Re: [PATCH 3/3] dp8393x: fix receiving buffer exhaustion

[Laurent Vivier <laurent@vivier.eu>]

Le 05/11/2019 à 21:45, Hervé Poussineau a écrit :
> Le 02/11/2019 à 18:15, Laurent Vivier a écrit :
>> The card is not able to exit from exhaustion state, because
>> while the drive consumes the buffers, the RRP is incremented
>> (when the driver clears the ISR RBE bit), so it stays equal
>> to RWP, and while RRP == RWP, the card thinks it is always
>> in exhaustion state. So the driver consumes all the buffers,
>> but the card cannot receive new ones.
>>
>> This patch fixes the problem by not incrementing RRP when
>> the driver clears the ISR RBE bit.
>>
>> Signed-off-by: Laurent Vivier <laurent@vivier.eu>
>> ---
>>   hw/net/dp8393x.c | 31 ++++++++++++++++---------------
>>   1 file changed, 16 insertions(+), 15 deletions(-)
> 
> I checked the DP83932C specification, available at
> https://www.eit.lth.se/fileadmin/eit/courses/datablad/Periphery/Communication/DP83932C.pdf
> 
> 
> In the Buffer Resources Exhausted section (page 20):
> "To continue reception after the last RBA is used, the system must supply
> additional RRA descriptor(s), update the RWP register, and clear the RBE
> bit in the ISR. The SONIC rereads the RRA after this bit is cleared."
> 
> If I understand correctly, if the OS updates first the RWP and then
> clear the RBE bit,
> then RRP should be different of RWP in dp8393x_do_read_rra() ? Or did I
> miss something?

No, I found that by debugging the problem, I didn't have the
documentation. I'll rework this patch, relying on the doc to better
understand the problem.

Thanks,
Laurent