Messages in this thread | | | Date | Mon, 22 Jun 2009 10:04:53 +0100 | From | Alan Cox <> | Subject | Re: BUG: scheduling while atomic |
| |
On Sat, 20 Jun 2009 21:54:52 +0300 Sergey Senozhatsky <sergey.senozhatsky@mail.by> wrote:
> On (06/20/09 23:13), Rabin Vincent wrote: > > On Fri, Jun 19, 2009 at 09:15:29PM +0300, Sergey Senozhatsky wrote: > > > Hello. > > > I see lots of "BUG: scheduling while atomic" with current kernels (started with 2.6.30-git1, 2.6.30 is ok). > > > For example (2.6.30-git14): > > > > Looks like the net softirq (via ppp) ends up calling tty_throttle(), and > > this causes problems because of the mutex_lock() added there by > > 38db8979. I've cc'd the tty maintainer.
Looks like we caught another naughty. It's not safe to call the throttle/unthrottle methods from an IRQ path, never really has been as they can be things like USB drivers that need to send USB messages and sleep when that happens.
Alan
| |