-
Here's something interesting out of the strace:
9767 stat64("/var/run/pam_mount/djk7wb", {st_mode=S_IFREG|0600, st_size=3, ...}) = 0
9767 read(1, "0x1", 19) = 3
9767 unlink("/var/run/pam_mount/djk7wb") = 0
9767 close(1) = 0
9767 fstat64(1, 0xbfb8988c) = -1 EBADF (Bad file descriptor)
9767 mmap2(NULL, 8192, PROT_READ|PROT_WRITE...
2009-02-11 14:57:16 UTC in pam_mount module
-
I can also note that pmvarrun is indeed executed, as the login count correctly shows zero after this test. Just nothing is passed back to pam_mount from the execution of pmvarrun.
2009-02-10 22:17:43 UTC in pam_mount module
-
Here's a better description from syslog:
Feb 10 16:11:09 r80djk7wb login: command: [pmvarrun] [-u] [djk7wb] [-o] [1]
Feb 10 16:11:09 r80djk7wb login: pam_mount(misc.c:38): set_myuid: (uid=0, euid=0, gid=0, egid=0)
Feb 10 16:11:09 r80djk7wb login: pam_mount(misc.c:38): set_myuid: (uid=0, euid=0, gid=0, egid=0)
Feb 10 16:11:09 r80djk7wb login: pam_mount(pam_mount.c:419): pmvarrun...
2009-02-10 22:16:35 UTC in pam_mount module
-
Sorry, that was me. Anyway, I'm assuming the latest changes in git are designed to give more debugging info when pam_mount has debugging enabled.--I'll be glad to try that today at work.
2009-02-10 13:48:16 UTC in pam_mount module
-
On login, pam_mount correctly reads the login count and reports it. On logout, however, pam_mount shows the error, "error reading login count from pmvarrun" and thus will not attempt unmounting. The *only* difference between the two modify_pm_count seems to be that the logout runs with the gid and egid of my user (but uid and euid of root), while the login runs with uid, euid, gid, and egid of...
2009-02-09 23:02:17 UTC in pam_mount module
-
pmvarrun will segfault when given no arguments. Seems to be that set_defaults() sets the user string to NULL, and when strlen is called on this, it segfaults. Setting the string to an empty string (just "") will resolve this problem.
2009-02-09 21:06:16 UTC in pam_mount module
-
The logic is still an issue in the latest pam_mount release:
(icase && strcasecmp(*wp, user)) == 0)
Should read:
(icase && (strcasecmp(*wp, user)==0)))
Current code effectively means if (!icase || !strcasecmp(*wp, user))
Sorry about any confusion...
2009-02-06 21:41:40 UTC in pam_mount module
-
The function user_in_sgrp returns the incorrect value (almost) all the time. The problem is with the string comparison (line 551, in rdconf1.c from pam_mount-1.0):
if (strcmp(*wp, user) ||
(icase && strcasecmp(*wp, user)) == 0)
return true;
Logically, this returns true if the group member is not a case-sensitive match to the username... or, if either case-insensitive matching is off,
2008-10-13 16:19:23 UTC in pam_mount module
-
dougk_ff7 committed patchset 3145 of module phpBB2 to the phpBB CVS repository, changing 3 files.
2002-11-29 06:58:37 UTC in phpBB
-
dougk_ff7 committed patchset 3144 of module phpBB2 to the phpBB CVS repository, changing 7 files.
2002-11-29 05:46:17 UTC in phpBB
