RE: [RFC] ATH9K: infinite loop in Tasklet

["voncken" <cedric.voncken@acksys.fr>]

	Yes, 

	I used an mpc8314 at 400Mhz, with 128 Mbit of RAM for my test.
	
	To find it, I measured the time elapsing in ath_rxbuf_alloc(..) with
the kernel function local_clock().

	With SLOB I found around 123 us (and I have an infinite loop in
ath9k tasklet, because when I have consumed one packet the next packet is
ready to rx process)
	With SLAB I found around 22 us
	With SLUB I found around 10 us

Cedric Voncken 

-----Message d'origine-----
De : adrian.chadd@gmail.com [mailto:adrian.chadd@gmail.com] De la part de
Adrian Chadd
Envoyé : mercredi 2 janvier 2013 17:22
À : voncken
Cc : Felix Fietkau; Mohammed Shafi; linux-wireless@vger.kernel.org
Objet : Re: [RFC] ATH9K: infinite loop in Tasklet

... interesting. The allocator choice is enough to cause your CPU to run too
far behind the workload?



Adrian

On 2 January 2013 05:17, voncken <cedric.voncken@acksys.fr> wrote:
>         Thanks for your answer.
>
>         I found a workaround for my problem.
>
>         In my Linux kernel, the default SLAB allocator was changed in 
> default settings.
>         SLOB  was used in place of SLAB.
>         I select the SLAB allocator and that fix the default because 
> it runs faster.
>
>         However  the infinite loop can still happen if the cpu is 
> heavily loaded with other tasklets.
>
>         Best regards.
>
> Cedric Voncken
> -----Message d'origine-----
> De : adrian.chadd@gmail.com [mailto:adrian.chadd@gmail.com] De la part 
> de Adrian Chadd Envoyé : mercredi 2 janvier 2013 00:41 À : Felix 
> Fietkau Cc : Mohammed Shafi; Cedric VONCKEN; 
> linux-wireless@vger.kernel.org Objet : Re: [RFC] ATH9K: infinite loop 
> in Tasklet
>
> On 1 January 2013 07:18, Felix Fietkau <nbd@openwrt.org> wrote:
>
>> I think the best way to properly fix this issue is to implement NAPI 
>> support (which mac80211 already supports).
>
> Sure, but there may be some underlying issues (maybe even on just that 
> particular platform) that need to be addressed.
>
> Migrating to NAPI / breaking out of the tasklet early should also be 
> done, but it's potentially orthogonal to the OP's problem.
>
>
> Adrian
>
> --
> To unsubscribe from this list: send the line "unsubscribe 
> linux-wireless" in the body of a message to majordomo@vger.kernel.org 
> More majordomo info at  http://vger.kernel.org/majordomo-info.html