This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: New x86/PE gas testsuite failure: i386 inval-crc32
- From: Jan Beulich <jbeulich at suse dot com>
- To: Nick Clifton <nickc at redhat dot com>
- Cc: binutils at sourceware dot org
- Date: Tue, 21 Jan 2020 12:56:07 +0100
- Subject: Re: New x86/PE gas testsuite failure: i386 inval-crc32
- References: <87h80pnj0r.fsf@redhat.com>
On 21.01.2020 12:31, Nick Clifton wrote:
> Hi Jan,
>
> I am seeing a new failure in the gas testsuite for various x86/PE
> based targets:
>
> regexp_diff match failure
> regexp "^[ ]*23[ ]+crc32 ax, ax$"
> line " 23 ???? 909090 crc32 ax, ax"
> extra lines in dump.out starting with "^**** Error: unsupported instruction `crc32'$"
> EOF from /work/sources/binutils/current/gas/testsuite/gas/i386/inval-crc32.l
> FAIL: i386 inval-crc32
>
> (And similarly for the x86_64-inval-crc32 test for 64-bit targets).
>
> This turned up in tests of these targets:
>
> x86_64-pc-cygwin
> x86_64-pc-mingw64
> i686-pc-mingw32
> i386-pc-go32
> i686-pc-cygwin
>
> Could you have a look please ?
I'll take a look, sure, but
- I didn't touch the source of the test at all, so I don't see
where the 0x90 (NOP) bytes would come from in the listing all
of the sudden,
- I think in general test cases shouldn't be as fragile, and
there shouldn't be a need to add funny alignment directives
(I guess that's what may be needed here, based on prior
experience); such should rather be taken care of by the
framework.
Jan