Problem is as follows: accept_mutex_init() uses popenf on the config
pool to open its file descriptor... which means that it gets closed when
the config pool is cleared on restarts. But the code is "clever" --- if
it has already opened the file, it doesn't reopen it. The fix (now
being tested) is not to be so clever:
*** http_main.c~ Thu Aug 17 11:41:05 1995
--- http_main.c Fri Aug 18 10:01:52 1995
***************
*** 152,159 ****
{
char lock_fname[19];
- if (lock_fd >= 0) return; /* Already initialised */
-
strcpy(lock_fname, "/usr/tmp/htlock.XXXXXX");
if (mktemp(lock_fname) == NULL || lock_fname[0] == '\0')
Assuming this works, I'll have a fixed 0.8.10 ready in half an hour, tops;
I see no reason not to release the result, but by strict formal rules, we
would need another 24 hours to vote on the above patch...
rst
pool to open its file descriptor... which means that it gets closed when
the config pool is cleared on restarts. But the code is "clever" --- if
it has already opened the file, it doesn't reopen it. The fix (now
being tested) is not to be so clever:
*** http_main.c~ Thu Aug 17 11:41:05 1995
--- http_main.c Fri Aug 18 10:01:52 1995
***************
*** 152,159 ****
{
char lock_fname[19];
- if (lock_fd >= 0) return; /* Already initialised */
-
strcpy(lock_fname, "/usr/tmp/htlock.XXXXXX");
if (mktemp(lock_fname) == NULL || lock_fname[0] == '\0')
Assuming this works, I'll have a fixed 0.8.10 ready in half an hour, tops;
I see no reason not to release the result, but by strict formal rules, we
would need another 24 hours to vote on the above patch...
rst