[media] dib8000: improve the message that reports per-layer locks
The message is currently highly encoded:
[70299.863521] DiB8000: Mpeg locks [ L0 : 0 | L1 : 1 | L2 : 0 ]
And doesn't properly reflect that some problems might have happened.
Instead, display it as:
[75160.822321] DiB8000: Not all ISDB-T layers locked in 32 ms: Layer A NOT LOCKED, Layer B locked, Layer C not enabled
In order to better reflect what's happening with layer locking.
Acked-By: Patrick Boettcher <pboettcher@kernellabs.com> Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
When debug is enabled:
- Report when frontend gets restarted;
- Be coherent on the displayed lines;
- Show the transmission mode;
- Hide unused layers.
No functional changes (except at the printk's).
Acked-By: Patrick Boettcher <pboettcher@kernellabs.com> Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
[media] dib0090: Fix the sleep time at the state machine
msleep() is not too precise: its precision depends on the
HZ config. As the driver selects precise timings for the
state machine, change it to usleep_range().
Acked-By: Patrick Boettcher <pboettcher@kernellabs.com> Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
[media] dib8000: Fix the sleep time at the state machine
msleep() is not too precise: its precision depends on the
HZ config. As the driver selects precise timings for the
state machine, change it to usleep_range().
Acked-By: Patrick Boettcher <pboettcher@kernellabs.com> Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
[media] dib8000: In auto-search, try first with partial reception enabled
TV broadcasters generally use partial reception. So, enable it by
default in auto-search mode. The driver will latter detect if the
transmission is on some other mode.
Acked-By: Patrick Boettcher <pboettcher@kernellabs.com> Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
[media] adv7604: Update recommended writes for the adv7611
Update the recommended writes to those mentioned in the Rev 1.5 version of the
ADV7611 Register Settings Recommendations document released by Analog Devices.
The document does not mention why the recommended settings have been updated,
but presumably those are more fine tuned settings that can enhance performance
in some cases.
Philipp Zabel [Mon, 26 May 2014 14:17:32 +0000 (11:17 -0300)]
[media] videobuf2-dma-contig: allow to vmap contiguous dma buffers
This allows drivers to vmap contiguous dma buffers so they can inspect the
buffer contents with the CPU. This will be needed for the CODA driver's JPEG
handling. On CODA960, the header parsing has to be done on the CPU. The
hardware modules can only process the entropy coded segment after all
registers and tables are set up.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de> Acked-by: Marek Szyprowski <m.szyprowski@samsung.com> Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Hans Verkuil [Mon, 16 Jun 2014 07:47:58 +0000 (04:47 -0300)]
[media] sn9c102: remove deprecated driver
During the media summit meeting in Edinburgh it was decided to move this
driver to staging as the first step to removing it altogether.
Most webcams covered by this driver are now supported by gspca. Nobody has the
hardware or is willing to convert the remaining devices to gspca.
This driver needs a major overhaul to have it conform to the latest frameworks
and compliancy tests.
Without hardware, however, this is next to impossible. Given the fact that
this driver seems to be pretty much unused (it has been removed from Fedora
several versions ago and nobody complained about that), we decided to drop
this driver.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Cc: Luca Risolia <luca.risolia@studio.unibo.it> Acked-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Paul Bolle [Thu, 22 May 2014 19:42:13 +0000 (16:42 -0300)]
[media] dm644x_ccdc: remove check for CONFIG_DM644X_VIDEO_PORT_ENABLE
A check for CONFIG_DM644X_VIDEO_PORT_ENABLE was added in v2.6.32. The
related Kconfig symbol was never added so this check has always
evaluated to false. Remove that check.
Signed-off-by: Paul Bolle <pebolle@tiscali.nl> Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
This patch remove variables that are initialized with a constant,
are never updated, and are only used as parameter of return.
Return the constant instead of using a variable.
Verified by compilation only.
The coccinelle script that find and fixes this issue is:
// <smpl>
@@
type T;
constant C;
identifier ret;
@@
- T ret = C;
... when != ret
when strict
return
- ret
+ C
;
// </smpl>
Signed-off-by: Peter Senna Tschudin <peter.senna@gmail.com> Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
This patch remove variables that are initialized with a constant,
are never updated, and are only used as parameter of return.
Return the constant instead of using a variable.
Verified by compilation only.
The coccinelle script that find and fixes this issue is:
// <smpl>
@@
type T;
constant C;
identifier ret;
@@
- T ret = C;
... when != ret
when strict
return
- ret
+ C
;
// </smpl>
Signed-off-by: Peter Senna Tschudin <peter.senna@gmail.com> Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Hans Verkuil [Tue, 27 May 2014 07:12:19 +0000 (04:12 -0300)]
[media] em28xx: add MSI Digivox Trio support
Add support for the MSI Digivox Trio device.
Signed-off-by: P. van Gaans <w3ird_n3rd@gmx.net> Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Antonio Ospite [Wed, 25 Jun 2014 09:27:57 +0000 (06:27 -0300)]
[media] gspca_kinect: add support for the depth stream
Add support for the depth stream at 10bpp, for now use a 'depth_mode'
command line parameter to switch between video and depth mode.
Signed-off-by: Alexander Sosna <alexander@xxor.de> Signed-off-by: Antonio Ospite <ao2@ao2.it> Signed-off-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Antonio Ospite [Wed, 25 Jun 2014 09:27:56 +0000 (06:27 -0300)]
[media] gspca: provide a mechanism to select a specific transfer endpoint
Currently gspca selects the first ISOC input endpoint as the input
transfer endpoint, however some devices can provide streams on endpoints
different then the first one, so some subdrivers (e.g. gspca_kinect) may
want to select a specific endpoint to use as a transfer endpoint.
Add an xfer_ep field to struct gspca_dev, and change alt_xfer() so that
it accepts a parameter which represents a specific endpoint address to
look for.
If a subdriver wants to specify a value for gspca_dev->xfer_ep it can do
that in its sd_config() callback.
Signed-off-by: Antonio Ospite <ao2@ao2.it> Signed-off-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
While this device doesn't have a scaler (or have it disabled),
the screen dimentions are a function of the standard. Ok, right
now, only 480 lines standards are implemented, although it
supports other ones. Yet, let's calculate the size, to make
easier to add more standards latter.
[media] au0828: Only alt setting logic when needed
It seems that there's a bug at au0828 hardware/firmware
related to alternate setting: when the device is already at
alt 5, a further call causes the URBs to receive -ESHUTDOWN.
I found two different encarnations of this issue:
1) at qv4l2, it fails the second time we try to open the
video screen;
2) at xawtv, when audio underrun occurs, with is very
frequent, at least on my test machine.
The fix is simple: just check if alt=5 before calling
set_usb_interface().
There are several issues on the current code:
1) msleep(1) is actually equivalent to msleep(10);
2) au8522_video_set() will set reg 0xa4 to the
proper value for SIF, CVBS or S-Video. No need
to force it to CVBS;
3) Let's not hardcode 0x9d for CBS on audio_set.
[media] au8522: be sure that we'll setup audio routing at the right time
Let's set the the audio routing also at stream start. With this change,
we don't risk enabling the analog demux while not streaming, reducing
the risk of interfering with a DVB demux that might be happening.
[media] au8522: be sure that the setup will happen at streamon time
The same demod is used on both analog and digital mode. We should
not let the commands for analog mode to happen while the device
is in digital mode. So, monitor it via streamon.
Alexander Shiyan [Sat, 24 May 2014 04:56:22 +0000 (01:56 -0300)]
[media] media: mx1_camera: Remove driver
That driver hasn't been really maintained for a long time. It doesn't
compile in any way, it includes non-existent headers, has no users,
and marked as "BROKEN" more than year. Due to these factors, mx1_camera
is now removed from the tree.
[media] V4L: soc-camera: explicitly free allocated managed memory on error
devm_kzalloc() allocations are freed when the device is unbound. But if a
certain path fails and the allocated memory cannot be used anyway it is
better to free it explicitly immediately. This patch does exactly this if
asynchronous group probing in scan_async_group() fails after memory has
been allocated.
For this callibration, I used adapter 0 (antenna 1), connected
via a normal cable to a DTF-2111 generator.
The same cabling and RF generator connected to adapter 1
(antenna 2) has a difference of +3dBm (signal was stronger
there).
Yet, changing the RF for a difference of, let's say, 6dB
reflected on a 6dB difference at the measured signal, with
is good.
So, the signal strengh can indeed be used to measure the
antenna gain, if the antenna is repositioned. It is
not precise to measure the absolute dBm value, however.
Adds DVBv5 stats support. For now, just mimic whatever dib8000
does, as they're very similar, with regards to statistics.
However, dib7000p_get_time_us() likely require some
adjustments, as I didn't actually reviewed the formula
for it to work with DVB-T. Still, better than nothing,
as latter patches can improve it.
This patch also doesn't show the signal strength in dB
yet. The code is already there, but it requires to be
callibrated.
A latter patch will do the calibration.
It seems that this patch is also a bug fix: Before this
patch, the frontend were not tuning with some userspace
tools. I suspect that dib7000p firmware or hardware
internally expects that the statistics to be collected,
in order for it to work. With this patch, the DVB core
will always retrive statistics, even if userspace doesn't
request. So, it makes the device work on all tested apps.
[media] dib8000: rename dib8000_attach to dib8000_init
Well, what we call as "foo_attach" is the method that should
be called by the dvb_attach() macro.
It should be noticed that the name "dvb_attach" is really a
bad name and don't express what it does.
dvb_attach() basically does three things, if the frontend is
compiled as a module:
- It lookups for the module that it is known to have the
given symbol name and requests such module;
- It increments the module usage (anonymously - so lsmod
doesn't print who loaded the module);
- after loading the module, it runs the function associated
with the dynamic symbol.
When compiled as builtin, it just calls the function given to it.
As dvb_attach() increments refcount, it can't be (easily)
called more than once for the same module, or the kernel
will deny to remove the module, because refcount will never
be zeroed.
In other words, the function name given to dvb_attach()
should be one single symbol that will always be called
before any other function on that module to be used.
For almost all DVB frontends, there's just one function,
but, on dib8000, there are several exported symbols.
We need to get rid of all those direct calls, because they
cause compilation breakages when bridge is builtin and
frontend is module, we'll need to add a new function that
will be the first one to be called, whatever initialization
is needed.
So, let's rename this function, in order to prepare for
a next patch that will add a new attach() function that
will be the only one exported by this module.
Exporting multiple symbols don't work as it causes compilation
breakages, due to the way dvb_attach() works.
This were reported several times, like:
drivers/built-in.o: In function `cxusb_dualdig4_rev2_tuner_attach':
>> cxusb.c:(.text+0x27d4b5): undefined reference to `dib7000p_get_i2c_master'
drivers/built-in.o: In function `dib7070_set_param_override':
cxusb.c:(.text+0x27d5a5): undefined reference to `dib0070_wbd_offset'
>> cxusb.c:(.text+0x27d5be): undefined reference to `dib7000p_set_wbd_ref'
drivers/built-in.o: In function `dib7070_tuner_reset':
>> cxusb.c:(.text+0x27d5f9): undefined reference to `dib7000p_set_gpio'
drivers/built-in.o: In function `cxusb_dualdig4_rev2_frontend_attach':
>> cxusb.c:(.text+0x27df5c): undefined reference to `dib7000p_i2c_enumeration'
In this specific report:
CONFIG_DVB_USB_CXUSB=y
CONFIG_DVB_DIB7000P=m
But the same type of bug can happen if:
CONFIG_DVB_DIB7000P=m
and one of the bridge drivers is compiled builtin (cxusb, cx23885-dvb
and/or dib0700).
As a bonus, dib7000p won't be loaded anymore if the device uses
a different frontend, reducing the memory footprint.
[media] dib7000p: rename dib7000p_attach to dib7000p_init
Well, what we call as "foo_attach" is the method that should
be called by the dvb_attach() macro.
It should be noticed that the name "dvb_attach" is really a
bad name and don't express what it does.
dvb_attach() basically does three things, if the frontend is
compiled as a module:
- It lookups for the module that it is known to have the
given symbol name and requests such module;
- It increments the module usage (anonymously - so lsmod
doesn't print who loaded the module);
- after loading the module, it runs the function associated
with the dynamic symbol.
When compiled as builtin, it just calls the function given to it.
As dvb_attach() increments refcount, it can't be (easily)
called more than once for the same module, or the kernel
will deny to remove the module, because refcount will never
be zeroed.
In other words, the function name given to dvb_attach()
should be one single symbol that will always be called
before any other function on that module to be used.
For almost all DVB frontends, there's just one function.
However, the dib7000p initialization can require up to 3
functions to be called:
- dib7000p_get_i2c_master;
- dib7000p_i2c_enumeration;
- dib7000p_init (before this patch dib7000_attach).
(plus a bunch of other functions that the bridge driver will
need to call).
As we need to get rid of all those direct calls, because they
cause compilation breakages when bridge is builtin and
frontend is module, we'll need to add a new function that
will be the first one to be called, whatever initialization
is needed.
So, let's rename the function that probes and init the hardware
to dib7000p_init.
A latter patch will add a new dib7000p_attach that will be
used as originally conceived by dvb_attach() way.
The dvb_attach() was unbalanced, as there was no dvb_dettach. Ok,
on current cases, the dettach is done by dvbdev, but that are some
future corner cases where we may need to do this before registering
the frontend.
So, add a dvb_detach() and use it at dvb_frontend.c.
Sakari Ailus [Mon, 12 May 2014 11:30:55 +0000 (08:30 -0300)]
[media] smiapp: I2C address is the last part of the subdev name
The I2C address of the sensor device was in the middle of the sub-device
name and not in the end as it should have been. The smiapp sub-device names
will change from e.g. "vs6555 1-0010 pixel array" to "vs6555 pixel array
1-0010".
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com> Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Lad, Prabhakar [Fri, 16 May 2014 13:33:47 +0000 (10:33 -0300)]
[media] media: davinci: vpif_capture: drop unneeded module params
Remove bogus 'numbuffers' and 'bufsize' module options. The number of buffers and
buffer sizes are determined by VIDIOC_REQBUFS and VIDIOC_S_FMT and the amount of
available memory (in the case of the MMAP stream I/O mode) and not by module
options.
These module params are a left-over from the original montavista code that used
these parameters to pre-allocate the memory needed for the buffers. The code that
allocated those buffers was never upstreamed since by the time the drivers were
added to the kernel the TI cmem module could be used in combination with the
USERPTR mode to reserve and pass physically contiguous memory pointers around.
These days of course CMA is used instead of cmem.
This patch removes these module options altogether since they no longer do what
they originally were designed for. They should never have been part of the
upstreamed code in the first place, so they've been pointless ever since
2.6.32 when this driver first appeared in the mainline kernel.
Laurent Pinchart [Wed, 21 May 2014 20:39:16 +0000 (17:39 -0300)]
[media] v4l: vsp1: Remove the unneeded vsp1_video_buffer video field
The field is assigned but never read, remove it.
This fixes a bug caused by the struct vb2_buffer field not being be the
very first field of the vsp1_video_buffer buffer structure as required
by videobuf2.
* tag 'v3.16-rc1': (11443 commits)
Linux 3.16-rc1
net: sctp: fix permissions for rto_alpha and rto_beta knobs
vxlan: Checksum fixes
net: add skb_pop_rcv_encapsulation
udp: call __skb_checksum_complete when doing full checksum
net: Fix save software checksum complete
net: Fix GSO constants to match NETIF flags
fix __swap_writepage() compile failure on old gcc versions
udp: ipv4: do not waste time in __udp4_lib_mcast_demux_lookup
vxlan: use dev->needed_headroom instead of dev->hard_header_len
MAINTAINERS: update cxgb4 maintainer
x86/vdso: Fix vdso_install
NVMe: Fix START_STOP_UNIT Scsi->NVMe translation.
btrfs: fix error handling in create_pending_snapshot
btrfs: fix use of uninit "ret" in end_extent_writepage()
btrfs: free ulist in qgroup_shared_accounting() error path
Btrfs: fix qgroups sanity test crash or hang
btrfs: prevent RCU warning when dereferencing radix tree slot
Btrfs: fix unfinished readahead thread for raid5/6 degraded mounting
btrfs: new ioctl TREE_SEARCH_V2
...
1) Fix checksumming regressions, from Tom Herbert.
2) Undo unintentional permissions changes for SCTP rto_alpha and
rto_beta sysfs knobs, from Denial Borkmann.
3) VXLAN, like other IP tunnels, should advertize it's encapsulation
size using dev->needed_headroom instead of dev->hard_header_len.
From Cong Wang.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
net: sctp: fix permissions for rto_alpha and rto_beta knobs
vxlan: Checksum fixes
net: add skb_pop_rcv_encapsulation
udp: call __skb_checksum_complete when doing full checksum
net: Fix save software checksum complete
net: Fix GSO constants to match NETIF flags
udp: ipv4: do not waste time in __udp4_lib_mcast_demux_lookup
vxlan: use dev->needed_headroom instead of dev->hard_header_len
MAINTAINERS: update cxgb4 maintainer
Linus Torvalds [Mon, 16 Jun 2014 02:02:20 +0000 (16:02 -1000)]
Merge tag 'clk-for-linus-3.16-part2' of git://git.linaro.org/people/mike.turquette/linux
Pull more clock framework updates from Mike Turquette:
"This contains the second half the of the clk changes for 3.16.
They are simply fixes and code refactoring for the OMAP clock drivers.
The sunxi clock driver changes include splitting out the one
mega-driver into several smaller pieces and adding support for the A31
SoC clocks"
* tag 'clk-for-linus-3.16-part2' of git://git.linaro.org/people/mike.turquette/linux: (25 commits)
clk: sunxi: document PRCM clock compatible strings
clk: sunxi: add PRCM (Power/Reset/Clock Management) clks support
clk: sun6i: Protect SDRAM gating bit
clk: sun6i: Protect CPU clock
clk: sunxi: Rework clock protection code
clk: sunxi: Move the GMAC clock to a file of its own
clk: sunxi: Move the 24M oscillator to a file of its own
clk: sunxi: Remove calls to clk_put
clk: sunxi: document new A31 USB clock compatible
clk: sunxi: Implement A31 USB clock
ARM: dts: OMAP5/DRA7: use omap5-mpu-dpll-clock capable of dealing with higher frequencies
CLK: TI: dpll: support OMAP5 MPU DPLL that need special handling for higher frequencies
ARM: OMAP5+: dpll: support Duty Cycle Correction(DCC)
CLK: TI: clk-54xx: Set the rate for dpll_abe_m2x2_ck
CLK: TI: Driver for DRA7 ATL (Audio Tracking Logic)
dt:/bindings: DRA7 ATL (Audio Tracking Logic) clock bindings
ARM: dts: dra7xx-clocks: Correct name for atl clkin3 clock
CLK: TI: gate: add composite interface clock to OMAP2 only build
ARM: OMAP2: clock: add DT boot support for cpufreq_ck
CLK: TI: OMAP2: add clock init support
...
Pull NVMe update from Matthew Wilcox:
"Mostly bugfixes again for the NVMe driver. I'd like to call out the
exported tracepoint in the block layer; I believe Keith has cleared
this with Jens.
We've had a few reports from people who're really pounding on NVMe
devices at scale, hence the timeout changes (and new module
parameters), hotplug cpu deadlock, tracepoints, and minor performance
tweaks"
[ Jens hadn't seen that tracepoint thing, but is ok with it - it will
end up going away when mq conversion happens ]
* git://git.infradead.org/users/willy/linux-nvme: (22 commits)
NVMe: Fix START_STOP_UNIT Scsi->NVMe translation.
NVMe: Use Log Page constants in SCSI emulation
NVMe: Define Log Page constants
NVMe: Fix hot cpu notification dead lock
NVMe: Rename io_timeout to nvme_io_timeout
NVMe: Use last bytes of f/w rev SCSI Inquiry
NVMe: Adhere to request queue block accounting enable/disable
NVMe: Fix nvme get/put queue semantics
NVMe: Delete NVME_GET_FEAT_TEMP_THRESH
NVMe: Make admin timeout a module parameter
NVMe: Make iod bio timeout a parameter
NVMe: Prevent possible NULL pointer dereference
NVMe: Fix the buffer size passed in GetLogPage(CDW10.NUMD)
NVMe: Update data structures for NVMe 1.2
NVMe: Enable BUILD_BUG_ON checks
NVMe: Update namespace and controller identify structures to the 1.1a spec
NVMe: Flush with data support
NVMe: Configure support for block flush
NVMe: Add tracepoints
NVMe: Protect against badly formatted CQEs
...
Daniel Borkmann [Sat, 14 Jun 2014 22:59:14 +0000 (00:59 +0200)]
net: sctp: fix permissions for rto_alpha and rto_beta knobs
Commit 3fd091e73b81 ("[SCTP]: Remove multiple levels of msecs
to jiffies conversions.") has silently changed permissions for
rto_alpha and rto_beta knobs from 0644 to 0444. The purpose of
this was to discourage users from tweaking rto_alpha and
rto_beta knobs in production environments since they are key
to correctly compute rtt/srtt.
RFC4960 under section 6.3.1. RTO Calculation says regarding
rto_alpha and rto_beta under rule C3 and C4:
[...]
C3) When a new RTT measurement R' is made, set
Note: The value of SRTT used in the update to RTTVAR
is its value before updating SRTT itself using the
second assignment. After the computation, update
RTO <- SRTT + 4 * RTTVAR.
C4) When data is in flight and when allowed by rule C5
below, a new RTT measurement MUST be made each round
trip. Furthermore, new RTT measurements SHOULD be
made no more than once per round trip for a given
destination transport address. There are two reasons
for this recommendation: First, it appears that
measuring more frequently often does not in practice
yield any significant benefit [ALLMAN99]; second,
if measurements are made more often, then the values
of RTO.Alpha and RTO.Beta in rule C3 above should be
adjusted so that SRTT and RTTVAR still adjust to
changes at roughly the same rate (in terms of how many
round trips it takes them to reflect new values) as
they would if making only one measurement per
round-trip and using RTO.Alpha and RTO.Beta as given
in rule C3. However, the exact nature of these
adjustments remains a research issue.
[...]
While it is discouraged to adjust rto_alpha and rto_beta
and not further specified how to adjust them, the RFC also
doesn't explicitly forbid it, but rather gives a RECOMMENDED
default value (rto_alpha=3, rto_beta=2). We have a couple
of users relying on the old permissions before they got
changed. That said, if someone really has the urge to adjust
them, we could allow it with a warning in the log.
Fixes: 3fd091e73b81 ("[SCTP]: Remove multiple levels of msecs to jiffies conversions.") Signed-off-by: Daniel Borkmann <dborkman@redhat.com> Cc: Vlad Yasevich <vyasevich@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Sun, 15 Jun 2014 08:00:56 +0000 (01:00 -0700)]
Merge branch 'csum_fixes'
Tom Herbert says:
====================
Fixes related to some recent checksum modifications.
- Fix GSO constants to match NETIF flags
- Fix logic in saving checksum complete in __skb_checksum_complete
- Call __skb_checksum_complete from UDP if we are checksumming over
whole packet in order to save checksum.
- Fixes to VXLAN to work correctly with checksum complete
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Tom Herbert [Sun, 15 Jun 2014 06:24:28 +0000 (23:24 -0700)]
net: add skb_pop_rcv_encapsulation
This function is used by UDP encapsulation protocols in RX when
crossing encapsulation boundary. If ip_summed is set to
CHECKSUM_UNNECESSARY and encapsulation is not set, change to
CHECKSUM_NONE since the checksum has not been validated within the
encapsulation. Clears csum_valid by the same rationale.
Signed-off-by: Tom Herbert <therbert@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Tom Herbert [Sun, 15 Jun 2014 06:24:20 +0000 (23:24 -0700)]
udp: call __skb_checksum_complete when doing full checksum
In __udp_lib_checksum_complete check if checksum is being done over all
the data (len is equal to skb->len) and if it is call
__skb_checksum_complete instead of __skb_checksum_complete_head. This
allows checksum to be saved in checksum complete.
Signed-off-by: Tom Herbert <therbert@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Tom Herbert [Sun, 15 Jun 2014 06:24:03 +0000 (23:24 -0700)]
net: Fix save software checksum complete
Geert reported issues regarding checksum complete and UDP.
The logic introduced in commit 7e3cead5172927732f51fde
("net: Save software checksum complete") is not correct.
This patch:
1) Restores code in __skb_checksum_complete_header except for setting
CHECKSUM_UNNECESSARY. This function may be calculating checksum on
something less than skb->len.
2) Adds saving checksum to __skb_checksum_complete. The full packet
checksum 0..skb->len is calculated without adding in pseudo header.
This value is saved in skb->csum and then the pseudo header is added
to that to derive the checksum for validation.
3) In both __skb_checksum_complete_header and __skb_checksum_complete,
set skb->csum_valid to whether checksum of zero was computed. This
allows skb_csum_unnecessary to return true without changing to
CHECKSUM_UNNECESSARY which was done previously.
4) Copy new csum related bits in __copy_skb_header.
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Tom Herbert <therbert@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Tom Herbert [Sun, 15 Jun 2014 06:23:52 +0000 (23:23 -0700)]
net: Fix GSO constants to match NETIF flags
Joseph Gasparakis reported that VXLAN GSO offload stopped working with
i40e device after recent UDP changes. The problem is that the
SKB_GSO_* bits are out of sync with the corresponding NETIF flags. This
patch fixes that. Also, we add BUILD_BUG_ONs in net_gso_ok for several
GSO constants that were missing to avoid the problem in the future.
Reported-by: Joseph Gasparakis <joseph.gasparakis@intel.com> Signed-off-by: Tom Herbert <therbert@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Linus Torvalds [Sun, 15 Jun 2014 00:49:48 +0000 (19:49 -0500)]
Merge tag 'scsi-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Pull more SCSI updates from James Bottomley:
"This is just a couple of drivers (hpsa and lpfc) that got left out for
further testing in linux-next. We also have one fix to a prior
submission (qla2xxx sparse)"
* tag 'scsi-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi: (36 commits)
qla2xxx: fix sparse warnings introduced by previous target mode t10-dif patch
lpfc: Update lpfc version to driver version 10.2.8001.0
lpfc: Fix ExpressLane priority setup
lpfc: mark old devices as obsolete
lpfc: Fix for initializing RRQ bitmap
lpfc: Fix for cleaning up stale ring flag and sp_queue_event entries
lpfc: Update lpfc version to driver version 10.2.8000.0
lpfc: Update Copyright on changed files from 8.3.45 patches
lpfc: Update Copyright on changed files
lpfc: Fixed locking for scsi task management commands
lpfc: Convert runtime references to old xlane cfg param to fof cfg param
lpfc: Fix FW dump using sysfs
lpfc: Fix SLI4 s abort loop to process all FCP rings and under ring_lock
lpfc: Fixed kernel panic in lpfc_abort_handler
lpfc: Fix locking for postbufq when freeing
lpfc: Fix locking for lpfc_hba_down_post
lpfc: Fix dynamic transitions of FirstBurst from on to off
hpsa: fix handling of hpsa_volume_offline return value
hpsa: return -ENOMEM not -1 on kzalloc failure in hpsa_get_device_id
hpsa: remove messages about volume status VPD inquiry page not supported
...
Linus Torvalds [Sun, 15 Jun 2014 00:48:43 +0000 (19:48 -0500)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs
Pull more btrfs updates from Chris Mason:
"This has a few fixes since our last pull and a new ioctl for doing
btree searches from userland. It's very similar to the existing
ioctl, but lets us return larger items back down to the app"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
btrfs: fix error handling in create_pending_snapshot
btrfs: fix use of uninit "ret" in end_extent_writepage()
btrfs: free ulist in qgroup_shared_accounting() error path
Btrfs: fix qgroups sanity test crash or hang
btrfs: prevent RCU warning when dereferencing radix tree slot
Btrfs: fix unfinished readahead thread for raid5/6 degraded mounting
btrfs: new ioctl TREE_SEARCH_V2
btrfs: tree_search, search_ioctl: direct copy to userspace
btrfs: new function read_extent_buffer_to_user
btrfs: tree_search, copy_to_sk: return needed size on EOVERFLOW
btrfs: tree_search, copy_to_sk: return EOVERFLOW for too small buffer
btrfs: tree_search, search_ioctl: accept varying buffer
btrfs: tree_search: eliminate redundant nr_items check
Al Viro [Sat, 14 Jun 2014 06:12:41 +0000 (07:12 +0100)]
fix __swap_writepage() compile failure on old gcc versions
Tetsuo Handa wrote:
"Commit 62a8067a7f35 ("bio_vec-backed iov_iter") introduced an unnamed
union inside a struct which gcc-4.4.7 cannot handle. Name the unnamed
union as u in order to fix build failure"
Let's do this instead: there is only one place in the entire tree that
steps into this breakage. Anon structs and unions work in older gcc
versions; as the matter of fact, we have those in the tree - see e.g.
struct ieee80211_tx_info in include/net/mac80211.h
What doesn't work is handling their initializers:
struct {
int a;
union {
int b;
char c;
};
} x[2] = {{.a = 1, .c = 'a'}, {.a = 0, .b = 1}};
is the obvious syntax for initializer, perfectly fine for C11 and
handled correctly by gcc-4.7 or later.
Earlier versions, though, break on it - declaration is fine and so's
access to fields (i.e. x[0].c = 'a'; would produce the right code), but
members of the anon structs and unions are not inserted into the right
namespace. Tellingly, those older versions will not barf on struct {int
a; struct {int a;};}; - looks like they just have it hacked up somewhere
around the handling of . and -> instead of doing the right thing.
The easiest way to deal with that crap is to turn initialization of
those fields (in the only place where we have such initializer of
iov_iter) into plain assignment.
Reported-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Reported-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>