[PATCH] B43: Handle DMA RX descriptor underrun

[Rafał Miłecki]

2013/5/5 Michael Büsch <m at bues.ch>:
> On Sun, 5 May 2013 18:31:20 +0200
> Rafał Miłecki <zajec5 at gmail.com> wrote:
>
>> Still worth considering is my previous e-mail. Why writing (for
>> example) 1 to RXSTOPINDEX doesn't stop firmware from using slot 1?
>
> What makes you think this register does not work?

Take a look at this:

[  327.224976] [DBG] old current:5      new current:6
[  327.224982] [DBG] reading slot 5
[  327.224997] [DBG] writing stop slot 6

In above ring->slot was 5, but IRQ was generated, and we read new
"current" using get_current_rxslot. It appeared to be 6. So we read
packet from slot 5 and then called
ops->set_current_rxslot(ring, 6);
AFAIU hardware shouldn't use slot 6, right? But take a look at what
happens next:

[  327.319582] [DBG] old current:6      new current:7
[  327.319590] [DBG] reading slot 6
[  327.319619] [DBG] writing stop slot 7

Hardware generated IRQ and we get_current_rxslot returned 7. It means
we're allowed to read slots up to 7 (excluding). It other words it
means firmware used slot 6... but 100ms earlier we forbid firmware to
use slot 6!

This is part I don't understand. It looks like firmware ignores what
we set with the ops->set_current_rxslot


> Do you write a 1 to the register, or do you mean "the offset that
> corresponds to slot 1"?

I mean "the offset that corresponds to slot 1". In code it's:
ops->set_current_rxslot(ring, 1);


>> Can it be because it's "too late"? For example:
>> 1) Firmware writes to slot 0, changes "curr descr" to 1
>> 2) Firmware grabs slot 1
>> 3) Firmware generates IRQ
>> 4) Driver reads slot 0
>> 5) Driver writes RX stop index 1
>
> The stop index is the index that the firmware does _not_ write to.
> If you set it _after_ it already wrote, you clearly are too late (for
> this ring wrapping).

Take a look at above log. First we forbid firmware to use slot 6:
[  327.224997] [DBG] writing stop slot 6
but 100ms later firmware used that slot:
[  327.319582] [DBG] old current:6      new current:7
(it filled slot 6 and bumped value in B43_DMA64_RXSTATUS)

-- 
Rafał