public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Vojtech Pavlik <vojtech@suse.cz>
To: Edward Stempel <eazstempel@cal009001.student.utwente.nl>
Cc: linux-kernel mailinglist <linux-kernel@vger.kernel.org>
Subject: Re: DMA conflicts with soundcard for ide driver via82cxxx
Date: Sat, 5 Jan 2002 18:19:42 +0100	[thread overview]
Message-ID: <20020105181942.A26258@suse.cz> (raw)
In-Reply-To: <20020105172336.B25998@suse.cz> <Pine.LNX.4.21.0201051758010.12020-100000@nieuw3.stempel.dhs.org>
In-Reply-To: <Pine.LNX.4.21.0201051758010.12020-100000@nieuw3.stempel.dhs.org>; from eazstempel@cal009001.student.utwente.nl on Sat, Jan 05, 2002 at 06:03:59PM +0100

On Sat, Jan 05, 2002 at 06:03:59PM +0100, Edward Stempel wrote:
> On Sat, 5 Jan 2002, Vojtech Pavlik wrote:
> 
> > On Sat, Jan 05, 2002 at 05:17:28PM +0100, Edward Stempel wrote:
> > > On Tue, 1 Jan 2002, Vojtech Pavlik wrote:
> > > 
> > > > On Tue, Jan 01, 2002 at 11:03:52PM +0100, Edward Stempel wrote:
> > > > > In my BIOS-setup there is an option "PCI latency" which defaults to
> > > > > 32. First I changed it to 128. That worked perfect. Later I have decreased
> > > > > it to 64, 48 and 40. But 40 still gave some problems (a little bit
> > > > > distortion of my sound). After that I tried 44, but the disk was to slow
> > > > > with this setting (30 MB/s instead of 40 MB/s). So now I have a PCI
> > > > > latency of 48. The strange thing is that it seems not to work
> > > > > directly. After the system boot when I do hdparm -t, I get about 5
> > > > > MB/s. But then when I switch using_dma off and on again using hdparm, I
> > > > > get the full 40 MB/s and no distortion of my es1371. Any ideas what might
> > > > > cause this?
> > > > 
> > > > You'd have to tell me all other settings in your setup first so I can
> > > > guess the cause. What options do you use? Is DMA enabled automatically?
> > > > Etc, etc ...
> > > 
> > > I am not sure which settings you need. But I compiled a kernel with:
> > > - Generic PCI IDE chipset support
> > > - Sharing PCI IDE interrupt support
> > > - Generic PCI bus-master DMA support
> > > - Use PCI DMA by default when available
> > > - VIA82CXXX chipset support
> > > 
> > > After booting this kernel hdparm -v reports that I am using dma, but
> > > hdparm -t only reports about 5 MB/s data transfer rate. When I switch the
> > > using of dma off and on with hdparm -d the transfer rate becomes about
> > > 40MB/s
> > 
> > What does hdparm -i say before and after? Is there any difference (the
> > asterisk)?
> 
> Sorry, I cannot reproduce this behaviour again. (I just did a reboot) I am
> not sure what happened, as I am using the same kernel and bios-setup. Kind
> of strange non-deterministic behaviour here. However, as far as  can
> remember (it has been some days ago) I did not see something different in
> hdparm -i before ore after the switching of and on. Both times udma-5 were
> reported by hdparm.
> 
> As I cannot reproduce this behaviour now, I suggest we forget about
> it. Maybe I did something wird when this happened. I will contact you
> whenever I notice the same behaviour again.

Ok; thanks.

> 
> 
> > 
> > > The PCI latency is set in the BIOS setup, and according to /proc/pci this
> > > latency is applied to all PCI devices except my matrox G550 video card,
> > > which has a latency of 64.
> > > 
> > > 
> > > I hope this explains something.
> > 
> > Not much yet.
> > 
> > > 
> > > 
> > > 
> > > > 
> > > > > 
> > > > > Regards,
> > > > > 
> > > > > Edward  
> > > > >  
> > > > > 
> > > > > 
> > > > > 
> > > > > ---------- Forwarded message ----------
> > > > > Date: Tue, 1 Jan 2002 18:14:41 +0100
> > > > > From: Vojtech Pavlik <vojtech@suse.cz>
> > > > > To: Edward Stempel <eazstempel@cal009001.student.utwente.nl>
> > > > > Cc: linux-kernel mailinglist <linux-kernel@vger.kernel.org>
> > > > > Subject: Re: DMA conflicts with soundcard for ide driver via82cxxx
> > > > > 
> > > > > On Fri, Dec 28, 2001 at 05:10:43PM +0100, Edward Stempel wrote:
> > > > > 
> > > > > > Excellent! That solved my problem.
> > > > > 
> > > > > What exactly did you have to change? It might be worth to include the
> > > > > changing of the latency setting in the kernel.
> > > > > 
> > > > > > 
> > > > > > Thankx
> > > > > > 
> > > > > > Edward
> > > > > > 
> > > > > > 
> > > > > > 
> > > > > > 
> > > > > > Date: Fri, 28 Dec 2001 10:15:05 +0100
> > > > > > From: Vojtech Pavlik <vojtech@suse.cz>
> > > > > > To: Edward Stempel <eazstempel@cal009001.student.utwente.nl>
> > > > > > Cc: linux-kernel mailinglist <linux-kernel@vger.kernel.org>
> > > > > > Subject: Re: DMA conflicts with soundcard for ide driver via82cxxx
> > > > > > 
> > > > > > On Fri, Dec 28, 2001 at 03:32:34AM +0100, Edward Stempel wrote:
> > > > > > 
> > > > > > > I have an Asus a7v266 mother board and an Ensoniq sound card in it. 
> > > > > > > The ide chipset is a VIA VT8233 that is capable of UDMA100. So I built a
> > > > > > > kernel with the es1371 sound driver and the via82cxxx ide driver
> > > > > > > configured in it. Actually I tried the kernel 2.4.17 first, and the latest
> > > > > > > I tried is 4.5.1 with the latest patch (patch-2.5.2-pre3) applied to it.
> > > > > > > I also tried the kernel 2.5.1 with Vojtech  patch (via-3.33.diff from
> > > > > > > his email dated 2001-12-23 23:20:48) applied to it, with the same
> > > > > > > (negative) results.
> > > > > > > 
> > > > > > > The good thing is that hdparm reports appr. 40 MB/sec when using DMA and
> > > > > > > about 6 MB/sec when not using DMA.
> > > > > > > Unfortunately using DMA for ide results in some ugly distortion of the
> > > > > > > sound from my soundcard whenever some IO to the disk is done.  :((
> > > > > > > 
> > > > > > > I have assigned different interrupts to the PCI-cards (ide is 
> > > > > > > on-board) and I even changed the sound card's PCI slot, so it shared
> > > > > > > its interrupt with another device (acpi instead of USB). It did not solve
> > > > > > > the problem. Because the problem only occurs when switching on using_dma
> > > > > > > on the ide driver, I think it is a DMA problem with the ide driver. It may
> > > > > > > be the es1371 driver as well off course, but I suspect it is the ide
> > > > > > > driver (or chipset).
> > > > > > >   
> > > > > > > Reading the list archive from linux-kernel, I discovered there have been
> > > > > > > more problems with DMA using this chipset, but I did not find anyone
> > > > > > > having the same problem as I have now.
> > > > > > > 
> > > > > > > Has someone also dealt with these problems, or can someone help me
> > > > > > > solving this problem? Please help!  
> > > > > > > 
> > > > > > > Below are some outputs using kernel 5.1 with patch-2.5.2-pre3.
> > > > > > 
> > > > > > You may try changing the PCI latency settings on either the IDE
> > > > > > controller or the sound card. Other than that, I don't know how to help.
> > > > > > 
> > > > > > -- 
> > > > > > Vojtech Pavlik
> > > > > > SuSE Labs
> > > > > 
> > > > > -- 
> > > > > Vojtech Pavlik
> > > > > SuSE Labs
> > > > 
> > > > 
> > > 
> > 
> > 

-- 
Vojtech Pavlik
SuSE Labs

  reply	other threads:[~2002-01-05 17:20 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2002-01-01 22:03 DMA conflicts with soundcard for ide driver via82cxxx Edward Stempel
2002-01-01 22:20 ` Vojtech Pavlik
2002-01-05 16:17   ` Edward Stempel
2002-01-05 16:23     ` Vojtech Pavlik
2002-01-05 17:03       ` Edward Stempel
2002-01-05 17:19         ` Vojtech Pavlik [this message]
  -- strict thread matches above, loose matches on Subject: below --
2001-12-28 16:10 Edward Stempel
2002-01-01 17:14 ` Vojtech Pavlik
2001-12-28  2:32 Edward Stempel
2001-12-28  9:15 ` Vojtech Pavlik

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20020105181942.A26258@suse.cz \
    --to=vojtech@suse.cz \
    --cc=eazstempel@cal009001.student.utwente.nl \
    --cc=linux-kernel@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox