[PATCH] ALSA: us122l: mark hwdep mappings VM_MIXEDMAP

[PATCH] ALSA: us122l: mark hwdep mappings VM_MIXEDMAP

[Lorenzo Stoakes (ARM)]

mm has recently moved to a strict model which requires any driver which
clears VMA_MAYWRITE_BIT to be mapping non-mm-managed memory
(VMA_PFNMAP_BIT or VMA_MIXEDMAP_BIT), failing any such mmap() with an
error and a kernel warning - see "mm/vma: only allow mmap to clear
VMA_MAYWRITE_BIT if not mm-managed".

However, simultaneous to this, commit 71c610aeb177 ("ALSA: us122l: Prevent
write upgrades for read mappings") updated us122l to clear
VMA_MAYWRITE_BIT without marking the region with either VMA_PFNMAP_BIT or
VMA_MIXEDMAP_BIT.

vm_insert_pages() and friends set this automatically, however this driver
maps its pre-allocated pages from a fault handler, so never calls these.

Simply set the flag in usb_stream_hwdep_mmap() to resolve the issue. This
correctly marks the mapping as one whose pages core mm does not manage,
which has no meaningful impact for this driver beyond preventing the mmap()
from failing.

Fixes: 71c610aeb177 ("ALSA: us122l: Prevent write upgrades for read mappings")
Reported-by: David Carlier <devnexen@gmail.com>
Closes: https://lore.kernel.org/20261018025905.201296-1-devnexen@gmail.com
Cc: Kazuki Hanai <hnkz.64@gmail.com>
Cc: Takashi Iwai <tiwai@suse.com>
Cc: Jaroslav Kysela <perex@perex.cz>
Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
---
Andrew - I think easiest for you to take this.

It should go in before my series ideally to avoid bisection hazard, I think
while it's a fix for the driver, that'll be the safest way.

It should be prioritised to be part of the 2nd batch of patches for this merge
window.

The commit it fixes is cc: stable but this does NOT need to be backported, it's
for 7.4 only.

Thanks!
---
 sound/usb/usx2y/us122l.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/sound/usb/usx2y/us122l.c b/sound/usb/usx2y/us122l.c
index a5db0d044ef9..2986a90ea335 100644
--- a/sound/usb/usx2y/us122l.c
+++ b/sound/usb/usx2y/us122l.c
@@ -193,7 +193,7 @@ static int usb_stream_hwdep_mmap(struct snd_hwdep *hw,
 	}
 
 	area->vm_ops = &usb_stream_hwdep_vm_ops;
-	vm_flags_set(area, VM_DONTDUMP);
+	vm_flags_set(area, VM_MIXEDMAP | VM_DONTDUMP);
 	if (!read)
 		vm_flags_set(area, VM_DONTEXPAND);
 	area->vm_private_data = us122l;

---
base-commit: 2bba28e98578de622ef307d00875aa8f8bca3d10
change-id: 20261023-fix-us122l-mixedmap-2d42f05ab421

Cheers,
-- 
Lorenzo Stoakes (ARM) <ljs@kernel.org>

Re: [PATCH] ALSA: us122l: mark hwdep mappings VM_MIXEDMAP

[Takashi Iwai]

On Fri, 23 Oct 2026 12:21:31 +0200,
Lorenzo Stoakes (ARM) wrote:
> 
> mm has recently moved to a strict model which requires any driver which
> clears VMA_MAYWRITE_BIT to be mapping non-mm-managed memory
> (VMA_PFNMAP_BIT or VMA_MIXEDMAP_BIT), failing any such mmap() with an
> error and a kernel warning - see "mm/vma: only allow mmap to clear
> VMA_MAYWRITE_BIT if not mm-managed".
> 
> However, simultaneous to this, commit 71c610aeb177 ("ALSA: us122l: Prevent
> write upgrades for read mappings") updated us122l to clear
> VMA_MAYWRITE_BIT without marking the region with either VMA_PFNMAP_BIT or
> VMA_MIXEDMAP_BIT.
> 
> vm_insert_pages() and friends set this automatically, however this driver
> maps its pre-allocated pages from a fault handler, so never calls these.
> 
> Simply set the flag in usb_stream_hwdep_mmap() to resolve the issue. This
> correctly marks the mapping as one whose pages core mm does not manage,
> which has no meaningful impact for this driver beyond preventing the mmap()
> from failing.
> 
> Fixes: 71c610aeb177 ("ALSA: us122l: Prevent write upgrades for read mappings")
> Reported-by: David Carlier <devnexen@gmail.com>
> Closes: https://lore.kernel.org/20261018025905.201296-1-devnexen@gmail.com
> Cc: Kazuki Hanai <hnkz.64@gmail.com>
> Cc: Takashi Iwai <tiwai@suse.com>
> Cc: Jaroslav Kysela <perex@perex.cz>
> Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
> ---
> Andrew - I think easiest for you to take this.
> 
> It should go in before my series ideally to avoid bisection hazard, I think
> while it's a fix for the driver, that'll be the safest way.

Yes, feel free to take:

Acked-by: Takashi Iwai <tiwai@suse.de>


thanks,

Takashi