summaryrefslogtreecommitdiff
path: root/daemon/auth.c
diff options
context:
space:
mode:
authorGeorge Lebl <jirka@5z.com>2003-07-28 21:13:11 +0000
committerGeorge Lebl <jirka@src.gnome.org>2003-07-28 21:13:11 +0000
commitc0ef449e5f8c9347683712073fcccbaa586fdf62 (patch)
tree94e8353dd6204c15f6d60e115a715fddf1a2df74 /daemon/auth.c
parent1ed4984a24db1e471ed9c560b2b99e12e868f939 (diff)
downloadgdm-c0ef449e5f8c9347683712073fcccbaa586fdf62.tar.gz
for all functions for which I found in the manpages that EINTR could be
Mon Jul 28 14:10:12 2003 George Lebl <jirka@5z.com> * daemon/gdm.h, daemon/*.c: for all functions for which I found in the manpages that EINTR could be returned on at least some systems, use a new IGNORE_EINTR macro which will ignore the EINTR errno. This should fix any bugs with signals coming at the wrong times, though that is quite unlikely I'd say. Apparently SVr4 can return EINTR even for execv which is realy anal since EINTR is pretty much useless. Why can't we all be sane like BSD. * daemon/misc.c, daemon/server.c: automatic vars have weird behaviour when using Setjmp, so use static (volatile wouldn't work here if I'm reading the docs right)
Diffstat (limited to 'daemon/auth.c')
-rw-r--r--daemon/auth.c2
1 files changed, 1 insertions, 1 deletions
diff --git a/daemon/auth.c b/daemon/auth.c
index a1d0a155..5e475282 100644
--- a/daemon/auth.c
+++ b/daemon/auth.c
@@ -580,7 +580,7 @@ gdm_auth_user_remove (GdmDisplay *d, uid_t user)
/* If we are using the fallback cookie location, simply nuke the
* cookie file */
if (d->authfb) {
- unlink (d->userauth);
+ IGNORE_EINTR (unlink (d->userauth));
g_free (d->userauth);
d->userauth = NULL;
return;