of operation for Shared memory. Caches are transparent to software for these locations and therefore no software maintenance is required to maintain coherency.
1: For the data cache, Normal Cacheable shared locations are treated as Writethrough. For the instruction cache, shared locations are treated as being Noncacheable. Programmed inner cacheability attributes are ignored. All writes are globally visible. Other memory agent updates are not visible to Cortex
®
-M7 software without suitable cache maintenance.
Useful for heterogeneous MP-systems where, for example, Cortex
®
-M7 processor is integrated on the Accelerator Coherency Port (ACP) interface on an MP-capable processor.
This bit is RAZ if the data cache is excluded.
PM0253 Rev 5 249/254
253
Cortex-M7 peripherals
4.9.4
PM0253
Disabling cache error checking and correction
If the cache error checking and correction is included in the processor it is enabled by default from reset. The following code example can be used to disable the feature. The operation is carried out by modifying the CACR.ECCEN field in the PPB memory region.
CACR EQU 0xE000EF9C
LDR r11, =CACR
LDR r0, [r11]
BFC r0, #0x1, #0x1 ; Clear CACR.ECCEN
STR r0, [r11]
DSB
ISB
Care must be taken when the software changes CACR.ECCEN. If CACR.ECCEN changes when the caches contain data, ECC information in the caches might not be correct for the new setting, resulting in unexpected errors and data loss. Therefore the software must only change CACR.ECCEN when both caches are turned off and both caches must be invalidated after the change.
AHB slave control register
The AHBSCR is used by software to control the priority of AHB slave traffic. The bit assignments are:
Figure 63. AHBSCR bit assignments
31
Reserved
15
INITCOUNT
11 10
TPRI
2 1 0
CTL
MSv39650V1
Bits
[31:16] -
Name
-
Table 108. AHBSCR bit assignments
Type Function
[15:11] INITCOUNT RW
Reserved.
Fairness counter initialization value. Use to demote access priority of the requestor selected by the AHBSCR.CTL field. The reset value is 0b01.
For round-robin mode set INITCOUNT to 0b01 and AHBSCR.CTL to 0b00 or
0b01.
INITCOUNT must not be set to 0b00 because the demoted requestor always takes priority when contention occurs, which can lead to livelock.
INITCOUNT is not used when AHBSCR.CTL is 0b11.
250/254 PM0253 Rev 5
PM0253 Cortex-M7 peripherals
Bits Name
[10:2] TPRI
[1:0] CTL
Table 108. AHBSCR bit assignments (continued)
Type Function
RW
RW
Threshold execution priority for AHBS traffic demotion.
0b0xxxxxxx: Priority is TPRI[7:0]. This is the same as the NVIC register encodings.
0b11111111: Priority of -1. This is the priority of the HardFault exception.
0b11111110: Priority of -2. This is the priority of the NMI exception.
AHBS prioritization control:
0b00: AHBS access priority demoted. This is the reset value.
0b01: Software access priority demoted.
0b10: AHBS access priority demoted by initializing the fairness counter to the
AHBSCR.INITCOUNT value when the software execution priority is higher than or equal to the threshold level programmed in AHBSCR.TPRI. When the software execution priority is below this value, the fairness counter is initialized with 1 (round-robin).
The threshold level encoding matches the NVIC encoding and uses arithmetically larger numbers to represent lower priority.
0b11: AHBSPRI signal has control of access priority.
4.9.5 Auxiliary bus fault status register
The ABFSR stores information on the source of asynchronous bus faults. The ASBFSR bit assignments are:
Figure 64. ABFSR bit assignments
31
Reserved
10 9 8 7
Reserved
4 3 2 1 0
AXIMTYPE EPPB
AXIM
AHBP
DTCM
ITCM
MSv39647V1
Bits
[31:10]
[9:8]
[7:5]
[4]
-
-
EPPB
Name
AXIMTYPE
Table 109. ABFSR bit assignments
Function
Reserved
Indicates the type of fault on the AXIM interface: b00: OKAY b01: EXOKAY b10: SLVERR b11: DECERR
Only valid when AXIM is 1.
Reserved
Asynchronous fault on EPPB interface.
PM0253 Rev 5 251/254
253
Cortex-M7 peripherals PM0253
[3]
[2]
[1]
[0]
Bits
AXIM
AHBP
DTCM
ITCM
Table 109. ABFSR bit assignments (continued)
Name Function
Asynchronous fault on AXIM interface.
Asynchronous fault on AHBP interface.
Asynchronous fault on DTCM interface.
Asynchronous fault on ITCM interface
In the bus-fault handler, the software reads the BFSR, and if an asynchronous fault occurs, the ABFSR is read to determine which interfaces are affected. The ABFSR[4:0] fields remains valid until cleared by writing to the ABFSR with any value.
For more information about the BFSR, see
BusFault status register on page 207 .
252/254 PM0253 Rev 5
PM0253 Revision history
Date
18-Dec-2015
22-Apr-2016
02-Feb-2017
14-Nov-2017
15-Jun-2019
Table 110. Document revision history
Revision Changes
1
2
3
4
5
Initial release.
Updated
Table 11: STM32F746xx/STM32F756xx Cortex ® -M7 configuration
title.
Added
Table 12: STM32F76xxx/STM32F77xxx Cortex ® -M7 configuration
.
In Section 2.3.2: Memory system ordering of memory accesses
updated link to section 2.3.4: software ordering of memory accesses .
Added
Table 13: STM32F72xxx/STM32F73xxx Cortex
.
Added STM32H7 Series in the whole document.
Added
Table 14: STM32H7 Series Cortex
Updated DREGION function description in
Table 85: TYPE bit assignments .
Added
Section 4.7.7: Enabling and clearing FPU exception interrupts
.
PM0253 Rev 5 253/254
253
PM0253
IMPORTANT NOTICE – PLEASE READ CAREFULLY
STMicroelectronics NV and its subsidiaries (“ST”) reserve the right to make changes, corrections, enhancements, modifications, and improvements to ST products and/or to this document at any time without notice. Purchasers should obtain the latest relevant information on
ST products before placing orders. ST products are sold pursuant to ST’s terms and conditions of sale in place at the time of order acknowledgement.
Purchasers are solely responsible for the choice, selection, and use of ST products and ST assumes no liability for application assistance or the design of Purchasers’ products.
No license, express or implied, to any intellectual property right is granted by ST herein.
Resale of ST products with provisions different from the information set forth herein shall void any warranty granted by ST for such product.
ST and the ST logo are trademarks of ST. For additional information about ST trademarks, please refer to www.st.com/trademarks . All other product or service names are the property of their respective owners.
Information in this document supersedes and replaces information previously supplied in any prior versions of this document.
© 2019 STMicroelectronics – All rights reserved
254/254 PM0253 Rev 5
Preuzimanje
Javna poveznica ažurirana
Javna poveznica na vaš razgovor je ažurirana. You may view and manage your chat links after registration here.
Javna poveznica dokumenta
Oglas
