Re: [GIT PULL] sound fixes for 6.1-rc5

[Takashi Iwai]

On Sat, 12 Nov 2022 02:28:19 +0100,
Linus Torvalds wrote:
> 
> On Fri, Nov 11, 2022 at 12:56 AM Takashi Iwai <tiwai@xxxxxxx> wrote:
> >
> > please pull sound fixes for v6.1-rc5 from:
> 
> Hmm. I don't know that this came in through this particular pull
> request, but I don't seem to have seen it before:
> 
>   Nov 11 14:32:37 xps13 kernel:
>         snd_hda_intel 0000:00:1f.3: Too many BDL entries:
> buffer=2097152, period=65536
> 
> there's six of those lines in my logs (two batches of three, five
> seconds apart).
> 
> I don't see any negative side effects aside from the messages, and
> sound seems to work fine, but since I don't think I've ever seen this
> one before I thought I'd just mention it.
> 
> I also haven't been using this laptop on a while - an arm64 Fedora
> uboot update broke my M2 boot, so I'm temporarily back to using my
> trusty old xps13.
> 
> So while it's new to me, the issue that introduced it may not be
> particularly new.
> 
> Google does show that the message itself has been happening for others
> for a long time. But I checked with 'journalctl', and it hasn't
> happened on this particular machine before.
> 
> It might be a non-kernel change that triggers it, of course.

Wooh, that must be the last change in the memalloc helper for
non-contiguous pages :-<  Admittedly the fix was a bit premature,
sorry for the mess.

Below is the quick workaround.  I'm going to send another PR soon
later.


thanks,

Takashi

-- 8< --
From: Takashi Iwai <tiwai@xxxxxxx>
Subject: [PATCH] ALSA: memalloc: Try dma_alloc_noncontiguous() at first

The latest fix for the non-contiguous memalloc helper changed the
allocation method for a non-IOMMU system to use only the fallback
allocator.  This should have worked, but it caused a problem sometimes
when too many non-contiguous pages are allocated that can't be treated
by HD-audio controller.

As a quirk workaround, go back to the original strategy: use
dma_alloc_noncontiguous() at first, and apply the fallback only when
it fails, but only for non-IOMMU case.

We'll need a better fix in the fallback code as well, but this
workaround should paper over most cases.

Fixes: 9736a325137b ("ALSA: memalloc: Don't fall back for SG-buffer with IOMMU")
Reported-by: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
Link: https://lore.kernel.org/r/CAHk-=wgSH5ubdvt76gNwa004ooZAEJL_1Q-Fyw5M2FDdqL==dg@xxxxxxxxxxxxxx
Signed-off-by: Takashi Iwai <tiwai@xxxxxxx>
---
 sound/core/memalloc.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/sound/core/memalloc.c b/sound/core/memalloc.c
index 6a81aaab25ab..ba095558b6d1 100644
--- a/sound/core/memalloc.c
+++ b/sound/core/memalloc.c
@@ -542,8 +542,10 @@ static void *snd_dma_noncontig_alloc(struct snd_dma_buffer *dmab, size_t size)
 	struct sg_table *sgt;
 	void *p;
 
+	sgt = dma_alloc_noncontiguous(dmab->dev.dev, size, dmab->dev.dir,
+				      DEFAULT_GFP, 0);
 #ifdef CONFIG_SND_DMA_SGBUF
-	if (!get_dma_ops(dmab->dev.dev)) {
+	if (!sgt && !get_dma_ops(dmab->dev.dev)) {
 		if (dmab->dev.type == SNDRV_DMA_TYPE_DEV_WC_SG)
 			dmab->dev.type = SNDRV_DMA_TYPE_DEV_WC_SG_FALLBACK;
 		else
@@ -551,9 +553,6 @@ static void *snd_dma_noncontig_alloc(struct snd_dma_buffer *dmab, size_t size)
 		return snd_dma_sg_fallback_alloc(dmab, size);
 	}
 #endif
-
-	sgt = dma_alloc_noncontiguous(dmab->dev.dev, size, dmab->dev.dir,
-				      DEFAULT_GFP, 0);
 	if (!sgt)
 		return NULL;
 
-- 
2.35.3