| Message ID | a78f7e5820a6b827c9d68362a94dcbf80a317dc9.1587299429.git.stefan@agner.ch (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | ARM: make use of UAL VFP mnemonics when possible | expand |
On Sun, Apr 19, 2020 at 02:35:49PM +0200, Stefan Agner wrote: > Explicit FPU selection has been introduced in commit 1a6be26d5b1a > ("[ARM] Enable VFP to be built when non-VFP capable CPUs are selected") > to make use of assembler mnemonics for VFP instructions. > > However, clang currently does not support passing assembler flags > like this and errors out with: > clang-10: error: the clang compiler does not support '-Wa,-mfpu=softvfp+vfp' > > Make use of the .fpu assembler directives to select the floating point > hardware selectively. Also use the new unified assembler language > mnemonics. This allows to build these procedures with Clang. > > Link: https://github.com/ClangBuiltLinux/linux/issues/762 > Signed-off-by: Stefan Agner <stefan@agner.ch> > --- > Changes in v2: > - Add link in commit message > > arch/arm/vfp/Makefile | 2 -- > arch/arm/vfp/vfphw.S | 30 +++++++++++++++++++----------- > 2 files changed, 19 insertions(+), 13 deletions(-) > > diff --git a/arch/arm/vfp/Makefile b/arch/arm/vfp/Makefile > index 9975b63ac3b0..749901a72d6d 100644 > --- a/arch/arm/vfp/Makefile > +++ b/arch/arm/vfp/Makefile > @@ -8,6 +8,4 @@ > # ccflags-y := -DDEBUG > # asflags-y := -DDEBUG > > -KBUILD_AFLAGS :=$(KBUILD_AFLAGS:-msoft-float=-Wa,-mfpu=softvfp+vfp -mfloat-abi=soft) > - > obj-y += vfpmodule.o entry.o vfphw.o vfpsingle.o vfpdouble.o > diff --git a/arch/arm/vfp/vfphw.S b/arch/arm/vfp/vfphw.S > index b2e560290860..e214007a20a2 100644 > --- a/arch/arm/vfp/vfphw.S > +++ b/arch/arm/vfp/vfphw.S > @@ -258,11 +258,13 @@ vfp_current_hw_state_address: > > ENTRY(vfp_get_float) > tbl_branch r0, r3, #3 > - .irp dr,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 > -1: mrc p10, 0, r0, c\dr, c0, 0 @ fmrs r0, s0 > + .irp dr,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15, Apart from the extraneous comma above, this looks fine, thanks. > +1: vmov r0, s\dr > ret lr > .org 1b + 8 > -1: mrc p10, 0, r0, c\dr, c0, 4 @ fmrs r0, s1 > + .endr > + .irp dr,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31 > +1: vmov r0, s\dr > ret lr > .org 1b + 8 > .endr > @@ -271,10 +273,12 @@ ENDPROC(vfp_get_float) > ENTRY(vfp_put_float) > tbl_branch r1, r3, #3 > .irp dr,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 > -1: mcr p10, 0, r0, c\dr, c0, 0 @ fmsr r0, s0 > +1: vmov s\dr, r0 > ret lr > .org 1b + 8 > -1: mcr p10, 0, r0, c\dr, c0, 4 @ fmsr r0, s1 > + .endr > + .irp dr,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31 > +1: vmov s\dr, r0 > ret lr > .org 1b + 8 > .endr > @@ -282,15 +286,17 @@ ENDPROC(vfp_put_float) > > ENTRY(vfp_get_double) > tbl_branch r0, r3, #3 > + .fpu vfpv2 > .irp dr,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 > -1: fmrrd r0, r1, d\dr > +1: vmov r0, r1, d\dr > ret lr > .org 1b + 8 > .endr > #ifdef CONFIG_VFPv3 > @ d16 - d31 registers > - .irp dr,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 > -1: mrrc p11, 3, r0, r1, c\dr @ fmrrd r0, r1, d\dr > + .fpu vfpv3 > + .irp dr,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31 > +1: vmov r0, r1, d\dr > ret lr > .org 1b + 8 > .endr > @@ -304,15 +310,17 @@ ENDPROC(vfp_get_double) > > ENTRY(vfp_put_double) > tbl_branch r2, r3, #3 > + .fpu vfpv2 > .irp dr,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 > -1: fmdrr d\dr, r0, r1 > +1: vmov d\dr, r0, r1 > ret lr > .org 1b + 8 > .endr > #ifdef CONFIG_VFPv3 > + .fpu vfpv3 > @ d16 - d31 registers > - .irp dr,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 > -1: mcrr p11, 3, r0, r1, c\dr @ fmdrr r0, r1, d\dr > + .irp dr,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31 > +1: vmov d\dr, r0, r1 > ret lr > .org 1b + 8 > .endr > -- > 2.25.1 > >
diff --git a/arch/arm/vfp/Makefile b/arch/arm/vfp/Makefile index 9975b63ac3b0..749901a72d6d 100644 --- a/arch/arm/vfp/Makefile +++ b/arch/arm/vfp/Makefile @@ -8,6 +8,4 @@ # ccflags-y := -DDEBUG # asflags-y := -DDEBUG -KBUILD_AFLAGS :=$(KBUILD_AFLAGS:-msoft-float=-Wa,-mfpu=softvfp+vfp -mfloat-abi=soft) - obj-y += vfpmodule.o entry.o vfphw.o vfpsingle.o vfpdouble.o diff --git a/arch/arm/vfp/vfphw.S b/arch/arm/vfp/vfphw.S index b2e560290860..e214007a20a2 100644 --- a/arch/arm/vfp/vfphw.S +++ b/arch/arm/vfp/vfphw.S @@ -258,11 +258,13 @@ vfp_current_hw_state_address: ENTRY(vfp_get_float) tbl_branch r0, r3, #3 - .irp dr,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 -1: mrc p10, 0, r0, c\dr, c0, 0 @ fmrs r0, s0 + .irp dr,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15, +1: vmov r0, s\dr ret lr .org 1b + 8 -1: mrc p10, 0, r0, c\dr, c0, 4 @ fmrs r0, s1 + .endr + .irp dr,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31 +1: vmov r0, s\dr ret lr .org 1b + 8 .endr @@ -271,10 +273,12 @@ ENDPROC(vfp_get_float) ENTRY(vfp_put_float) tbl_branch r1, r3, #3 .irp dr,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 -1: mcr p10, 0, r0, c\dr, c0, 0 @ fmsr r0, s0 +1: vmov s\dr, r0 ret lr .org 1b + 8 -1: mcr p10, 0, r0, c\dr, c0, 4 @ fmsr r0, s1 + .endr + .irp dr,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31 +1: vmov s\dr, r0 ret lr .org 1b + 8 .endr @@ -282,15 +286,17 @@ ENDPROC(vfp_put_float) ENTRY(vfp_get_double) tbl_branch r0, r3, #3 + .fpu vfpv2 .irp dr,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 -1: fmrrd r0, r1, d\dr +1: vmov r0, r1, d\dr ret lr .org 1b + 8 .endr #ifdef CONFIG_VFPv3 @ d16 - d31 registers - .irp dr,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 -1: mrrc p11, 3, r0, r1, c\dr @ fmrrd r0, r1, d\dr + .fpu vfpv3 + .irp dr,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31 +1: vmov r0, r1, d\dr ret lr .org 1b + 8 .endr @@ -304,15 +310,17 @@ ENDPROC(vfp_get_double) ENTRY(vfp_put_double) tbl_branch r2, r3, #3 + .fpu vfpv2 .irp dr,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 -1: fmdrr d\dr, r0, r1 +1: vmov d\dr, r0, r1 ret lr .org 1b + 8 .endr #ifdef CONFIG_VFPv3 + .fpu vfpv3 @ d16 - d31 registers - .irp dr,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 -1: mcrr p11, 3, r0, r1, c\dr @ fmdrr r0, r1, d\dr + .irp dr,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31 +1: vmov d\dr, r0, r1 ret lr .org 1b + 8 .endr
Explicit FPU selection has been introduced in commit 1a6be26d5b1a ("[ARM] Enable VFP to be built when non-VFP capable CPUs are selected") to make use of assembler mnemonics for VFP instructions. However, clang currently does not support passing assembler flags like this and errors out with: clang-10: error: the clang compiler does not support '-Wa,-mfpu=softvfp+vfp' Make use of the .fpu assembler directives to select the floating point hardware selectively. Also use the new unified assembler language mnemonics. This allows to build these procedures with Clang. Link: https://github.com/ClangBuiltLinux/linux/issues/762 Signed-off-by: Stefan Agner <stefan@agner.ch> --- Changes in v2: - Add link in commit message arch/arm/vfp/Makefile | 2 -- arch/arm/vfp/vfphw.S | 30 +++++++++++++++++++----------- 2 files changed, 19 insertions(+), 13 deletions(-)