Re: [Patch} to fix oops in olympic token ring driver on media disconnect

[Neil Horman <nhorman@redhat.com>]

Jeff Garzik wrote:

> Pete Zaitcev wrote:
>
>> On Fri, 13 Aug 2004 23:07:16 +0100
>> Alan Cox <alan@lxorguk.ukuu.org.uk> wrote:
>>
>>
>>> On Gwe, 2004-08-13 at 20:11, Neil Horman wrote:
>>>
>>>> the olympic_close routine was waiting on.  This patch cleans that up.
>>>>
>>>> Tested by me, on 2.4 and 2.6 with good, working results, and no 
>>>> more oopses.
>>>
>>>
>>> Should it not be blocking the IRQs on the chip as well ?
>>
>>
>>
>> I assumed that old olympic_close() did, but perhaps it didn't after all.
>> There is nothing like the following in it:
>>
>> +#define DISABLE_IRQS(base_addr) do { \
>> +   writel(LISR_LIE,(base_addr)+LISR_RWM);\
>> +   writel(SISR_MI,(base_addr)+SISR_RWM);\
>> +} while(0)
>>
>> This is curious. If something was never used, how do we know if it 
>> works?
>> Maybe it's safer just to leave things as Neil did.
>
>
>
> Well, regardless, Neil's patch is IMO a good first step.
>
> There is plenty of work in olympic for any motivated person :)
>
>     Jeff
>
>
Thanks :) I figure if someone reports system hangs or lock-ups with this 
card in place, I just re-add teh irq disable stuff.
Neil