jimzah/openafs
1
0
Fork 0
mirror of https://git.openafs.org/openafs.git synced 2025-01-22 17:00:15 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

vol: Add timeouts to SYNC server select() calls

Browse Source 
Normally *SYNC server processes wait indefinitely for activity to
occur on one of the SYNC sockets. On some Linux kernels, there exists
a race condition where data can come in on a socket, but the select()
call continues to wait. To ensure that we do not hang forever in such
a scenario, add a timeout to the select() call, which will ensure we
notice the new data within 10 seconds. Raise the timeout on non-Linux
to reduce impact elsewhere.

The Linux kernel bug is tracked in
<https://bugzilla.redhat.com/show_bug.cgi?id=494404>, though that bug
report may not represent all affected kernels.

Change-Id: I3250eb53d59610ccbcffe9e8e283984d5ae0e2b4
Reviewed-on: http://gerrit.openafs.org/4377
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>
This commit is contained in:
Andrew Deason 2011-03-29 12:28:46 -05:00 committed by Derrick Brashear
parent e2cd861cdf
commit 920a00e075
3 changed files with 17 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
src/vol/daemon_com.h
View File

@ -87,6 +87,15 @@ enum SYNCReasonCode {
afs_int64 _##buf##_l[SYNC_PROTO_MAX_LEN/sizeof(afs_int64)]; \
char * buf = (char *)(_##buf##_l)
#ifdef AFS_LINUX26_ENV
/* Some Linux kernels have a bug where we are not woken up immediately from a
* select() when data is available. Work around this by having a low select()
* timeout, so we don't hang in those situations. */
# define SYNC_SELECT_TIMEOUT 10
#else
# define SYNC_SELECT_TIMEOUT 86400
#endif
#ifdef USE_UNIX_SOCKETS
#include <afs/afsutil.h>
#include <sys/un.h>

5
src/vol/fssync-server.c
View File

@ -310,12 +310,15 @@ FSYNC_sync(void * args)
CallHandler(FSYNC_readfds, nfds, POLLIN|POLLPRI);
#else
int maxfd;
struct timeval s_timeout;
GetHandler(&FSYNC_readfds, &maxfd);
s_timeout.tv_sec = SYNC_SELECT_TIMEOUT;
s_timeout.tv_usec = 0;
/* Note: check for >= 1 below is essential since IOMGR_select
* doesn't have exactly same semantics as select.
*/
#ifdef AFS_PTHREAD_ENV
if (select(maxfd + 1, &FSYNC_readfds, NULL, NULL, NULL) >= 1)
if (select(maxfd + 1, &FSYNC_readfds, NULL, NULL, &s_timeout) >= 1)
#else /* AFS_PTHREAD_ENV */
if (IOMGR_Select(maxfd + 1, &FSYNC_readfds, NULL, NULL, NULL) >= 1)
#endif /* AFS_PTHREAD_ENV */

5
src/vol/salvsync-server.c
View File

@ -342,11 +342,14 @@ SALVSYNC_syncThread(void * args)
for (;;) {
int maxfd;
struct timeval s_timeout;
GetHandler(&SALVSYNC_readfds, &maxfd);
s_timeout.tv_sec = SYNC_SELECT_TIMEOUT;
s_timeout.tv_usec = 0;
/* Note: check for >= 1 below is essential since IOMGR_select
* doesn't have exactly same semantics as select.
*/
if (select(maxfd + 1, &SALVSYNC_readfds, NULL, NULL, NULL) >= 1)
if (select(maxfd + 1, &SALVSYNC_readfds, NULL, NULL, &s_timeout) >= 1)
CallHandler(&SALVSYNC_readfds);
}