[Buildroot] [Bug 15491] New: segmentation fault in busybox/glibc due to CONFIG_STACK_OPTIMIZATION_386

* [Buildroot] [Bug 15491] New: segmentation fault in busybox/glibc due to CONFIG_STACK_OPTIMIZATION_386
@ 2023-03-25 21:10 bugzilla
  2023-03-25 21:12 ` [Buildroot] [Bug 15491] " bugzilla
                   ` (2 more replies)
  0 siblings, 3 replies; 4+ messages in thread
From: bugzilla @ 2023-03-25 21:10 UTC (permalink / raw)
  To: buildroot

https://bugs.busybox.net/show_bug.cgi?id=15491

            Bug ID: 15491
           Summary: segmentation fault in busybox/glibc due to
                    CONFIG_STACK_OPTIMIZATION_386
           Product: buildroot
           Version: 2023.02
          Hardware: All
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P5
         Component: Other
          Assignee: unassigned@buildroot.uclibc.org
          Reporter: fesc2000@mailbox.org
                CC: buildroot@uclibc.org
  Target Milestone: ---

Hi,

i was facing the issue that busybox (ls appled) was generating a random
segmentation fault in libc/ctime() (i386 platform, glibc 2.30).
The actual problem was that CONFIG_STACK_OPTIMIZATION_386 was set by default,
causing an 8 byte aligned stack, and glibc uses a movaps instruction on it
(which requires 16 byte alignment).

Obviously i could solve this by disabling CONFIG_STACK_OPTIMIZATION_386, but it
took me quite a while to find it out.

This really should be disabled by default. After some digging it seems that the
ABI nowadays mandates 16 byte stack alignment even for i386 (see
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=38496).

-- 
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

^ permalink raw reply	[flat|nested] 4+ messages in thread