On Saturday 24 February 2007 16:22, Sergey Temerkhanov wrote:
> On Saturday 24 February 2007 16:14, you wrote:
> I've tried to send files 2 times - first ime panic occured after about 900M
> and 2nd - after 20M.
> I'll try to insert some checks in NdisFreePacket and look what exactly is
> going on.
> > --- Sergey Temerkhanov <temerkhanov@...> wrote:
> > > Still panics (log attached).
> > You are expected to give details - just the log won't help.
> > Did it crash right away? Managed to send a file? Better or worse than
> > before?
> > It looks to me crash happens in NdisFreePacket. If you can sprinkle some
> > trace messages (with INFO macro), to locate where exactly it is crashing,
> > it will help.
> > Giri
> > _________________________________________________________________________
> >__ _________ Do you Yahoo!?
> > Everyone is raving about the all-new Yahoo! Mail beta.
> > http://new.mail.yahoo.com
Looks like something nasty happens to "pool" variable.
I've tried to insert printk's into the NdisFreePacket and found out that crash
occurs after "irql = nt_spin_lock_irql(&pool->lock, DISPATCH_LEVEL);" line.
Btw, adding printk's makes crash much more hard to reproduce as it happens
much later (after 1G).
Then I loaded ndiswrapper.ko into gdb and examined it with "list
*(NdisFreePacket+0x9d)" (address where panic occurs). It gave me this output:
0x5ded is in NdisFreePacket
852 #ifdef CONFIG_SMP
854 static inline void nt_spin_lock(volatile NT_SPIN_LOCK *lock)
856 __asm__ __volatile__(
859 " xchgl %1, %0\n\t"
860 " cmpl %2, %1\n\t"
Of course, I'll try with vanilla-sources (BTW, will gentoo-sources suffice?)
but I don't think it's somehow involved.