[Bug nscd/10402] New: nscd locking severely broken

[Bug nscd/10402] New: nscd locking severely broken

[pasky at suse dot cz]

The commit 20e498bd removes the pthread_mutex_rdlock() calls, but not the
corresponding pthread_mutex_unlock() calls. Also, the database lock is never
unlocked in one branch of the mempool_alloc() if.

I think unreproducible random assert(dh->usable) crashes in prune_cache() were
caused by this. But an easy way to make nscd threads hang with the broken
locking was

  while true; do id root; id tttest; nscd -i passwd; done

-- 
           Summary: nscd locking severely broken
           Product: glibc
           Version: 2.10
            Status: NEW
          Severity: critical
          Priority: P2
         Component: nscd
        AssignedTo: drepper at redhat dot com
        ReportedBy: pasky at suse dot cz
                CC: glibc-bugs at sources dot redhat dot com


http://sourceware.org/bugzilla/show_bug.cgi?id=10402

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug nscd/10402] nscd locking severely broken

[pasky at suse dot cz]

------- Additional Comments From pasky at suse dot cz  2009-07-16 14:38 -------
Created an attachment (id=4056)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=4056&action=view)
proposed patch


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=10402

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug nscd/10402] nscd locking severely broken

[drepper at redhat dot com]

------- Additional Comments From drepper at redhat dot com  2009-07-16 17:11 -------
Applied to upstream.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


http://sourceware.org/bugzilla/show_bug.cgi?id=10402

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug nscd/10402] nscd locking severely broken

[drepper at redhat dot com]

------- Additional Comments From drepper at redhat dot com  2009-07-17 14:51 -------
Actually, the patch is wrong.  The unlocking deliberately doesn't happen at
those times.  Keeping the lock is the key to handling memory allocations in a
thread-safe way.

I reverted the patch.

What is the problem you were trying to solve?  Always provide that information.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |


http://sourceware.org/bugzilla/show_bug.cgi?id=10402

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug nscd/10402] nscd locking severely broken

[drepper at redhat dot com]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |WAITING


http://sourceware.org/bugzilla/show_bug.cgi?id=10402

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug nscd/10402] nscd locking severely broken

[pasky at suse dot cz]

------- Additional Comments From pasky at suse dot cz  2009-07-21 10:17 -------
You are right, I forgot the locking scheme since I've reviewed your patch that
adds it; too bad there isn't a single comment in the code alerting the reader of
it. :(

I have already described the testcase in my first comment; is it not hanging for
you?

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |NEW


http://sourceware.org/bugzilla/show_bug.cgi?id=10402

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug nscd/10402] nscd locking severely broken

[jakub at redhat dot com]

------- Additional Comments From jakub at redhat dot com  2009-07-21 10:46 -------
That reproducer is quite vague.  You haven't provided your /etc/nsswitch.conf,
nor details about root and ttest user, etc.
Assuming only files for passwd and group databases and root
root:x:0:0:root:/root:/bin/bash
and non-existing ttest user, I can't reproduce this (at least not within 100000
cycles of that loop on a quad core box).

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |WAITING


http://sourceware.org/bugzilla/show_bug.cgi?id=10402

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug nscd/10402] nscd locking severely broken

[pasky at suse dot cz]

------- Additional Comments From pasky at suse dot cz  2009-08-25 14:37 -------
I'm sorry, I wasn't indeed careful enough when testing - it turns out to be
specific to passwd: compat and +:::::: entry in /etc/passwd.

I'm debugging this further.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |NEW


http://sourceware.org/bugzilla/show_bug.cgi?id=10402

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug nscd/10402] nscd locking severely broken

[pasky at suse dot cz]

------- Additional Comments From pasky at suse dot cz  2009-08-25 22:46 -------
I'm sorry for the false bugreport, it turned out to be a trivial messup in a
local patch we have that introduces proper support for negtimeout == 0 - I
severely underestimated how much of a havoc can it cause. :-(

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |WORKSFORME


http://sourceware.org/bugzilla/show_bug.cgi?id=10402

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug nscd/10402] nscd locking severely broken

[fweimer at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=10402

Florian Weimer <fweimer at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
              Flags|                            |security-

-- 
You are receiving this mail because:
You are on the CC list for the bug.