Close some races at disconnection of a USB audio device by adding the
chip->shutdown_mutex and chip->shutdown check at appropriate places.
The spots to put bandaids are:
- PCM prepare, hw_params and hw_free
- where the usb device is accessed for communication or get speed, in
mixer.c and others; the device speed is now cached in subs->speed
instead of accessing to chip->dev
The accesses in PCM open and close don't need the mutex protection
because these are already handled in the core PCM disconnection code.
The autosuspend/autoresume codes are still uncovered by this patch
because of possible mutex deadlocks. They'll be covered by the
upcoming change to rwsem.
Also the mixer codes are untouched, too. These will be fixed in
another patch, too.
Reported-by: Matthieu CASTET <[email protected]>
Cc: <[email protected]>
Signed-off-by: Takashi Iwai <[email protected]>
Fix races at PCM disconnection:
- while a PCM device is being opened or closed
- while the PCM state is being changed without lock in prepare,
hw_params, hw_free ops
Reported-by: Matthieu CASTET <[email protected]>
Cc: <[email protected]>
Signed-off-by: Takashi Iwai <[email protected]>
xfstests has always had random failures of tests due to loop devices
failing to be torn down and hence leaving filesytems that cannot be
unmounted. This causes test runs to immediately stop.
Over the past 6 or 7 years we've added hacks like explicit unmount
-d commands for loop mounts, losetup -d after unmount -d fails, etc,
but still the problems persist. Recently, the frequency of loop
related failures increased again to the point that xfstests 259 will
reliably fail with a stray loop device that was not torn down.
That is despite the fact the test is above as simple as it gets -
loop 5 or 6 times running mkfs.xfs with different paramters:
lofile=$(losetup -f)
losetup $lofile "$testfile"
"$MKFS_XFS_PROG" -b size=512 $lofile >/dev/null || echo "mkfs failed!"
sync
losetup -d $lofile
And losteup -d $lofile is failing with EBUSY on 1-3 of these loops
every time the test is run.
Turns out that blkid is running simultaneously with losetup -d, and
so it sees an elevated reference count and returns EBUSY. But why
is blkid running? It's obvious, isn't it? udev has decided to try
and find out what is on the block device as a result of a creation
notification. And it is racing with mkfs, so might still be scanning
the device when mkfs finishes and we try to tear it down.
So, make losetup -d force autoremove behaviour. That is, when the
last reference goes away, tear down the device. xfstests wants it
*gone*, not causing random teardown failures when we know that all
the operations the tests have specifically run on the device have
completed and are no longer referencing the loop device.
Signed-off-by: Dave Chinner <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
Added appropriate timeout value for secure erase based on identify device data
Signed-off-by: Asai Thambi S P <[email protected]>
Signed-off-by: Selvan Mani <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
Changing the type of bdev parameters to be unsigned int :1, rather than bool.
This is more consistent with the types of other features in the block drivers.
Signed-off-by: Oliver Chick <[email protected]>
Signed-off-by: Konrad Rzeszutek Wilk <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
The patch cciss-use-check_signature.patch in -mm tree introduced
a build error:
drivers/built-in.o: In function `CISS_signature_present':
drivers/block/cciss.c:4270: undefined reference to `check_signature'
Add missing CONFIG_CHECK_SIGNATURE to fix this issue.
Reported-by: Fengguang Wu <[email protected]>
Signed-off-by: Akinobu Mita <[email protected]>
Cc: Fengguang Wu <[email protected]>
Cc: Mike Miller <[email protected]>
Cc: Jens Axboe <[email protected]>
Acked-by: "Stephen M. Cameron" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
Using kmem_cache_zalloc() instead of kmem_cache_alloc() and memset().
spatch with a semantic match is used to found this problem.
(http://coccinelle.lip6.fr/)
Signed-off-by: Wei Yongjun <[email protected]>
Signed-off-by: Konrad Rzeszutek Wilk <[email protected]>
Peter is not going to maintain the driver any more. I have the
hardware.
Acked-by: Peter Osterlund <[email protected]>
Signed-off-by: Jiri Kosina <[email protected]>
This is a small cleanup, that also may turn error handling of
unitialized disks more readable. We don't need a separate variable to
track allocated disks, remove dr and reuse drive variable instead.
Signed-off-by: Herton Ronaldo Krzesinski <[email protected]>
Signed-off-by: Jiri Kosina <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
The same checks to see if a drive can be or is registered are
repeated through the code, factor out the checks in a common function
and replace the repeated checks with it.
Signed-off-by: Herton Ronaldo Krzesinski <[email protected]>
Signed-off-by: Jiri Kosina <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
On floppy initialization, if something failed inside the loop we call
add_disk, there was no cleanup of previous iterations in the error
handling.
Cc: [email protected]
Signed-off-by: Herton Ronaldo Krzesinski <[email protected]>
Signed-off-by: Jiri Kosina <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
If blk_init_queue fails, we do not call put_disk on the current dr
(dr is decremented first in the error handling loop).
Cc: [email protected]
Reviewed-by: Ben Hutchings <[email protected]>
Signed-off-by: Herton Ronaldo Krzesinski <[email protected]>
Signed-off-by: Jiri Kosina <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
Since commit 070ad7e ("floppy: convert to delayed work and single-thread
wq"), we end up calling alloc_ordered_workqueue multiple times inside
the loop, which shouldn't be intended. Besides the leak, other side
effect in the current code is if blk_init_queue fails, we would end up
calling unregister_blkdev even if we didn't call yet register_blkdev.
Just moved the allocation of floppy_wq before the loop, and adjusted the
code accordingly.
Cc: [email protected] # 3.5+
Acked-by: Vivek Goyal <[email protected]>
Reviewed-by: Ben Hutchings <[email protected]>
Signed-off-by: Herton Ronaldo Krzesinski <[email protected]>
Signed-off-by: Jiri Kosina <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
drivers/block/xen-blkback/xenbus.c:260:5: warning: symbol 'xenvbd_sysfs_addif' was not declared. Should it be static?
drivers/block/xen-blkback/xenbus.c:284:6: warning: symbol 'xenvbd_sysfs_delif' was not declared. Should it be static?
Signed-off-by: Konrad Rzeszutek Wilk <[email protected]>
This adds a "select" dependency of KEYBOARD_LPC32XX on INPUT_MATRIXKMAP,
as the other drivers are doing in this regard. This fixes the following
compile error if KEYBOARD_LPC32XX is enabled but INPUT_MATRIXKMAP is not:
drivers/input/keyboard/lpc32xx-keys.c:230: undefined reference to
`matrix_keypad_build_keymap'
Signed-off-by: Roland Stigge <[email protected]>
Signed-off-by: Dmitry Torokhov <[email protected]>
Bootloader can leave interrupt bit pending, and it confuses driver.
Signed-off-by: Vasily Khoruzhick <[email protected]>
Acked-by: Robert Jarzmik <[email protected]>
Signed-off-by: Dmitry Torokhov <[email protected]>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.18 (GNU/Linux)
iQEcBAABAgAGBQJQjYaSAAoJEHm+PkMAQRiGIlYH+wbwj7i5CcUG46VRFT5WlG7W
iYaojL68FKKIUCBh4GOvZxE2D907KgqHQI6rePE/xDhK493d0Nyi4PGQJ1JoTE+l
JQq++C3CoX9mm/zCRW5JyOcemE4t2t1/poi2ege9Kf9E9N5Cw0P4r8Cz1urCFKL6
EO2/gA5d6cBQpg+gVq9T88SnYgiJGs7d8AYjFeC7SrqIwDvG6tfg+YF4IRxC4QAc
0zbBgej/kbuiLbGl2ltU0I9HTOwlTgp53tdYe810cs3CdzNC2ykdF391Q24tloeo
uxmCSQrA2upHA9aILqY+T5f+cMOpVS4T8Ica3N9OzVEIPXSh2K9d4tESGPikMEE=
=yiDr
-----END PGP SIGNATURE-----
Merge tag 'v3.7-rc3' into for-linus to sync up with recent USB changes
bf->bf_next is only while buffers are chained as part of an A-MPDU
in the tx queue. When a tid queue is flushed (e.g. on tearing down
an aggregation session), frames can be enqueued again as normal
transmission, without bf_next being cleared. This can lead to the
old pointer being dereferenced again later.
This patch might fix crashes and "Failed to stop TX DMA!" messages.
Signed-off-by: Felix Fietkau <[email protected]>
Cc: [email protected]
Signed-off-by: John W. Linville <[email protected]>
Pull Ceph fixes form Sage Weil:
"There are two fixes in the messenger code, one that can trigger a NULL
dereference, and one that error in refcounting (extra put). There is
also a trivial fix that in the fs client code that is triggered by NFS
reexport."
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
ceph: fix dentry reference leak in encode_fh()
libceph: avoid NULL kref_put when osd reset races with alloc_msg
rbd: reset BACKOFF if unable to re-queue
Split the omap4_hdmi_mux_pads() function into two parts, one handles the
tpd12s015 gpio muxing, the other handles the hdmi pins.
This is clearer, as hdmi and tpd12s015 are separate devices, and it also
allows us to mux those separately with DT.
Signed-off-by: Tomi Valkeinen <[email protected]>
Cc: Ricardo Neri <[email protected]>
Acked-by: Tony Lindgren <[email protected]>
The dispc error handler tries to "fix" issues by disabling and enabling
panel. This is problematic, as we're trying to remove the dependency
from omapdss to the omap_dss_devices. It's also racy, and doesn't really
fix anything.
This patch removes the use of omap_dss_device from the error handler,
and just disables and enables the associated overlay manager. This
should produce similar results as the previous solution, without using
dssdev.
However, the error handling is still horrible. But the problem boils
down to one question, to which I don't have a clear answer: what to do
when a HW error happens?
Signed-off-by: Tomi Valkeinen <[email protected]>
The dispc's error handler has a loop inside another loop, and both use
the same loop variable. This is clearly wrong, and this patch makes a
new variable for the inner loop.
Signed-off-by: Tomi Valkeinen <[email protected]>
The current omapfb code goes over all the modes found from the monitors
EDID data, and searches for a mode that is compatible with the DSS
hardware and has the highest x-res.
While this works ok as such, it proves problematic when using DSI PLL
for pixel clock. Calculating DSI PLL dividers is not the fastest of the
operations, and while doing it for one mode is usually ok, doing it for
20 modes is noticable.
Also, the first mode given in the EDID data should be the native mode of
the monitor, and thus also the best mode, so if that can be used, no
need to look further.
This patch changes the code to use the first mode that is compatible
with the DSS hardware.
Signed-off-by: Tomi Valkeinen <[email protected]>
It seems that using the second EDID block causes more problems than is
of any help. The first mode in the extended block will get
FB_MODE_IS_FIRST set, which will override the first mode from the first
EDID block, thus making the default videomode selection not to work
properly.
This patch removes the use of the extended edid block for now.
Signed-off-by: Tomi Valkeinen <[email protected]>
The hdmi driver tries to find the given video timings from its static
list of timings, to find the required ID for the mode. The check tries
to find exact match for the pixel clock, among other checks.
with omapfb driver there can be some amount of error in the give pixel
clock, as the pixel clock is converted between Hz and ps, thus the
hdmi's check fails to find the mode.
This patch makes the check more allowing, by rounding the pixel clocks
to nearest MHz.
Signed-off-by: Tomi Valkeinen <[email protected]>
Cc: Ricardo Neri <[email protected]>
This patch makes use of the hdmi_power_[on|off]_core() functions added
in the previous patch. The functions are used when reading EDID or
detecting if a monitor is connected.
Signed-off-by: Tomi Valkeinen <[email protected]>
Cc: Ricardo Neri <[email protected]>
There's currently just one power-on function for HDMI, which enables the
IP and the video output. When reading EDID or detecting if a monitor is
connected, we don't need the video output.
Enabling the video output for these operations is not a big problem in
itself, but the quick enable/disable cycles caused by the operations
seem to cause sync lost errors from time to time. Also, this makes it
possible to read the EDID before the full video path has been set up.
This patch splits the hdmi_power_on into two parts, hdmi_power_on_core
and hdmi_power_on_full. The "full" version does what hdmi_power_on does
currently, and hdmi_power_on_core only enables the core IP. Similar
changes are made for power_off.
Note that these don't allow the HDMI IP to be first enabled, and later
enable the video output, but the HDMI IP will first need to be powered
off before calling the full version. So this is rather limited
implementation, but it fills the needs for reading EDID.
Signed-off-by: Tomi Valkeinen <[email protected]>
Cc: Ricardo Neri <[email protected]>
DISPC irqs need to be handled from the compat layer and also in the
future by the omapdrm. To make this possible, this patchs adds a set of
helper functions, so that the irqs can be managed without direct
register reads/writes.
The following functions are added, and all the current direct reg
reads/writes are changed to use these.
u32 dispc_read_irqstatus(void);
void dispc_clear_irqstatus(u32 mask);
u32 dispc_read_irqenable(void);
void dispc_write_irqenable(u32 mask);
Signed-off-by: Tomi Valkeinen <[email protected]>
Add new dispc function, dispc_ovl_enabled(). This returns if the overlay
enable bit is set in the registers.
Signed-off-by: Tomi Valkeinen <[email protected]>
We need a low level manager-enable function for omapdrm. We have that
function as dispc internal func, _enable_mgr_out().
This patch exposes that function, and renames it to dispc_mgr_enable().
Signed-off-by: Tomi Valkeinen <[email protected]>
The current dispc_mgr_enable/disable function are blocking, and do a bit
too much for omapdrm. We'll expose new enable & disable functions that
will just set the bits in the registers in the following patches.
This patch renames the current functions to *_sync, to make it clear
that they are blocking, and also to free up the dispc_mgr_enable/disable
names for these new functions.
Signed-off-by: Tomi Valkeinen <[email protected]>
Use dss_feat_get_num_ovls() in dispc.c instead of
omap_dss_get_num_overlays() to remove the dependency to overlay.c. Note
that we still have uses of omap_dss_get_num_overlays() in dispc.c, but
these will be moved out in the future patches.
Signed-off-by: Tomi Valkeinen <[email protected]>
Currently omapdss driver sets up the initial connections between
overlays, overlay manager and a panel, based on default display
parameter coming from the board file or via module parameters.
This is unnecessary, as it's the higher level component that should
decide what display to use and how. This patch removes the code from
omapdss, and implements similar code to omapfb.
The def_disp module parameter and the default display platform_data
parameter are kept in omapdss, but omapdss doesn't do anything with
them. It will just return the default display name with
dss_get_default_display_name() call, which omapfb uses. This is done to
keep the backward compatibility.
Signed-off-by: Tomi Valkeinen <[email protected]>
Export dss_get_def_display_name() with the name of
omapdss_get_def_display_name() so that omapfb can use it after the next
patch which moves default display handling to omapfb.
Signed-off-by: Tomi Valkeinen <[email protected]>
commit 119c0d4460 changed
ext4_new_inode() such that the inode bitmap was being modified
outside a transaction, which could lead to corruption, and was
discovered when journal_checksum found a bad checksum in the
journal during log replay.
Nix ran into this when using the journal_async_commit mount
option, which enables journal checksumming. The ensuing
journal replay failures due to the bad checksums led to
filesystem corruption reported as the now infamous
"Apparent serious progressive ext4 data corruption bug"
[ Changed by tytso to only call ext4_journal_get_write_access() only
when we're fairly certain that we're going to allocate the inode. ]
I've tested this by mounting with journal_checksum and
running fsstress then dropping power; I've also tested by
hacking DM to create snapshots w/o first quiescing, which
allows me to test journal replay repeatedly w/o actually
power-cycling the box. Without the patch I hit a journal
checksum error every time. With this fix it survives
many iterations.
Reported-by: Nix <[email protected]>
Signed-off-by: Eric Sandeen <[email protected]>
Signed-off-by: "Theodore Ts'o" <[email protected]>
Cc: [email protected]
WARNING: net/ipv6/netfilter/nf_defrag_ipv6.o(.text+0xe0): Section mismatch in
reference from the function nf_ct_net_init() to the function
.init.text:nf_ct_frag6_sysctl_register()
The function nf_ct_net_init() references the function
__init nf_ct_frag6_sysctl_register().
In case nf_conntrack_ipv6 is compiled as a module, nf_ct_net_init could be
called after the init code and data are unloaded. Therefore remove the
"__net_init" annotation from nf_ct_frag6_sysctl_register().
Signed-off-by: Hein Tibosch <[email protected]>
Acked-by: Cong Wang <[email protected]>
Signed-off-by: Pablo Neira Ayuso <[email protected]>
ICMP tuples have id in src and type/code in dst.
So comparing src.u.all with dst.u.all will always fail here
and ip_xfrm_me_harder() is called for every ICMP packet,
even if there was no NAT.
Signed-off-by: Ulrich Weber <[email protected]>
Signed-off-by: Pablo Neira Ayuso <[email protected]>
Pull i2c subsystem fixes from Jean Delvare.
* 'i2c-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jdelvare/staging:
i2c-i801: Fix comment
i2c-i801: Simplify dependency towards GPIOLIB
i2c-stub: Move to drivers/i2c
Arbitrarily selecting GPIOLIB causes trouble on some architectures,
so don't do that. Instead, just make the optional multiplexing code
depend on CONFIG_I2C_MUX_GPIO instead of CONFIG_I2C_MUX for now. We
can revisit if the i2c-i801 driver ever supports other multiplexing
flavors.
Also make that optional code depend on DMI, as it won't do anything
without that.
Signed-off-by: Jean Delvare <[email protected]>
Cc: Fengguang Wu <[email protected]>
Move the i2c-stub driver to drivers/i2c, to match the Kconfig entry.
This is less confusing that way.
I also fixed all checkpatch warnings and errors.
Signed-off-by: Jean Delvare <[email protected]>
Cc: Peter Huewe <[email protected]>
were not being brought up. I found that the modules for the
network was not being installed. This was due to the config
CONFIG_MODULES_USE_ELF_RELA that came before CONFIG_MODULES, and
confused ktest in thinking that CONFIG_MODULES=y was not found.
Ktest needs to test all configs and not just stop if something starts
with CONFIG_MODULES.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)
iQEcBAABAgAGBQJQig8hAAoJEOdOSU1xswtMkFgIALLcnba79RHsPdGCTX247Hcg
UdteytZgyd1XayDSPLOVAR5f1vJeZ/6/L5dwWqZpf+j6wUTBwdUTc4DlBwHNpi8V
XDKbwAYWAQp4BVaQkKcrxKZZepE791NWxCelG7T7S0d7jIkwFTA4BhT4+8+QFztZ
H5IDL+HA73Jvehfv3gpJW6yDQ/QSyUjIK4QCsJS+wodB9nDzhAEiZ6ZKflSXFGq4
J+Fl/UfRfnA+j0aP75ecL7hewfdiLOmK67vKvW3l8wZ7s0y3NnIsxymmaa6sTIAQ
lIAsmSPdqOzXExIKLBHnsHCog6UW4a91MmEqM05IDpt+AcCnwDbk4EfbJEXa8ug=
=vl1t
-----END PGP SIGNATURE-----
Merge tag 'ktest-v3.7-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-ktest
Pull ktest confusion fix from Steven Rostedt:
"With the v3.7-rc2 kernel, the network cards on my target boxes were
not being brought up.
I found that the modules for the network was not being installed.
This was due to the config CONFIG_MODULES_USE_ELF_RELA that came
before CONFIG_MODULES, and confused ktest in thinking that
CONFIG_MODULES=y was not found.
Ktest needs to test all configs and not just stop if something starts
with CONFIG_MODULES."
* tag 'ktest-v3.7-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-ktest:
ktest: Fix ktest confusion with CONFIG_MODULES_USE_ELF_RELA
