lkml.org 
[lkml]   [2021]   [Oct]   [11]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Date
SubjectRe: clang-13: s390/kernel/head64.S:24:17: error: invalid operand for instruction
+ Nemanja


On Mon, Oct 11, 2021 at 3:01 AM Heiko Carstens <hca@linux.ibm.com> wrote:
>
> On Mon, Oct 11, 2021 at 11:47:42AM +0530, Naresh Kamboju wrote:
> > [Please ignore this email if it is already reported ]
> >
> > Following s390 builds failed due to warnings / errors.
> >
> > metadata:
> > git_describe: v5.15-rc5
> > git_repo: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
> > git_short_log: 64570fbc14f8 (\"Linux 5.15-rc5\")
> > target_arch: s390
> > toolchain: clang-13
> >
> >
> > Fail (2861 errors) s390 (tinyconfig) with clang-nightly
> > @ https://builds.tuxbuild.com/1zL35IUSGhDGeVuyIrAp7eyzEUi/
> > Fail (2861 errors) s390 (tinyconfig) with clang-13
> > @ https://builds.tuxbuild.com/1zL35Hn7wjErKsLDM6zAgh27BYJ/
> > Fail (4112 errors) s390 (allnoconfig) with clang-13
> > @ https://builds.tuxbuild.com/1zL35HR60hSFvBmAcYJvKHm8Lko/
> > Fail (4112 errors) s390 (allnoconfig) with clang-nightly
> > @ https://builds.tuxbuild.com/1zL35DTlrX9qRGCtGqgtmmMDjnQ/
> > Fail (23048 errors) s390 (defconfig) with clang-13
> > @ https://builds.tuxbuild.com/1zL35DE2KWQUPxbbXeTbwIJaWXS/
> > Fail (23045 errors) s390 (defconfig) with clang-nightly
> > @ https://builds.tuxbuild.com/1zL35EgeQfWQDXDupp4itkUO5At/
> >
> >
> > Build errors log:
> > ----------------
> > arch/s390/kernel/head64.S:24:17: error: invalid operand for instruction
> > lctlg %c0,%c15,.Lctl-.LPG1(%r13) # load control registers
> > ^
> > arch/s390/kernel/head64.S:40:8: error: invalid operand for instruction
> > lpswe .Ldw-.(%r13) # load disabled wait psw
>
> You need to pass LLVM_IAS=0 on the make command line on s390 since
> commit f12b034afeb3 ("scripts/Makefile.clang: default to LLVM_IAS=1").
>
> LLVM's integrated assembler doesn't seem to work well when compiling
> the kernel for s390 yet.

Dunno if you folks have cycles to work on it, but if you do, we have
just a handful of bugs left in Clang's assembler (or kernel source
issues) preventing us from assembling the kernel with clang for
SystemZ and PPC.

s390:
https://github.com/ClangBuiltLinux/linux/issues?q=is%3Aopen+is%3Aissue+label%3A%22%5BARCH%5D+s390%22+label%3A%22%5BTOOL%5D+integrated-as%22

ppc:
https://github.com/ClangBuiltLinux/linux/issues?q=is%3Aopen+is%3Aissue+label%3A%22%5BTOOL%5D+integrated-as%22+label%3A%22%5BARCH%5D+powerpc%22

I'm happy to hop on a call to discuss these more. We have all other
architectures working here, in the sense that if we can build them
with clang, we're using the integrated assembler by default.
--
Thanks,
~Nick Desaulniers

\
 
 \ /
  Last update: 2021-10-11 19:32    [W:3.305 / U:0.028 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site