[uClibc]realloc broken (Was: Re: pthread implementation broken with -D_GNU_SOURCE)

[Ronald Wahl]

On 28 Jun 2002 17:08:44 +0200, Ronald Wahl wrote:

> On 27 Jun 2002 16:14:28 +0200, Ronald Wahl wrote:
>> Hallo,
>> I just noticed that some routines like pthread_mutex_lock fail if the
>> calling program was compiled with -D_GNU_SOURCE. The reason is that
>> these functions do not check all types of mutexes - in my case
>> PTHREAD_MUTEX_TIMED_NP (the standard mutex type) falls through the
>> switch statement and so pthread_mutex_lock returns EINVAL. Maybe
>> the pthread library should be synced with a newer one (these include
>> the missing code). I hope this can be fixed soon.

> ... and am I right when I say that because of this bug parts of uClibc
> (these that use the statical PTHREAD_MUTEX_INITIALIZER) are currently
> not thread-safe? Since the return value of pthread_mutex_lock is not
> checked the bug remains hidden.

Everything is broken ;-( Now realloc - if used with _working_
mutexes. I have two fixes - one to get the locks working and one to fix
the malloc problem. Get them here and incorporate them into uClibc:

        http://www.peppercon.de/~rwa/malloc-nodeadlock.patch

        http://www.peppercon.de/~rwa/libpthread.patch

Maybe the mutex patch uncovers some other hidden bugs...

thanx,
ron

-- 
\\      Dipl.-Inf. Ronald Wahl   |   Peppercon AG                  //
 \\\         rwa at peppercon.com  |||  http://www.peppercon.com/   /// 
  \OO  -----------------------  OOO  -------------------------  OO/ 
   OO  GnuPG/PGP key available  OOO  Keep Systems running       OO