AR9462 PCIe1x card: endless messages in dmesg

[Adrian Chadd]

On 22 January 2015 at 07:19, Alexey Dokuchaev <danfe at nsu.ru> wrote:
> On Wed, Jan 21, 2015 at 08:06:09AM -0800, Adrian Chadd wrote:
>> On 21 January 2015 at 01:20, Alexey Dokuchaev <danfe at nsu.ru> wrote:
>> > Just installed this AR9462-based PCIe (1x) card into my work i386 desktop
>> > running fortnight-old -CURRENT (r276691).  So far so good -- it delivers
>> > pretty stable, lagless Internet experience, but keeps shitting in kernel
>> > buffer very quickly with these (tons of them):
>> >
>> >     ath0: ath_edma_rxbuf_alloc: nothing on rxbuf?!
>> >     ath0: ath_edma_rxfifo_alloc: Q1: alloc failed: i=0, nbufs=128?
>>
>> Put ATH_ENABLE_11N in your kernel config.
>
> This helped, thank you.  Message storm is gone.  It seems I've been stripping
> GENERIC overly aggressively.  Would it make sense to add these lines to the
> /usr/src/sys/modules/ath_pci/Makefile:
>
>   .if !defined(KERNBUILDDIR)
>   CFLAGS+= -DATH_ENABLE_11N <... maybe some other useful defines ...>
>   .endif
>
> ?  This way, in-kernel ath_pci(4) would obey kernel config, yet module will
> always have them by default.  As a bonus, this would allow to build it with
> simply calling make(1) from the module directory.

I may eventually do that. I'm just wary of it all because various ABI
changes occur with different options and I don't want people bitten by
that.

(I really should just up the default set of tx/rx buffers; that'd at
least fix the error.)

-adrian