public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Ingo Molnar <mingo@elte.hu>
To: Takashi Iwai <tiwai@suse.de>
Cc: Rene Herman <rene.herman@keyaccess.nl>,
	Glauber Costa <gcosta@redhat.com>,
	Thomas Gleixner <tglx@linutronix.de>,
	Pete Clements <clem@clem.clem-digital.net>,
	Linux Kernel <linux-kernel@vger.kernel.org>,
	ALSA devel <alsa-devel@alsa-project.org>,
	Jesse Barnes <jbarnes@virtuousgeek.org>,
	"H. Peter Anvin" <hpa@zytor.com>
Subject: Re: 2.6.26-rc1 regression: ISA DMA broken (bisected)
Date: Fri, 9 May 2008 10:58:24 +0200	[thread overview]
Message-ID: <20080509085824.GF2592@elte.hu> (raw)
In-Reply-To: <20080509085511.GE2592@elte.hu>


* Ingo Molnar <mingo@elte.hu> wrote:

> good catch! Queued it up for testing. Jesse, do you concur?

here's the patch below, tidied up.

	Ingo

------------------------------->
Subject: x86/pci: fix broken ISA DMA
From: Takashi Iwai <tiwai@suse.de>
Date: Fri, 09 May 2008 08:06:55 +0200

Rene Herman reported:

> commit 8779f2fc3b84ebb6c5181fb13d702e9944c16069
>
> "x86: don't try to allocate from DMA zone at first"
>
> breaks all of ISA DMA. Or all of ALSA ISA DMA at least. All
> ISA soundcards are silent following that commit -- no error
> messages, everything appears fine, just silence.

That patch is buggy. We had an implicit assumption that
dev = NULL for ISA devices that require 24bit DMA.

The recent work on x86 dma_alloc_coherent() breaks the ISA DMA buffer
allocation, which is represented by "dev = NULL" and requires 24bit
DMA implicitly.

Bisected-by: Rene Herman <rene.herman@keyaccess.nl>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
---
 arch/x86/kernel/pci-dma.c |    8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

Index: linux-x86.q/arch/x86/kernel/pci-dma.c
===================================================================
--- linux-x86.q.orig/arch/x86/kernel/pci-dma.c
+++ linux-x86.q/arch/x86/kernel/pci-dma.c
@@ -386,11 +386,13 @@ dma_alloc_coherent(struct device *dev, s
 	if (dma_alloc_from_coherent_mem(dev, size, dma_handle, &memory))
 		return memory;
 
-	if (!dev)
+	if (!dev) {
 		dev = &fallback_dev;
+		gfp |= GFP_DMA;
+	}
 	dma_mask = dev->coherent_dma_mask;
 	if (dma_mask == 0)
-		dma_mask = DMA_32BIT_MASK;
+		dma_mask = (gfp & GFP_DMA) ? DMA_24BIT_MASK : DMA_32BIT_MASK;
 
 	/* Device not DMA able */
 	if (dev->dma_mask == NULL)
@@ -404,7 +406,7 @@ dma_alloc_coherent(struct device *dev, s
 	   larger than 16MB and in this case we have a chance of
 	   finding fitting memory in the next higher zone first. If
 	   not retry with true GFP_DMA. -AK */
-	if (dma_mask <= DMA_32BIT_MASK)
+	if (dma_mask <= DMA_32BIT_MASK && !(gfp & GFP_DMA))
 		gfp |= GFP_DMA32;
 #endif
 

  reply	other threads:[~2008-05-09  8:58 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-05-09  1:37 2.6.26-rc1 regression: ISA DMA broken (bisected) Rene Herman
2008-05-09  6:06 ` Takashi Iwai
2008-05-09  8:55   ` Ingo Molnar
2008-05-09  8:58     ` Ingo Molnar [this message]
2008-05-09 17:20       ` Jesse Barnes
2008-05-09 12:03   ` Rene Herman
2008-05-09 12:28     ` Ingo Molnar
2008-05-09 23:00       ` Rene Herman
2008-05-13 14:36         ` Ingo Molnar
2008-05-13 15:26           ` Rene Herman
2008-05-09 12:29     ` Pete Clements
2008-05-09 12:48   ` Glauber Costa
2008-05-13 16:59   ` Bjorn Helgaas
2008-05-13 17:01     ` Alan Cox
2008-05-13 17:33       ` Rene Herman
2008-05-13 23:18         ` Bjorn Helgaas
2008-05-14  9:25           ` Takashi Iwai
2008-05-14 12:46           ` Rene Herman
2008-05-14 13:01             ` Takashi Iwai
2008-05-14 15:40               ` Rene Herman
2008-05-14 15:53                 ` Takashi Iwai
2008-05-14 18:41                 ` Rene Herman
2008-05-14 18:50                   ` Bjorn Helgaas
2008-05-14 19:09                     ` Rene Herman
2008-05-30 21:15                       ` [PATCH] " Rene Herman
2008-05-30 21:28                         ` [DEVICE MODEL] dev->dma_mask Rene Herman
2008-05-30 21:43                         ` [PATCH] Re: 2.6.26-rc1 regression: ISA DMA broken (bisected) Bjorn Helgaas
2008-05-30 22:11                           ` Rene Herman
2008-05-30 22:37                             ` [PATCH] ISA: set 24-bit dma_mask for ISA devices Rene Herman
2008-05-30 22:55                               ` Andrew Morton
2008-05-30 23:50                                 ` Rene Herman
2008-05-30 23:54                             ` [PATCH] PNP: set the pnp_card dma_mask for use by ISAPnP cards Rene Herman
2008-05-31  8:55                               ` Takashi Iwai
2008-05-30 23:55                             ` [PATCH] ISA: set 24-bit dma_mask for ISA devices Rene Herman
2008-05-31  8:56                               ` Takashi Iwai
2008-05-14 15:26             ` 2.6.26-rc1 regression: ISA DMA broken (bisected) Bjorn Helgaas

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=20080509085824.GF2592@elte.hu \
    --to=mingo@elte.hu \
    --cc=alsa-devel@alsa-project.org \
    --cc=clem@clem.clem-digital.net \
    --cc=gcosta@redhat.com \
    --cc=hpa@zytor.com \
    --cc=jbarnes@virtuousgeek.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=rene.herman@keyaccess.nl \
    --cc=tglx@linutronix.de \
    --cc=tiwai@suse.de \
    /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