diff options
| author | Pasha <pasha@member.fsf.org> | 2024-02-29 19:30:30 +0000 |
|---|---|---|
| committer | Pasha <pasha@member.fsf.org> | 2024-02-29 19:30:30 +0000 |
| commit | 84d199451cf33734003c00c043a1480920f5563b (patch) | |
| tree | 4655f03b3637184bfb363b4e86d376756e85c9e3 /patches/glibc/unsubmitted-prof-eintr.diff | |
| download | cross-hurd-84d199451cf33734003c00c043a1480920f5563b.tar.gz cross-hurd-84d199451cf33734003c00c043a1480920f5563b.tar.bz2 | |
initial commit
Diffstat (limited to 'patches/glibc/unsubmitted-prof-eintr.diff')
| -rw-r--r-- | patches/glibc/unsubmitted-prof-eintr.diff | 21 |
1 files changed, 21 insertions, 0 deletions
diff --git a/patches/glibc/unsubmitted-prof-eintr.diff b/patches/glibc/unsubmitted-prof-eintr.diff new file mode 100644 index 0000000..1b0808c --- /dev/null +++ b/patches/glibc/unsubmitted-prof-eintr.diff @@ -0,0 +1,21 @@ +When profiling ext2fs, the shutdown stops all RPC, which thus interrupts this +open, and we have to retry here. + +TODO: is open really supposed to expose such EINTR? Should the generic gmon +loop around EINTR (since there might be signals happening indeed) + +diff --git a/gmon/gmon.c b/gmon/gmon.c +index 6439ed1caa..507ad0c9fc 100644 +--- a/gmon/gmon.c ++++ b/gmon/gmon.c +@@ -390,8 +390,10 @@ write_gmon (void) + + if (fd == -1) + { ++ do + fd = __open_nocancel ("gmon.out", O_CREAT | O_TRUNC | O_WRONLY + | O_NOFOLLOW | O_CLOEXEC, 0666); ++ while (fd < 0 && errno == EINTR); + if (fd < 0) + { + char buf[300]; |