We have currently two global locks, a rwlock and a rwsem, that are
used for managing linking the PCM streams. Due to these global locks,
once when a linked stream is used, the lock granularity suffers a
lot.
This patch attempts to eliminate the former global lock for atomic
ops. The latter rwsem needs remaining because of the loosy way of the
loop calls in snd_pcm_action_nonatomic(), as well as for avoiding the
deadlock at linking. However, these are used far rarely, actually
only by two actions (prepare and reset), where both are no timing
critical ones. So this can be still seen as a good improvement.
The basic strategy to eliminate the rwlock is to assure group->lock at
adding or removing a stream to / from the group. Since we already
takes the group lock whenever taking the all substream locks under the
group, this shouldn't be a big problem. The reference to group
pointer in snd_pcm_substream object is protected by the stream lock
itself.
However, there are still pitfalls: a race window at re-locking and the
lifecycle of group object. The former is a small race window for
dereferencing the substream group object opened while snd_pcm_action()
performs re-locking to avoid ABBA deadlocks. This includes the unlink
of group during that window, too. And the latter is the kfree
performed after all streams are removed from the group while it's
still dereferenced.
For addressing these corner cases, two new tricks are introduced:
- After re-locking, the group assigned to the stream is checked again;
if the group is changed, we retry the whole procedure.
- Introduce a refcount to snd_pcm_group object, so that it's freed
only when it's empty and really no one refers to it.
(Some readers might wonder why not RCU for the latter. RCU in this
case would cost more than refcounting, unfortunately. We take the
group lock sooner or later, hence the performance improvement by RCU
would be negligible. Meanwhile, because we need to deal with
schedulable context depending on the pcm->nonatomic flag, it'll become
dynamic RCU/SRCU switch, and the grace period may become too long.)
Along with these changes, there are a significant amount of code
refactoring. The complex group re-lock & ref code is factored out to
snd_pcm_stream_group_ref() function, for example.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The snd_pcm_group_for_each_entry() loop found in snd_pcm_unlink() is
only for taking the first list entry. Use list_first_entry() to make
clearer.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Make a common helper to re-assign the PCM link using list_move() instead
of open code with manual list_del() and list_add_tail(). This assures
the consistency and we can get rid of snd_pcm_group.count field -- its
purpose is only to check whether the list is singular, and we can know
it by list_is_singular() call now.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
There are multiple open codes that initialize the same object.
Create a common helper function instead.
Also, use kzalloc() to be safer at creating a group object, and move
the initialization out of the critical section.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The snd_card_unref() call in snd_pcm_link() looks suspicious through a
quick glance, but it's a correct usage; this is needed just because
the file descriptor check in is_pcm_file() calls the helper
snd_lookup_minor_data() that keeps the card refcount.
Despite of the correctness, the code still looks confusing.
Basically, keeping the card ref for the whole code isn't needed
as fdget() blocks the release of the opened file. Hence it's more
understandable if snd_card_unref() is moved into is_pcm_file(), then
the caller doesn't have to take care after the call.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Nobody has actually used the type (VERIFY_READ vs VERIFY_WRITE) argument
of the user address range verification function since we got rid of the
old racy i386-only code to walk page tables by hand.
It existed because the original 80386 would not honor the write protect
bit when in kernel mode, so you had to do COW by hand before doing any
user access. But we haven't supported that in a long time, and these
days the 'type' argument is a purely historical artifact.
A discussion about extending 'user_access_begin()' to do the range
checking resulted this patch, because there is no way we're going to
move the old VERIFY_xyz interface to that model. And it's best done at
the end of the merge window when I've done most of my merges, so let's
just get this done once and for all.
This patch was mostly done with a sed-script, with manual fix-ups for
the cases that weren't of the trivial 'access_ok(VERIFY_xyz' form.
There were a couple of notable cases:
- csky still had the old "verify_area()" name as an alias.
- the iter_iov code had magical hardcoded knowledge of the actual
values of VERIFY_{READ,WRITE} (not that they mattered, since nothing
really used it)
- microblaze used the type argument for a debug printout
but other than those oddities this should be a total no-op patch.
I tried to fix up all architectures, did fairly extensive grepping for
access_ok() uses, and the changes are trivial, but I may have missed
something. Any missed conversion should be trivially fixable, though.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Not much work on the core this time around but we've seen quite a bit of
driver work, including on the generic DT drivers. There's also a large
part of the diff from a merge of the DaVinci and OMAP directories, along
with some active development there:
- Preparatory work from Morimoto-san for merging the audio-graph and
audio-graph-scu cards.
- A merge of the TI OMAP and DaVinci directories, the OMAP product line
has been merged into the DaVinci product line so there is now a lot
of IP sharing which meant that the split directories just got in the
way. This has pulled in a few architecture changes as well.
- A big cleanup of the Maxim MAX9867 driver from Ladislav Michl.
- Support for Asahi Kaesi AKM4118, AMD ACP3x, Intel platforms with
RT5660, Meson AXG S/PDIF inputs, several Qualcomm IPs and Xilinx I2S
controllers.
-----BEGIN PGP SIGNATURE-----
iQFHBAABCgAxFiEEreZoqmdXGLWf4p/qJNaLcl1Uh9AFAlwY7KsTHGJyb29uaWVA
a2VybmVsLm9yZwAKCRAk1otyXVSH0MFqB/4m9nlfUeXTpiSD1FgVIlMmdPNUg4V2
6Ybztaw4kRc5LuQN8PXQmaFLx020yAnLvI7Zzj7l3K8r6a9lfyFs+pKL0wtBnNK2
9QEFmOVQ3QFpt31Yb2IQeO4dfNbiyKeczjLau4mXWTl0j5dc/UH+HasE1dRZOxsC
rqJ8IsdibIVxVtQ7ZmcnU+y6XK0inBHAAh6ksMehsufShGrfrLs/nRBaXRZcRqJg
ciSFY5uYRYkDxTgogTpNRfVy4nr17N10+0sgrQ3RtaaqgG3gBXsHca1meyxKYW3Q
TssOJGIl3+uGiAMNyZqzxe5pAwwuGhZ3hAAAODtfYJQtAuAOW3/45Wqh
=phWF
-----END PGP SIGNATURE-----
Merge tag 'asoc-v4.21' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound into for-next
ASoC: Updates for v4.21
Not much work on the core this time around but we've seen quite a bit of
driver work, including on the generic DT drivers. There's also a large
part of the diff from a merge of the DaVinci and OMAP directories, along
with some active development there:
- Preparatory work from Morimoto-san for merging the audio-graph and
audio-graph-scu cards.
- A merge of the TI OMAP and DaVinci directories, the OMAP product line
has been merged into the DaVinci product line so there is now a lot
of IP sharing which meant that the split directories just got in the
way. This has pulled in a few architecture changes as well.
- A big cleanup of the Maxim MAX9867 driver from Ladislav Michl.
- Support for Asahi Kaesi AKM4118, AMD ACP3x, Intel platforms with
RT5660, Meson AXG S/PDIF inputs, several Qualcomm IPs and Xilinx I2S
controllers.
Default copy function uses kmalloc to allocate buffers, lets check
if the runtime buffers are setup before making this allocations.
This can be useful if the buffers are dma buffers.
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Acked-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
stream is indirectly controlled by user-space, hence leading to
a potential exploitation of the Spectre variant 1 vulnerability.
This issue was detected with the help of Smatch:
sound/core/pcm.c:140 snd_pcm_control_ioctl() warn: potential spectre issue 'pcm->streams' [r] (local cap)
Fix this by sanitizing stream before using it to index pcm->streams
Notice that given that speculation windows are large, the policy is
to kill the speculation on the first load and not worry if it can be
completed with a dependent load/store [1].
[1] https://marc.info/?l=linux-kernel&m=152449131114778&w=2
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Cc: stable@vger.kernel.org
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Commit 67ec1072b0 ("ALSA: pcm: Fix rwsem deadlock for non-atomic PCM
stream") fixes deadlock for non-atomic PCM stream. But, This patch
causes antother stuck.
If writer is RT thread and reader is a normal thread, the reader
thread will be difficult to get scheduled. It may not give chance to
release readlocks and writer gets stuck for a long time if they are
pinned to single cpu.
The deadlock described in the previous commit is because the linux
rwsem queues like a FIFO. So, we might need non-FIFO writelock, not
non-block one.
My suggestion is that the writer gives reader a chance to be scheduled
by using the minimum msleep() instaed of spinning without blocking by
writer. Also, The *_nonblock may be changed to *_nonfifo appropriately
to this concept.
In terms of performance, when trylock is failed, this minimum periodic
msleep will have the same performance as the tick-based
schedule()/wake_up_q().
[ Although this has a fairly high performance penalty, the relevant
code path became already rare due to the previous commit ("ALSA:
pcm: Call snd_pcm_unlink() conditionally at closing"). That is, now
this unconditional msleep appears only when using linked streams,
and this must be a rare case. So we accept this as a quick
workaround until finding a more suitable one -- tiwai ]
Fixes: 67ec1072b0 ("ALSA: pcm: Fix rwsem deadlock for non-atomic PCM stream")
Suggested-by: Wonmin Jung <wonmin.jung@lge.com>
Signed-off-by: Chanho Min <chanho.min@lge.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Currently the PCM core calls snd_pcm_unlink() always unconditionally
at closing a stream. However, since snd_pcm_unlink() invokes the
global rwsem down, the lock can be easily contended. More badly, when
a thread runs in a high priority RT-FIFO, it may stall at spinning.
Basically the call of snd_pcm_unlink() is required only for the linked
streams that are already rare occasion. For normal use cases, this
code path is fairly superfluous.
As an optimization (and also as a workaround for the RT problem
above in normal situations without linked streams), this patch adds a
check before calling snd_pcm_unlink() and calls it only when needed.
Reported-by: Chanho Min <chanho.min@lge.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Both snd_ctl_add() and snd_ctl_replace() process the things in a
fairly similar way, and indeed the most of the codes can be unified.
This patch is a refactoring to consolidate the both functions to call
a single helper with an extra "mode" argument. There should be no
functional difference, except for one additional sanity check applied
now to snd_ctl_replace() (which was rather overlooking, IMO), too.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The procedure for adding a user control element has some window opened
for race against the concurrent removal of a user element. This was
caught by syzkaller, hitting a KASAN use-after-free error.
This patch addresses the bug by wrapping the whole procedure to add a
user control element with the card->controls_rwsem, instead of only
around the increment of card->user_ctl_count.
This required a slight code refactoring, too. The function
snd_ctl_add() is split to two parts: a core function to add the
control element and a part calling it. The former is called from the
function for adding a user control element inside the controls_rwsem.
One change to be noted is that snd_ctl_notify() for adding a control
element gets called inside the controls_rwsem as well while it was
called outside the rwsem. But this should be OK, as snd_ctl_notify()
takes another (finer) rwlock instead of rwsem, and the call of
snd_ctl_notify() inside rwsem is already done in another code path.
Reported-by: syzbot+dc09047bce3820621ba2@syzkaller.appspotmail.com
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
PCM OSS layer may allocate a few temporary buffers, one for the core
read/write and another for the conversions via plugins. Currently
both are allocated via vmalloc(). But as the allocation size is
equivalent with the PCM period size, the required size might be quite
small, depending on the application.
This patch replaces these vmalloc() calls with kvzalloc() for covering
small period sizes better. Also, we use "z"-alloc variant here for
addressing the possible uninitialized access reported by syzkaller.
Reported-by: syzbot+1cb36954e127c98dd037@syzkaller.appspotmail.com
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
As a preparatory patch for the upcoming -Wimplicit-fallthrough
compiler checks, replace with the standard "fall through" annotation.
Unfortunately gcc doesn't understand a chattier text.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
As a preparatory patch for the upcoming -Wimplicit-fallthrough
compiler checks, add the "fall through" annotation in
snd_dma_alloc_pages(). Note that this seems necessary to be put
exactly before the next label, so it's outside the ifdef block.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
For discarding the pending bytes on rawmidi, we process with a loop of
snd_rawmidi_transmit() which is just a waste of CPU power.
Implement a lightweight API function to discard the pending bytes and
the proceed the ring buffer instantly, and use it instead of open
codes.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This ensures the transfer loop won't waste a run to read
the few frames (if any) between start and hw_ptr update.
It will wait for the next interrupt with wait_for_avail().
Signed-off-by: Ricardo Biehl Pasquali <pasqualirb@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
syzbot reported the uninitialized value exposure in certain situations
using virmidi loop. It's likely a very small race at writing and
reading, and the influence is almost negligible. But it's safer to
paper over this just by replacing the existing kvmalloc() with
kvzalloc().
Reported-by: syzbot+194dffdb8b22fc5d207a@syzkaller.appspotmail.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
In some cases (mainly for x86), we need the DMA coherent buffer with
non-cached pages. Although this has been done in each driver side
like HD-audio and intel8x0, it can be done cleaner in the core memory
allocator.
This patch adds the new types, SNDRV_DMA_TYPE_DEV_UC and
SNDRV_DMA_TYPE_DEV_UC_SG, for allocating such non-cached buffer
pages. On non-x86 architectures, they work as same as the standard
SNDRV_DMA_TYPE_DEV and *_SG.
One additional change by this move is that we can assure to pass the
non-cached pgprot to the vmapped buffer, too. It eventually fixes the
case like non-snoop mode without mmap access on HD-audio.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
snd_malloc_dev_pages() and snd_free_dev_pages() are local functions
and the parameters passed there are all contained in snd_dma_buffer
object. As a code-simplification, pass snd_dma_buffer object and
assign the address there like other allocators do (except for
snd_malloc_pages() which is called from outside, hence we can't change
easily).
Only code refactoring, no functional changes.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The size passed to dma_alloc_coherent() doesn't have to be aligned
with power-of-two, rather it should be the raw size. As a minor
optimization, remove the size adjustment in the current code.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
snd_seq_system_client_init() doesn't check the errors returned from
its port creations. Let's do it properly and handle the error paths.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Static checkers complain that snd_seq_create_kernel_client() can return
-EBUSY here so we need to have some error handling.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The "frames" variable is unsigned so the error handling doesn't work
properly.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
In __snd_pcm_lib_xfer(), when capture, if state is PREPARED
and size is less than start_threshold nothing can be done.
As there is no error, 0 is returned.
Signed-off-by: Ricardo Biehl Pasquali <pasqualirb@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The recent change to move the virmidi output processing to a work
slightly modified the code to discard the unsubscribed outputs so that
it works without a temporary buffer. However, this is actually buggy,
and may spew a kernel warning due to the unexpected call of
snd_rawmidi_transmit_ack(), as triggered by syzbot.
This patch takes back to the original code in that part, use a
temporary buffer and simply repeat snd_rawmidi_transmit(), in order to
address the regression.
Fixes: f7debfe540 ("ALSA: seq: virmidi: Offload the output event processing")
Reported-by: syzbot+ec5f605c91812d200367@syzkaller.appspotmail.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.
Warning level 2 was used: -Wimplicit-fallthrough=2
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.
Notice that in this particular case, I replaced the code comment with
a proper "fall through" annotation, which is what GCC is expecting
to find.
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
For a sake of code simplification, remove the init and the exit
entries that do nothing.
Notes for readers: actually it's OK to remove *both* init and exit,
but not OK to remove the exit entry. By removing only the exit while
keeping init, the module becomes permanently loaded; i.e. you cannot
unload it any longer!
Signed-off-by: Takashi Iwai <tiwai@suse.de>
All usages of mutex in ALSA sequencer core would take too long, hence
we don't have to care about the user interruption that makes things
complicated. Let's replace them with simpler mutex_lock().
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The sequencer core module doesn't call some destructors in the error
path of the init code, which may leave some resources.
This patch mainly fix these leaks by calling the destructors
appropriately at alsa_seq_init(). Also the patch brings a few
cleanups along with it, namely:
- Expand the old "if ((err = xxx) < 0)" coding style
- Get rid of empty seq_queue_init() and its caller
- Change snd_seq_info_done() to void
Last but not least, a couple of functions lose __exit annotation since
they are called also in alsa_seq_init().
No functional changes but minor code cleanups.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
There are a few functions that have been commented out for ages.
And also there are functions that do nothing but placeholders.
Let's kill them.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
snd_midi_event_encode_byte() can never fail, and it can return rather
true/false. Change the return type to bool, adjust the argument to
receive a MIDI byte as unsigned char, and adjust the comment
accordingly. This allows callers to drop error checks, which
simplifies the code.
Meanwhile, snd_midi_event_encode() helper is used only in seq_midi.c,
and it can be better folded into it. This will reduce the total
amount of lines in the end.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.
Addresses-Coverity-ID: 1357375 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The trigger flag in vmidi object can be referred in different contexts
concurrently, hence it's better to be put with READ_ONCE() and
WRITE_ONCE() macros to assure the accesses.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The virmidi sequencer stuff tries to translate the rawmidi bytes to
sequencer events and deliver the packets at trigger callback. The
amount of the whole process of these translations and deliveries
depends on the incoming rawmidi bytes, and we have no limit for that;
this was the cause of a CPU soft lockup that had been reported and
fixed recently.
Although we've fixed the soft lockup by putting the temporary unlock
and cond_resched(), it's rather a quick band aid. In this patch,
meanwhile, the event parsing and delivery process is offloaded to a
dedicated work, and the trigger callback just kicks it off. It has
three merits, at least:
- The processing is always done in a sleepable context, which can
assure the event delivery with non-atomic flag without hackish
is_atomic() usage.
- Other relevant codes can be simplified, reducing the lines
- It makes me happier
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The PCM format type is with __bitwise, hence it needs the explicit
cast with __force. It's ugly, but there is a reason for that cost...
This fixes the sparse warning:
sound/core/oss/pcm_oss.c:1854:55: warning: incorrect type in argument 1 (different base types)
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The virmidi output trigger tries to parse the all available bytes and
process sequencer events as much as possible. In a normal situation,
this is supposed to be relatively short, but a program may give a huge
buffer and it'll take a long time in a single spin lock, which may
eventually lead to a soft lockup.
This patch simply adds a workaround, a cond_resched() call in the loop
if applicable. A better solution would be to move the event processor
into a work, but let's put a duct-tape quickly at first.
Reported-and-tested-by: Dae R. Jeong <threeearcat@gmail.com>
Reported-by: syzbot+619d9f40141d826b097e@syzkaller.appspotmail.com
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The sanity checks in ALSA sequencer and OSS sequencer emulation codes
return falsely -ENXIO from poll callback. They should be EPOLLERR
instead.
This was caught thanks to the recent change to the return value.
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
snd_dma_alloc_pages_fallback() tries to allocate pages again when the
allocation fails with reduced size. But the first try actually
*increases* the size to power-of-two, which may give back a larger
chunk than the requested size. This confuses the callers, e.g. sgbuf
assumes that the size is equal or less, and it may result in a bad
loop due to the underflow and eventually lead to Oops.
The code of this function seems incorrectly assuming the usage of
get_order(). We need to decrease at first, then align to
power-of-two.
Reported-and-tested-by: he, bo <bo.he@intel.com>
Reported-by: zhang jun <jun.zhang@intel.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
A timer object for the classes SNDRV_TIMER_CLASS_CARD and
SNDRV_TIMER_CLASS_PCM has to be associated with a card object, but we
have no check at creation time. Such a timer object with NULL card
causes various unexpected problems, e.g. NULL dereference at reading
the sound timer proc file.
So as preventive measure while the creating the sound timer object is
created the card information availability is checked for the mentioned
entries and returned error if its NULL.
Signed-off-by: Srikanth K H <srikanth.h@samsung.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The size of in-kernel rawmidi buffers may be big up to 1MB, and it can
be specified freely by user-space; which implies that user-space may
trigger kmalloc() errors frequently.
This patch replaces the buffer allocation via kvmalloc() for dealing
with bigger buffers gracefully.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Unify a few open codes with helper functions to improve the
readability. Minor behavior changes (rather fixes) are:
- runtime->drain clearance is done within lock
- active_sensing is updated before resizing buffer in
SNDRV_RAWMIDI_IOCTL_PARAMS ioctl.
Other than that, simply code cleanups.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
