Bug 679430 – sys-devel/gcc-8.2.0 - kernel on PPC - In file included from at arch/powerpc/kernel/ptrace.c:619:9: ./include/linux/regset.h:295:4: error: 'memcpy' offset [-527, -529] is out of the bounds [0, 16] of object 'vrsave' with type 'union <anonymous>'

• Product: Gentoo Linux
• Status: RESOLVED UPSTREAM

[Description Matt Turner 2019-03-04 17:33:25 UTC]

gcc-8 regression causes the 32-bit powerpc kernel build to fail. This is currently preventing ppc autobuilds from succeeding.

The regression has been fixed in the gcc-9 branch (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88273) but it does not seem to have been backported to gcc-8 (yet?).


In parallel, a fix has gone into powerpc next:

https://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux.git/commit/?id=ca6d5149d2ad0a8d2f9c28cbe3798022

I'll let toolchain@ and kernel@ decide how to resolve this.

[Comment 1 Sergei Trofimovich (RETIRED) 2019-03-04 23:07:37 UTC]

bug #675334 should backport a fix to kernel as well.

[Comment 2 Sergei Trofimovich (RETIRED) 2019-03-04 23:20:34 UTC]

(In reply to Sergei Trofimovich from comment #1)
> bug #675334 should backport a fix to kernel as well.

Sorry. That's unrelated.

[Comment 3 Jeroen Roovers (RETIRED) 2019-03-05 05:08:21 UTC]

*** Bug 679454 has been marked as a duplicate of this bug. ***

[Comment 4 Matt Turner 2019-04-06 19:43:44 UTC]

This is now fixed in gcc-8.4 (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88273#c15)

[Comment 5 Matt Turner 2019-07-17 03:09:38 UTC]

Fixed in upcoming gcc-8.4.0, but I'm not sure we really care at this point.

[Comment 6 Sergei Trofimovich (RETIRED) 2019-08-01 22:52:29 UTC]

Let's close as RESO/UPSTREAM.