Re: [alsa-devel] [PATCH] mips/ad1843: convert time to jiffies HZ independent

From: Takashi Iwai
Date: Mon Jun 15 2015 - 14:40:32 EST


At Mon, 15 Jun 2015 18:27:27 +0200,
Nicholas Mc Guire wrote:
>
> On Mon, 15 Jun 2015, Takashi Iwai wrote:
>
> > At Sun, 14 Jun 2015 19:16:59 +0200,
> > Nicholas Mc Guire wrote:
> > >
> > > On Sun, 14 Jun 2015, Takashi Iwai wrote:
> > >
> > > > At Sun, 14 Jun 2015 10:20:59 +0200,
> > > > Nicholas Mc Guire wrote:
> > > > >
> > > > > PI compliance scanning with coccinelle flagged:
> > > > > ./sound/mips/ad1843.c:503:2-32: WARNING:
> > > > > timeout (5) seems HZ dependent
> > > > >
> > > > > This was introduced in 'commit 862c2c0a61c5 ("ALSA: ALSA driver for SGI O2
> > > > > audio board")'. schedule_timeout_interruptible() expects a timeout in
> > > > > jiffies so the numeric constant makes the effective timeout HZ dependent.
> > > > > Simply put it through msecs_to_jiffies() to make it HZ independent.
> > > > >
> > > > > Signed-off-by: Nicholas Mc Guire <hofrat@xxxxxxxxx>
> > > > > ---
> > > > >
> > > > > The loop here is bounded by 500 jiffies and it does not seem reasonable that
> > > > > it would loop 10 times faster on HZ=1000 configs over HZ=100 configs. The
> > > > > conversion via msecs_to_jiffies(5) ensures that it will loop at most 100
> > > > > times for all configs of HZ which seems reasonable (to me) but needs to be
> > > > > checked by someone that knows the details of this driver.
> > > >
> > > > Using 5ms looks reasonable for this case. However,
> > > > schedule_timeout_interruptible() isn't. This is a loop that won't
> > > > break. So, the best would be to replace it with msleep() or its
> > > > friends.
> > > >
> > > > > Note that the CONFIG_SND_PCM_OSS dependency was not listed in the
> > > > > Depends on: in Kconfig (which it probably should)
> > > >
> > > > Why at all? It has nothing to do with OSS.
> > > >
> > > the procedure to see the problem I used was
> > > (linux-next 20150612)
> > > make ip32_defconfig ARCH=mips
> > > make menuconfig ARCH=mips
> > > Device Drivers
> > > ...
> > > <M> Sound card support --->
> > > <M> Advanced Linux Sound Architecture --->
> > > [*] MIPS sound devices --->
> > > <M> SGI O2 Audio
> > >
> > > compiling fails with:
> > > MODPOST 95 modules
> > > ERROR: "snd_pcm_period_elapsed" [sound/mips/snd-sgi-o2.ko] undefined!
> > > ERROR: "snd_pcm_set_ops" [sound/mips/snd-sgi-o2.ko] undefined!
> > > ERROR: "snd_pcm_lib_get_vmalloc_page" [sound/mips/snd-sgi-o2.ko] undefined!
> > > ERROR: "snd_pcm_lib_ioctl" [sound/mips/snd-sgi-o2.ko] undefined!
> > > ERROR: "_snd_pcm_lib_alloc_vmalloc_buffer" [sound/mips/snd-sgi-o2.ko] undefined!
> > > ERROR: "snd_pcm_new" [sound/mips/snd-sgi-o2.ko] undefined!
> > > ERROR: "snd_pcm_lib_free_vmalloc_buffer" [sound/mips/snd-sgi-o2.ko] undefined!
> > > make[1]: *** [__modpost] Error 1
> > > make: *** [modules] Error 2
> > >
> > > as the pcm_lib functions depend on SND_PCM I added that - So
> > > if I add the below patch the problem goes away and the module
> > > build correclty - not saying that this is the right fix though
> > > (did not look into it in more detail yet)
> >
> > Yes, this dependency fix is correct, and I queued your patch now.
> > But the dependency on CONFIG_SND_PCM_OSS is wrong.
> >
>
> Not too suprised I got this wrong - which is also why I did not submit
> this patch - but if you took care of it all the better !
>
> > A similar fix is needed for HAL2, and I fixed it, too.
> >
>
> thx!
> hofrat

Oh, the original patch to fix jiffies value wasn't applied yet, since
I prefer replacing the call with msleep(). Could you respin?


thanks,

Takashi
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/