buggy GFP_KERNEL allocators

[Rik van Riel]

On Wed, 26 Jan 2000, Ben LaHaise wrote:
> On Wed, 26 Jan 2000 kuznet en ms2.inr.ac.ru wrote:
> 
> > Hello!
> > 
> > > +				current->state = TASK_RUNNING;	/* We *must* do this before touching userspace! */
> > 
> > This smells suspiciously. Page fault code should not stuck only because
> > task state is not running. It contaradicts all the idea behind wait queues.
> > 
> > Essentially, fault is like signal. It must move task state to running.
> 
> Applying the change to the page fault handler might just be the
> right thing to do, and in fact it would make sense in 2.3 to have
> the fault handler also switch to lazy tlb mode if the task has to
> perform io.  The fault handler does neither of these things right
> now, and in fact this was causing occasional stuck tasks if the
> fault ended up calling schedule before changing current->state.

Won't we confuse lots of pieces of code when their
TASK_UNINTERRUPTIBLE process is returned to them as
a TASK_RUNNING one?

I'm somewhat afraid that we will break code whichever
way we go. That's why I've added a no-sleep workaround
to __get_free_pages() (hey, this isn't happening often
enough to run us out of memory) and added the printk()
to warn us about potentially dangerous code.

regards,

Rik
--
The Internet is not a network of computers. It is a network
of people. That is its real strength.


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo en vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/