[Bug gold/26541] gold makes incorrect usage of fallocate in fallback path
https://sourceware.org/bugzilla/show_bug.cgi?id=26541 holger.berger at googlemail dot com changed: What|Removed |Added CC||holger.berger at googlemail dot co ||m --- Comment #1 from holger.berger at googlemail dot com --- I see the same thing, and it actually leads to failure of ld.gold on a certain version of lustre filesystem. ld.gold: fatal error: a.out: Unknown error 524 -- You are receiving this mail because: You are on the CC list for the bug.
[Bug gold/26541] gold makes incorrect usage of fallocate in fallback path
https://sourceware.org/bugzilla/show_bug.cgi?id=26541
--- Comment #3 from holger.berger at googlemail dot com ---
for me this patch works, but I think lustre should not leak it's internal error
code to teh client side, so the lustre part should probably not be part of an
official fix, and I think later versions of lustre fix that issue.
diff --git a/gold/output.cc b/gold/output.cc
index b7505ff..23a24ce 100644
--- a/gold/output.cc
+++ b/gold/output.cc
@@ -134,16 +134,17 @@ gold_fallocate(int o, off_t offset, off_t len)
if (parameters->options().posix_fallocate())
{
int err = ::posix_fallocate(o, offset, len);
- if (err != EINVAL && err != ENOSYS && err != EOPNOTSUPP)
+ if (err != EINVAL && err != ENOSYS && err != EOPNOTSUPP && err != 524 /*
LUSTRE_ENOTSUPP == 524 Operation is not supported */ )
return err;
}
#endif // defined(HAVE_POSIX_FALLOCATE)
#ifdef HAVE_FALLOCATE
{
+errno = 0;
int err = ::fallocate(o, 0, offset, len);
-if (err != EINVAL && err != ENOSYS && err != EOPNOTSUPP)
- return err;
+if (err == -1 && (errno != EINVAL && errno != ENOSYS && errno !=
EOPNOTSUPP && errno != 524 /* LUSTRE_ENOTSUPP == 524 Operation is not supported
*/ ))
+ return errno;
}
#endif // defined(HAVE_FALLOCATE)
--
You are receiving this mail because:
You are on the CC list for the bug.
[Bug gold/26541] gold makes incorrect usage of fallocate in fallback path
https://sourceware.org/bugzilla/show_bug.cgi?id=26541 --- Comment #6 from holger.berger at googlemail dot com --- correct... the proper patch would not contain the 524 thing, but some sort of patch is needed. Am Fr., 19. Feb. 2021 um 18:56 Uhr schrieb bugdal at aerifal dot cx < sourceware-bugzi...@sourceware.org>: > https://sourceware.org/bugzilla/show_bug.cgi?id=26541 > > --- Comment #5 from Rich Felker --- > Ah no, at least according to your patch the original gold bug seems to be > there > still too. > > -- > You are receiving this mail because: > You are on the CC list for the bug. -- You are receiving this mail because: You are on the CC list for the bug.
