kernel/linux.git/include/sound/rawmidi.h, branch linux-6.12.y

ALSA: rawmidi: Don't embed device

2023-08-17T07:24:08+00:00

This patch detaches the struct device from the snd_rawmidi object by allocating via snd_device_alloc(), just like done for other devices. Reviewed-by: Jaroslav Kysela Signed-off-by: Curtis Malainey Tested-by: Curtis Malainey Link: https://lore.kernel.org/r/20230816160252.23396-6-tiwai@suse.de Signed-off-by: Takashi Iwai

ALSA: ump: Additional proc output

2023-05-23T10:10:59+00:00

UMP devices may have more interesting information than the traditional rawmidi. Extend the rawmidi_global_ops to allow the optional proc info output and show some more bits in the proc file for UMP. Note that the "Groups" field shows the first and the last UMP Groups, and both numbers are 1-based (i.e. the first group is 1). Reviewed-by: Jaroslav Kysela Link: https://lore.kernel.org/r/20230523075358.9672-7-tiwai@suse.de Signed-off-by: Takashi Iwai

ALSA: rawmidi: UMP support

2023-05-23T10:10:54+00:00

This patch adds the support helpers for UMP (Universal MIDI Packet) in ALSA core. The basic design is that a rawmidi instance is assigned to each UMP Endpoint. A UMP Endpoint provides a UMP stream, typically bidirectional (but can be also uni-directional, too), which may hold up to 16 UMP Groups, where each UMP (input/output) Group corresponds to the traditional MIDI I/O Endpoint. Additionally, the ALSA UMP abstraction provides the multiple UMP Blocks that can be assigned to each UMP Endpoint. A UMP Block is a metadata to hold the UMP Group clusters, and can represent the functions assigned to each UMP Group. A typical implementation of UMP Block is the Group Terminal Blocks of USB MIDI 2.0 specification. For distinguishing from the legacy byte-stream MIDI device, a new device "umpC*D*" will be created, instead of the standard (MIDI 1.0) devices "midiC*D*". The UMP instance can be identified by the new rawmidi info bit SNDRV_RAWMIDI_INFO_UMP, too. A UMP rawmidi device reads/writes only in 4-bytes words alignment, stored in CPU native endianness. The transmit and receive functions take care of the input/out data alignment, and may return zero or aligned size, and the params ioctl may return -EINVAL when the given input/output buffer size isn't aligned. A few new UMP-specific ioctls are added for obtaining the new UMP endpoint and block information. As of this commit, no ALSA sequencer instance is attached to UMP devices yet. They will be supported by later patches. Along with those changes, the protocol version for rawmidi is bumped to 2.0.3. Reviewed-by: Jaroslav Kysela Link: https://lore.kernel.org/r/20230523075358.9672-4-tiwai@suse.de Signed-off-by: Takashi Iwai

ALSA: rawmidi: Add ioctl callback to snd_rawmidi_global_ops

2023-05-23T10:10:53+00:00

A new callback, ioctl, is added to snd_rawmidi_global_ops for allowing the driver to deal with the own ioctls. This is another preparation patch for the upcoming UMP support. Reviewed-by: Jaroslav Kysela Link: https://lore.kernel.org/r/20230523075358.9672-3-tiwai@suse.de Signed-off-by: Takashi Iwai

ALSA: rawmidi: Pass rawmidi directly to snd_rawmidi_kernel_open()

2023-05-23T10:10:52+00:00

snd_rawmidi_kernel_open() is used only internally from ALSA sequencer, so far, and parsing the card / device matching table at each open is redundant, as each sequencer client already gets the rawmidi object beforehand. This patch optimizes the path by passing the rawmidi object directly at snd_rawmidi_kernel_open(). This is also a preparation for the upcoming UMP rawmidi I/O support. Reviewed-by: Jaroslav Kysela Link: https://lore.kernel.org/r/20230523075358.9672-2-tiwai@suse.de Signed-off-by: Takashi Iwai

ALSA: rawmidi: Move lock to snd_rawmidi_substream

2022-06-20T07:36:02+00:00

Having a lock in snd_rawmidi_runtime can be a problem especially when a substream is accessed from the outside, as the runtime creation might be racy with the external calls. As a first step for hardening, move the spinlock from snd_rawmidi_runtime to snd_rawmidi_substream. This patch just replaces the lock calls, no real functional change is put yet. Link: https://lore.kernel.org/r/20220617144051.18985-3-tiwai@suse.de Signed-off-by: Takashi Iwai

ALSA: rawmidi: Make internal functions local static

2022-06-20T07:36:01+00:00

__snd_rawmidi_transmit_peek() and __snd_rawmidi_transmit_ack() are never called from the outside. Let's make them local static and unexport them. Link: https://lore.kernel.org/r/20220617144051.18985-2-tiwai@suse.de Signed-off-by: Takashi Iwai

ALSA: rawmidi: introduce SNDRV_RAWMIDI_IOCTL_USER_PVERSION

2021-09-23T07:26:40+00:00

The new framing mode causes the user space regression, because the alsa-lib code does not initialize the reserved space in the params structure when the device is opened. This change adds SNDRV_RAWMIDI_IOCTL_USER_PVERSION like we do for the PCM interface for the protocol acknowledgment. Cc: David Henningsson Cc: Fixes: 08fdced60ca0 ("ALSA: rawmidi: Add framing mode") BugLink: https://github.com/alsa-project/alsa-lib/issues/178 Signed-off-by: Jaroslav Kysela Link: https://lore.kernel.org/r/20210920171850.154186-1-perex@perex.cz Signed-off-by: Takashi Iwai

ALSA: rawmidi: Add framing mode

2021-05-17T14:02:44+00:00

This commit adds a new framing mode that frames all MIDI data into 32-byte frames with a timestamp. The main benefit is that we can get accurate timestamps even if userspace wakeup and processing is not immediate. Testing on a Celeron N3150 with this mode has a max jitter of 2.8 ms, compared to the in-kernel seq implementation which has a max jitter of 5 ms during idle and much worse when running scheduler stress tests in parallel. Signed-off-by: David Henningsson Reviewed-by: Jaroslav Kysela Link: https://lore.kernel.org/r/20210515071533.55332-1-coding@diwic.se Signed-off-by: Takashi Iwai

ALSA: rawmidi: Fix racy buffer resize under concurrent accesses

2020-05-07T20:29:14+00:00

The rawmidi core allows user to resize the runtime buffer via ioctl, and this may lead to UAF when performed during concurrent reads or writes: the read/write functions unlock the runtime lock temporarily during copying form/to user-space, and that's the race window. This patch fixes the hole by introducing a reference counter for the runtime buffer read/write access and returns -EBUSY error when the resize is performed concurrently against read/write. Note that the ref count field is a simple integer instead of refcount_t here, since the all contexts accessing the buffer is basically protected with a spinlock, hence we need no expensive atomic ops. Also, note that this busy check is needed only against read / write functions, and not in receive/transmit callbacks; the race can happen only at the spinlock hole mentioned in the above, while the whole function is protected for receive / transmit callbacks. Reported-by: butt3rflyh4ck Cc: Link: https://lore.kernel.org/r/CAFcO6XMWpUVK_yzzCpp8_XP7+=oUpQvuBeCbMffEDkpe8jWrfg@mail.gmail.com Link: https://lore.kernel.org/r/s5heerw3r5z.wl-tiwai@suse.de Signed-off-by: Takashi Iwai