[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Bug ld/16139] ld/mips: abort in _bfd_merged_section_offset
From: |
amodra at gmail dot com |
Subject: |
[Bug ld/16139] ld/mips: abort in _bfd_merged_section_offset |
Date: |
Fri, 29 Nov 2013 01:18:05 +0000 |
https://sourceware.org/bugzilla/show_bug.cgi?id=16139
Alan Modra <amodra at gmail dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |amodra at gmail dot com
--- Comment #3 from Alan Modra <amodra at gmail dot com> ---
I'm inclined to close this bug as likely to be due to a hardware fault, but
I'll leave it open for the time being. The occasional bit-flip in memory is
known to trigger this internal error in bfd/merge.c. I've seen bit-flips
myself, and am 100% sure that the cases I saw were really hardware errors and
not some wild pointer write corrupting the buffer that bfd/merge.c was
operating on.
I've also seen this error with projects that have Makefile dependency problems.
Using make -j can lead to ld trying to link an object file before it has been
completely written.. There are of course other possibilities that might lead
to a make -j race, eg. a kernel failing to invalidate cache or returning too
early from a dma operation.
If these problems are not reproducible when running make without -j, and
running ld under valgrind doesn't show uninitialized memory reads (I've done
that), then it's probably not a binutils bug.
--
You are receiving this mail because:
You are on the CC list for the bug.