The current code deals with the stream start / stop solely via
line6_pcm_acquire() and line6_pcm_release(). This was (supposedly)
intended to avoid the races, but it doesn't work as expected. The
concurrent acquire and release calls can be performed without proper
protections, thus this might result in memory corruption.
Furthermore, we can't take a mutex to protect the whole function
because it can be called from the PCM trigger callback that is an
atomic context. Also spinlock isn't appropriate because the function
allocates with kmalloc with GFP_KERNEL. That is, these function just
lead to singular problems.
This is an attempt to reduce the existing races. First off, separate
both the stream buffer management and the stream URB management. The
former is protected via a newly introduced state_mutex while the
latter is protected via each line6_pcm_stream lock.
Secondly, the stream state are now managed in opened and running bit
flags of each line6_pcm_stream. Not only this a bit clearer than
previous combined bit flags, this also gives a better abstraction.
These rewrites allows us to make common hw_params and hw_free
callbacks for both playback and capture directions.
For the monitor and impulse operations, still line6_pcm_acquire() and
line6_pcm_release() are used. They call internally the corresponding
functions for both playback and capture streams with proper lock or
mutex. Unlike the previous versions, these function don't take the
bit masks but the only single type value. Also they are supposed to
be applied only as duplex operations.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Clearing prev_fsize in line6_pcm_acquire() is pretty racy.
This can be called at any time while the stream is being played.
Rather better to clear prev_fbuf and prev_fsize at the proper place
like the stream stop for capture, and just after copying the monitor /
impulse data inside the spinlock.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The impulse and monitor handling in submit_audio_out_urb() isn't
protected thus this can be racy with the capture stream handling.
This patch extends the range to protect via each stream's spinlock
(now the whole submit_audio_*_urb() are covered), and take the capture
stream lock additionally for the impulse and monitor handling part.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Move the check of multi configurations before snd_card_new() as a
short path, and reduce superfluous pointer references.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Instead of allocating the private data individually in each driver's
probe at first, let snd_card_new() allocate the data that is called in
line6_probe(). This simplifies the primary probe functions.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The interface argument is used just for retrieving the assigned
device, which can be already found in line6->ifcdev. Drop them from
the callbacks. Also, pass the usb id to private_init so that the
driver can deal with it there. This is a preliminary work for the
further cleanup to move the whole allocation into driver.c.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
A minor optimization; while pausing, the driver just copies the zero
that doesn't need any volume changes.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The PCM stream buffer allocation and free are identical for both
playback and capture streams. Provide single helper functions.
These are used only in pcm.c, thus they can be even static.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The codes to unlink and sync URBs are identical for both playback and
capture streams. Consolidate to single helper functions.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Introduce a new line6_pcm_stream structure and group individual
fields of snd_line6_pcm struct to playback and capture groups.
This patch itself just does rename and nothing else. More
meaningful cleanups based on these fields shuffling will follow.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
If the problem still really remains, we should fix it instead of
papering over it like this...
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Using a decremental loop without particular reasons worsens the
readability a lot. Use incremental loops instead.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The trigger callback is already spinlocked, so we need no more lock
here (even for the linked substreams). Let's drop it.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
line6_pcm_acquire() tries to restore the newly obtained resources at
the error path. But some flags aren't recorded and released properly
when the corresponding buffer is already present. These bits have to
be cleared in the error recovery, too.
Also, "flags_final" can be initialized to zero since we pass only the
subset of "channels" bits.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Fixed a few places using bits OR wrongly for condition checks.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The midi_transmit_lock is used always inside the send_urb_lock, thus
it doesn't play any role. Let's kill it. Also, rename
"send_urb_lock" as a more simple name "lock" since this is the only
lock for midi.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The function isn't used any longer after rewriting from sysfs to leds
class in toneport.c.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Fix memory leak at probe error path by rearranging the call order in
line6_destruct() so that the common destructor is always called.
Also this simplifies the error path to a single goto label.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Instead of non-standard sysfs, reimplement the LED controls on
TonePort as LED class devices.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Currently disconnect callback is used as a driver's destructor, and
this has to be called not only at the disconnection time but also at
the error paths during probe.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
... so that timer_del_sync() in the destructor can be called safely at
any time. Also move the mod_timer() call in toneport_setup(), which
is a bit clearer place.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The interface and driver objects are always set when callbacks are
called. Drop such superfluous NULL checks in init and disconnect
calls of each driver.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
It's utterly unsafe to proceed further the disconnect procedure if the
assigned usbdev is inconsistent with the expected object. Better to
put a WARN_ON() for more cautions and abort immediately.
Tested-by: Chris Rorvick <chris@rorvick.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Pull drm fixes from Dave Airlie:
"This feels larger than I'd like but its for three reasons.
a) amdkfd finalising the API more, this is a new feature introduced
last merge window, and I'd prefer to make the tweaks to the API
before it first gets into a stable release.
b) radeon regression required splitting an internal API to fix
properly, so it just changed a few more lines
c) vmwgfx fix changes a lock from a mutex->spin lock, this is fallout
from the new sleep checking.
Otherwise there is just some tda998x fixes"
* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
drm/radeon: Remove rdev->gart.pages_addr array
drm/radeon: Restore GART table contents after pinning it in VRAM v3
drm/radeon: Split off gart_get_page_entry ASIC hook from set_page_entry
drm/amdkfd: Fix bug in call to init_pipelines()
drm/amdkfd: Fix bug in pipelines initialization
drm/radeon: Don't increment pipe_id in kgd_init_pipeline
drm/i2c: tda998x: set the CEC I2C address based on the slave I2C address
drm/vmwgfx: Replace the hw mutex with a hw spinlock
drm/amdkfd: Allow user to limit only queues per device
drm/amdkfd: PQM handle queue creation fault
drm: tda998x: Fix EDID read timeout on HDMI connect
drm: tda998x: Protect the page register
The xfstests btrfs/072 reports uncorrectable read errors in dmesg,
because scrub forgets to use commit_root for parity scrub routine
and scrub attempts to scrub those extents items whose contents are
not fully on disk.
To fix it, we just add the @search_commit_root flag back.
Signed-off-by: Gui Hecheng <guihc.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Miao Xie <miaoxie@huawei.com>
Signed-off-by: Chris Mason <clm@fb.com>
Pull networking fixes from David Miller:
1) Don't OOPS on socket AIO, from Christoph Hellwig.
2) Scheduled scans should be aborted upon RFKILL, from Emmanuel
Grumbach.
3) Fix sleep in atomic context in kvaser_usb, from Ahmed S Darwish.
4) Fix RCU locking across copy_to_user() in bpf code, from Alexei
Starovoitov.
5) Lots of crash, memory leak, short TX packet et al bug fixes in
sh_eth from Ben Hutchings.
6) Fix memory corruption in SCTP wrt. INIT collitions, from Daniel
Borkmann.
7) Fix return value logic for poll handlers in netxen, enic, and bnx2x.
From Eric Dumazet and Govindarajulu Varadarajan.
8) Header length calculation fix in mac80211 from Fred Chou.
9) mv643xx_eth doesn't handle highmem correctly in non-TSO code paths.
From Ezequiel Garcia.
10) udp_diag has bogus logic in it's hash chain skipping, copy same fix
tcp diag used. From Herbert Xu.
11) amd-xgbe programs wrong rx flow control register, from Thomas
Lendacky.
12) Fix race leading to use after free in ping receive path, from Subash
Abhinov Kasiviswanathan.
13) Cache redirect routes otherwise we can get a heavy backlog of rcu
jobs liberating DST_NOCACHE entries. From Hannes Frederic Sowa.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (48 commits)
net: don't OOPS on socket aio
stmmac: prevent probe drivers to crash kernel
bnx2x: fix napi poll return value for repoll
ipv6: replacing a rt6_info needs to purge possible propagated rt6_infos too
sh_eth: Fix DMA-API usage for RX buffers
sh_eth: Check for DMA mapping errors on transmit
sh_eth: Ensure DMA engines are stopped before freeing buffers
sh_eth: Remove RX overflow log messages
ping: Fix race in free in receive path
udp_diag: Fix socket skipping within chain
can: kvaser_usb: Fix state handling upon BUS_ERROR events
can: kvaser_usb: Retry the first bulk transfer on -ETIMEDOUT
can: kvaser_usb: Send correct context to URB completion
can: kvaser_usb: Do not sleep in atomic context
ipv4: try to cache dst_entries which would cause a redirect
samples: bpf: relax test_maps check
bpf: rcu lock must not be held when calling copy_to_user()
net: sctp: fix slab corruption from use after free on INIT collisions
net: mv643xx_eth: Fix highmem support in non-TSO egress path
sh_eth: Fix serialisation of interrupt disable with interrupt & NAPI handlers
...
In the case when alloc_netdev fails we return NULL to a caller. But there is no
check for NULL in the probe drivers. This patch changes NULL to an error
pointer. The function description is amended to reflect what we may get
returned.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We need hold lock each time updating shirm registers, otherwise,
we may set unexpected values to them when they are set in
different thread at different time sequence.
The notification work will be scheduled in global work queue,
which won't hold this sst->spinlock itself, so here we need
change to use the lock version to update shim registers.
Signed-off-by: Jie Yang <yang.jie@intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
The I2C init path forgot to init the mutex, leading to an oops when
controls are accessed.
Signed-off-by: Manuel Lauss <manuel.lauss@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
As the clock can be get from TK/RK pin, so remove the comments.
Signed-off-by: Bo Shen <voice.shen@atmel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
According to the I2S specification information as following:
- WS = 0, channel 1 (left)
- WS = 1, channel 2 (right)
So, the start event should be TF/RF falling edge.
Reported-by: Songjun Wu <songjun.wu@atmel.com>
Signed-off-by: Bo Shen <voice.shen@atmel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
Save a bit of scheduling by coalescing all the output power up delays
into a single delay.
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Give the output disable sequence a chance to fully complete, otherwise
there is a danger we may remove the clock before it is finished
resulting in a pop noise. The delay for each output must be cumulative
and these are coalesced into a single delay.
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
- Add description for the driver
- Add dependency on the I2C module
Signed-off-by: Thomas Niederprüm <niederp@physik.uni-kl.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
