diff options
| author | Takashi Iwai <tiwai@suse.de> | 2025-09-17 15:09:01 +0200 |
|---|---|---|
| committer | Greg Kroah-Hartman <gregkh@linuxfoundation.org> | 2025-09-25 11:16:52 +0200 |
| commit | 09b473a80c1ce5e0d21680374b58d80278fbb1a4 (patch) | |
| tree | 36fa38c55dd26c58a0d2237d51f42ed868fdf6f3 /include/net/aligned_data.h | |
| parent | b45cabfa717dd2fb1c0e85ad157f02513f576961 (diff) | |
ALSA: usb: qcom: Fix false-positive address space check
[ Upstream commit 44499ecb4f2817743c37d861bdb3e95f37d3d9cd ]
The sanity check previously added to uaudio_transfer_buffer_setup()
assumed the allocated buffer being linear-mapped. But the buffer
allocated via usb_alloc_coherent() isn't always so, rather to be used
with (SG-)DMA API. This leaded to a false-positive warning and the
driver failed to work.
Actually uaudio_transfer_buffer_setup() deals only with the DMA-API
addresses for MEM_XFER_BUF type, while other callers of
uaudio_iommu_map() are with pages with physical addresses for
MEM_EVENT_RING and MEM_XFER_RING types. So this patch splits the
mapping helper function to two different ones, uaudio_iommu_map() for
the DMA pages and uaudio_iommu_map_pa() for the latter, in order to
handle mapping differently for each type. Along with it, the
unnecessary address check that caused probe error is dropped, too.
Fixes: 3335a1bbd624 ("ALSA: qc_audio_offload: try to reduce address space confusion")
Suggested-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Reported-and-tested-by: Luca Weiss <luca.weiss@fairphone.com>
Closes: https://lore.kernel.org/DBR2363A95M1.L9XBNC003490@fairphone.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Diffstat (limited to 'include/net/aligned_data.h')
0 files changed, 0 insertions, 0 deletions
