Hi Andrew,
Your patch breaks LTO bootstrap for, at least, ARM and AArch64. Would you
please investigate?
--
Maxim Kuvyrkov
https://www.linaro.org
> On 14 Sep 2021, at 07:47, ci_not...@linaro.org wrote:
>
> Identified regression caused by
> *gcc:9e58de3ce00fc2385c9efb7faf321e0c601f0b0c*:
> co
[TCWG CI] Regression caused by binutils: Automatic date update in version.in:
commit dc746ef741993a7aed1f7fc0083cd7a9636481a3
Author: GDB Administrator
Automatic date update in version.in
Results regressed to
# reset_artifacts:
-10
# build_abe binutils:
-9
# build_abe stage1:
-5
# build_abe
00:09:17 ./libtool: eval: line 1118: `nm .libs/lto-plugin.o | |
/usr/bin/sed 's/.* //' | sort | uniq > .libs/liblto_plugin.exp'
Normally it does:
libtool: link: nm .libs/lto-plugin.o | sed -n -e 's/^.*[
]\([ABCDGIRSTW][ABCDGIRSTW]*\)[ ][
]*\([_A-Za-z][_A-Za-z0-9]*\)$/
I filed https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102389 for the problem.
I will look into fixing it either later today or tomorrow. It might just need
to export nm and such.
From: linaro-toolchain on behalf of
Andrew Pinski
Sent: Friday, September
Note after my fix, GCC LTO bootstrap still require binutils 2.34.x or newer due
to https://sourceware.org/PR25355 . There is nothing I can workaround in gcc's
configure as this is a standard test for libtool. I will be updating the
installation documentation anyways as it was already required
Patch submitted:
https://gcc.gnu.org/pipermail/gcc-patches/2021-September/579731.html
Note you might need to use a newer binutils too. I did not check your CI to
see which one is being used. There is no workaround for that really as
binutils was broken for LTO usage before 2.34.x really; just
