To more accurately calculate overhead for "bsd" style
df reporting, we should count the journal blocks as
overhead as well.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Tested-by: Eric Whitney <enwlinux@gmail.com>
Although I put this in, I now think it was a bad decision. For most
users, there is very little to be done in this case. They get the
message, once per day, with no real context or proposed action. TBH,
it generates support calls when it probably does not need to; the
message sounds more dire than the situation really is.
Just nuke it. Normal investigation via blktrace or whatnot can
reveal poor IO patterns if bad performance is encountered.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
i_mutex is not held when ->sync_file is called.
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
We cannot wait for transaction commit in journal_unmap_buffer()
because we hold page lock which ranks below transaction start. We
solve the issue by bailing out of journal_unmap_buffer() and
jbd2_journal_invalidatepage() with -EBUSY. Caller is then responsible
for waiting for transaction commit to finish and try invalidation
again. Since the issue can happen only for page stradding i_size, it
is simple enough to manually call jbd2_journal_invalidatepage() for
such page from ext4_setattr(), check the return value and wait if
necessary.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
In data=journal mode we don't need delalloc or DIO handling in invalidatepage
and similarly in other modes we don't need the journal handling. So split
invalidatepage implementations.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
On the imx23-olinuxino board GPIO2_1 is connected to the LED and GPIO0_17
is the USB PHY reset.
So make the IOMUX assignment properly.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
The sequence:
unshare(CLONE_NEWPID)
clone(CLONE_THREAD|CLONE_SIGHAND|CLONE_VM)
Creates a new process in the new pid namespace without setting
pid_ns->child_reaper. After forking this results in a NULL
pointer dereference.
Avoid this and other nonsense scenarios that can show up after
creating a new pid namespace with unshare by adding a new
check in copy_prodcess.
Pointed-out-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Sedat reported the following commit caused a regression:
commit 9650388b5c
Author: Eric Dumazet <edumazet@google.com>
Date: Fri Dec 21 07:32:10 2012 +0000
ipv4: arp: fix a lockdep splat in arp_solicit
This is due to the 6th parameter of arp_send() needs to be NULL
for the broadcast case, the above commit changed it to an all-zero
array by mistake.
Reported-by: Sedat Dilek <sedat.dilek@gmail.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Cc: Sedat Dilek <sedat.dilek@gmail.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If of_get_gpio_flags() returns an error (as in case when GPIO probe is
deferred) the driver would attempt to claim invalid GPIO. It should
propagate the error code up the stack instead so that the probe either
fails or will be retried later (in case of -EPROBE_DEFER).
Cc: stable@vger.kernel.org
Reported-by: Gabor Juhos <juhosg@openwrt.org>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
If GPIO probing is deferred, the driver tries to claim an invalid GPIO line
which leads to an error message like this:
gpio-keys-polled buttons.2: unable to claim gpio 4294966779, err=-22
gpio-keys-polled: probe of buttons.2 failed with error -22
We should make sure that error code returned by of_get_gpio_flags (including
-EPROBE_DEFER) is propagated up the stack.
Cc: stable@vger.kernel.org
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Report only the position of the first finger as absolute non-MT coordinates,
instead of reporting both fingers alternatively. Actual MT events are
unaffected.
This fixes horizontal and improves vertical scrolling with the touchpad.
Cc: stable@vger.kernel.org
Signed-off-by: Christophe TORDEUX <christophe@tordeux.net>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
The new kernel module syscall appraises kernel modules based
on policy. If the IMA policy requires kernel module checking,
fallback to module signature enforcing for the existing syscall.
Without CONFIG_MODULE_SIG_FORCE enabled, the kernel module's
integrity is unknown, return -EACCES.
Changelog v1:
- Fix ima_module_check() return result (Tetsuo Handa)
Reported-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Reviewed-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: Mimi Zohar <zohar@us.ibm.com>
Florian Westphal reported that the removal of the NOTRACK target
(9655050 netfilter: remove xt_NOTRACK) is breaking some existing
setups.
That removal was scheduled for removal since long time ago as
described in Documentation/feature-removal-schedule.txt
What: xt_NOTRACK
Files: net/netfilter/xt_NOTRACK.c
When: April 2011
Why: Superseded by xt_CT
Still, people may have not notice / may have decided to stick to an
old iptables version. I agree with him in that some more conservative
approach by spotting some printk to warn users for some time is less
agressive.
Current iptables 1.4.16.3 already contains the aliasing support
that makes it point to the CT target, so upgrading would fix it.
Still, the policy so far has been to avoid pushing our users to
upgrade.
As a solution, this patch recovers the NOTRACK target inside the CT
target and it now spots a warning.
Reported-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
I kept the count as the hardware default with dvb-usb-v2, with 5, users
can still run in to trouble with Video PIDs.
I have traced it to an incorrect endpoint size when the PID filter
is enabled. It also affected USB 2.0 with the filter on.
Reported-by: Antti Palosaari <crope@iki.fi>
Tested-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Malcolm Priestley <tvboxspy@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This series add remote control support for MyGica X8507.
I test for 2 month under OpenSuse(X64) 11.4 and 12.2 with
kernel 3.4, 3.5, 3.6 also 3.7-rc2 and rc3.
[mchehab@redhat.com: fixed whitespacing - it seems that Alfredo's emailer mangled
it]
Signed-off-by: Alfredo J. Delaiti <alfredodelaiti@netscape.net>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Using bulk mode allows more than one webcam, as the maximum fps
is low at 640x480 resolution. So, prefer it, if the device is
a webcam.
Tested with Silvercrest 1.3 Mpixel webcam (em2710) on both bulk and isoc
modes.
Tested analog with HVR-950 model 65201/A1C0 (em2883), where only ISOC
endpoints are available for both DVB and Analog.
Tested on Hauppauge WinTV USB 2 (em2840) on both bulk and isoc modes.
It should be noticed that enabling bulk mode by default with TV boards
is a bad idea; what happens is that, while with ISOC the USB logic will
prevent the concurrent usage of two devices that spends more than 100%
of the USB2 traffic, it doesn't care with bulk transfers.
On my tests, I started two streams, one with a WinTV at 640x480x30fps
and the other one with a Silvercrest webcam at 640x480, on a lower fps)
both on bulk mode. One of the streams always silently failed.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
In order to make easier to analize the logs when multiple devices
are plugged, change the device name accordingly with the chip
version.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Pull i2c __dev* attribute removal from Wolfram Sang:
"The squashed patches from Bill to get rid of the __dev* annotations in
the i2c subsystem. I couldn't include it in my previous pull request
due to some dependency with the mfd subsystem. I had this patch in
linux-next for two days before rc1 and nothing popped up."
* 'i2c-embedded/for-next' of git://git.pengutronix.de/git/wsa/linux:
i2c: remove __dev* attributes from subsystem
Teach pgdat_balanced() about order-0 allocations so that we can simplify
code in a few places in vmstat.c.
Suggested-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Zlatko Calusic <zlatko.calusic@iskon.hr>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit f0ed2ce840 ("[media] uvcvideo: Set error_idx properly for
extended controls API failures") causes user space to behave incorrectly
on one of my test machines (there is no sound under KDE 4.9.4 using
pulseaudio and there is a knotify4 process occupying one of the CPU
cores 100% of the time). Reverting that commit entirely fixes the
problem for me.
However, commit f0ed2ce840 appears to do more than it follows from its
changelog, because the changelog only says about the changes related to
ctrls->error_idx, while the commit additionally changes error codes
returned by various functions in uvc_ctrl.c and uvc_v4l2.c. It turns
out that the changes of the returned error codes confuse the user spce,
so it is sufficient to revert the part of commit f0ed2ce840 not
mentioned in its changelog to fix the problem.
[ 'ENOENT' is not a valid error return from an ioctl to begin with, and
I don't understand how anybody ever even thought it would be. - Linus ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Mauro Carvalho Chehab <mchehab@redhat.com>
Cc: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit 68136b10 ("ARM: sunxi: Change device tree naming scheme for
sunxi") changed the naming scheme and the compatible strings used in the
device trees related to the sunXi platform, but forgot to change the
compatible string in the DT machine definition.
This prevents the kernel from booting on these boards.
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
As both bulk and isoc modes can be available, display what it
was found for both DVB and analog.
While here, also displays if audio is provided via USB Audio
Class or via vendor's extension.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
As the bulk mode is set at device's probe, it is not possible
to change it later. So, change the parameter to be read only
after modprobing.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Modesetting seems to work alright, as does graphics (using binary driver
fuc from nve7...).
Lots to be done no doubt, but this'll get an image on the screen for
people.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
The code in em28xx_vbi_copy can be simplified a lot.
Also rename some variables to something more meaningful and fix+add the
function descriptions.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
em28xx_urb_data_copy() actually consists of two parts:
USB urb processing (checks, data extraction) and frame data packet processing.
Move the latter to a separate function and call it from em28xx_urb_data_copy()
for each data packet.
The em25xx, em2760, em2765 (and likely em277x) chip variants are using a
different frame data format, for which support will be added later with
another function.
This reduces the size of em28xx_urb_data_copy() and makes the code much more
readable. While we're at it, clean up the code a bit (rename some variables to
something more meaningful, improve some comments etc.)
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Reduce code duplication by moving the duplicate code for dev->capture_type=0
(vbi start) and dev->capture_type=2 (video start) to a function.
The same function will also be called by the (not yet existing) em25xx frame
data processing code.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
In the current code em28xx_urb_data_copy() caches the pointer to the vmalloc
memory in videobuf locally.
The alternative would be to call videobuf_to_vmalloc() for each processed USB
data packet (isoc USB transfers => 64 times per URB) in the em28xx_copy_*()
functions.
With the next commits, the data processing code will be split into functions
for serveral reasons:
- em28xx_urb_data_copy() is generally way to long, making it less readable
- there is code duplication between VBI and video data processing
- support for em25xx data processing (uses a different header and frame
end signaling mechanism) will be added
This would require extensive usage of pointer-pointers, which usually makes the
code less readable and prone to bugs.
The better solution is to cache the pointer in struct em28xx_buffer.
This also improves consistency, because we already track the buffer fill count there.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
When a new frame header is detected in em28xx_urb_data_copy() and the data
packet contains both, VBI data and video data, the prevoius VBI buffer doesn't
get finished and is overwritten with the new VBI data.
This bug is not triggered with isochronous USB transfers, because the data
packetes are much smaller than the VBI data size.
But when using USB bulk transfers, the whole data of an URB is treated as
single packet, which is usually much larger then the VBI data size.
Refactor the VBI data processing code to fix this bug, but also to simplify the
code and make it similar to the video data processing code part (which allows
further code abstraction/unification in the future).
The changes have been tested with device "Hauppauge HVR-900".
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This field is used to keep track of the current memory position in the buffer,
not in the dma queue, so move it to right place.
This also allows us to get rid of the struct em28xx_dmaqueue pointer parameter
in functions em28xx_copy_video() and em28xx_copy_vbi().
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
get_next_buf() and vbi_get_next_buf() do exactly the same just with a
different dma queue and buffer. Saving the new buffer pointer back to the
device struct in em28xx_urb_data_copy() instead of doing this from inside
these functions makes it possible to get rid of one of them.
Also refactor the function parameters and return type:
- pass a pointer to struct em28xx as parameter (instead of obtaining the
pointer from the dma queue pointer with the container_of macro) like we do
it in all other functions
- instead of using a pointer-pointer, return the pointer to the new buffer
as return value of the function
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
em28xx_urb_data_copy_vbi() is actually an extended version of
em28xx_urb_data_copy(). With the preceding fixes and improvements, it works
fine with both, vbi and non-vbi data streams without performance impacts.
So rename em28xx_urb_data_copy_vbi() to em28xx_urb_data_copy(), delete the
the old implementation of em28xx_urb_data_copy() and change the code to use
this function for both data stream types.
Tested with "SilverCrest 1.3 MPix webcam" (progressive, non-vbi) and
"Hauppauge HVR-900 (65008/A1C0)" (interlaced, vbi enabled and disabled).
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Set capture type to 1 (video start) when the video frame start header is
detected. This bug didn't cause any trouble, because this type of header is
never received in vbi mode.
Fix it, because we want to use this function with disabled vbi in the future.
Also start with capture type -1 to avoid processing of corrupted/incomplete
frame data which is usually received at streaming start (especially when
USB bulk transfers are used).
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The header check/removal code at the end of function em28xx_urb_data_copy_vbi()
is obsolete, because this is already done earlier in this function.
In fact it is incomplete (doesn't check for vbi header) and causes trouble
when the first data bytes are the same as header bytes (which is fortunately
very unlikely).
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
By default, isoc transfers are used if possible.
With the new module parameter, bulk can be selected as the
preferred USB transfer type.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The current enpoint logic ignores all bulk endpoints and uses
a fixed mapping between endpint addresses and the supported
data stream types (analog/audio/DVB):
Ep 0x82, isoc => analog
Ep 0x83, isoc => audio
Ep 0x84, isoc => DVB
Now that the code can also do bulk transfers, the endpoint
logic has to be extended to also consider bulk endpoints.
The new logic preserves backwards compatibility and reflects
the endpoint configurations we have seen so far:
Ep 0x82, isoc => analog
Ep 0x82, bulk => analog
Ep 0x83, isoc* => audio
Ep 0x84, isoc => digital
Ep 0x84, bulk => analog or digital**
(*: audio should always be isoc)
(**: analog, if ep 0x82 is isoc, otherwise digital)
[mchehab@redhat.com: Fix a CodingStyle issue: don't break strings
into separate lines]
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
