[PATCH] riscv: Fix ALT_THEAD_PMA's asm parameters

[Palmer Dabbelt]

On Mon, 13 Jun 2022 13:26:17 PDT (-0700), nathan at kernel.org wrote:
> On Wed, May 25, 2022 at 07:40:37PM -0700, Nathan Chancellor wrote:
>> Small ping on this and https://lore.kernel.org/20220516214520.3252074-1-nathan@kernel.org/.
>>
>> Our builds on -next have been broken for a week now. Hopefully these can
>> make the first RISC-V pull request to avoid mainline being broken in the
>> same fashion.
>
> One more small ping. The patch linked above made it into 5.19-rc1 but
> this one has not been applied so our builds are still broken.

Sorry I dropped the ball on this one, it's in fixes.  Thanks!

> Cheers,
> Nathan
>
>> On Wed, May 18, 2022 at 11:45:29AM -0700, Nathan Chancellor wrote:
>> > After commit a35707c3d850 ("riscv: add memory-type errata for T-Head"),
>> > builds with LLVM's integrated assembler fail like:
>> >
>> >   In file included from arch/riscv/kernel/asm-offsets.c:10:
>> >   In file included from ./include/linux/mm.h:29:
>> >   In file included from ./include/linux/pgtable.h:6:
>> >   In file included from ./arch/riscv/include/asm/pgtable.h:114:
>> >   ./arch/riscv/include/asm/pgtable-64.h:210:2: error: invalid input constraint '0' in asm
>> >           ALT_THEAD_PMA(prot_val);
>> >           ^
>> >   ./arch/riscv/include/asm/errata_list.h:88:4: note: expanded from macro 'ALT_THEAD_PMA'
>> >           : "0"(_val),                                                    \
>> >             ^
>> >
>> > This was reported upstream to LLVM where Jessica pointed out a couple of
>> > issues with the existing implementation of ALT_THEAD_PMA:
>> >
>> > * t3 is modified but not listed in the clobbers list.
>> >
>> > * "+r"(_val) marks _val as both an input and output of the asm but then
>> >   "0"(_val) marks _val as an input matching constraint, which does not
>> >   make much sense in this situation, as %1 is not actually used in the
>> >   asm and matching constraints are designed to be used for different
>> >   inputs that need to use the same register.
>> >
>> > Drop the matching contraint and shift all the operands by one, as %1 is
>> > unused, and mark t3 as clobbered. This resolves the build error and goes
>> > not cause any problems with GNU as.
>> >
>> > Fixes: a35707c3d850 ("riscv: add memory-type errata for T-Head")
>> > Link: https://github.com/ClangBuiltLinux/linux/issues/1641
>> > Link: https://github.com/llvm/llvm-project/issues/55514
>> > Link: https://gcc.gnu.org/onlinedocs/gcc/Simple-Constraints.html
>> > Suggested-by: Jessica Clarke <jrtc27 at jrtc27.com>
>> > Signed-off-by: Nathan Chancellor <nathan at kernel.org>
>> > ---
>> >  arch/riscv/include/asm/errata_list.h | 14 +++++++-------
>> >  1 file changed, 7 insertions(+), 7 deletions(-)
>> >
>> > diff --git a/arch/riscv/include/asm/errata_list.h b/arch/riscv/include/asm/errata_list.h
>> > index 9e2888dbb5b1..416ead0f9a65 100644
>> > --- a/arch/riscv/include/asm/errata_list.h
>> > +++ b/arch/riscv/include/asm/errata_list.h
>> > @@ -75,20 +75,20 @@ asm volatile(ALTERNATIVE(						\
>> >  	"nop\n\t"							\
>> >  	"nop\n\t"							\
>> >  	"nop",								\
>> > -	"li      t3, %2\n\t"						\
>> > -	"slli    t3, t3, %4\n\t"					\
>> > +	"li      t3, %1\n\t"						\
>> > +	"slli    t3, t3, %3\n\t"					\
>> >  	"and     t3, %0, t3\n\t"					\
>> >  	"bne     t3, zero, 2f\n\t"					\
>> > -	"li      t3, %3\n\t"						\
>> > -	"slli    t3, t3, %4\n\t"					\
>> > +	"li      t3, %2\n\t"						\
>> > +	"slli    t3, t3, %3\n\t"					\
>> >  	"or      %0, %0, t3\n\t"					\
>> >  	"2:",  THEAD_VENDOR_ID,						\
>> >  		ERRATA_THEAD_PBMT, CONFIG_ERRATA_THEAD_PBMT)		\
>> >  	: "+r"(_val)							\
>> > -	: "0"(_val),							\
>> > -	  "I"(_PAGE_MTMASK_THEAD >> ALT_THEAD_PBMT_SHIFT),		\
>> > +	: "I"(_PAGE_MTMASK_THEAD >> ALT_THEAD_PBMT_SHIFT),		\
>> >  	  "I"(_PAGE_PMA_THEAD >> ALT_THEAD_PBMT_SHIFT),			\
>> > -	  "I"(ALT_THEAD_PBMT_SHIFT))
>> > +	  "I"(ALT_THEAD_PBMT_SHIFT)					\
>> > +	: "t3")
>> >  #else
>> >  #define ALT_THEAD_PMA(_val)
>> >  #endif
>> >
>> > base-commit: 93c0651617a62a69717299f1464dda798af8bebb
>> > --
>> > 2.36.1
>> >
>> >
>> > _______________________________________________
>> > linux-riscv mailing list
>> > linux-riscv at lists.infradead.org
>> > http://lists.infradead.org/mailman/listinfo/linux-riscv
>>