Re: Kernel panic in PCI driver

[jasjit singh <singh.jasjit@xxxxxxxxxxx>]

Hi

>Hi..
>Not an expert...but trying to help...
>Anyway, I think you're sending HTML based message here...so it that's
>true, please use plain text next time..

On Thu, Apr 24, 2008 at 1:57 PM, jasjit singh <singh.jasjit@xxxxxxxxxxx> wrote:
>>
>> Hi
>>
> >I am using following functions in my driver for a PCI device.
>>
>> wait_event_interruptible and wake_up_interruptible.
>>
> >my algorithm goes like this
>>
> >fun1()
>>{
>> waits for an event whose source is an interrupt from the PCI device; /*i.e
> >calls wait_event_interruptible*/
> >}
>>
>>fun2()
>> {
>> Check if there is an interrupt from the PCI device;
>> if so, give wake up call to queue that is waited upon in fun1;/*i.e calls
>> wake_up_interruptible*/
>> }

>Maybe that's the problem.. It's awaken too many times, then it ate kernel stack.

Correct me if i am misinterpreting your point. I am not able to get how waking up is eating the stack and that too in an accumulated fashion.

>I suggest to redesign your code so the "event handler" just do the
>quick jobs such as ACK-ing the interrupt...and for the rest of lengthy
>jobs, put in on workqueue.

I will look into my code. Thanks for replying

>regards,

>Mulyadi.

---

Sent from Yahoo! Mail.
A Smarter Email.