Re: [PATCH v5] x86/crypto: aesni: Fix build with LLVM_IAS=1
From: Sedat Dilek
Date: Fri Jul 10 2020 - 05:00:26 EST
On Thu, Jul 9, 2020 at 2:56 PM Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx> wrote:
>
> On Fri, Jul 03, 2020 at 04:32:06PM +0200, Sedat Dilek wrote:
> > When building with LLVM_IAS=1 means using Clang's Integrated Assembly (IAS)
> > from LLVM/Clang >= v10.0.1-rc1+ instead of GNU/as from GNU/binutils
> > I see the following breakage in Debian/testing AMD64:
> >
> > <instantiation>:15:74: error: too many positional arguments
> > PRECOMPUTE 8*3+8(%rsp), %xmm1, %xmm2, %xmm3, %xmm4, %xmm5, %xmm6, %xmm7,
> > ^
> > arch/x86/crypto/aesni-intel_asm.S:1598:2: note: while in macro instantiation
> > GCM_INIT %r9, 8*3 +8(%rsp), 8*3 +16(%rsp), 8*3 +24(%rsp)
> > ^
> > <instantiation>:47:2: error: unknown use of instruction mnemonic without a size suffix
> > GHASH_4_ENCRYPT_4_PARALLEL_dec %xmm9, %xmm10, %xmm11, %xmm12, %xmm13, %xmm14, %xmm0, %xmm1, %xmm2, %xmm3, %xmm4, %xmm5, %xmm6, %xmm7, %xmm8, enc
> > ^
> > arch/x86/crypto/aesni-intel_asm.S:1599:2: note: while in macro instantiation
> > GCM_ENC_DEC dec
> > ^
> > <instantiation>:15:74: error: too many positional arguments
> > PRECOMPUTE 8*3+8(%rsp), %xmm1, %xmm2, %xmm3, %xmm4, %xmm5, %xmm6, %xmm7,
> > ^
> > arch/x86/crypto/aesni-intel_asm.S:1686:2: note: while in macro instantiation
> > GCM_INIT %r9, 8*3 +8(%rsp), 8*3 +16(%rsp), 8*3 +24(%rsp)
> > ^
> > <instantiation>:47:2: error: unknown use of instruction mnemonic without a size suffix
> > GHASH_4_ENCRYPT_4_PARALLEL_enc %xmm9, %xmm10, %xmm11, %xmm12, %xmm13, %xmm14, %xmm0, %xmm1, %xmm2, %xmm3, %xmm4, %xmm5, %xmm6, %xmm7, %xmm8, enc
> > ^
> > arch/x86/crypto/aesni-intel_asm.S:1687:2: note: while in macro instantiation
> > GCM_ENC_DEC enc
> >
> > Craig Topper suggested me in ClangBuiltLinux issue #1050:
> >
> > > I think the "too many positional arguments" is because the parser isn't able
> > > to handle the trailing commas.
> > >
> > > The "unknown use of instruction mnemonic" is because the macro was named
> > > GHASH_4_ENCRYPT_4_PARALLEL_DEC but its being instantiated with
> > > GHASH_4_ENCRYPT_4_PARALLEL_dec I guess gas ignores case on the
> > > macro instantiation, but llvm doesn't.
> >
> > First, I removed the trailing comma in the PRECOMPUTE line.
> >
> > Second, I substituted:
> > 1. GHASH_4_ENCRYPT_4_PARALLEL_DEC -> GHASH_4_ENCRYPT_4_PARALLEL_dec
> > 2. GHASH_4_ENCRYPT_4_PARALLEL_ENC -> GHASH_4_ENCRYPT_4_PARALLEL_enc
> >
> > With these changes I was able to build with LLVM_IAS=1 and boot on bare metal.
> >
> > I confirmed that this works with Linux-kernel v5.7.5 final.
> >
> > NOTE: This patch is on top of Linux v5.7 final.
> >
> > Thanks to Craig and especially Nick for double-checking and his comments.
> >
> > Suggested-by: Craig Topper <craig.topper@xxxxxxxxx>
> > Suggested-by: Craig Topper <craig.topper@xxxxxxxxx>
> > Suggested-by: Nick Desaulniers <ndesaulniers@xxxxxxxxxx>
> > Reviewed-by: Nick Desaulniers <ndesaulniers@xxxxxxxxxx>
> > Cc: "ClangBuiltLinux" <clang-built-linux@xxxxxxxxxxxxxxxx>
> > Link: https://github.com/ClangBuiltLinux/linux/issues/1050
> > Link: https://bugs.llvm.org/show_bug.cgi?id=24494
> > Signed-off-by: Sedat Dilek <sedat.dilek@xxxxxxxxx>
> > ---
> > Changes v4->v5:
> > - Drop "5.7" tag from subject line as requested by Herbert Xu
> > - Add Link to LLVM bug #24494 (thanks Nick)
> >
> > Changes v3->v4:
> > - Add <> around email address as desired by Nick
> > - Add Nick's Reviewed-by
> >
> > Changes v2->v3:
> > - Add this Changelog
> >
> > Changes v1->v2:
> > - Replace Cc by Suggested-by for Craig
> > - Replace Cc by Suggested-by for Nick (dropped Cc as desired)
> > - Really follow the suggestions of Craig
> > - Drop unneeded comments for my build-environment and Links
> >
> > arch/x86/crypto/aesni-intel_asm.S | 6 +++---
> > 1 file changed, 3 insertions(+), 3 deletions(-)
>
> Patch applied. Thanks.
[ CC Arnd ]
Thanks Herbert.
Last Monday I switched over from Linux v5.7.y to v5.8-rc4 and my patch
applied cleanly - I was able to build and boot on bare metal.
One request :
We (ClangBuiltLinux folks) need at least Arnd's patch (see [1])...
"x86: crypto: fix building crc32c with clang ias"
...to be full compliant with LLVM_IAS=1.
We had some discussion internally (see [2]-[4]).
Looking at Arnd's patch again: It was also sent to linux-crypto ML
(May 27, 2020, 2:17 p.m. UTC).
If you pick this directly, feel free to add my:
Reported-by: Sedat Dilek <sedat.dilek@xxxxxxxxx>
Tested-by: Sedat Dilek <sedat.dilek@xxxxxxxxx>
If you have any other wishes/requirements, please let me know.
Regards,
- Sedat -
P.S.: If you are interested in the status of LLVM_IAS=1 see [5] - it
needs to be updated to my recent experiments with Linux v5.8-rc4.
P.S.S.: BTW, a recent version of llvm-toolchain v11.0.0-git is full
compliant with LLVM=1 - no GNU/binutils/tools used just LLVM/utilities
(beyond GNU/make) :-).
[1] https://lore.kernel.org/patchwork/patch/1248401/
[2] https://github.com/ClangBuiltLinux/linux/issues/1010
[3] https://github.com/ClangBuiltLinux/linux/issues/1010#issuecomment-653759336
[4] https://github.com/ClangBuiltLinux/linux/issues/1010#issuecomment-656406876
[5] https://github.com/ClangBuiltLinux/linux/issues/1049
ADDENDUM: Demonstrate/show used Clang's integrated assembler (IAS):
$ llvm-dwarfdump-11 arch/x86/crypto/crc32c-intel.o
...
0x0000adc3: DW_TAG_compile_unit
DW_AT_stmt_list (0x00000b94)
DW_AT_ranges (0x00000040
[0x0000000000000350, 0x0000000000001430)
[0x0000000000000000, 0x000000000000000a))
DW_AT_name ("arch/x86/crypto/crc32c-pcl-intel-asm_64.S")
DW_AT_comp_dir ("/home/dileks/src/linux-kernel/git")
DW_AT_producer ("Debian clang version
11.0.0-++20200701093119+ffee8040534-1~exp1 ")
DW_AT_language (DW_LANG_Mips_Assembler)
...
- EOT -