Re: OHCI1394:sleeping function called from invalid context [need idea for fixing]

[kernel-stuff]

Okay - So the dma_pool_destroy problem can be solved by schedule_work() (that's what I think allows to create and schedule work queues, correct me if I am wrong). For allocation - I think it cannot be scheduled as the driver needs memory then and there.  How can that be handled? (dma_pool_create doesn't seem to be taking any flags like SLAB_ATOMIC - which would have solved the problem..)

Thanks!

Parry


> Am Mittwoch, 1. Dezember 2004 06:22 schrieb kernel-stuff@xxxxxxxxxxx:
> > I am trying to debug  a problem with the OHCI1394 module. Basically i get 
> "sleeping function called from invalid context" in dmesg whenever I capture 
> video from my camcorder. Looking at the stack trace - dma_pool_create is called 
> from within IRQ handler - or so it seems. Similarly the ohci1394 module also 
> calls dma_pool_destroy in IRQ. This happens when I connect and disconnect the 
> camcorder.
> > I am thinking of ways to fix this - and have no clue so far. I fixed couple 
> other similar but easy-to-fix ones in ohci1394.c - by changing from GFP_KERNEL 
> to GFP_ATOMIC but this one looks more complicated - DMA Pool needs to be freed 
> when device disconnects - how to do this outside of the IRQ? More complicated 
> (to me of course :) is the allocaction part - driver needs a DMA pool on device 
> connect - no idea how to get it outside of the IRQ path.
> > 
> > Can anyone give an idea how to go about fixing this? Normally what's the 
> correct place for a module to call dma_pool_create/destroy?
> 
> You might switch to work queues.
> 
> 	HTH
> 		Oliver
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/