RE: Patch for Apani Nortel VPN Client to build against kernel 2.6.22 help/review

!MAILaRCHIVE_VOTE_RePLACE
Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]
To: Nick Bowler <nbowler@...>
Cc: <linux-kernel@...>, <samuel.gelineau@...>
Date: Tuesday, September 11, 2007 - 12:57 pm

> -----Message d'origine-----

(Forwarding info by Samuel to lkml...)

Thanks for your comments, here is a new patch (bowler.patch) reflecting =
the changes you suggested. Unfortunately, and unsurprisingly for such =
cosmetic changes, the code still hangs.

I have added a great many debugging statements in order to pinpoint the =
problem (backtrace.patch); surprisingly, my custom backtraces didn't =
show up in the console, but a kernel call trace did (call_trace).

Here is the contents of the offending function. The {enter,exit}_func() =
calls are my debugging statements, and I have verified that only the =
else branch was compiled. As you can see, there are only function calls =
here, no memory manipulation nor dereferencing at all. What could =
possibly fail in this function? Shouldn`t the backtrace include one of =
the subcalls?

  void nl_spin_lock_irqsave(spinlock_t *lockptr, DWORD *flagptr)
  {
  enter_func(__LINE__);
  #if (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,7))
          assert(lockptr);
          assert(flagptr);
          spin_lock_irqsave((spinlock_t *)lockptr, (*flagptr));
  #else
          spin_lock_irq((spinlock_t *)lockptr);
  #endif
  exit_func(__LINE__);
  }

- Samuel G=E9lineau
Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]

Messages in current thread:
Patch for Apani Nortel VPN Client to build against kernel 2...., Fortier,Vincent [Montreal]..., (Wed Aug 22, 11:38 am)
RE: Patch for Apani Nortel VPN Client to build against kerne..., Fortier,Vincent [Montreal]..., (Tue Sep 11, 12:57 pm)