* [PATCH 0/7] MIPS: Add asm macros for unsupported instructions @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle From: James Hogan <jhogan@kernel.org> This patch series implements Ralf's idea[1] for better supporting older assembler versions which don't implement newer instructions, in inline assembly. It has a number of benefits which are described in the individual patches, but its also particularly beneficial for Marcin's CRC patchset. So far I've converted the VZ, XPA, and MSA abstractions to use it. Old style abstractions remain for MT and DSP instructions, which are left for future patches. There are also other abstraction macros in <asm/asmmacro.h> for use in .S files (particularly for MSA), which similarly don't handle $n register naming and use the $at move trick. Cleaning these up to use something like parse_r is also left for future patches. [1] https://www.linux-mips.org/archives/linux-mips/2017-10/msg00043.html Patch overview: Patch 1 adds some helpers to streamline the construction of assembler macros for instructions unimplemented in the current assembler. It is needed for the other patches. Patches 2-3 update the VZ helpers to make use of the new assembler macro helpers, and make the VZ helpers consistent with the normal c0 register helpers now that it is possible to do so. Patches 4-6 update the XPA helpers to make use of the new assembler macro helpers, and to use the instructions provided by the assembler if they are available. These are also changed to be consistent with the normal register helpers now that it is possible to do so. Patch 7 updates the MSA control register helpers to make use of the new assembler macro helpers. Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Marcin Nowakowski <marcin.nowakowski@mips.com> Cc: linux-mips@linux-mips.org James Hogan (7): MIPS: Add helpers for assembler macro instructions MIPS: VZ: Update helpers to use new asm macros MIPS: VZ: Pass GC0 register names in $n format MIPS: XPA: Use XPA instructions in assembly MIPS: XPA: Allow use of $0 (zero) to MTHC0 MIPS: XPA: Standardise readx/writex accessors MIPS: MSA: Update helpers to use new asm macros arch/mips/Makefile | 6 +- arch/mips/include/asm/mipsregs.h | 663 +++++++++++++++----------------- arch/mips/include/asm/msa.h | 63 +--- 3 files changed, 356 insertions(+), 376 deletions(-) base-commit: e0c5f36b2a638fc3298200c385af7f196d3b5cd4 -- git-series 0.9.1 ^ permalink raw reply [flat|nested] 17+ messages in thread
* [PATCH 0/7] MIPS: Add asm macros for unsupported instructions @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle From: James Hogan <jhogan@kernel.org> This patch series implements Ralf's idea[1] for better supporting older assembler versions which don't implement newer instructions, in inline assembly. It has a number of benefits which are described in the individual patches, but its also particularly beneficial for Marcin's CRC patchset. So far I've converted the VZ, XPA, and MSA abstractions to use it. Old style abstractions remain for MT and DSP instructions, which are left for future patches. There are also other abstraction macros in <asm/asmmacro.h> for use in .S files (particularly for MSA), which similarly don't handle $n register naming and use the $at move trick. Cleaning these up to use something like parse_r is also left for future patches. [1] https://www.linux-mips.org/archives/linux-mips/2017-10/msg00043.html Patch overview: Patch 1 adds some helpers to streamline the construction of assembler macros for instructions unimplemented in the current assembler. It is needed for the other patches. Patches 2-3 update the VZ helpers to make use of the new assembler macro helpers, and make the VZ helpers consistent with the normal c0 register helpers now that it is possible to do so. Patches 4-6 update the XPA helpers to make use of the new assembler macro helpers, and to use the instructions provided by the assembler if they are available. These are also changed to be consistent with the normal register helpers now that it is possible to do so. Patch 7 updates the MSA control register helpers to make use of the new assembler macro helpers. Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Marcin Nowakowski <marcin.nowakowski@mips.com> Cc: linux-mips@linux-mips.org James Hogan (7): MIPS: Add helpers for assembler macro instructions MIPS: VZ: Update helpers to use new asm macros MIPS: VZ: Pass GC0 register names in $n format MIPS: XPA: Use XPA instructions in assembly MIPS: XPA: Allow use of $0 (zero) to MTHC0 MIPS: XPA: Standardise readx/writex accessors MIPS: MSA: Update helpers to use new asm macros arch/mips/Makefile | 6 +- arch/mips/include/asm/mipsregs.h | 663 +++++++++++++++----------------- arch/mips/include/asm/msa.h | 63 +--- 3 files changed, 356 insertions(+), 376 deletions(-) base-commit: e0c5f36b2a638fc3298200c385af7f196d3b5cd4 -- git-series 0.9.1 ^ permalink raw reply [flat|nested] 17+ messages in thread
* [PATCH 1/7] MIPS: Add helpers for assembler macro instructions @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle From: James Hogan <jhogan@kernel.org> Implement a parse_r assembler macro in asm/mipsregs.h to parse a register in $n form, and a few C macros for defining assembler macro instructions. These can be used to more transparently support older binutils versions which don't support for example the msa, virt, xpa, or crc instructions. In particular they overcome the difficulty of turning a register name in $n form into an instruction encoding suitable for giving to .word / .hword, which is particularly problematic when needed from inline assembly where the compiler is responsible for register allocation. Traditionally this had required the use of $at and an extra MOV instruction, but for CRC instructions with multiple GP register operands that approach becomes more difficult. Three assembler macro creation helpers are added: - _ASM_MACRO_0(OP, ENC) This is to define an assembler macro for an instruction which has no operands, for example the VZ TLBGR instruction. - _ASM_MACRO_2R(OP, R1, R2, ENC) This is to define an assembler macro for an instruction which has 2 register operands, for example the CFCMSA instruction. - _ASM_MACRO_3R(OP, R1, R2, R3, ENC) This is to define an assembler macro for an instruction which has 3 register operands, for example the crc32 instructions. - _ASM_MACRO_2R_1S(OP, R1, R2, SEL3, ENC) This is to define an assembler macro for a Cop0 move instruction, with 2 register operands and an optional register select operand which defaults to 0, for example the VZ MFGC0 instruction. Suggested-by: Ralf Baechle <ralf@linux-mips.org> Signed-off-by: James Hogan <jhogan@kernel.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Marcin Nowakowski <marcin.nowakowski@mips.com> Cc: linux-mips@linux-mips.org --- arch/mips/include/asm/mipsregs.h | 83 +++++++++++++++++++++++++++++++++- 1 file changed, 83 insertions(+) diff --git a/arch/mips/include/asm/mipsregs.h b/arch/mips/include/asm/mipsregs.h index 6b1f1ad0542c..972698557b4b 100644 --- a/arch/mips/include/asm/mipsregs.h +++ b/arch/mips/include/asm/mipsregs.h @@ -1181,6 +1181,89 @@ static inline int mm_insn_16bit(u16 insn) #endif /* + * parse_r var, r - Helper assembler macro for parsing register names. + * + * This converts the register name in $n form provided in \r to the + * corresponding register number, which is assigned to the variable \var. It is + * needed to allow explicit encoding of instructions in inline assembly where + * registers are chosen by the compiler in $n form, allowing us to avoid using + * fixed register numbers. + * + * It also allows newer instructions (not implemented by the assembler) to be + * transparently implemented using assembler macros, instead of needing separate + * cases depending on toolchain support. + * + * Simple usage example: + * __asm__ __volatile__("parse_r __rt, %0\n\t" + * ".insn\n\t" + * "# di %0\n\t" + * ".word (0x41606000 | (__rt << 16))" + * : "=r" (status); + */ + +/* Match an individual register number and assign to \var */ +#define _IFC_REG(n) \ + ".ifc \\r, $" #n "\n\t" \ + "\\var = " #n "\n\t" \ + ".endif\n\t" + +__asm__(".macro parse_r var r\n\t" + "\\var = -1\n\t" + _IFC_REG(0) _IFC_REG(1) _IFC_REG(2) _IFC_REG(3) + _IFC_REG(4) _IFC_REG(5) _IFC_REG(6) _IFC_REG(7) + _IFC_REG(8) _IFC_REG(9) _IFC_REG(10) _IFC_REG(11) + _IFC_REG(12) _IFC_REG(13) _IFC_REG(14) _IFC_REG(15) + _IFC_REG(16) _IFC_REG(17) _IFC_REG(18) _IFC_REG(19) + _IFC_REG(20) _IFC_REG(21) _IFC_REG(22) _IFC_REG(23) + _IFC_REG(24) _IFC_REG(25) _IFC_REG(26) _IFC_REG(27) + _IFC_REG(28) _IFC_REG(29) _IFC_REG(30) _IFC_REG(31) + ".iflt \\var\n\t" + ".error \"Unable to parse register name \\r\"\n\t" + ".endif\n\t" + ".endm"); + +#undef _IFC_REG + +/* + * C macros for generating assembler macros for common instruction formats. + * + * The names of the operands can be chosen by the caller, and the encoding of + * register operand \<Rn> is assigned to __<Rn> where it can be accessed from + * the ENC encodings. + */ + +/* Instructions with no operands */ +#define _ASM_MACRO_0(OP, ENC) \ + __asm__(".macro " #OP "\n\t" \ + ENC \ + ".endm") + +/* Instructions with 2 register operands */ +#define _ASM_MACRO_2R(OP, R1, R2, ENC) \ + __asm__(".macro " #OP " " #R1 ", " #R2 "\n\t" \ + "parse_r __" #R1 ", \\" #R1 "\n\t" \ + "parse_r __" #R2 ", \\" #R2 "\n\t" \ + ENC \ + ".endm") + +/* Instructions with 3 register operands */ +#define _ASM_MACRO_3R(OP, R1, R2, R3, ENC) \ + __asm__(".macro " #OP " " #R1 ", " #R2 ", " #R3 "\n\t" \ + "parse_r __" #R1 ", \\" #R1 "\n\t" \ + "parse_r __" #R2 ", \\" #R2 "\n\t" \ + "parse_r __" #R3 ", \\" #R3 "\n\t" \ + ENC \ + ".endm") + +/* Instructions with 2 register operands and 1 optional select operand */ +#define _ASM_MACRO_2R_1S(OP, R1, R2, SEL3, ENC) \ + __asm__(".macro " #OP " " #R1 ", " #R2 ", " #SEL3 " = 0\n\t" \ + "parse_r __" #R1 ", \\" #R1 "\n\t" \ + "parse_r __" #R2 ", \\" #R2 "\n\t" \ + ENC \ + ".endm") + +/* * TLB Invalidate Flush */ static inline void tlbinvf(void) -- git-series 0.9.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 1/7] MIPS: Add helpers for assembler macro instructions @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle From: James Hogan <jhogan@kernel.org> Implement a parse_r assembler macro in asm/mipsregs.h to parse a register in $n form, and a few C macros for defining assembler macro instructions. These can be used to more transparently support older binutils versions which don't support for example the msa, virt, xpa, or crc instructions. In particular they overcome the difficulty of turning a register name in $n form into an instruction encoding suitable for giving to .word / .hword, which is particularly problematic when needed from inline assembly where the compiler is responsible for register allocation. Traditionally this had required the use of $at and an extra MOV instruction, but for CRC instructions with multiple GP register operands that approach becomes more difficult. Three assembler macro creation helpers are added: - _ASM_MACRO_0(OP, ENC) This is to define an assembler macro for an instruction which has no operands, for example the VZ TLBGR instruction. - _ASM_MACRO_2R(OP, R1, R2, ENC) This is to define an assembler macro for an instruction which has 2 register operands, for example the CFCMSA instruction. - _ASM_MACRO_3R(OP, R1, R2, R3, ENC) This is to define an assembler macro for an instruction which has 3 register operands, for example the crc32 instructions. - _ASM_MACRO_2R_1S(OP, R1, R2, SEL3, ENC) This is to define an assembler macro for a Cop0 move instruction, with 2 register operands and an optional register select operand which defaults to 0, for example the VZ MFGC0 instruction. Suggested-by: Ralf Baechle <ralf@linux-mips.org> Signed-off-by: James Hogan <jhogan@kernel.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Marcin Nowakowski <marcin.nowakowski@mips.com> Cc: linux-mips@linux-mips.org --- arch/mips/include/asm/mipsregs.h | 83 +++++++++++++++++++++++++++++++++- 1 file changed, 83 insertions(+) diff --git a/arch/mips/include/asm/mipsregs.h b/arch/mips/include/asm/mipsregs.h index 6b1f1ad0542c..972698557b4b 100644 --- a/arch/mips/include/asm/mipsregs.h +++ b/arch/mips/include/asm/mipsregs.h @@ -1181,6 +1181,89 @@ static inline int mm_insn_16bit(u16 insn) #endif /* + * parse_r var, r - Helper assembler macro for parsing register names. + * + * This converts the register name in $n form provided in \r to the + * corresponding register number, which is assigned to the variable \var. It is + * needed to allow explicit encoding of instructions in inline assembly where + * registers are chosen by the compiler in $n form, allowing us to avoid using + * fixed register numbers. + * + * It also allows newer instructions (not implemented by the assembler) to be + * transparently implemented using assembler macros, instead of needing separate + * cases depending on toolchain support. + * + * Simple usage example: + * __asm__ __volatile__("parse_r __rt, %0\n\t" + * ".insn\n\t" + * "# di %0\n\t" + * ".word (0x41606000 | (__rt << 16))" + * : "=r" (status); + */ + +/* Match an individual register number and assign to \var */ +#define _IFC_REG(n) \ + ".ifc \\r, $" #n "\n\t" \ + "\\var = " #n "\n\t" \ + ".endif\n\t" + +__asm__(".macro parse_r var r\n\t" + "\\var = -1\n\t" + _IFC_REG(0) _IFC_REG(1) _IFC_REG(2) _IFC_REG(3) + _IFC_REG(4) _IFC_REG(5) _IFC_REG(6) _IFC_REG(7) + _IFC_REG(8) _IFC_REG(9) _IFC_REG(10) _IFC_REG(11) + _IFC_REG(12) _IFC_REG(13) _IFC_REG(14) _IFC_REG(15) + _IFC_REG(16) _IFC_REG(17) _IFC_REG(18) _IFC_REG(19) + _IFC_REG(20) _IFC_REG(21) _IFC_REG(22) _IFC_REG(23) + _IFC_REG(24) _IFC_REG(25) _IFC_REG(26) _IFC_REG(27) + _IFC_REG(28) _IFC_REG(29) _IFC_REG(30) _IFC_REG(31) + ".iflt \\var\n\t" + ".error \"Unable to parse register name \\r\"\n\t" + ".endif\n\t" + ".endm"); + +#undef _IFC_REG + +/* + * C macros for generating assembler macros for common instruction formats. + * + * The names of the operands can be chosen by the caller, and the encoding of + * register operand \<Rn> is assigned to __<Rn> where it can be accessed from + * the ENC encodings. + */ + +/* Instructions with no operands */ +#define _ASM_MACRO_0(OP, ENC) \ + __asm__(".macro " #OP "\n\t" \ + ENC \ + ".endm") + +/* Instructions with 2 register operands */ +#define _ASM_MACRO_2R(OP, R1, R2, ENC) \ + __asm__(".macro " #OP " " #R1 ", " #R2 "\n\t" \ + "parse_r __" #R1 ", \\" #R1 "\n\t" \ + "parse_r __" #R2 ", \\" #R2 "\n\t" \ + ENC \ + ".endm") + +/* Instructions with 3 register operands */ +#define _ASM_MACRO_3R(OP, R1, R2, R3, ENC) \ + __asm__(".macro " #OP " " #R1 ", " #R2 ", " #R3 "\n\t" \ + "parse_r __" #R1 ", \\" #R1 "\n\t" \ + "parse_r __" #R2 ", \\" #R2 "\n\t" \ + "parse_r __" #R3 ", \\" #R3 "\n\t" \ + ENC \ + ".endm") + +/* Instructions with 2 register operands and 1 optional select operand */ +#define _ASM_MACRO_2R_1S(OP, R1, R2, SEL3, ENC) \ + __asm__(".macro " #OP " " #R1 ", " #R2 ", " #SEL3 " = 0\n\t" \ + "parse_r __" #R1 ", \\" #R1 "\n\t" \ + "parse_r __" #R2 ", \\" #R2 "\n\t" \ + ENC \ + ".endm") + +/* * TLB Invalidate Flush */ static inline void tlbinvf(void) -- git-series 0.9.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* Re: [PATCH 1/7] MIPS: Add helpers for assembler macro instructions 2017-11-22 11:30 ` James Hogan (?) @ 2018-05-24 19:47 ` Hauke Mehrtens -1 siblings, 0 replies; 17+ messages in thread From: Hauke Mehrtens @ 2018-05-24 19:47 UTC (permalink / raw) To: James Hogan, linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle Hi James, On 11/22/2017 12:30 PM, James Hogan wrote: > From: James Hogan <jhogan@kernel.org> > > Implement a parse_r assembler macro in asm/mipsregs.h to parse a > register in $n form, and a few C macros for defining assembler macro > instructions. These can be used to more transparently support older > binutils versions which don't support for example the msa, virt, xpa, or > crc instructions. > > In particular they overcome the difficulty of turning a register name in > $n form into an instruction encoding suitable for giving to .word / > .hword, which is particularly problematic when needed from inline > assembly where the compiler is responsible for register allocation. > Traditionally this had required the use of $at and an extra MOV > instruction, but for CRC instructions with multiple GP register operands > that approach becomes more difficult. > > Three assembler macro creation helpers are added: > > - _ASM_MACRO_0(OP, ENC) > This is to define an assembler macro for an instruction which has no > operands, for example the VZ TLBGR instruction. > > - _ASM_MACRO_2R(OP, R1, R2, ENC) > This is to define an assembler macro for an instruction which has 2 > register operands, for example the CFCMSA instruction. > > - _ASM_MACRO_3R(OP, R1, R2, R3, ENC) > This is to define an assembler macro for an instruction which has 3 > register operands, for example the crc32 instructions. > > - _ASM_MACRO_2R_1S(OP, R1, R2, SEL3, ENC) > This is to define an assembler macro for a Cop0 move instruction, > with 2 register operands and an optional register select operand > which defaults to 0, for example the VZ MFGC0 instruction. > > Suggested-by: Ralf Baechle <ralf@linux-mips.org> > Signed-off-by: James Hogan <jhogan@kernel.org> > Cc: Ralf Baechle <ralf@linux-mips.org> > Cc: Marcin Nowakowski <marcin.nowakowski@mips.com> > Cc: linux-mips@linux-mips.org > --- > arch/mips/include/asm/mipsregs.h | 83 +++++++++++++++++++++++++++++++++- > 1 file changed, 83 insertions(+) > > diff --git a/arch/mips/include/asm/mipsregs.h b/arch/mips/include/asm/mipsregs.h > index 6b1f1ad0542c..972698557b4b 100644 > --- a/arch/mips/include/asm/mipsregs.h > +++ b/arch/mips/include/asm/mipsregs.h > @@ -1181,6 +1181,89 @@ static inline int mm_insn_16bit(u16 insn) > #endif > > /* > + * parse_r var, r - Helper assembler macro for parsing register names. > + * > + * This converts the register name in $n form provided in \r to the > + * corresponding register number, which is assigned to the variable \var. It is > + * needed to allow explicit encoding of instructions in inline assembly where > + * registers are chosen by the compiler in $n form, allowing us to avoid using > + * fixed register numbers. > + * > + * It also allows newer instructions (not implemented by the assembler) to be > + * transparently implemented using assembler macros, instead of needing separate > + * cases depending on toolchain support. > + * > + * Simple usage example: > + * __asm__ __volatile__("parse_r __rt, %0\n\t" > + * ".insn\n\t" > + * "# di %0\n\t" > + * ".word (0x41606000 | (__rt << 16))" > + * : "=r" (status); > + */ > + > +/* Match an individual register number and assign to \var */ > +#define _IFC_REG(n) \ > + ".ifc \\r, $" #n "\n\t" \ > + "\\var = " #n "\n\t" \ > + ".endif\n\t" > + > +__asm__(".macro parse_r var r\n\t" > + "\\var = -1\n\t" > + _IFC_REG(0) _IFC_REG(1) _IFC_REG(2) _IFC_REG(3) > + _IFC_REG(4) _IFC_REG(5) _IFC_REG(6) _IFC_REG(7) > + _IFC_REG(8) _IFC_REG(9) _IFC_REG(10) _IFC_REG(11) > + _IFC_REG(12) _IFC_REG(13) _IFC_REG(14) _IFC_REG(15) > + _IFC_REG(16) _IFC_REG(17) _IFC_REG(18) _IFC_REG(19) > + _IFC_REG(20) _IFC_REG(21) _IFC_REG(22) _IFC_REG(23) > + _IFC_REG(24) _IFC_REG(25) _IFC_REG(26) _IFC_REG(27) > + _IFC_REG(28) _IFC_REG(29) _IFC_REG(30) _IFC_REG(31) > + ".iflt \\var\n\t" > + ".error \"Unable to parse register name \\r\"\n\t" > + ".endif\n\t" > + ".endm"); This parse_r __asm__ macro causes problems when compiling the kernel with Link time optimization (LTO) and I do not know how to fix this. I am getting many errors complaining about parser_r being redefined while linking the kernel, this looks like this: ./ccPboPK5.s:96693: Error: Macro `parse_r' was already defined When using LTO the compiler puts all the objects together in the link time and the parse_r macro is defined in the global scope of all files which include this header file. All these definitions are then conflicting. An option to solve this is to put this into a function in the header file and other would be to move this into a source file. Functions which are defined multiple times are probably getting removed automatically. Isn't this an extended ASM statement? The GCC documentation says this: Note that extended asm statements must be inside a function. Only basic asm may be outside functions (see Basic Asm). Functions declared with the naked attribute also require basic asm (see Function Attributes). https://gcc.gnu.org/onlinedocs/gcc/Extended-Asm.html I am not very familiar with asm code included in c code. I plan to debug the LTO kernel further and will probably send some initial patches which are fixing some problems soon, but not the general LTO build support. Hauke ^ permalink raw reply [flat|nested] 17+ messages in thread
* [PATCH 2/7] MIPS: VZ: Update helpers to use new asm macros @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle From: James Hogan <jhogan@kernel.org> Update VZ guest register & guest TLB access helpers to use the new assembly macros for parsing register names and creating custom assembly macro instructions, which has a number of advantages: - Better code can be generated on toolchains which don't support VZ, more closely matching those which do, since there is no need to bounce values via the $at register. Some differences still remain due to the inability to safely fill branch delay slots and R6 compact branch forbidden slots with explicitly encoded instructions, resulting in some extra NOPs added by the assembler. - Some code duplication between toolchains which do and don't support VZ instructions is removed, since the helpers are only implemented once. When the toolchain doesn't implement the instruction an assembly macro implements it instead. - Instruction encodings are kept together in the source. On a generic kernel with KVM VZ support enabled this change saves about 2.5KiB of kernel code when TOOLCHAIN_SUPPORTS_VIRT=n, bringing it down to about 0.5KiB more than when TOOLCHAIN_SUPPORTS_VIRT=y on r6, and just 68 bytes more on r2. Signed-off-by: James Hogan <jhogan@kernel.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: linux-mips@linux-mips.org --- arch/mips/include/asm/mipsregs.h | 164 +++++++------------------------- 1 file changed, 37 insertions(+), 127 deletions(-) diff --git a/arch/mips/include/asm/mipsregs.h b/arch/mips/include/asm/mipsregs.h index 972698557b4b..361c35243366 100644 --- a/arch/mips/include/asm/mipsregs.h +++ b/arch/mips/include/asm/mipsregs.h @@ -1913,14 +1913,40 @@ do { \ * Macros to access the guest system control coprocessor */ -#ifdef TOOLCHAIN_SUPPORTS_VIRT +#ifndef TOOLCHAIN_SUPPORTS_VIRT +_ASM_MACRO_2R_1S(mfgc0, rt, rs, sel, + _ASM_INSN_IF_MIPS(0x40600000 | __rt << 16 | __rs << 11 | \\sel) + _ASM_INSN32_IF_MM(0x000004fc | __rt << 21 | __rs << 16 | \\sel << 11)); +_ASM_MACRO_2R_1S(dmfgc0, rt, rs, sel, + _ASM_INSN_IF_MIPS(0x40600100 | __rt << 16 | __rs << 11 | \\sel) + _ASM_INSN32_IF_MM(0x580004fc | __rt << 21 | __rs << 16 | \\sel << 11)); +_ASM_MACRO_2R_1S(mtgc0, rt, rd, sel, + _ASM_INSN_IF_MIPS(0x40600200 | __rt << 16 | __rd << 11 | \\sel) + _ASM_INSN32_IF_MM(0x000006fc | __rt << 21 | __rd << 16 | \\sel << 11)); +_ASM_MACRO_2R_1S(dmtgc0, rt, rd, sel, + _ASM_INSN_IF_MIPS(0x40600300 | __rt << 16 | __rd << 11 | \\sel) + _ASM_INSN32_IF_MM(0x580006fc | __rt << 21 | __rd << 16 | \\sel << 11)); +_ASM_MACRO_0(tlbgp, _ASM_INSN_IF_MIPS(0x42000010) + _ASM_INSN32_IF_MM(0x0000017c)); +_ASM_MACRO_0(tlbgr, _ASM_INSN_IF_MIPS(0x42000009) + _ASM_INSN32_IF_MM(0x0000117c)); +_ASM_MACRO_0(tlbgwi, _ASM_INSN_IF_MIPS(0x4200000a) + _ASM_INSN32_IF_MM(0x0000217c)); +_ASM_MACRO_0(tlbgwr, _ASM_INSN_IF_MIPS(0x4200000e) + _ASM_INSN32_IF_MM(0x0000317c)); +_ASM_MACRO_0(tlbginvf, _ASM_INSN_IF_MIPS(0x4200000c) + _ASM_INSN32_IF_MM(0x0000517c)); +#define _ASM_SET_VIRT "" +#else /* !TOOLCHAIN_SUPPORTS_VIRT */ +#define _ASM_SET_VIRT ".set\tvirt\n\t" +#endif #define __read_32bit_gc0_register(source, sel) \ ({ int __res; \ __asm__ __volatile__( \ ".set\tpush\n\t" \ ".set\tmips32r2\n\t" \ - ".set\tvirt\n\t" \ + _ASM_SET_VIRT \ "mfgc0\t%0, $%1, %2\n\t" \ ".set\tpop" \ : "=r" (__res) \ @@ -1933,8 +1959,8 @@ do { \ __asm__ __volatile__( \ ".set\tpush\n\t" \ ".set\tmips64r2\n\t" \ - ".set\tvirt\n\t" \ - "dmfgc0\t%0, $%1, %2\n\t" \ + _ASM_SET_VIRT \ + "dmfgc0\t%0, $%1, %2\n\t" \ ".set\tpop" \ : "=r" (__res) \ : "i" (source), "i" (sel)); \ @@ -1946,7 +1972,7 @@ do { \ __asm__ __volatile__( \ ".set\tpush\n\t" \ ".set\tmips32r2\n\t" \ - ".set\tvirt\n\t" \ + _ASM_SET_VIRT \ "mtgc0\t%z0, $%1, %2\n\t" \ ".set\tpop" \ : : "Jr" ((unsigned int)(value)), \ @@ -1958,75 +1984,13 @@ do { \ __asm__ __volatile__( \ ".set\tpush\n\t" \ ".set\tmips64r2\n\t" \ - ".set\tvirt\n\t" \ + _ASM_SET_VIRT \ "dmtgc0\t%z0, $%1, %2\n\t" \ ".set\tpop" \ : : "Jr" (value), \ "i" (register), "i" (sel)); \ } while (0) -#else /* TOOLCHAIN_SUPPORTS_VIRT */ - -#define __read_32bit_gc0_register(source, sel) \ -({ int __res; \ - __asm__ __volatile__( \ - ".set\tpush\n\t" \ - ".set\tnoat\n\t" \ - "# mfgc0\t$1, $%1, %2\n\t" \ - _ASM_INSN_IF_MIPS(0x40610000 | %1 << 11 | %2) \ - _ASM_INSN32_IF_MM(0x002004fc | %1 << 16 | %2 << 11) \ - "move\t%0, $1\n\t" \ - ".set\tpop" \ - : "=r" (__res) \ - : "i" (source), "i" (sel)); \ - __res; \ -}) - -#define __read_64bit_gc0_register(source, sel) \ -({ unsigned long long __res; \ - __asm__ __volatile__( \ - ".set\tpush\n\t" \ - ".set\tnoat\n\t" \ - "# dmfgc0\t$1, $%1, %2\n\t" \ - _ASM_INSN_IF_MIPS(0x40610100 | %1 << 11 | %2) \ - _ASM_INSN32_IF_MM(0x582004fc | %1 << 16 | %2 << 11) \ - "move\t%0, $1\n\t" \ - ".set\tpop" \ - : "=r" (__res) \ - : "i" (source), "i" (sel)); \ - __res; \ -}) - -#define __write_32bit_gc0_register(register, sel, value) \ -do { \ - __asm__ __volatile__( \ - ".set\tpush\n\t" \ - ".set\tnoat\n\t" \ - "move\t$1, %z0\n\t" \ - "# mtgc0\t$1, $%1, %2\n\t" \ - _ASM_INSN_IF_MIPS(0x40610200 | %1 << 11 | %2) \ - _ASM_INSN32_IF_MM(0x002006fc | %1 << 16 | %2 << 11) \ - ".set\tpop" \ - : : "Jr" ((unsigned int)(value)), \ - "i" (register), "i" (sel)); \ -} while (0) - -#define __write_64bit_gc0_register(register, sel, value) \ -do { \ - __asm__ __volatile__( \ - ".set\tpush\n\t" \ - ".set\tnoat\n\t" \ - "move\t$1, %z0\n\t" \ - "# dmtgc0\t$1, $%1, %2\n\t" \ - _ASM_INSN_IF_MIPS(0x40610300 | %1 << 11 | %2) \ - _ASM_INSN32_IF_MM(0x582006fc | %1 << 16 | %2 << 11) \ - ".set\tpop" \ - : : "Jr" (value), \ - "i" (register), "i" (sel)); \ -} while (0) - -#endif /* !TOOLCHAIN_SUPPORTS_VIRT */ - #define __read_ulong_gc0_register(reg, sel) \ ((sizeof(unsigned long) == 4) ? \ (unsigned long) __read_32bit_gc0_register(reg, sel) : \ @@ -2664,8 +2628,6 @@ static inline void tlb_write_random(void) ".set reorder"); } -#ifdef TOOLCHAIN_SUPPORTS_VIRT - /* * Guest TLB operations. * @@ -2676,7 +2638,7 @@ static inline void guest_tlb_probe(void) __asm__ __volatile__( ".set push\n\t" ".set noreorder\n\t" - ".set virt\n\t" + _ASM_SET_VIRT "tlbgp\n\t" ".set pop"); } @@ -2686,7 +2648,7 @@ static inline void guest_tlb_read(void) __asm__ __volatile__( ".set push\n\t" ".set noreorder\n\t" - ".set virt\n\t" + _ASM_SET_VIRT "tlbgr\n\t" ".set pop"); } @@ -2696,7 +2658,7 @@ static inline void guest_tlb_write_indexed(void) __asm__ __volatile__( ".set push\n\t" ".set noreorder\n\t" - ".set virt\n\t" + _ASM_SET_VIRT "tlbgwi\n\t" ".set pop"); } @@ -2706,7 +2668,7 @@ static inline void guest_tlb_write_random(void) __asm__ __volatile__( ".set push\n\t" ".set noreorder\n\t" - ".set virt\n\t" + _ASM_SET_VIRT "tlbgwr\n\t" ".set pop"); } @@ -2719,63 +2681,11 @@ static inline void guest_tlbinvf(void) __asm__ __volatile__( ".set push\n\t" ".set noreorder\n\t" - ".set virt\n\t" + _ASM_SET_VIRT "tlbginvf\n\t" ".set pop"); } -#else /* TOOLCHAIN_SUPPORTS_VIRT */ - -/* - * Guest TLB operations. - * - * It is responsibility of the caller to take care of any TLB hazards. - */ -static inline void guest_tlb_probe(void) -{ - __asm__ __volatile__( - "# tlbgp\n\t" - _ASM_INSN_IF_MIPS(0x42000010) - _ASM_INSN32_IF_MM(0x0000017c)); -} - -static inline void guest_tlb_read(void) -{ - __asm__ __volatile__( - "# tlbgr\n\t" - _ASM_INSN_IF_MIPS(0x42000009) - _ASM_INSN32_IF_MM(0x0000117c)); -} - -static inline void guest_tlb_write_indexed(void) -{ - __asm__ __volatile__( - "# tlbgwi\n\t" - _ASM_INSN_IF_MIPS(0x4200000a) - _ASM_INSN32_IF_MM(0x0000217c)); -} - -static inline void guest_tlb_write_random(void) -{ - __asm__ __volatile__( - "# tlbgwr\n\t" - _ASM_INSN_IF_MIPS(0x4200000e) - _ASM_INSN32_IF_MM(0x0000317c)); -} - -/* - * Guest TLB Invalidate Flush - */ -static inline void guest_tlbinvf(void) -{ - __asm__ __volatile__( - "# tlbginvf\n\t" - _ASM_INSN_IF_MIPS(0x4200000c) - _ASM_INSN32_IF_MM(0x0000517c)); -} - -#endif /* !TOOLCHAIN_SUPPORTS_VIRT */ - /* * Manipulate bits in a register. */ -- git-series 0.9.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 2/7] MIPS: VZ: Update helpers to use new asm macros @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle From: James Hogan <jhogan@kernel.org> Update VZ guest register & guest TLB access helpers to use the new assembly macros for parsing register names and creating custom assembly macro instructions, which has a number of advantages: - Better code can be generated on toolchains which don't support VZ, more closely matching those which do, since there is no need to bounce values via the $at register. Some differences still remain due to the inability to safely fill branch delay slots and R6 compact branch forbidden slots with explicitly encoded instructions, resulting in some extra NOPs added by the assembler. - Some code duplication between toolchains which do and don't support VZ instructions is removed, since the helpers are only implemented once. When the toolchain doesn't implement the instruction an assembly macro implements it instead. - Instruction encodings are kept together in the source. On a generic kernel with KVM VZ support enabled this change saves about 2.5KiB of kernel code when TOOLCHAIN_SUPPORTS_VIRT=n, bringing it down to about 0.5KiB more than when TOOLCHAIN_SUPPORTS_VIRT=y on r6, and just 68 bytes more on r2. Signed-off-by: James Hogan <jhogan@kernel.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: linux-mips@linux-mips.org --- arch/mips/include/asm/mipsregs.h | 164 +++++++------------------------- 1 file changed, 37 insertions(+), 127 deletions(-) diff --git a/arch/mips/include/asm/mipsregs.h b/arch/mips/include/asm/mipsregs.h index 972698557b4b..361c35243366 100644 --- a/arch/mips/include/asm/mipsregs.h +++ b/arch/mips/include/asm/mipsregs.h @@ -1913,14 +1913,40 @@ do { \ * Macros to access the guest system control coprocessor */ -#ifdef TOOLCHAIN_SUPPORTS_VIRT +#ifndef TOOLCHAIN_SUPPORTS_VIRT +_ASM_MACRO_2R_1S(mfgc0, rt, rs, sel, + _ASM_INSN_IF_MIPS(0x40600000 | __rt << 16 | __rs << 11 | \\sel) + _ASM_INSN32_IF_MM(0x000004fc | __rt << 21 | __rs << 16 | \\sel << 11)); +_ASM_MACRO_2R_1S(dmfgc0, rt, rs, sel, + _ASM_INSN_IF_MIPS(0x40600100 | __rt << 16 | __rs << 11 | \\sel) + _ASM_INSN32_IF_MM(0x580004fc | __rt << 21 | __rs << 16 | \\sel << 11)); +_ASM_MACRO_2R_1S(mtgc0, rt, rd, sel, + _ASM_INSN_IF_MIPS(0x40600200 | __rt << 16 | __rd << 11 | \\sel) + _ASM_INSN32_IF_MM(0x000006fc | __rt << 21 | __rd << 16 | \\sel << 11)); +_ASM_MACRO_2R_1S(dmtgc0, rt, rd, sel, + _ASM_INSN_IF_MIPS(0x40600300 | __rt << 16 | __rd << 11 | \\sel) + _ASM_INSN32_IF_MM(0x580006fc | __rt << 21 | __rd << 16 | \\sel << 11)); +_ASM_MACRO_0(tlbgp, _ASM_INSN_IF_MIPS(0x42000010) + _ASM_INSN32_IF_MM(0x0000017c)); +_ASM_MACRO_0(tlbgr, _ASM_INSN_IF_MIPS(0x42000009) + _ASM_INSN32_IF_MM(0x0000117c)); +_ASM_MACRO_0(tlbgwi, _ASM_INSN_IF_MIPS(0x4200000a) + _ASM_INSN32_IF_MM(0x0000217c)); +_ASM_MACRO_0(tlbgwr, _ASM_INSN_IF_MIPS(0x4200000e) + _ASM_INSN32_IF_MM(0x0000317c)); +_ASM_MACRO_0(tlbginvf, _ASM_INSN_IF_MIPS(0x4200000c) + _ASM_INSN32_IF_MM(0x0000517c)); +#define _ASM_SET_VIRT "" +#else /* !TOOLCHAIN_SUPPORTS_VIRT */ +#define _ASM_SET_VIRT ".set\tvirt\n\t" +#endif #define __read_32bit_gc0_register(source, sel) \ ({ int __res; \ __asm__ __volatile__( \ ".set\tpush\n\t" \ ".set\tmips32r2\n\t" \ - ".set\tvirt\n\t" \ + _ASM_SET_VIRT \ "mfgc0\t%0, $%1, %2\n\t" \ ".set\tpop" \ : "=r" (__res) \ @@ -1933,8 +1959,8 @@ do { \ __asm__ __volatile__( \ ".set\tpush\n\t" \ ".set\tmips64r2\n\t" \ - ".set\tvirt\n\t" \ - "dmfgc0\t%0, $%1, %2\n\t" \ + _ASM_SET_VIRT \ + "dmfgc0\t%0, $%1, %2\n\t" \ ".set\tpop" \ : "=r" (__res) \ : "i" (source), "i" (sel)); \ @@ -1946,7 +1972,7 @@ do { \ __asm__ __volatile__( \ ".set\tpush\n\t" \ ".set\tmips32r2\n\t" \ - ".set\tvirt\n\t" \ + _ASM_SET_VIRT \ "mtgc0\t%z0, $%1, %2\n\t" \ ".set\tpop" \ : : "Jr" ((unsigned int)(value)), \ @@ -1958,75 +1984,13 @@ do { \ __asm__ __volatile__( \ ".set\tpush\n\t" \ ".set\tmips64r2\n\t" \ - ".set\tvirt\n\t" \ + _ASM_SET_VIRT \ "dmtgc0\t%z0, $%1, %2\n\t" \ ".set\tpop" \ : : "Jr" (value), \ "i" (register), "i" (sel)); \ } while (0) -#else /* TOOLCHAIN_SUPPORTS_VIRT */ - -#define __read_32bit_gc0_register(source, sel) \ -({ int __res; \ - __asm__ __volatile__( \ - ".set\tpush\n\t" \ - ".set\tnoat\n\t" \ - "# mfgc0\t$1, $%1, %2\n\t" \ - _ASM_INSN_IF_MIPS(0x40610000 | %1 << 11 | %2) \ - _ASM_INSN32_IF_MM(0x002004fc | %1 << 16 | %2 << 11) \ - "move\t%0, $1\n\t" \ - ".set\tpop" \ - : "=r" (__res) \ - : "i" (source), "i" (sel)); \ - __res; \ -}) - -#define __read_64bit_gc0_register(source, sel) \ -({ unsigned long long __res; \ - __asm__ __volatile__( \ - ".set\tpush\n\t" \ - ".set\tnoat\n\t" \ - "# dmfgc0\t$1, $%1, %2\n\t" \ - _ASM_INSN_IF_MIPS(0x40610100 | %1 << 11 | %2) \ - _ASM_INSN32_IF_MM(0x582004fc | %1 << 16 | %2 << 11) \ - "move\t%0, $1\n\t" \ - ".set\tpop" \ - : "=r" (__res) \ - : "i" (source), "i" (sel)); \ - __res; \ -}) - -#define __write_32bit_gc0_register(register, sel, value) \ -do { \ - __asm__ __volatile__( \ - ".set\tpush\n\t" \ - ".set\tnoat\n\t" \ - "move\t$1, %z0\n\t" \ - "# mtgc0\t$1, $%1, %2\n\t" \ - _ASM_INSN_IF_MIPS(0x40610200 | %1 << 11 | %2) \ - _ASM_INSN32_IF_MM(0x002006fc | %1 << 16 | %2 << 11) \ - ".set\tpop" \ - : : "Jr" ((unsigned int)(value)), \ - "i" (register), "i" (sel)); \ -} while (0) - -#define __write_64bit_gc0_register(register, sel, value) \ -do { \ - __asm__ __volatile__( \ - ".set\tpush\n\t" \ - ".set\tnoat\n\t" \ - "move\t$1, %z0\n\t" \ - "# dmtgc0\t$1, $%1, %2\n\t" \ - _ASM_INSN_IF_MIPS(0x40610300 | %1 << 11 | %2) \ - _ASM_INSN32_IF_MM(0x582006fc | %1 << 16 | %2 << 11) \ - ".set\tpop" \ - : : "Jr" (value), \ - "i" (register), "i" (sel)); \ -} while (0) - -#endif /* !TOOLCHAIN_SUPPORTS_VIRT */ - #define __read_ulong_gc0_register(reg, sel) \ ((sizeof(unsigned long) == 4) ? \ (unsigned long) __read_32bit_gc0_register(reg, sel) : \ @@ -2664,8 +2628,6 @@ static inline void tlb_write_random(void) ".set reorder"); } -#ifdef TOOLCHAIN_SUPPORTS_VIRT - /* * Guest TLB operations. * @@ -2676,7 +2638,7 @@ static inline void guest_tlb_probe(void) __asm__ __volatile__( ".set push\n\t" ".set noreorder\n\t" - ".set virt\n\t" + _ASM_SET_VIRT "tlbgp\n\t" ".set pop"); } @@ -2686,7 +2648,7 @@ static inline void guest_tlb_read(void) __asm__ __volatile__( ".set push\n\t" ".set noreorder\n\t" - ".set virt\n\t" + _ASM_SET_VIRT "tlbgr\n\t" ".set pop"); } @@ -2696,7 +2658,7 @@ static inline void guest_tlb_write_indexed(void) __asm__ __volatile__( ".set push\n\t" ".set noreorder\n\t" - ".set virt\n\t" + _ASM_SET_VIRT "tlbgwi\n\t" ".set pop"); } @@ -2706,7 +2668,7 @@ static inline void guest_tlb_write_random(void) __asm__ __volatile__( ".set push\n\t" ".set noreorder\n\t" - ".set virt\n\t" + _ASM_SET_VIRT "tlbgwr\n\t" ".set pop"); } @@ -2719,63 +2681,11 @@ static inline void guest_tlbinvf(void) __asm__ __volatile__( ".set push\n\t" ".set noreorder\n\t" - ".set virt\n\t" + _ASM_SET_VIRT "tlbginvf\n\t" ".set pop"); } -#else /* TOOLCHAIN_SUPPORTS_VIRT */ - -/* - * Guest TLB operations. - * - * It is responsibility of the caller to take care of any TLB hazards. - */ -static inline void guest_tlb_probe(void) -{ - __asm__ __volatile__( - "# tlbgp\n\t" - _ASM_INSN_IF_MIPS(0x42000010) - _ASM_INSN32_IF_MM(0x0000017c)); -} - -static inline void guest_tlb_read(void) -{ - __asm__ __volatile__( - "# tlbgr\n\t" - _ASM_INSN_IF_MIPS(0x42000009) - _ASM_INSN32_IF_MM(0x0000117c)); -} - -static inline void guest_tlb_write_indexed(void) -{ - __asm__ __volatile__( - "# tlbgwi\n\t" - _ASM_INSN_IF_MIPS(0x4200000a) - _ASM_INSN32_IF_MM(0x0000217c)); -} - -static inline void guest_tlb_write_random(void) -{ - __asm__ __volatile__( - "# tlbgwr\n\t" - _ASM_INSN_IF_MIPS(0x4200000e) - _ASM_INSN32_IF_MM(0x0000317c)); -} - -/* - * Guest TLB Invalidate Flush - */ -static inline void guest_tlbinvf(void) -{ - __asm__ __volatile__( - "# tlbginvf\n\t" - _ASM_INSN_IF_MIPS(0x4200000c) - _ASM_INSN32_IF_MM(0x0000517c)); -} - -#endif /* !TOOLCHAIN_SUPPORTS_VIRT */ - /* * Manipulate bits in a register. */ -- git-series 0.9.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 3/7] MIPS: VZ: Pass GC0 register names in $n format @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle From: James Hogan <jhogan@kernel.org> Now that we are using assembler macros to implement VZ instructions on toolchains which don't support them, pass VZ guest Cop0 register names to the __{read,write}_{32bit,ulong,64bit}_gc0_register macros in $n format rather than register numbers. This is to make them consistent with the normal root Cop0 register access macros which they were originally based on. Signed-off-by: James Hogan <jhogan@kernel.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: linux-mips@linux-mips.org --- arch/mips/include/asm/mipsregs.h | 376 ++++++++++++++++---------------- 1 file changed, 188 insertions(+), 188 deletions(-) diff --git a/arch/mips/include/asm/mipsregs.h b/arch/mips/include/asm/mipsregs.h index 361c35243366..000530b18f2e 100644 --- a/arch/mips/include/asm/mipsregs.h +++ b/arch/mips/include/asm/mipsregs.h @@ -1947,10 +1947,10 @@ _ASM_MACRO_0(tlbginvf, _ASM_INSN_IF_MIPS(0x4200000c) ".set\tpush\n\t" \ ".set\tmips32r2\n\t" \ _ASM_SET_VIRT \ - "mfgc0\t%0, $%1, %2\n\t" \ + "mfgc0\t%0, " #source ", %1\n\t" \ ".set\tpop" \ : "=r" (__res) \ - : "i" (source), "i" (sel)); \ + : "i" (sel)); \ __res; \ }) @@ -1960,10 +1960,10 @@ _ASM_MACRO_0(tlbginvf, _ASM_INSN_IF_MIPS(0x4200000c) ".set\tpush\n\t" \ ".set\tmips64r2\n\t" \ _ASM_SET_VIRT \ - "dmfgc0\t%0, $%1, %2\n\t" \ + "dmfgc0\t%0, " #source ", %1\n\t" \ ".set\tpop" \ : "=r" (__res) \ - : "i" (source), "i" (sel)); \ + : "i" (sel)); \ __res; \ }) @@ -1973,10 +1973,10 @@ do { \ ".set\tpush\n\t" \ ".set\tmips32r2\n\t" \ _ASM_SET_VIRT \ - "mtgc0\t%z0, $%1, %2\n\t" \ + "mtgc0\t%z0, " #register ", %1\n\t" \ ".set\tpop" \ : : "Jr" ((unsigned int)(value)), \ - "i" (register), "i" (sel)); \ + "i" (sel)); \ } while (0) #define __write_64bit_gc0_register(register, sel, value) \ @@ -1985,10 +1985,10 @@ do { \ ".set\tpush\n\t" \ ".set\tmips64r2\n\t" \ _ASM_SET_VIRT \ - "dmtgc0\t%z0, $%1, %2\n\t" \ + "dmtgc0\t%z0, " #register ", %1\n\t" \ ".set\tpop" \ : : "Jr" (value), \ - "i" (register), "i" (sel)); \ + "i" (sel)); \ } while (0) #define __read_ulong_gc0_register(reg, sel) \ @@ -2004,207 +2004,207 @@ do { \ __write_64bit_gc0_register(reg, sel, val); \ } while (0) -#define read_gc0_index() __read_32bit_gc0_register(0, 0) -#define write_gc0_index(val) __write_32bit_gc0_register(0, 0, val) +#define read_gc0_index() __read_32bit_gc0_register($0, 0) +#define write_gc0_index(val) __write_32bit_gc0_register($0, 0, val) -#define read_gc0_entrylo0() __read_ulong_gc0_register(2, 0) -#define write_gc0_entrylo0(val) __write_ulong_gc0_register(2, 0, val) +#define read_gc0_entrylo0() __read_ulong_gc0_register($2, 0) +#define write_gc0_entrylo0(val) __write_ulong_gc0_register($2, 0, val) -#define read_gc0_entrylo1() __read_ulong_gc0_register(3, 0) -#define write_gc0_entrylo1(val) __write_ulong_gc0_register(3, 0, val) +#define read_gc0_entrylo1() __read_ulong_gc0_register($3, 0) +#define write_gc0_entrylo1(val) __write_ulong_gc0_register($3, 0, val) -#define read_gc0_context() __read_ulong_gc0_register(4, 0) -#define write_gc0_context(val) __write_ulong_gc0_register(4, 0, val) +#define read_gc0_context() __read_ulong_gc0_register($4, 0) +#define write_gc0_context(val) __write_ulong_gc0_register($4, 0, val) -#define read_gc0_contextconfig() __read_32bit_gc0_register(4, 1) -#define write_gc0_contextconfig(val) __write_32bit_gc0_register(4, 1, val) +#define read_gc0_contextconfig() __read_32bit_gc0_register($4, 1) +#define write_gc0_contextconfig(val) __write_32bit_gc0_register($4, 1, val) -#define read_gc0_userlocal() __read_ulong_gc0_register(4, 2) -#define write_gc0_userlocal(val) __write_ulong_gc0_register(4, 2, val) +#define read_gc0_userlocal() __read_ulong_gc0_register($4, 2) +#define write_gc0_userlocal(val) __write_ulong_gc0_register($4, 2, val) -#define read_gc0_xcontextconfig() __read_ulong_gc0_register(4, 3) -#define write_gc0_xcontextconfig(val) __write_ulong_gc0_register(4, 3, val) +#define read_gc0_xcontextconfig() __read_ulong_gc0_register($4, 3) +#define write_gc0_xcontextconfig(val) __write_ulong_gc0_register($4, 3, val) -#define read_gc0_pagemask() __read_32bit_gc0_register(5, 0) -#define write_gc0_pagemask(val) __write_32bit_gc0_register(5, 0, val) +#define read_gc0_pagemask() __read_32bit_gc0_register($5, 0) +#define write_gc0_pagemask(val) __write_32bit_gc0_register($5, 0, val) -#define read_gc0_pagegrain() __read_32bit_gc0_register(5, 1) -#define write_gc0_pagegrain(val) __write_32bit_gc0_register(5, 1, val) +#define read_gc0_pagegrain() __read_32bit_gc0_register($5, 1) +#define write_gc0_pagegrain(val) __write_32bit_gc0_register($5, 1, val) -#define read_gc0_segctl0() __read_ulong_gc0_register(5, 2) -#define write_gc0_segctl0(val) __write_ulong_gc0_register(5, 2, val) +#define read_gc0_segctl0() __read_ulong_gc0_register($5, 2) +#define write_gc0_segctl0(val) __write_ulong_gc0_register($5, 2, val) -#define read_gc0_segctl1() __read_ulong_gc0_register(5, 3) -#define write_gc0_segctl1(val) __write_ulong_gc0_register(5, 3, val) +#define read_gc0_segctl1() __read_ulong_gc0_register($5, 3) +#define write_gc0_segctl1(val) __write_ulong_gc0_register($5, 3, val) -#define read_gc0_segctl2() __read_ulong_gc0_register(5, 4) -#define write_gc0_segctl2(val) __write_ulong_gc0_register(5, 4, val) +#define read_gc0_segctl2() __read_ulong_gc0_register($5, 4) +#define write_gc0_segctl2(val) __write_ulong_gc0_register($5, 4, val) -#define read_gc0_pwbase() __read_ulong_gc0_register(5, 5) -#define write_gc0_pwbase(val) __write_ulong_gc0_register(5, 5, val) +#define read_gc0_pwbase() __read_ulong_gc0_register($5, 5) +#define write_gc0_pwbase(val) __write_ulong_gc0_register($5, 5, val) -#define read_gc0_pwfield() __read_ulong_gc0_register(5, 6) -#define write_gc0_pwfield(val) __write_ulong_gc0_register(5, 6, val) +#define read_gc0_pwfield() __read_ulong_gc0_register($5, 6) +#define write_gc0_pwfield(val) __write_ulong_gc0_register($5, 6, val) -#define read_gc0_pwsize() __read_ulong_gc0_register(5, 7) -#define write_gc0_pwsize(val) __write_ulong_gc0_register(5, 7, val) +#define read_gc0_pwsize() __read_ulong_gc0_register($5, 7) +#define write_gc0_pwsize(val) __write_ulong_gc0_register($5, 7, val) -#define read_gc0_wired() __read_32bit_gc0_register(6, 0) -#define write_gc0_wired(val) __write_32bit_gc0_register(6, 0, val) +#define read_gc0_wired() __read_32bit_gc0_register($6, 0) +#define write_gc0_wired(val) __write_32bit_gc0_register($6, 0, val) -#define read_gc0_pwctl() __read_32bit_gc0_register(6, 6) -#define write_gc0_pwctl(val) __write_32bit_gc0_register(6, 6, val) - -#define read_gc0_hwrena() __read_32bit_gc0_register(7, 0) -#define write_gc0_hwrena(val) __write_32bit_gc0_register(7, 0, val) - -#define read_gc0_badvaddr() __read_ulong_gc0_register(8, 0) -#define write_gc0_badvaddr(val) __write_ulong_gc0_register(8, 0, val) - -#define read_gc0_badinstr() __read_32bit_gc0_register(8, 1) -#define write_gc0_badinstr(val) __write_32bit_gc0_register(8, 1, val) - -#define read_gc0_badinstrp() __read_32bit_gc0_register(8, 2) -#define write_gc0_badinstrp(val) __write_32bit_gc0_register(8, 2, val) - -#define read_gc0_count() __read_32bit_gc0_register(9, 0) - -#define read_gc0_entryhi() __read_ulong_gc0_register(10, 0) -#define write_gc0_entryhi(val) __write_ulong_gc0_register(10, 0, val) - -#define read_gc0_compare() __read_32bit_gc0_register(11, 0) -#define write_gc0_compare(val) __write_32bit_gc0_register(11, 0, val) - -#define read_gc0_status() __read_32bit_gc0_register(12, 0) -#define write_gc0_status(val) __write_32bit_gc0_register(12, 0, val) - -#define read_gc0_intctl() __read_32bit_gc0_register(12, 1) -#define write_gc0_intctl(val) __write_32bit_gc0_register(12, 1, val) - -#define read_gc0_cause() __read_32bit_gc0_register(13, 0) -#define write_gc0_cause(val) __write_32bit_gc0_register(13, 0, val) - -#define read_gc0_epc() __read_ulong_gc0_register(14, 0) -#define write_gc0_epc(val) __write_ulong_gc0_register(14, 0, val) - -#define read_gc0_prid() __read_32bit_gc0_register(15, 0) - -#define read_gc0_ebase() __read_32bit_gc0_register(15, 1) -#define write_gc0_ebase(val) __write_32bit_gc0_register(15, 1, val) - -#define read_gc0_ebase_64() __read_64bit_gc0_register(15, 1) -#define write_gc0_ebase_64(val) __write_64bit_gc0_register(15, 1, val) - -#define read_gc0_config() __read_32bit_gc0_register(16, 0) -#define read_gc0_config1() __read_32bit_gc0_register(16, 1) -#define read_gc0_config2() __read_32bit_gc0_register(16, 2) -#define read_gc0_config3() __read_32bit_gc0_register(16, 3) -#define read_gc0_config4() __read_32bit_gc0_register(16, 4) -#define read_gc0_config5() __read_32bit_gc0_register(16, 5) -#define read_gc0_config6() __read_32bit_gc0_register(16, 6) -#define read_gc0_config7() __read_32bit_gc0_register(16, 7) -#define write_gc0_config(val) __write_32bit_gc0_register(16, 0, val) -#define write_gc0_config1(val) __write_32bit_gc0_register(16, 1, val) -#define write_gc0_config2(val) __write_32bit_gc0_register(16, 2, val) -#define write_gc0_config3(val) __write_32bit_gc0_register(16, 3, val) -#define write_gc0_config4(val) __write_32bit_gc0_register(16, 4, val) -#define write_gc0_config5(val) __write_32bit_gc0_register(16, 5, val) -#define write_gc0_config6(val) __write_32bit_gc0_register(16, 6, val) -#define write_gc0_config7(val) __write_32bit_gc0_register(16, 7, val) - -#define read_gc0_lladdr() __read_ulong_gc0_register(17, 0) -#define write_gc0_lladdr(val) __write_ulong_gc0_register(17, 0, val) - -#define read_gc0_watchlo0() __read_ulong_gc0_register(18, 0) -#define read_gc0_watchlo1() __read_ulong_gc0_register(18, 1) -#define read_gc0_watchlo2() __read_ulong_gc0_register(18, 2) -#define read_gc0_watchlo3() __read_ulong_gc0_register(18, 3) -#define read_gc0_watchlo4() __read_ulong_gc0_register(18, 4) -#define read_gc0_watchlo5() __read_ulong_gc0_register(18, 5) -#define read_gc0_watchlo6() __read_ulong_gc0_register(18, 6) -#define read_gc0_watchlo7() __read_ulong_gc0_register(18, 7) -#define write_gc0_watchlo0(val) __write_ulong_gc0_register(18, 0, val) -#define write_gc0_watchlo1(val) __write_ulong_gc0_register(18, 1, val) -#define write_gc0_watchlo2(val) __write_ulong_gc0_register(18, 2, val) -#define write_gc0_watchlo3(val) __write_ulong_gc0_register(18, 3, val) -#define write_gc0_watchlo4(val) __write_ulong_gc0_register(18, 4, val) -#define write_gc0_watchlo5(val) __write_ulong_gc0_register(18, 5, val) -#define write_gc0_watchlo6(val) __write_ulong_gc0_register(18, 6, val) -#define write_gc0_watchlo7(val) __write_ulong_gc0_register(18, 7, val) - -#define read_gc0_watchhi0() __read_32bit_gc0_register(19, 0) -#define read_gc0_watchhi1() __read_32bit_gc0_register(19, 1) -#define read_gc0_watchhi2() __read_32bit_gc0_register(19, 2) -#define read_gc0_watchhi3() __read_32bit_gc0_register(19, 3) -#define read_gc0_watchhi4() __read_32bit_gc0_register(19, 4) -#define read_gc0_watchhi5() __read_32bit_gc0_register(19, 5) -#define read_gc0_watchhi6() __read_32bit_gc0_register(19, 6) -#define read_gc0_watchhi7() __read_32bit_gc0_register(19, 7) -#define write_gc0_watchhi0(val) __write_32bit_gc0_register(19, 0, val) -#define write_gc0_watchhi1(val) __write_32bit_gc0_register(19, 1, val) -#define write_gc0_watchhi2(val) __write_32bit_gc0_register(19, 2, val) -#define write_gc0_watchhi3(val) __write_32bit_gc0_register(19, 3, val) -#define write_gc0_watchhi4(val) __write_32bit_gc0_register(19, 4, val) -#define write_gc0_watchhi5(val) __write_32bit_gc0_register(19, 5, val) -#define write_gc0_watchhi6(val) __write_32bit_gc0_register(19, 6, val) -#define write_gc0_watchhi7(val) __write_32bit_gc0_register(19, 7, val) - -#define read_gc0_xcontext() __read_ulong_gc0_register(20, 0) -#define write_gc0_xcontext(val) __write_ulong_gc0_register(20, 0, val) - -#define read_gc0_perfctrl0() __read_32bit_gc0_register(25, 0) -#define write_gc0_perfctrl0(val) __write_32bit_gc0_register(25, 0, val) -#define read_gc0_perfcntr0() __read_32bit_gc0_register(25, 1) -#define write_gc0_perfcntr0(val) __write_32bit_gc0_register(25, 1, val) -#define read_gc0_perfcntr0_64() __read_64bit_gc0_register(25, 1) -#define write_gc0_perfcntr0_64(val) __write_64bit_gc0_register(25, 1, val) -#define read_gc0_perfctrl1() __read_32bit_gc0_register(25, 2) -#define write_gc0_perfctrl1(val) __write_32bit_gc0_register(25, 2, val) -#define read_gc0_perfcntr1() __read_32bit_gc0_register(25, 3) -#define write_gc0_perfcntr1(val) __write_32bit_gc0_register(25, 3, val) -#define read_gc0_perfcntr1_64() __read_64bit_gc0_register(25, 3) -#define write_gc0_perfcntr1_64(val) __write_64bit_gc0_register(25, 3, val) -#define read_gc0_perfctrl2() __read_32bit_gc0_register(25, 4) -#define write_gc0_perfctrl2(val) __write_32bit_gc0_register(25, 4, val) -#define read_gc0_perfcntr2() __read_32bit_gc0_register(25, 5) -#define write_gc0_perfcntr2(val) __write_32bit_gc0_register(25, 5, val) -#define read_gc0_perfcntr2_64() __read_64bit_gc0_register(25, 5) -#define write_gc0_perfcntr2_64(val) __write_64bit_gc0_register(25, 5, val) -#define read_gc0_perfctrl3() __read_32bit_gc0_register(25, 6) -#define write_gc0_perfctrl3(val) __write_32bit_gc0_register(25, 6, val) -#define read_gc0_perfcntr3() __read_32bit_gc0_register(25, 7) -#define write_gc0_perfcntr3(val) __write_32bit_gc0_register(25, 7, val) -#define read_gc0_perfcntr3_64() __read_64bit_gc0_register(25, 7) -#define write_gc0_perfcntr3_64(val) __write_64bit_gc0_register(25, 7, val) - -#define read_gc0_errorepc() __read_ulong_gc0_register(30, 0) -#define write_gc0_errorepc(val) __write_ulong_gc0_register(30, 0, val) - -#define read_gc0_kscratch1() __read_ulong_gc0_register(31, 2) -#define read_gc0_kscratch2() __read_ulong_gc0_register(31, 3) -#define read_gc0_kscratch3() __read_ulong_gc0_register(31, 4) -#define read_gc0_kscratch4() __read_ulong_gc0_register(31, 5) -#define read_gc0_kscratch5() __read_ulong_gc0_register(31, 6) -#define read_gc0_kscratch6() __read_ulong_gc0_register(31, 7) -#define write_gc0_kscratch1(val) __write_ulong_gc0_register(31, 2, val) -#define write_gc0_kscratch2(val) __write_ulong_gc0_register(31, 3, val) -#define write_gc0_kscratch3(val) __write_ulong_gc0_register(31, 4, val) -#define write_gc0_kscratch4(val) __write_ulong_gc0_register(31, 5, val) -#define write_gc0_kscratch5(val) __write_ulong_gc0_register(31, 6, val) -#define write_gc0_kscratch6(val) __write_ulong_gc0_register(31, 7, val) +#define read_gc0_pwctl() __read_32bit_gc0_register($6, 6) +#define write_gc0_pwctl(val) __write_32bit_gc0_register($6, 6, val) + +#define read_gc0_hwrena() __read_32bit_gc0_register($7, 0) +#define write_gc0_hwrena(val) __write_32bit_gc0_register($7, 0, val) + +#define read_gc0_badvaddr() __read_ulong_gc0_register($8, 0) +#define write_gc0_badvaddr(val) __write_ulong_gc0_register($8, 0, val) + +#define read_gc0_badinstr() __read_32bit_gc0_register($8, 1) +#define write_gc0_badinstr(val) __write_32bit_gc0_register($8, 1, val) + +#define read_gc0_badinstrp() __read_32bit_gc0_register($8, 2) +#define write_gc0_badinstrp(val) __write_32bit_gc0_register($8, 2, val) + +#define read_gc0_count() __read_32bit_gc0_register($9, 0) + +#define read_gc0_entryhi() __read_ulong_gc0_register($10, 0) +#define write_gc0_entryhi(val) __write_ulong_gc0_register($10, 0, val) + +#define read_gc0_compare() __read_32bit_gc0_register($11, 0) +#define write_gc0_compare(val) __write_32bit_gc0_register($11, 0, val) + +#define read_gc0_status() __read_32bit_gc0_register($12, 0) +#define write_gc0_status(val) __write_32bit_gc0_register($12, 0, val) + +#define read_gc0_intctl() __read_32bit_gc0_register($12, 1) +#define write_gc0_intctl(val) __write_32bit_gc0_register($12, 1, val) + +#define read_gc0_cause() __read_32bit_gc0_register($13, 0) +#define write_gc0_cause(val) __write_32bit_gc0_register($13, 0, val) + +#define read_gc0_epc() __read_ulong_gc0_register($14, 0) +#define write_gc0_epc(val) __write_ulong_gc0_register($14, 0, val) + +#define read_gc0_prid() __read_32bit_gc0_register($15, 0) + +#define read_gc0_ebase() __read_32bit_gc0_register($15, 1) +#define write_gc0_ebase(val) __write_32bit_gc0_register($15, 1, val) + +#define read_gc0_ebase_64() __read_64bit_gc0_register($15, 1) +#define write_gc0_ebase_64(val) __write_64bit_gc0_register($15, 1, val) + +#define read_gc0_config() __read_32bit_gc0_register($16, 0) +#define read_gc0_config1() __read_32bit_gc0_register($16, 1) +#define read_gc0_config2() __read_32bit_gc0_register($16, 2) +#define read_gc0_config3() __read_32bit_gc0_register($16, 3) +#define read_gc0_config4() __read_32bit_gc0_register($16, 4) +#define read_gc0_config5() __read_32bit_gc0_register($16, 5) +#define read_gc0_config6() __read_32bit_gc0_register($16, 6) +#define read_gc0_config7() __read_32bit_gc0_register($16, 7) +#define write_gc0_config(val) __write_32bit_gc0_register($16, 0, val) +#define write_gc0_config1(val) __write_32bit_gc0_register($16, 1, val) +#define write_gc0_config2(val) __write_32bit_gc0_register($16, 2, val) +#define write_gc0_config3(val) __write_32bit_gc0_register($16, 3, val) +#define write_gc0_config4(val) __write_32bit_gc0_register($16, 4, val) +#define write_gc0_config5(val) __write_32bit_gc0_register($16, 5, val) +#define write_gc0_config6(val) __write_32bit_gc0_register($16, 6, val) +#define write_gc0_config7(val) __write_32bit_gc0_register($16, 7, val) + +#define read_gc0_lladdr() __read_ulong_gc0_register($17, 0) +#define write_gc0_lladdr(val) __write_ulong_gc0_register($17, 0, val) + +#define read_gc0_watchlo0() __read_ulong_gc0_register($18, 0) +#define read_gc0_watchlo1() __read_ulong_gc0_register($18, 1) +#define read_gc0_watchlo2() __read_ulong_gc0_register($18, 2) +#define read_gc0_watchlo3() __read_ulong_gc0_register($18, 3) +#define read_gc0_watchlo4() __read_ulong_gc0_register($18, 4) +#define read_gc0_watchlo5() __read_ulong_gc0_register($18, 5) +#define read_gc0_watchlo6() __read_ulong_gc0_register($18, 6) +#define read_gc0_watchlo7() __read_ulong_gc0_register($18, 7) +#define write_gc0_watchlo0(val) __write_ulong_gc0_register($18, 0, val) +#define write_gc0_watchlo1(val) __write_ulong_gc0_register($18, 1, val) +#define write_gc0_watchlo2(val) __write_ulong_gc0_register($18, 2, val) +#define write_gc0_watchlo3(val) __write_ulong_gc0_register($18, 3, val) +#define write_gc0_watchlo4(val) __write_ulong_gc0_register($18, 4, val) +#define write_gc0_watchlo5(val) __write_ulong_gc0_register($18, 5, val) +#define write_gc0_watchlo6(val) __write_ulong_gc0_register($18, 6, val) +#define write_gc0_watchlo7(val) __write_ulong_gc0_register($18, 7, val) + +#define read_gc0_watchhi0() __read_32bit_gc0_register($19, 0) +#define read_gc0_watchhi1() __read_32bit_gc0_register($19, 1) +#define read_gc0_watchhi2() __read_32bit_gc0_register($19, 2) +#define read_gc0_watchhi3() __read_32bit_gc0_register($19, 3) +#define read_gc0_watchhi4() __read_32bit_gc0_register($19, 4) +#define read_gc0_watchhi5() __read_32bit_gc0_register($19, 5) +#define read_gc0_watchhi6() __read_32bit_gc0_register($19, 6) +#define read_gc0_watchhi7() __read_32bit_gc0_register($19, 7) +#define write_gc0_watchhi0(val) __write_32bit_gc0_register($19, 0, val) +#define write_gc0_watchhi1(val) __write_32bit_gc0_register($19, 1, val) +#define write_gc0_watchhi2(val) __write_32bit_gc0_register($19, 2, val) +#define write_gc0_watchhi3(val) __write_32bit_gc0_register($19, 3, val) +#define write_gc0_watchhi4(val) __write_32bit_gc0_register($19, 4, val) +#define write_gc0_watchhi5(val) __write_32bit_gc0_register($19, 5, val) +#define write_gc0_watchhi6(val) __write_32bit_gc0_register($19, 6, val) +#define write_gc0_watchhi7(val) __write_32bit_gc0_register($19, 7, val) + +#define read_gc0_xcontext() __read_ulong_gc0_register($20, 0) +#define write_gc0_xcontext(val) __write_ulong_gc0_register($20, 0, val) + +#define read_gc0_perfctrl0() __read_32bit_gc0_register($25, 0) +#define write_gc0_perfctrl0(val) __write_32bit_gc0_register($25, 0, val) +#define read_gc0_perfcntr0() __read_32bit_gc0_register($25, 1) +#define write_gc0_perfcntr0(val) __write_32bit_gc0_register($25, 1, val) +#define read_gc0_perfcntr0_64() __read_64bit_gc0_register($25, 1) +#define write_gc0_perfcntr0_64(val) __write_64bit_gc0_register($25, 1, val) +#define read_gc0_perfctrl1() __read_32bit_gc0_register($25, 2) +#define write_gc0_perfctrl1(val) __write_32bit_gc0_register($25, 2, val) +#define read_gc0_perfcntr1() __read_32bit_gc0_register($25, 3) +#define write_gc0_perfcntr1(val) __write_32bit_gc0_register($25, 3, val) +#define read_gc0_perfcntr1_64() __read_64bit_gc0_register($25, 3) +#define write_gc0_perfcntr1_64(val) __write_64bit_gc0_register($25, 3, val) +#define read_gc0_perfctrl2() __read_32bit_gc0_register($25, 4) +#define write_gc0_perfctrl2(val) __write_32bit_gc0_register($25, 4, val) +#define read_gc0_perfcntr2() __read_32bit_gc0_register($25, 5) +#define write_gc0_perfcntr2(val) __write_32bit_gc0_register($25, 5, val) +#define read_gc0_perfcntr2_64() __read_64bit_gc0_register($25, 5) +#define write_gc0_perfcntr2_64(val) __write_64bit_gc0_register($25, 5, val) +#define read_gc0_perfctrl3() __read_32bit_gc0_register($25, 6) +#define write_gc0_perfctrl3(val) __write_32bit_gc0_register($25, 6, val) +#define read_gc0_perfcntr3() __read_32bit_gc0_register($25, 7) +#define write_gc0_perfcntr3(val) __write_32bit_gc0_register($25, 7, val) +#define read_gc0_perfcntr3_64() __read_64bit_gc0_register($25, 7) +#define write_gc0_perfcntr3_64(val) __write_64bit_gc0_register($25, 7, val) + +#define read_gc0_errorepc() __read_ulong_gc0_register($30, 0) +#define write_gc0_errorepc(val) __write_ulong_gc0_register($30, 0, val) + +#define read_gc0_kscratch1() __read_ulong_gc0_register($31, 2) +#define read_gc0_kscratch2() __read_ulong_gc0_register($31, 3) +#define read_gc0_kscratch3() __read_ulong_gc0_register($31, 4) +#define read_gc0_kscratch4() __read_ulong_gc0_register($31, 5) +#define read_gc0_kscratch5() __read_ulong_gc0_register($31, 6) +#define read_gc0_kscratch6() __read_ulong_gc0_register($31, 7) +#define write_gc0_kscratch1(val) __write_ulong_gc0_register($31, 2, val) +#define write_gc0_kscratch2(val) __write_ulong_gc0_register($31, 3, val) +#define write_gc0_kscratch3(val) __write_ulong_gc0_register($31, 4, val) +#define write_gc0_kscratch4(val) __write_ulong_gc0_register($31, 5, val) +#define write_gc0_kscratch5(val) __write_ulong_gc0_register($31, 6, val) +#define write_gc0_kscratch6(val) __write_ulong_gc0_register($31, 7, val) /* Cavium OCTEON (cnMIPS) */ -#define read_gc0_cvmcount() __read_ulong_gc0_register(9, 6) -#define write_gc0_cvmcount(val) __write_ulong_gc0_register(9, 6, val) +#define read_gc0_cvmcount() __read_ulong_gc0_register($9, 6) +#define write_gc0_cvmcount(val) __write_ulong_gc0_register($9, 6, val) -#define read_gc0_cvmctl() __read_64bit_gc0_register(9, 7) -#define write_gc0_cvmctl(val) __write_64bit_gc0_register(9, 7, val) +#define read_gc0_cvmctl() __read_64bit_gc0_register($9, 7) +#define write_gc0_cvmctl(val) __write_64bit_gc0_register($9, 7, val) -#define read_gc0_cvmmemctl() __read_64bit_gc0_register(11, 7) -#define write_gc0_cvmmemctl(val) __write_64bit_gc0_register(11, 7, val) +#define read_gc0_cvmmemctl() __read_64bit_gc0_register($11, 7) +#define write_gc0_cvmmemctl(val) __write_64bit_gc0_register($11, 7, val) -#define read_gc0_cvmmemctl2() __read_64bit_gc0_register(16, 6) -#define write_gc0_cvmmemctl2(val) __write_64bit_gc0_register(16, 6, val) +#define read_gc0_cvmmemctl2() __read_64bit_gc0_register($16, 6) +#define write_gc0_cvmmemctl2(val) __write_64bit_gc0_register($16, 6, val) /* * Macros to access the floating point coprocessor control registers -- git-series 0.9.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 3/7] MIPS: VZ: Pass GC0 register names in $n format @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle From: James Hogan <jhogan@kernel.org> Now that we are using assembler macros to implement VZ instructions on toolchains which don't support them, pass VZ guest Cop0 register names to the __{read,write}_{32bit,ulong,64bit}_gc0_register macros in $n format rather than register numbers. This is to make them consistent with the normal root Cop0 register access macros which they were originally based on. Signed-off-by: James Hogan <jhogan@kernel.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: linux-mips@linux-mips.org --- arch/mips/include/asm/mipsregs.h | 376 ++++++++++++++++---------------- 1 file changed, 188 insertions(+), 188 deletions(-) diff --git a/arch/mips/include/asm/mipsregs.h b/arch/mips/include/asm/mipsregs.h index 361c35243366..000530b18f2e 100644 --- a/arch/mips/include/asm/mipsregs.h +++ b/arch/mips/include/asm/mipsregs.h @@ -1947,10 +1947,10 @@ _ASM_MACRO_0(tlbginvf, _ASM_INSN_IF_MIPS(0x4200000c) ".set\tpush\n\t" \ ".set\tmips32r2\n\t" \ _ASM_SET_VIRT \ - "mfgc0\t%0, $%1, %2\n\t" \ + "mfgc0\t%0, " #source ", %1\n\t" \ ".set\tpop" \ : "=r" (__res) \ - : "i" (source), "i" (sel)); \ + : "i" (sel)); \ __res; \ }) @@ -1960,10 +1960,10 @@ _ASM_MACRO_0(tlbginvf, _ASM_INSN_IF_MIPS(0x4200000c) ".set\tpush\n\t" \ ".set\tmips64r2\n\t" \ _ASM_SET_VIRT \ - "dmfgc0\t%0, $%1, %2\n\t" \ + "dmfgc0\t%0, " #source ", %1\n\t" \ ".set\tpop" \ : "=r" (__res) \ - : "i" (source), "i" (sel)); \ + : "i" (sel)); \ __res; \ }) @@ -1973,10 +1973,10 @@ do { \ ".set\tpush\n\t" \ ".set\tmips32r2\n\t" \ _ASM_SET_VIRT \ - "mtgc0\t%z0, $%1, %2\n\t" \ + "mtgc0\t%z0, " #register ", %1\n\t" \ ".set\tpop" \ : : "Jr" ((unsigned int)(value)), \ - "i" (register), "i" (sel)); \ + "i" (sel)); \ } while (0) #define __write_64bit_gc0_register(register, sel, value) \ @@ -1985,10 +1985,10 @@ do { \ ".set\tpush\n\t" \ ".set\tmips64r2\n\t" \ _ASM_SET_VIRT \ - "dmtgc0\t%z0, $%1, %2\n\t" \ + "dmtgc0\t%z0, " #register ", %1\n\t" \ ".set\tpop" \ : : "Jr" (value), \ - "i" (register), "i" (sel)); \ + "i" (sel)); \ } while (0) #define __read_ulong_gc0_register(reg, sel) \ @@ -2004,207 +2004,207 @@ do { \ __write_64bit_gc0_register(reg, sel, val); \ } while (0) -#define read_gc0_index() __read_32bit_gc0_register(0, 0) -#define write_gc0_index(val) __write_32bit_gc0_register(0, 0, val) +#define read_gc0_index() __read_32bit_gc0_register($0, 0) +#define write_gc0_index(val) __write_32bit_gc0_register($0, 0, val) -#define read_gc0_entrylo0() __read_ulong_gc0_register(2, 0) -#define write_gc0_entrylo0(val) __write_ulong_gc0_register(2, 0, val) +#define read_gc0_entrylo0() __read_ulong_gc0_register($2, 0) +#define write_gc0_entrylo0(val) __write_ulong_gc0_register($2, 0, val) -#define read_gc0_entrylo1() __read_ulong_gc0_register(3, 0) -#define write_gc0_entrylo1(val) __write_ulong_gc0_register(3, 0, val) +#define read_gc0_entrylo1() __read_ulong_gc0_register($3, 0) +#define write_gc0_entrylo1(val) __write_ulong_gc0_register($3, 0, val) -#define read_gc0_context() __read_ulong_gc0_register(4, 0) -#define write_gc0_context(val) __write_ulong_gc0_register(4, 0, val) +#define read_gc0_context() __read_ulong_gc0_register($4, 0) +#define write_gc0_context(val) __write_ulong_gc0_register($4, 0, val) -#define read_gc0_contextconfig() __read_32bit_gc0_register(4, 1) -#define write_gc0_contextconfig(val) __write_32bit_gc0_register(4, 1, val) +#define read_gc0_contextconfig() __read_32bit_gc0_register($4, 1) +#define write_gc0_contextconfig(val) __write_32bit_gc0_register($4, 1, val) -#define read_gc0_userlocal() __read_ulong_gc0_register(4, 2) -#define write_gc0_userlocal(val) __write_ulong_gc0_register(4, 2, val) +#define read_gc0_userlocal() __read_ulong_gc0_register($4, 2) +#define write_gc0_userlocal(val) __write_ulong_gc0_register($4, 2, val) -#define read_gc0_xcontextconfig() __read_ulong_gc0_register(4, 3) -#define write_gc0_xcontextconfig(val) __write_ulong_gc0_register(4, 3, val) +#define read_gc0_xcontextconfig() __read_ulong_gc0_register($4, 3) +#define write_gc0_xcontextconfig(val) __write_ulong_gc0_register($4, 3, val) -#define read_gc0_pagemask() __read_32bit_gc0_register(5, 0) -#define write_gc0_pagemask(val) __write_32bit_gc0_register(5, 0, val) +#define read_gc0_pagemask() __read_32bit_gc0_register($5, 0) +#define write_gc0_pagemask(val) __write_32bit_gc0_register($5, 0, val) -#define read_gc0_pagegrain() __read_32bit_gc0_register(5, 1) -#define write_gc0_pagegrain(val) __write_32bit_gc0_register(5, 1, val) +#define read_gc0_pagegrain() __read_32bit_gc0_register($5, 1) +#define write_gc0_pagegrain(val) __write_32bit_gc0_register($5, 1, val) -#define read_gc0_segctl0() __read_ulong_gc0_register(5, 2) -#define write_gc0_segctl0(val) __write_ulong_gc0_register(5, 2, val) +#define read_gc0_segctl0() __read_ulong_gc0_register($5, 2) +#define write_gc0_segctl0(val) __write_ulong_gc0_register($5, 2, val) -#define read_gc0_segctl1() __read_ulong_gc0_register(5, 3) -#define write_gc0_segctl1(val) __write_ulong_gc0_register(5, 3, val) +#define read_gc0_segctl1() __read_ulong_gc0_register($5, 3) +#define write_gc0_segctl1(val) __write_ulong_gc0_register($5, 3, val) -#define read_gc0_segctl2() __read_ulong_gc0_register(5, 4) -#define write_gc0_segctl2(val) __write_ulong_gc0_register(5, 4, val) +#define read_gc0_segctl2() __read_ulong_gc0_register($5, 4) +#define write_gc0_segctl2(val) __write_ulong_gc0_register($5, 4, val) -#define read_gc0_pwbase() __read_ulong_gc0_register(5, 5) -#define write_gc0_pwbase(val) __write_ulong_gc0_register(5, 5, val) +#define read_gc0_pwbase() __read_ulong_gc0_register($5, 5) +#define write_gc0_pwbase(val) __write_ulong_gc0_register($5, 5, val) -#define read_gc0_pwfield() __read_ulong_gc0_register(5, 6) -#define write_gc0_pwfield(val) __write_ulong_gc0_register(5, 6, val) +#define read_gc0_pwfield() __read_ulong_gc0_register($5, 6) +#define write_gc0_pwfield(val) __write_ulong_gc0_register($5, 6, val) -#define read_gc0_pwsize() __read_ulong_gc0_register(5, 7) -#define write_gc0_pwsize(val) __write_ulong_gc0_register(5, 7, val) +#define read_gc0_pwsize() __read_ulong_gc0_register($5, 7) +#define write_gc0_pwsize(val) __write_ulong_gc0_register($5, 7, val) -#define read_gc0_wired() __read_32bit_gc0_register(6, 0) -#define write_gc0_wired(val) __write_32bit_gc0_register(6, 0, val) +#define read_gc0_wired() __read_32bit_gc0_register($6, 0) +#define write_gc0_wired(val) __write_32bit_gc0_register($6, 0, val) -#define read_gc0_pwctl() __read_32bit_gc0_register(6, 6) -#define write_gc0_pwctl(val) __write_32bit_gc0_register(6, 6, val) - -#define read_gc0_hwrena() __read_32bit_gc0_register(7, 0) -#define write_gc0_hwrena(val) __write_32bit_gc0_register(7, 0, val) - -#define read_gc0_badvaddr() __read_ulong_gc0_register(8, 0) -#define write_gc0_badvaddr(val) __write_ulong_gc0_register(8, 0, val) - -#define read_gc0_badinstr() __read_32bit_gc0_register(8, 1) -#define write_gc0_badinstr(val) __write_32bit_gc0_register(8, 1, val) - -#define read_gc0_badinstrp() __read_32bit_gc0_register(8, 2) -#define write_gc0_badinstrp(val) __write_32bit_gc0_register(8, 2, val) - -#define read_gc0_count() __read_32bit_gc0_register(9, 0) - -#define read_gc0_entryhi() __read_ulong_gc0_register(10, 0) -#define write_gc0_entryhi(val) __write_ulong_gc0_register(10, 0, val) - -#define read_gc0_compare() __read_32bit_gc0_register(11, 0) -#define write_gc0_compare(val) __write_32bit_gc0_register(11, 0, val) - -#define read_gc0_status() __read_32bit_gc0_register(12, 0) -#define write_gc0_status(val) __write_32bit_gc0_register(12, 0, val) - -#define read_gc0_intctl() __read_32bit_gc0_register(12, 1) -#define write_gc0_intctl(val) __write_32bit_gc0_register(12, 1, val) - -#define read_gc0_cause() __read_32bit_gc0_register(13, 0) -#define write_gc0_cause(val) __write_32bit_gc0_register(13, 0, val) - -#define read_gc0_epc() __read_ulong_gc0_register(14, 0) -#define write_gc0_epc(val) __write_ulong_gc0_register(14, 0, val) - -#define read_gc0_prid() __read_32bit_gc0_register(15, 0) - -#define read_gc0_ebase() __read_32bit_gc0_register(15, 1) -#define write_gc0_ebase(val) __write_32bit_gc0_register(15, 1, val) - -#define read_gc0_ebase_64() __read_64bit_gc0_register(15, 1) -#define write_gc0_ebase_64(val) __write_64bit_gc0_register(15, 1, val) - -#define read_gc0_config() __read_32bit_gc0_register(16, 0) -#define read_gc0_config1() __read_32bit_gc0_register(16, 1) -#define read_gc0_config2() __read_32bit_gc0_register(16, 2) -#define read_gc0_config3() __read_32bit_gc0_register(16, 3) -#define read_gc0_config4() __read_32bit_gc0_register(16, 4) -#define read_gc0_config5() __read_32bit_gc0_register(16, 5) -#define read_gc0_config6() __read_32bit_gc0_register(16, 6) -#define read_gc0_config7() __read_32bit_gc0_register(16, 7) -#define write_gc0_config(val) __write_32bit_gc0_register(16, 0, val) -#define write_gc0_config1(val) __write_32bit_gc0_register(16, 1, val) -#define write_gc0_config2(val) __write_32bit_gc0_register(16, 2, val) -#define write_gc0_config3(val) __write_32bit_gc0_register(16, 3, val) -#define write_gc0_config4(val) __write_32bit_gc0_register(16, 4, val) -#define write_gc0_config5(val) __write_32bit_gc0_register(16, 5, val) -#define write_gc0_config6(val) __write_32bit_gc0_register(16, 6, val) -#define write_gc0_config7(val) __write_32bit_gc0_register(16, 7, val) - -#define read_gc0_lladdr() __read_ulong_gc0_register(17, 0) -#define write_gc0_lladdr(val) __write_ulong_gc0_register(17, 0, val) - -#define read_gc0_watchlo0() __read_ulong_gc0_register(18, 0) -#define read_gc0_watchlo1() __read_ulong_gc0_register(18, 1) -#define read_gc0_watchlo2() __read_ulong_gc0_register(18, 2) -#define read_gc0_watchlo3() __read_ulong_gc0_register(18, 3) -#define read_gc0_watchlo4() __read_ulong_gc0_register(18, 4) -#define read_gc0_watchlo5() __read_ulong_gc0_register(18, 5) -#define read_gc0_watchlo6() __read_ulong_gc0_register(18, 6) -#define read_gc0_watchlo7() __read_ulong_gc0_register(18, 7) -#define write_gc0_watchlo0(val) __write_ulong_gc0_register(18, 0, val) -#define write_gc0_watchlo1(val) __write_ulong_gc0_register(18, 1, val) -#define write_gc0_watchlo2(val) __write_ulong_gc0_register(18, 2, val) -#define write_gc0_watchlo3(val) __write_ulong_gc0_register(18, 3, val) -#define write_gc0_watchlo4(val) __write_ulong_gc0_register(18, 4, val) -#define write_gc0_watchlo5(val) __write_ulong_gc0_register(18, 5, val) -#define write_gc0_watchlo6(val) __write_ulong_gc0_register(18, 6, val) -#define write_gc0_watchlo7(val) __write_ulong_gc0_register(18, 7, val) - -#define read_gc0_watchhi0() __read_32bit_gc0_register(19, 0) -#define read_gc0_watchhi1() __read_32bit_gc0_register(19, 1) -#define read_gc0_watchhi2() __read_32bit_gc0_register(19, 2) -#define read_gc0_watchhi3() __read_32bit_gc0_register(19, 3) -#define read_gc0_watchhi4() __read_32bit_gc0_register(19, 4) -#define read_gc0_watchhi5() __read_32bit_gc0_register(19, 5) -#define read_gc0_watchhi6() __read_32bit_gc0_register(19, 6) -#define read_gc0_watchhi7() __read_32bit_gc0_register(19, 7) -#define write_gc0_watchhi0(val) __write_32bit_gc0_register(19, 0, val) -#define write_gc0_watchhi1(val) __write_32bit_gc0_register(19, 1, val) -#define write_gc0_watchhi2(val) __write_32bit_gc0_register(19, 2, val) -#define write_gc0_watchhi3(val) __write_32bit_gc0_register(19, 3, val) -#define write_gc0_watchhi4(val) __write_32bit_gc0_register(19, 4, val) -#define write_gc0_watchhi5(val) __write_32bit_gc0_register(19, 5, val) -#define write_gc0_watchhi6(val) __write_32bit_gc0_register(19, 6, val) -#define write_gc0_watchhi7(val) __write_32bit_gc0_register(19, 7, val) - -#define read_gc0_xcontext() __read_ulong_gc0_register(20, 0) -#define write_gc0_xcontext(val) __write_ulong_gc0_register(20, 0, val) - -#define read_gc0_perfctrl0() __read_32bit_gc0_register(25, 0) -#define write_gc0_perfctrl0(val) __write_32bit_gc0_register(25, 0, val) -#define read_gc0_perfcntr0() __read_32bit_gc0_register(25, 1) -#define write_gc0_perfcntr0(val) __write_32bit_gc0_register(25, 1, val) -#define read_gc0_perfcntr0_64() __read_64bit_gc0_register(25, 1) -#define write_gc0_perfcntr0_64(val) __write_64bit_gc0_register(25, 1, val) -#define read_gc0_perfctrl1() __read_32bit_gc0_register(25, 2) -#define write_gc0_perfctrl1(val) __write_32bit_gc0_register(25, 2, val) -#define read_gc0_perfcntr1() __read_32bit_gc0_register(25, 3) -#define write_gc0_perfcntr1(val) __write_32bit_gc0_register(25, 3, val) -#define read_gc0_perfcntr1_64() __read_64bit_gc0_register(25, 3) -#define write_gc0_perfcntr1_64(val) __write_64bit_gc0_register(25, 3, val) -#define read_gc0_perfctrl2() __read_32bit_gc0_register(25, 4) -#define write_gc0_perfctrl2(val) __write_32bit_gc0_register(25, 4, val) -#define read_gc0_perfcntr2() __read_32bit_gc0_register(25, 5) -#define write_gc0_perfcntr2(val) __write_32bit_gc0_register(25, 5, val) -#define read_gc0_perfcntr2_64() __read_64bit_gc0_register(25, 5) -#define write_gc0_perfcntr2_64(val) __write_64bit_gc0_register(25, 5, val) -#define read_gc0_perfctrl3() __read_32bit_gc0_register(25, 6) -#define write_gc0_perfctrl3(val) __write_32bit_gc0_register(25, 6, val) -#define read_gc0_perfcntr3() __read_32bit_gc0_register(25, 7) -#define write_gc0_perfcntr3(val) __write_32bit_gc0_register(25, 7, val) -#define read_gc0_perfcntr3_64() __read_64bit_gc0_register(25, 7) -#define write_gc0_perfcntr3_64(val) __write_64bit_gc0_register(25, 7, val) - -#define read_gc0_errorepc() __read_ulong_gc0_register(30, 0) -#define write_gc0_errorepc(val) __write_ulong_gc0_register(30, 0, val) - -#define read_gc0_kscratch1() __read_ulong_gc0_register(31, 2) -#define read_gc0_kscratch2() __read_ulong_gc0_register(31, 3) -#define read_gc0_kscratch3() __read_ulong_gc0_register(31, 4) -#define read_gc0_kscratch4() __read_ulong_gc0_register(31, 5) -#define read_gc0_kscratch5() __read_ulong_gc0_register(31, 6) -#define read_gc0_kscratch6() __read_ulong_gc0_register(31, 7) -#define write_gc0_kscratch1(val) __write_ulong_gc0_register(31, 2, val) -#define write_gc0_kscratch2(val) __write_ulong_gc0_register(31, 3, val) -#define write_gc0_kscratch3(val) __write_ulong_gc0_register(31, 4, val) -#define write_gc0_kscratch4(val) __write_ulong_gc0_register(31, 5, val) -#define write_gc0_kscratch5(val) __write_ulong_gc0_register(31, 6, val) -#define write_gc0_kscratch6(val) __write_ulong_gc0_register(31, 7, val) +#define read_gc0_pwctl() __read_32bit_gc0_register($6, 6) +#define write_gc0_pwctl(val) __write_32bit_gc0_register($6, 6, val) + +#define read_gc0_hwrena() __read_32bit_gc0_register($7, 0) +#define write_gc0_hwrena(val) __write_32bit_gc0_register($7, 0, val) + +#define read_gc0_badvaddr() __read_ulong_gc0_register($8, 0) +#define write_gc0_badvaddr(val) __write_ulong_gc0_register($8, 0, val) + +#define read_gc0_badinstr() __read_32bit_gc0_register($8, 1) +#define write_gc0_badinstr(val) __write_32bit_gc0_register($8, 1, val) + +#define read_gc0_badinstrp() __read_32bit_gc0_register($8, 2) +#define write_gc0_badinstrp(val) __write_32bit_gc0_register($8, 2, val) + +#define read_gc0_count() __read_32bit_gc0_register($9, 0) + +#define read_gc0_entryhi() __read_ulong_gc0_register($10, 0) +#define write_gc0_entryhi(val) __write_ulong_gc0_register($10, 0, val) + +#define read_gc0_compare() __read_32bit_gc0_register($11, 0) +#define write_gc0_compare(val) __write_32bit_gc0_register($11, 0, val) + +#define read_gc0_status() __read_32bit_gc0_register($12, 0) +#define write_gc0_status(val) __write_32bit_gc0_register($12, 0, val) + +#define read_gc0_intctl() __read_32bit_gc0_register($12, 1) +#define write_gc0_intctl(val) __write_32bit_gc0_register($12, 1, val) + +#define read_gc0_cause() __read_32bit_gc0_register($13, 0) +#define write_gc0_cause(val) __write_32bit_gc0_register($13, 0, val) + +#define read_gc0_epc() __read_ulong_gc0_register($14, 0) +#define write_gc0_epc(val) __write_ulong_gc0_register($14, 0, val) + +#define read_gc0_prid() __read_32bit_gc0_register($15, 0) + +#define read_gc0_ebase() __read_32bit_gc0_register($15, 1) +#define write_gc0_ebase(val) __write_32bit_gc0_register($15, 1, val) + +#define read_gc0_ebase_64() __read_64bit_gc0_register($15, 1) +#define write_gc0_ebase_64(val) __write_64bit_gc0_register($15, 1, val) + +#define read_gc0_config() __read_32bit_gc0_register($16, 0) +#define read_gc0_config1() __read_32bit_gc0_register($16, 1) +#define read_gc0_config2() __read_32bit_gc0_register($16, 2) +#define read_gc0_config3() __read_32bit_gc0_register($16, 3) +#define read_gc0_config4() __read_32bit_gc0_register($16, 4) +#define read_gc0_config5() __read_32bit_gc0_register($16, 5) +#define read_gc0_config6() __read_32bit_gc0_register($16, 6) +#define read_gc0_config7() __read_32bit_gc0_register($16, 7) +#define write_gc0_config(val) __write_32bit_gc0_register($16, 0, val) +#define write_gc0_config1(val) __write_32bit_gc0_register($16, 1, val) +#define write_gc0_config2(val) __write_32bit_gc0_register($16, 2, val) +#define write_gc0_config3(val) __write_32bit_gc0_register($16, 3, val) +#define write_gc0_config4(val) __write_32bit_gc0_register($16, 4, val) +#define write_gc0_config5(val) __write_32bit_gc0_register($16, 5, val) +#define write_gc0_config6(val) __write_32bit_gc0_register($16, 6, val) +#define write_gc0_config7(val) __write_32bit_gc0_register($16, 7, val) + +#define read_gc0_lladdr() __read_ulong_gc0_register($17, 0) +#define write_gc0_lladdr(val) __write_ulong_gc0_register($17, 0, val) + +#define read_gc0_watchlo0() __read_ulong_gc0_register($18, 0) +#define read_gc0_watchlo1() __read_ulong_gc0_register($18, 1) +#define read_gc0_watchlo2() __read_ulong_gc0_register($18, 2) +#define read_gc0_watchlo3() __read_ulong_gc0_register($18, 3) +#define read_gc0_watchlo4() __read_ulong_gc0_register($18, 4) +#define read_gc0_watchlo5() __read_ulong_gc0_register($18, 5) +#define read_gc0_watchlo6() __read_ulong_gc0_register($18, 6) +#define read_gc0_watchlo7() __read_ulong_gc0_register($18, 7) +#define write_gc0_watchlo0(val) __write_ulong_gc0_register($18, 0, val) +#define write_gc0_watchlo1(val) __write_ulong_gc0_register($18, 1, val) +#define write_gc0_watchlo2(val) __write_ulong_gc0_register($18, 2, val) +#define write_gc0_watchlo3(val) __write_ulong_gc0_register($18, 3, val) +#define write_gc0_watchlo4(val) __write_ulong_gc0_register($18, 4, val) +#define write_gc0_watchlo5(val) __write_ulong_gc0_register($18, 5, val) +#define write_gc0_watchlo6(val) __write_ulong_gc0_register($18, 6, val) +#define write_gc0_watchlo7(val) __write_ulong_gc0_register($18, 7, val) + +#define read_gc0_watchhi0() __read_32bit_gc0_register($19, 0) +#define read_gc0_watchhi1() __read_32bit_gc0_register($19, 1) +#define read_gc0_watchhi2() __read_32bit_gc0_register($19, 2) +#define read_gc0_watchhi3() __read_32bit_gc0_register($19, 3) +#define read_gc0_watchhi4() __read_32bit_gc0_register($19, 4) +#define read_gc0_watchhi5() __read_32bit_gc0_register($19, 5) +#define read_gc0_watchhi6() __read_32bit_gc0_register($19, 6) +#define read_gc0_watchhi7() __read_32bit_gc0_register($19, 7) +#define write_gc0_watchhi0(val) __write_32bit_gc0_register($19, 0, val) +#define write_gc0_watchhi1(val) __write_32bit_gc0_register($19, 1, val) +#define write_gc0_watchhi2(val) __write_32bit_gc0_register($19, 2, val) +#define write_gc0_watchhi3(val) __write_32bit_gc0_register($19, 3, val) +#define write_gc0_watchhi4(val) __write_32bit_gc0_register($19, 4, val) +#define write_gc0_watchhi5(val) __write_32bit_gc0_register($19, 5, val) +#define write_gc0_watchhi6(val) __write_32bit_gc0_register($19, 6, val) +#define write_gc0_watchhi7(val) __write_32bit_gc0_register($19, 7, val) + +#define read_gc0_xcontext() __read_ulong_gc0_register($20, 0) +#define write_gc0_xcontext(val) __write_ulong_gc0_register($20, 0, val) + +#define read_gc0_perfctrl0() __read_32bit_gc0_register($25, 0) +#define write_gc0_perfctrl0(val) __write_32bit_gc0_register($25, 0, val) +#define read_gc0_perfcntr0() __read_32bit_gc0_register($25, 1) +#define write_gc0_perfcntr0(val) __write_32bit_gc0_register($25, 1, val) +#define read_gc0_perfcntr0_64() __read_64bit_gc0_register($25, 1) +#define write_gc0_perfcntr0_64(val) __write_64bit_gc0_register($25, 1, val) +#define read_gc0_perfctrl1() __read_32bit_gc0_register($25, 2) +#define write_gc0_perfctrl1(val) __write_32bit_gc0_register($25, 2, val) +#define read_gc0_perfcntr1() __read_32bit_gc0_register($25, 3) +#define write_gc0_perfcntr1(val) __write_32bit_gc0_register($25, 3, val) +#define read_gc0_perfcntr1_64() __read_64bit_gc0_register($25, 3) +#define write_gc0_perfcntr1_64(val) __write_64bit_gc0_register($25, 3, val) +#define read_gc0_perfctrl2() __read_32bit_gc0_register($25, 4) +#define write_gc0_perfctrl2(val) __write_32bit_gc0_register($25, 4, val) +#define read_gc0_perfcntr2() __read_32bit_gc0_register($25, 5) +#define write_gc0_perfcntr2(val) __write_32bit_gc0_register($25, 5, val) +#define read_gc0_perfcntr2_64() __read_64bit_gc0_register($25, 5) +#define write_gc0_perfcntr2_64(val) __write_64bit_gc0_register($25, 5, val) +#define read_gc0_perfctrl3() __read_32bit_gc0_register($25, 6) +#define write_gc0_perfctrl3(val) __write_32bit_gc0_register($25, 6, val) +#define read_gc0_perfcntr3() __read_32bit_gc0_register($25, 7) +#define write_gc0_perfcntr3(val) __write_32bit_gc0_register($25, 7, val) +#define read_gc0_perfcntr3_64() __read_64bit_gc0_register($25, 7) +#define write_gc0_perfcntr3_64(val) __write_64bit_gc0_register($25, 7, val) + +#define read_gc0_errorepc() __read_ulong_gc0_register($30, 0) +#define write_gc0_errorepc(val) __write_ulong_gc0_register($30, 0, val) + +#define read_gc0_kscratch1() __read_ulong_gc0_register($31, 2) +#define read_gc0_kscratch2() __read_ulong_gc0_register($31, 3) +#define read_gc0_kscratch3() __read_ulong_gc0_register($31, 4) +#define read_gc0_kscratch4() __read_ulong_gc0_register($31, 5) +#define read_gc0_kscratch5() __read_ulong_gc0_register($31, 6) +#define read_gc0_kscratch6() __read_ulong_gc0_register($31, 7) +#define write_gc0_kscratch1(val) __write_ulong_gc0_register($31, 2, val) +#define write_gc0_kscratch2(val) __write_ulong_gc0_register($31, 3, val) +#define write_gc0_kscratch3(val) __write_ulong_gc0_register($31, 4, val) +#define write_gc0_kscratch4(val) __write_ulong_gc0_register($31, 5, val) +#define write_gc0_kscratch5(val) __write_ulong_gc0_register($31, 6, val) +#define write_gc0_kscratch6(val) __write_ulong_gc0_register($31, 7, val) /* Cavium OCTEON (cnMIPS) */ -#define read_gc0_cvmcount() __read_ulong_gc0_register(9, 6) -#define write_gc0_cvmcount(val) __write_ulong_gc0_register(9, 6, val) +#define read_gc0_cvmcount() __read_ulong_gc0_register($9, 6) +#define write_gc0_cvmcount(val) __write_ulong_gc0_register($9, 6, val) -#define read_gc0_cvmctl() __read_64bit_gc0_register(9, 7) -#define write_gc0_cvmctl(val) __write_64bit_gc0_register(9, 7, val) +#define read_gc0_cvmctl() __read_64bit_gc0_register($9, 7) +#define write_gc0_cvmctl(val) __write_64bit_gc0_register($9, 7, val) -#define read_gc0_cvmmemctl() __read_64bit_gc0_register(11, 7) -#define write_gc0_cvmmemctl(val) __write_64bit_gc0_register(11, 7, val) +#define read_gc0_cvmmemctl() __read_64bit_gc0_register($11, 7) +#define write_gc0_cvmmemctl(val) __write_64bit_gc0_register($11, 7, val) -#define read_gc0_cvmmemctl2() __read_64bit_gc0_register(16, 6) -#define write_gc0_cvmmemctl2(val) __write_64bit_gc0_register(16, 6, val) +#define read_gc0_cvmmemctl2() __read_64bit_gc0_register($16, 6) +#define write_gc0_cvmmemctl2(val) __write_64bit_gc0_register($16, 6, val) /* * Macros to access the floating point coprocessor control registers -- git-series 0.9.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 4/7] MIPS: XPA: Use XPA instructions in assembly @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle From: James Hogan <jhogan@kernel.org> Utilise XPA instructions MFHC0 & MTHC0 in inline assembly instead of directly encoding them with the _ASM_INSN* macros, and transparently implement these instructions as assembler macros if the toolchain doesn't support them natively, using the recently introduced assembler macro helpers. The old direct encodings were restricted to using the register $at, so this allows the extra register moves to go away (saving a grand total of 24 bytes). Signed-off-by: James Hogan <jhogan@kernel.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: linux-mips@linux-mips.org --- arch/mips/Makefile | 6 ++++++ arch/mips/include/asm/mipsregs.h | 26 ++++++++++++++++---------- 2 files changed, 22 insertions(+), 10 deletions(-) diff --git a/arch/mips/Makefile b/arch/mips/Makefile index 9f6a26d72f9f..cbd560aeb5a6 100644 --- a/arch/mips/Makefile +++ b/arch/mips/Makefile @@ -216,6 +216,12 @@ cflags-$(toolchain-msa) += -DTOOLCHAIN_SUPPORTS_MSA endif toolchain-virt := $(call cc-option-yn,$(mips-cflags) -mvirt) cflags-$(toolchain-virt) += -DTOOLCHAIN_SUPPORTS_VIRT +# For -mmicromips, use -Wa,-fatal-warnings to catch unsupported -mxpa which +# only warns +xpa-cflags-y := $(mips-cflags) +xpa-cflags-$(micromips-ase) += -mmicromips -Wa$(comma)-fatal-warnings +toolchain-xpa := $(call cc-option-yn,$(xpa-cflags-y) -mxpa) +cflags-$(toolchain-xpa) += -DTOOLCHAIN_SUPPORTS_XPA # # Firmware support diff --git a/arch/mips/include/asm/mipsregs.h b/arch/mips/include/asm/mipsregs.h index 000530b18f2e..02f8ae1fac04 100644 --- a/arch/mips/include/asm/mipsregs.h +++ b/arch/mips/include/asm/mipsregs.h @@ -1487,18 +1487,27 @@ do { \ local_irq_restore(__flags); \ } while (0) +#ifndef TOOLCHAIN_SUPPORTS_XPA +_ASM_MACRO_2R_1S(mfhc0, rt, rs, sel, + _ASM_INSN_IF_MIPS(0x40400000 | __rt << 16 | __rs << 11 | \\sel) + _ASM_INSN32_IF_MM(0x000000f4 | __rt << 21 | __rs << 16 | \\sel << 11)); +_ASM_MACRO_2R_1S(mthc0, rt, rd, sel, + _ASM_INSN_IF_MIPS(0x40c00000 | __rt << 16 | __rd << 11 | \\sel) + _ASM_INSN32_IF_MM(0x000002f4 | __rt << 21 | __rd << 16 | \\sel << 11)); +#define _ASM_SET_XPA "" +#else /* !TOOLCHAIN_SUPPORTS_XPA */ +#define _ASM_SET_XPA ".set\txpa\n\t" +#endif + #define __readx_32bit_c0_register(source) \ ({ \ unsigned int __res; \ \ __asm__ __volatile__( \ " .set push \n" \ - " .set noat \n" \ " .set mips32r2 \n" \ - " # mfhc0 $1, %1 \n" \ - _ASM_INSN_IF_MIPS(0x40410000 | ((%1 & 0x1f) << 11)) \ - _ASM_INSN32_IF_MM(0x002000f4 | ((%1 & 0x1f) << 16)) \ - " move %0, $1 \n" \ + _ASM_SET_XPA \ + " mfhc0 %0, $%1 \n" \ " .set pop \n" \ : "=r" (__res) \ : "i" (source)); \ @@ -1509,12 +1518,9 @@ do { \ do { \ __asm__ __volatile__( \ " .set push \n" \ - " .set noat \n" \ " .set mips32r2 \n" \ - " move $1, %0 \n" \ - " # mthc0 $1, %1 \n" \ - _ASM_INSN_IF_MIPS(0x40c10000 | ((%1 & 0x1f) << 11)) \ - _ASM_INSN32_IF_MM(0x002002f4 | ((%1 & 0x1f) << 16)) \ + _ASM_SET_XPA \ + " mthc0 %0, $%1 \n" \ " .set pop \n" \ : \ : "r" (value), "i" (register)); \ -- git-series 0.9.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 4/7] MIPS: XPA: Use XPA instructions in assembly @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle From: James Hogan <jhogan@kernel.org> Utilise XPA instructions MFHC0 & MTHC0 in inline assembly instead of directly encoding them with the _ASM_INSN* macros, and transparently implement these instructions as assembler macros if the toolchain doesn't support them natively, using the recently introduced assembler macro helpers. The old direct encodings were restricted to using the register $at, so this allows the extra register moves to go away (saving a grand total of 24 bytes). Signed-off-by: James Hogan <jhogan@kernel.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: linux-mips@linux-mips.org --- arch/mips/Makefile | 6 ++++++ arch/mips/include/asm/mipsregs.h | 26 ++++++++++++++++---------- 2 files changed, 22 insertions(+), 10 deletions(-) diff --git a/arch/mips/Makefile b/arch/mips/Makefile index 9f6a26d72f9f..cbd560aeb5a6 100644 --- a/arch/mips/Makefile +++ b/arch/mips/Makefile @@ -216,6 +216,12 @@ cflags-$(toolchain-msa) += -DTOOLCHAIN_SUPPORTS_MSA endif toolchain-virt := $(call cc-option-yn,$(mips-cflags) -mvirt) cflags-$(toolchain-virt) += -DTOOLCHAIN_SUPPORTS_VIRT +# For -mmicromips, use -Wa,-fatal-warnings to catch unsupported -mxpa which +# only warns +xpa-cflags-y := $(mips-cflags) +xpa-cflags-$(micromips-ase) += -mmicromips -Wa$(comma)-fatal-warnings +toolchain-xpa := $(call cc-option-yn,$(xpa-cflags-y) -mxpa) +cflags-$(toolchain-xpa) += -DTOOLCHAIN_SUPPORTS_XPA # # Firmware support diff --git a/arch/mips/include/asm/mipsregs.h b/arch/mips/include/asm/mipsregs.h index 000530b18f2e..02f8ae1fac04 100644 --- a/arch/mips/include/asm/mipsregs.h +++ b/arch/mips/include/asm/mipsregs.h @@ -1487,18 +1487,27 @@ do { \ local_irq_restore(__flags); \ } while (0) +#ifndef TOOLCHAIN_SUPPORTS_XPA +_ASM_MACRO_2R_1S(mfhc0, rt, rs, sel, + _ASM_INSN_IF_MIPS(0x40400000 | __rt << 16 | __rs << 11 | \\sel) + _ASM_INSN32_IF_MM(0x000000f4 | __rt << 21 | __rs << 16 | \\sel << 11)); +_ASM_MACRO_2R_1S(mthc0, rt, rd, sel, + _ASM_INSN_IF_MIPS(0x40c00000 | __rt << 16 | __rd << 11 | \\sel) + _ASM_INSN32_IF_MM(0x000002f4 | __rt << 21 | __rd << 16 | \\sel << 11)); +#define _ASM_SET_XPA "" +#else /* !TOOLCHAIN_SUPPORTS_XPA */ +#define _ASM_SET_XPA ".set\txpa\n\t" +#endif + #define __readx_32bit_c0_register(source) \ ({ \ unsigned int __res; \ \ __asm__ __volatile__( \ " .set push \n" \ - " .set noat \n" \ " .set mips32r2 \n" \ - " # mfhc0 $1, %1 \n" \ - _ASM_INSN_IF_MIPS(0x40410000 | ((%1 & 0x1f) << 11)) \ - _ASM_INSN32_IF_MM(0x002000f4 | ((%1 & 0x1f) << 16)) \ - " move %0, $1 \n" \ + _ASM_SET_XPA \ + " mfhc0 %0, $%1 \n" \ " .set pop \n" \ : "=r" (__res) \ : "i" (source)); \ @@ -1509,12 +1518,9 @@ do { \ do { \ __asm__ __volatile__( \ " .set push \n" \ - " .set noat \n" \ " .set mips32r2 \n" \ - " move $1, %0 \n" \ - " # mthc0 $1, %1 \n" \ - _ASM_INSN_IF_MIPS(0x40c10000 | ((%1 & 0x1f) << 11)) \ - _ASM_INSN32_IF_MM(0x002002f4 | ((%1 & 0x1f) << 16)) \ + _ASM_SET_XPA \ + " mthc0 %0, $%1 \n" \ " .set pop \n" \ : \ : "r" (value), "i" (register)); \ -- git-series 0.9.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 5/7] MIPS: XPA: Allow use of $0 (zero) to MTHC0 @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle From: James Hogan <jhogan@kernel.org> Tweak __writex_32bit_c0_register() to allow the compiler to use $0 (the zero register) as an input to the mthc0 instruction. Signed-off-by: James Hogan <jhogan@kernel.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: linux-mips@linux-mips.org --- arch/mips/include/asm/mipsregs.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/mips/include/asm/mipsregs.h b/arch/mips/include/asm/mipsregs.h index 02f8ae1fac04..f2b9af887e83 100644 --- a/arch/mips/include/asm/mipsregs.h +++ b/arch/mips/include/asm/mipsregs.h @@ -1520,10 +1520,10 @@ do { \ " .set push \n" \ " .set mips32r2 \n" \ _ASM_SET_XPA \ - " mthc0 %0, $%1 \n" \ + " mthc0 %z0, $%1 \n" \ " .set pop \n" \ : \ - : "r" (value), "i" (register)); \ + : "Jr" (value), "i" (register)); \ } while (0) #define read_c0_index() __read_32bit_c0_register($0, 0) -- git-series 0.9.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 5/7] MIPS: XPA: Allow use of $0 (zero) to MTHC0 @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle From: James Hogan <jhogan@kernel.org> Tweak __writex_32bit_c0_register() to allow the compiler to use $0 (the zero register) as an input to the mthc0 instruction. Signed-off-by: James Hogan <jhogan@kernel.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: linux-mips@linux-mips.org --- arch/mips/include/asm/mipsregs.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/mips/include/asm/mipsregs.h b/arch/mips/include/asm/mipsregs.h index 02f8ae1fac04..f2b9af887e83 100644 --- a/arch/mips/include/asm/mipsregs.h +++ b/arch/mips/include/asm/mipsregs.h @@ -1520,10 +1520,10 @@ do { \ " .set push \n" \ " .set mips32r2 \n" \ _ASM_SET_XPA \ - " mthc0 %0, $%1 \n" \ + " mthc0 %z0, $%1 \n" \ " .set pop \n" \ : \ - : "r" (value), "i" (register)); \ + : "Jr" (value), "i" (register)); \ } while (0) #define read_c0_index() __read_32bit_c0_register($0, 0) -- git-series 0.9.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 6/7] MIPS: XPA: Standardise readx/writex accessors @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle From: James Hogan <jhogan@kernel.org> Now that we are using assembler macros to implement XPA instructions on toolchains which don't support them, pass Cop0 register names to the __{readx,writex}_32bit_c0_register macros in $n format rather than register numbers. Also pass a register select which may be useful in future (for example for MemoryMapID field of WatchHi registers on I6500). This is to make them consistent with the normal Cop0 register access macros which they were originally based on. Signed-off-by: James Hogan <jhogan@kernel.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: linux-mips@linux-mips.org --- arch/mips/include/asm/mipsregs.h | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-) diff --git a/arch/mips/include/asm/mipsregs.h b/arch/mips/include/asm/mipsregs.h index f2b9af887e83..5b901deab52a 100644 --- a/arch/mips/include/asm/mipsregs.h +++ b/arch/mips/include/asm/mipsregs.h @@ -1499,7 +1499,7 @@ _ASM_MACRO_2R_1S(mthc0, rt, rd, sel, #define _ASM_SET_XPA ".set\txpa\n\t" #endif -#define __readx_32bit_c0_register(source) \ +#define __readx_32bit_c0_register(source, sel) \ ({ \ unsigned int __res; \ \ @@ -1507,23 +1507,23 @@ _ASM_MACRO_2R_1S(mthc0, rt, rd, sel, " .set push \n" \ " .set mips32r2 \n" \ _ASM_SET_XPA \ - " mfhc0 %0, $%1 \n" \ + " mfhc0 %0, " #source ", %1 \n" \ " .set pop \n" \ : "=r" (__res) \ - : "i" (source)); \ + : "i" (sel)); \ __res; \ }) -#define __writex_32bit_c0_register(register, value) \ +#define __writex_32bit_c0_register(register, sel, value) \ do { \ __asm__ __volatile__( \ " .set push \n" \ " .set mips32r2 \n" \ _ASM_SET_XPA \ - " mthc0 %z0, $%1 \n" \ + " mthc0 %z0, " #register ", %1 \n" \ " .set pop \n" \ : \ - : "Jr" (value), "i" (register)); \ + : "Jr" (value), "i" (sel)); \ } while (0) #define read_c0_index() __read_32bit_c0_register($0, 0) @@ -1535,14 +1535,14 @@ do { \ #define read_c0_entrylo0() __read_ulong_c0_register($2, 0) #define write_c0_entrylo0(val) __write_ulong_c0_register($2, 0, val) -#define readx_c0_entrylo0() __readx_32bit_c0_register(2) -#define writex_c0_entrylo0(val) __writex_32bit_c0_register(2, val) +#define readx_c0_entrylo0() __readx_32bit_c0_register($2, 0) +#define writex_c0_entrylo0(val) __writex_32bit_c0_register($2, 0, val) #define read_c0_entrylo1() __read_ulong_c0_register($3, 0) #define write_c0_entrylo1(val) __write_ulong_c0_register($3, 0, val) -#define readx_c0_entrylo1() __readx_32bit_c0_register(3) -#define writex_c0_entrylo1(val) __writex_32bit_c0_register(3, val) +#define readx_c0_entrylo1() __readx_32bit_c0_register($3, 0) +#define writex_c0_entrylo1(val) __writex_32bit_c0_register($3, 0, val) #define read_c0_conf() __read_32bit_c0_register($3, 0) #define write_c0_conf(val) __write_32bit_c0_register($3, 0, val) -- git-series 0.9.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 6/7] MIPS: XPA: Standardise readx/writex accessors @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle From: James Hogan <jhogan@kernel.org> Now that we are using assembler macros to implement XPA instructions on toolchains which don't support them, pass Cop0 register names to the __{readx,writex}_32bit_c0_register macros in $n format rather than register numbers. Also pass a register select which may be useful in future (for example for MemoryMapID field of WatchHi registers on I6500). This is to make them consistent with the normal Cop0 register access macros which they were originally based on. Signed-off-by: James Hogan <jhogan@kernel.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: linux-mips@linux-mips.org --- arch/mips/include/asm/mipsregs.h | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-) diff --git a/arch/mips/include/asm/mipsregs.h b/arch/mips/include/asm/mipsregs.h index f2b9af887e83..5b901deab52a 100644 --- a/arch/mips/include/asm/mipsregs.h +++ b/arch/mips/include/asm/mipsregs.h @@ -1499,7 +1499,7 @@ _ASM_MACRO_2R_1S(mthc0, rt, rd, sel, #define _ASM_SET_XPA ".set\txpa\n\t" #endif -#define __readx_32bit_c0_register(source) \ +#define __readx_32bit_c0_register(source, sel) \ ({ \ unsigned int __res; \ \ @@ -1507,23 +1507,23 @@ _ASM_MACRO_2R_1S(mthc0, rt, rd, sel, " .set push \n" \ " .set mips32r2 \n" \ _ASM_SET_XPA \ - " mfhc0 %0, $%1 \n" \ + " mfhc0 %0, " #source ", %1 \n" \ " .set pop \n" \ : "=r" (__res) \ - : "i" (source)); \ + : "i" (sel)); \ __res; \ }) -#define __writex_32bit_c0_register(register, value) \ +#define __writex_32bit_c0_register(register, sel, value) \ do { \ __asm__ __volatile__( \ " .set push \n" \ " .set mips32r2 \n" \ _ASM_SET_XPA \ - " mthc0 %z0, $%1 \n" \ + " mthc0 %z0, " #register ", %1 \n" \ " .set pop \n" \ : \ - : "Jr" (value), "i" (register)); \ + : "Jr" (value), "i" (sel)); \ } while (0) #define read_c0_index() __read_32bit_c0_register($0, 0) @@ -1535,14 +1535,14 @@ do { \ #define read_c0_entrylo0() __read_ulong_c0_register($2, 0) #define write_c0_entrylo0(val) __write_ulong_c0_register($2, 0, val) -#define readx_c0_entrylo0() __readx_32bit_c0_register(2) -#define writex_c0_entrylo0(val) __writex_32bit_c0_register(2, val) +#define readx_c0_entrylo0() __readx_32bit_c0_register($2, 0) +#define writex_c0_entrylo0(val) __writex_32bit_c0_register($2, 0, val) #define read_c0_entrylo1() __read_ulong_c0_register($3, 0) #define write_c0_entrylo1(val) __write_ulong_c0_register($3, 0, val) -#define readx_c0_entrylo1() __readx_32bit_c0_register(3) -#define writex_c0_entrylo1(val) __writex_32bit_c0_register(3, val) +#define readx_c0_entrylo1() __readx_32bit_c0_register($3, 0) +#define writex_c0_entrylo1(val) __writex_32bit_c0_register($3, 0, val) #define read_c0_conf() __read_32bit_c0_register($3, 0) #define write_c0_conf(val) __write_32bit_c0_register($3, 0, val) -- git-series 0.9.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 7/7] MIPS: MSA: Update helpers to use new asm macros @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle, Paul Burton From: James Hogan <jhogan@kernel.org> Update MSA control register access helpers to use the new helpers for parsing register names and creating custom assembly macro instructions. This allows the move via $at to be dropped (saving a total of about 20 bytes of kernel code). Note, this does not alter the equivalent code in .S files, which still uses the $at trick. Signed-off-by: James Hogan <jhogan@kernel.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Paul Burton <paul.burton@mips.com> Cc: linux-mips@linux-mips.org --- arch/mips/include/asm/msa.h | 63 +++++++++++--------------------------- 1 file changed, 19 insertions(+), 44 deletions(-) diff --git a/arch/mips/include/asm/msa.h b/arch/mips/include/asm/msa.h index 8967b475ab10..93fa7133811b 100644 --- a/arch/mips/include/asm/msa.h +++ b/arch/mips/include/asm/msa.h @@ -160,7 +160,23 @@ static inline void init_msa_upper(void) _init_msa_upper(); } -#ifdef TOOLCHAIN_SUPPORTS_MSA +#ifndef TOOLCHAIN_SUPPORTS_MSA +/* + * Define assembler macros using .word for the c[ft]cmsa instructions in order + * to allow compilation with toolchains that do not support MSA. Once all + * toolchains in use support MSA these can be removed. + */ +_ASM_MACRO_2R(cfcmsa, rd, cs, + _ASM_INSN_IF_MIPS(0x787e0019 | __cs << 11 | __rd << 6) + _ASM_INSN32_IF_MM(0x587e0016 | __cs << 11 | __rd << 6)); +_ASM_MACRO_2R(ctcmsa, cd, rs, + _ASM_INSN_IF_MIPS(0x783e0019 | __rs << 11 | __cd << 6) + _ASM_INSN32_IF_MM(0x583e0016 | __rs << 11 | __cd << 6)); +#define _ASM_SET_MSA "" +#else /* TOOLCHAIN_SUPPORTS_MSA */ +#define _ASM_SET_MSA ".set\tfp=64\n\t" \ + ".set\tmsa\n\t" +#endif #define __BUILD_MSA_CTL_REG(name, cs) \ static inline unsigned int read_msa_##name(void) \ @@ -168,8 +184,7 @@ static inline unsigned int read_msa_##name(void) \ unsigned int reg; \ __asm__ __volatile__( \ " .set push\n" \ - " .set fp=64\n" \ - " .set msa\n" \ + _ASM_SET_MSA \ " cfcmsa %0, $" #cs "\n" \ " .set pop\n" \ : "=r"(reg)); \ @@ -180,52 +195,12 @@ static inline void write_msa_##name(unsigned int val) \ { \ __asm__ __volatile__( \ " .set push\n" \ - " .set fp=64\n" \ - " .set msa\n" \ + _ASM_SET_MSA \ " ctcmsa $" #cs ", %0\n" \ " .set pop\n" \ : : "r"(val)); \ } -#else /* !TOOLCHAIN_SUPPORTS_MSA */ - -/* - * Define functions using .word for the c[ft]cmsa instructions in order to - * allow compilation with toolchains that do not support MSA. Once all - * toolchains in use support MSA these can be removed. - */ - -#define __BUILD_MSA_CTL_REG(name, cs) \ -static inline unsigned int read_msa_##name(void) \ -{ \ - unsigned int reg; \ - __asm__ __volatile__( \ - " .set push\n" \ - " .set noat\n" \ - " # cfcmsa $1, $%1\n" \ - _ASM_INSN_IF_MIPS(0x787e0059 | %1 << 11) \ - _ASM_INSN32_IF_MM(0x587e0056 | %1 << 11) \ - " move %0, $1\n" \ - " .set pop\n" \ - : "=r"(reg) : "i"(cs)); \ - return reg; \ -} \ - \ -static inline void write_msa_##name(unsigned int val) \ -{ \ - __asm__ __volatile__( \ - " .set push\n" \ - " .set noat\n" \ - " move $1, %0\n" \ - " # ctcmsa $%1, $1\n" \ - _ASM_INSN_IF_MIPS(0x783e0819 | %1 << 6) \ - _ASM_INSN32_IF_MM(0x583e0816 | %1 << 6) \ - " .set pop\n" \ - : : "r"(val), "i"(cs)); \ -} - -#endif /* !TOOLCHAIN_SUPPORTS_MSA */ - __BUILD_MSA_CTL_REG(ir, 0) __BUILD_MSA_CTL_REG(csr, 1) __BUILD_MSA_CTL_REG(access, 2) -- git-series 0.9.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 7/7] MIPS: MSA: Update helpers to use new asm macros @ 2017-11-22 11:30 ` James Hogan 0 siblings, 0 replies; 17+ messages in thread From: James Hogan @ 2017-11-22 11:30 UTC (permalink / raw) To: linux-mips; +Cc: Marcin Nowakowski, James Hogan, Ralf Baechle, Paul Burton From: James Hogan <jhogan@kernel.org> Update MSA control register access helpers to use the new helpers for parsing register names and creating custom assembly macro instructions. This allows the move via $at to be dropped (saving a total of about 20 bytes of kernel code). Note, this does not alter the equivalent code in .S files, which still uses the $at trick. Signed-off-by: James Hogan <jhogan@kernel.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Paul Burton <paul.burton@mips.com> Cc: linux-mips@linux-mips.org --- arch/mips/include/asm/msa.h | 63 +++++++++++--------------------------- 1 file changed, 19 insertions(+), 44 deletions(-) diff --git a/arch/mips/include/asm/msa.h b/arch/mips/include/asm/msa.h index 8967b475ab10..93fa7133811b 100644 --- a/arch/mips/include/asm/msa.h +++ b/arch/mips/include/asm/msa.h @@ -160,7 +160,23 @@ static inline void init_msa_upper(void) _init_msa_upper(); } -#ifdef TOOLCHAIN_SUPPORTS_MSA +#ifndef TOOLCHAIN_SUPPORTS_MSA +/* + * Define assembler macros using .word for the c[ft]cmsa instructions in order + * to allow compilation with toolchains that do not support MSA. Once all + * toolchains in use support MSA these can be removed. + */ +_ASM_MACRO_2R(cfcmsa, rd, cs, + _ASM_INSN_IF_MIPS(0x787e0019 | __cs << 11 | __rd << 6) + _ASM_INSN32_IF_MM(0x587e0016 | __cs << 11 | __rd << 6)); +_ASM_MACRO_2R(ctcmsa, cd, rs, + _ASM_INSN_IF_MIPS(0x783e0019 | __rs << 11 | __cd << 6) + _ASM_INSN32_IF_MM(0x583e0016 | __rs << 11 | __cd << 6)); +#define _ASM_SET_MSA "" +#else /* TOOLCHAIN_SUPPORTS_MSA */ +#define _ASM_SET_MSA ".set\tfp=64\n\t" \ + ".set\tmsa\n\t" +#endif #define __BUILD_MSA_CTL_REG(name, cs) \ static inline unsigned int read_msa_##name(void) \ @@ -168,8 +184,7 @@ static inline unsigned int read_msa_##name(void) \ unsigned int reg; \ __asm__ __volatile__( \ " .set push\n" \ - " .set fp=64\n" \ - " .set msa\n" \ + _ASM_SET_MSA \ " cfcmsa %0, $" #cs "\n" \ " .set pop\n" \ : "=r"(reg)); \ @@ -180,52 +195,12 @@ static inline void write_msa_##name(unsigned int val) \ { \ __asm__ __volatile__( \ " .set push\n" \ - " .set fp=64\n" \ - " .set msa\n" \ + _ASM_SET_MSA \ " ctcmsa $" #cs ", %0\n" \ " .set pop\n" \ : : "r"(val)); \ } -#else /* !TOOLCHAIN_SUPPORTS_MSA */ - -/* - * Define functions using .word for the c[ft]cmsa instructions in order to - * allow compilation with toolchains that do not support MSA. Once all - * toolchains in use support MSA these can be removed. - */ - -#define __BUILD_MSA_CTL_REG(name, cs) \ -static inline unsigned int read_msa_##name(void) \ -{ \ - unsigned int reg; \ - __asm__ __volatile__( \ - " .set push\n" \ - " .set noat\n" \ - " # cfcmsa $1, $%1\n" \ - _ASM_INSN_IF_MIPS(0x787e0059 | %1 << 11) \ - _ASM_INSN32_IF_MM(0x587e0056 | %1 << 11) \ - " move %0, $1\n" \ - " .set pop\n" \ - : "=r"(reg) : "i"(cs)); \ - return reg; \ -} \ - \ -static inline void write_msa_##name(unsigned int val) \ -{ \ - __asm__ __volatile__( \ - " .set push\n" \ - " .set noat\n" \ - " move $1, %0\n" \ - " # ctcmsa $%1, $1\n" \ - _ASM_INSN_IF_MIPS(0x783e0819 | %1 << 6) \ - _ASM_INSN32_IF_MM(0x583e0816 | %1 << 6) \ - " .set pop\n" \ - : : "r"(val), "i"(cs)); \ -} - -#endif /* !TOOLCHAIN_SUPPORTS_MSA */ - __BUILD_MSA_CTL_REG(ir, 0) __BUILD_MSA_CTL_REG(csr, 1) __BUILD_MSA_CTL_REG(access, 2) -- git-series 0.9.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
end of thread, other threads:[~2018-05-24 19:48 UTC | newest] Thread overview: 17+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2017-11-22 11:30 [PATCH 0/7] MIPS: Add asm macros for unsupported instructions James Hogan 2017-11-22 11:30 ` James Hogan 2017-11-22 11:30 ` [PATCH 1/7] MIPS: Add helpers for assembler macro instructions James Hogan 2017-11-22 11:30 ` James Hogan 2018-05-24 19:47 ` Hauke Mehrtens 2017-11-22 11:30 ` [PATCH 2/7] MIPS: VZ: Update helpers to use new asm macros James Hogan 2017-11-22 11:30 ` James Hogan 2017-11-22 11:30 ` [PATCH 3/7] MIPS: VZ: Pass GC0 register names in $n format James Hogan 2017-11-22 11:30 ` James Hogan 2017-11-22 11:30 ` [PATCH 4/7] MIPS: XPA: Use XPA instructions in assembly James Hogan 2017-11-22 11:30 ` James Hogan 2017-11-22 11:30 ` [PATCH 5/7] MIPS: XPA: Allow use of $0 (zero) to MTHC0 James Hogan 2017-11-22 11:30 ` James Hogan 2017-11-22 11:30 ` [PATCH 6/7] MIPS: XPA: Standardise readx/writex accessors James Hogan 2017-11-22 11:30 ` James Hogan 2017-11-22 11:30 ` [PATCH 7/7] MIPS: MSA: Update helpers to use new asm macros James Hogan 2017-11-22 11:30 ` James Hogan
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.