Commit graph

982,081 commits

Author SHA1 Message Date
Jacopo Mondi
87c6f1b57f media: rcar-csi2: Handle per-SoC number of channels
The R-Car CSI-2 interface has a number of selectable 'channels' that
provides pixel data to the VINs during image acquisition.

Each channel can be used to match a CSI-2 data type and a CSI-2 virtual
channel to be routed to output path.

Different SoCs have different number of channels, with R-Car E3 being the
notable exception supporting only 2 of them.

Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:41:51 -05:00
Jacopo Mondi
10c08812fe media: rcar: rcar-csi2: Update V3M/E3 PHTW tables
Update PHTW tables for V3M and E3 SoCs to the latest datasheet release
(R-Car Series, 3rd Generation manual rev1.00 20181017).

Signed-off-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:41:16 -05:00
Jacopo Mondi
6511459d2a media: rcar-csi2: Add R8A77990 support
Add support for R-Car E3 R8A77965 to R-Car CSI-2 driver.
Based on the experimental patch from Magnus Damm.

Signed-off-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:40:46 -05:00
Jacopo Mondi
72a8914ff0 media: dt-bindings: rcar-csi2: Add R8A77990
Add compatible string for R-Car E3 R8A77990 to the list of supported SoCs.

Signed-off-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:40:09 -05:00
Jacopo Mondi
fad6d6dd4e media: rcar-vin: Add support for R-Car R8A77990
Add R-Car E3 R8A77990 SoC to the rcar-vin supported ones.
Based on the experimental patch from Magnus Damm.

Signed-off-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:39:49 -05:00
Jacopo Mondi
cc4e51b838 media: dt-bindings: rcar-vin: Add R8A77990 support
Add compatible string for R-Car E3 R8A77990 to the list of SoCs supported by
rcar-vin driver.

Signed-off-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:39:10 -05:00
Fabio Estevam
3431ebe5a6 media: imx-pxp: Improve pxp_soft_reset() error message
Improve the pxp_soft_reset() error message by moving it to the
caller function, associating it with a proper device and also
by displaying the error code.

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:38:50 -05:00
Fabio Estevam
e03eb36210 media: imx-pxp: Check for pxp_soft_reset() error
pxp_soft_reset() may fail with a timeout, so it is better to propagate
the error in this case.

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:38:17 -05:00
Fabio Estevam
efceb765ff media: imx-pxp: Check the return value from clk_prepare_enable()
clk_prepare_enable() may fail, so we should better check its return value
and propagate it in the case of error.

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:36:58 -05:00
Philipp Zabel
c229f5c746 media: coda: debug output when setting visible size via crop selection
In addition to the S_FMT debug output, S_SELECTION (SEL_TGT_CROP) is
relevant to determine encoded size. Add debug output for it.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:36:35 -05:00
Philipp Zabel
8f90d15e6d media: coda: normalise debug output
Consistently add the context index to debug output, which otherwise is
impossible to make sense of when two contexts are running concurrently.
For this purpose, add a convenience macro coda_dbg(). Use the function
name with the coda_ prefix stripped as keyword where applicable, and
consistently use vid-out and vid-cap names for the queues. Add sequence
counters to the decoder job finished message and correctly indicate B
frames. Add a start streaming message to complement the stop streaming
message and a start encoding message to complement the existing start
decoding message.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:36:07 -05:00
Philipp Zabel
299cc3181b media: coda: improve queue busy error message
Use v4l2_type_names to indicate which of the two queues is busy.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:31:50 -05:00
Michael Tretter
01a2d72149 media: coda: print SEQ_INIT error code as hex value
The error code looks much more like a bit field than an error value.
Print it as hex rather than decimal.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:31:25 -05:00
Philipp Zabel
7e47c38414 media: coda: fail S_SELECTION for read-only targets
v4l2-compliance complains if S_SELECTION returns 0 for read-only targets.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:31:06 -05:00
Philipp Zabel
cf13135cf5 media: coda: never set infinite timeperframe
v4l2-compliance complains if G_PARM returns 0 in the denominator.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:30:28 -05:00
Philipp Zabel
07b6080d4e media: coda: implement ENUM_FRAMEINTERVALS
v4l2-compliance complains about S_PARM being supported, but not
ENUM_FRAMEINTERVALS.
Report a continuous frame interval even though the hardware only
supports 16-bit numerator and denominator, with min/max values
that can be programmed into the mailbox registers.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:30:03 -05:00
Philipp Zabel
4c5ce24bca media: coda: set V4L2_CAP_TIMEPERFRAME flag in coda_s_parm
The flag is already set in coda_g_parm, but v4l2-compliance complains
about it not being set during S_PARM.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:29:17 -05:00
Lucas Stach
68d66a9310 media: coda: don't disable IRQs across buffer meta handling
The CODA driver uses threaded IRQs only, so there is nothing happening
in hardirq context that could interfere with the buffer meta handling.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:28:51 -05:00
Philipp Zabel
cdd87d3a2c media: coda: remove unused instances list
The per-device instance list is unused, remove it.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:27:59 -05:00
Philipp Zabel
074e8db02c media: coda: reduce minimum frame size to 48x16 pixels.
Three macroblocks seem to be the minimum resolution that can be encoded
and decoded by the CODA960 h.264 codec. Picture run commands fail for
smaller resolutions.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:27:25 -05:00
Lucas Stach
c3d996fb03 media: coda: limit queueing into internal bitstream buffer
The ringbuffer used to hold the bitstream is very conservatively sized,
as keyframes can get very large and still need to fit into this buffer.
This means that the buffer is way oversized for the average stream to
the extend that it will hold a few hundred frames when the video data
is compressing well.

The current strategy of queueing as much bitstream data as possible
leads to large delays when draining the decoder. In order to keep the
drain latency to a reasonable bound, try to only queue a full reorder
window of buffers. We can't always hit this low target for very well
compressible video data, as we might end up with less than the minimum
amount of data that needs to be available to the bitstream prefetcher,
so we must take this into account and allow more buffers to be queued
in this case.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:27:05 -05:00
Philipp Zabel
51407c2da0 media: coda: always hold back decoder jobs until we have enough bitstream payload
The bitstream prefetch unit reads data in 256 byte blocks with some kind
of queueing. For the decoder to see data up to a desired position in the
next run, the bitstream has to be filled for 2 256 byte blocks past that
position aligned up to the next 256 byte boundary.
This should make sure we never run into a buffer underrun condition if
userspace does not supply new input buffers fast enough.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:26:24 -05:00
Philipp Zabel
6c0f5d236f media: coda: store unmasked fifo position in meta
Storing the unmasked kfifo->in position as meta->start and ->end allows
to more easily compare a point past meta->end with the current
kfifo->in.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:21:37 -05:00
Philipp Zabel
649cfc2bdf media: coda: fix memory corruption in case more than 32 instances are opened
The ffz() return value is undefined if the instance mask does not
contain any zeros. If it returned 32, the following set_bit would
corrupt the debugfs_root pointer.
Switch to IDA for context index allocation. This also removes the
artificial 32 instance limit for all except CodaDx6.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:19:39 -05:00
Rui Miguel Silva
c45fbdf24c media: ov2680: fix null dereference at power on
Swapping the order between v4l2 subdevice registration and checking chip
id in b7a417628a ("media: ov2680: don't register the v4l2 subdevice
before checking chip ID") makes the mode restore to use the sensor
controls before they are set, so move the mode restore call to s_power
after the handler setup for controls is done.

This remove also the need for the error code path in power on function.

Fixes: b7a417628a ("media: ov2680: don't register the v4l2 subdevice before checking chip ID")

Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:16:29 -05:00
Sakari Ailus
5f229dcabc media: v4l: uAPI doc: Changing frame interval won't change format
Document that changing the frame interval has no effect on frame size.
While this was the assumption in the API, it was not documented as such.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:16:29 -05:00
Maxime Ripard
482ac2aa36 media: sun6i: Add A31 compatible
The first device that used that IP was the A31. Add it to our list of
compatibles.

Signed-off-by: Maxime Ripard <maxime.ripard@bootlin.com>
Reviewed-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:16:29 -05:00
Maxime Ripard
b7eadaa3a0 media: dt-bindings: media: sun6i: Add A31 and H3 compatibles
The H3 has a slightly different CSI controller (no BT656, no CCI) which
looks a lot like the original A31 controller. Add a compatible for the A31,
and more specific compatible the for the H3 to be used in combination for
the A31.

Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Maxime Ripard <maxime.ripard@bootlin.com>
Reviewed-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:16:28 -05:00
Mauro Carvalho Chehab
71bfeb4279 media: sun6i: manually fix other coding style issues
There are a few other coding style issues reported by checkpatch
while in --strict mode. Fix the ones that make sense.

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:16:24 -05:00
Mauro Carvalho Chehab
34d833a91a media: sum6i: Fix a few coding style issues
Make checkpatch.pl happier by running it on strict mode and
using the --fix-inline to solve some issues.

Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:16:18 -05:00
Yong Deng
5cc7522d89 media: sun6i: Add support for Allwinner CSI V3s
Allwinner V3s SoC features a CSI module with parallel interface.

This patch implement a v4l2 framework driver for it.

Reviewed-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Maxime Ripard <maxime.ripard@bootlin.com>
Tested-by: Maxime Ripard <maxime.ripard@bootlin.com>
Signed-off-by: Yong Deng <yong.deng@magewell.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
2018-11-23 05:15:59 -05:00
Ricardo Reis Marques Silva
16c44c57ec staging: vboxvideo: Rename uint32_t type to u32
Issue found by checkpatch.

CHECK: Prefer kernel type 'u32' over 'uint32_t'
+static const uint32_t vbox_cursor_plane_formats[] = {

CHECK: Prefer kernel type 'u32' over 'uint32_t'
+static const uint32_t vbox_primary_plane_formats[] = {

CHECK: Prefer kernel type 'u32' over 'uint32_t'
+       const uint32_t *formats;

Signed-off-by: Ricardo Reis Marques Silva <ricardormsilva93@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:58:47 +01:00
Michael Straube
b44beecab8 staging: rtl8188eu: remove variable from rtl8188eu_xmit_tasklet()
The local variable ret is only used to test the return value of the
call to rtl8188eu_xmitframe_complete(). Use the function directly in
the if test and remove the variable ret.

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:58:45 +01:00
Michael Straube
fdf60cd7e6 staging: rtl8188eu: cleanup declarations in usb_ops_linux.c
Replace tabs with spaces and/or remove spaces in declarations
to cleanup whitespace.

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:58:45 +01:00
Michael Straube
aa8591db2b staging: rtl8188eu: add spaces around '>>' and '&'
Add spaces around '>>' and '&' to follow kernel coding style.
Reported by checkpatch.

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:58:45 +01:00
Michael Straube
718d2523b9 staging: rtl8188eu: remove unnecessary parentheses
Remove unnecessary parentheses in usb_ops_linux.c.
Reported by checkpatch.

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:58:45 +01:00
Michael Straube
d6ad6d389d staging: rtl8188eu: correct spelling mistake in a comment
Correct spelling mistake in a comment reported by checkpatch.
checksumed -> checksummed

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:58:45 +01:00
Michael Straube
fa50658ec3 staging: rtl8188eu: remove braces from single if else statement
Remove braces from single line if else statement.
Reported by checkpatch.

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:58:45 +01:00
Michael Straube
c9ad9c49cc staging: rtl8188eu: cleanup line ending with a '('
Cleanup line ending with a '(' to follow kernel coding style.
Reported by checkpatch.

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:58:45 +01:00
Michael Straube
c03b0ff72e staging: rtl8188eu: use __func__ in usb_ops_linux.c
Use __func__ instead of hardcoded function names.
Reported by checkpatch.

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:58:45 +01:00
Gao Xiang
f0c519fc26 staging: erofs: rename strange variable names in z_erofs_vle_frontend
Previously, 2 members called `initial' and `cachedzone_la' are used
for applying caching policy (whether the workgroup is at either end),
which are hard to understand, rename them to `backmost' and `headoffset'.

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:53:08 +01:00
Gao Xiang
2d9b5dcd99 staging: erofs: decompress asynchronously if PG_readahead page at first
For the case of nr_to_read == lookahead_size, it is better to
decompress asynchronously as well since no page will be needed immediately.

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:53:08 +01:00
Gao Xiang
23edf3abe7 staging: erofs: locked before registering for all new workgroups
Let's make sure that the one registering a workgroup will also
take the primary work lock at first for two reasons:
  1) There's no need to introduce such a race window (and consequently
     overhead) between registering and locking, other tasks could break
     in by chance, and the race seems unnecessary (no benefit at all);

  2) It's better to take the primary work when a workgroup
     is registered to apply the cache managed policy, for example,
     if some other tasks break in, it could turn into the in-place
     decompression rather than use as the cached decompression.

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:53:08 +01:00
Gao Xiang
48d4bf3b05 staging: erofs: separate into init_once / always
`z_erofs_vle_workgroup' is heavily generated in the decompression,
for example, it resets 32 bytes redundantly for 64-bit platforms
even through Z_EROFS_VLE_INLINE_PAGEVECS + Z_EROFS_CLUSTER_MAX_PAGES,
default 4, pages are stored in `z_erofs_vle_workgroup'.

As an another example, `struct mutex' takes 72 bytes for our kirin
64-bit platforms, it's unnecessary to be reseted at first and
be initialized each time.

Let's avoid filling all `z_erofs_vle_workgroup' with 0 at first
since most fields are reinitialized to meaningful values later,
and pagevec is no need to initialized at all.

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:53:08 +01:00
Gao Xiang
948bbdb181 staging: erofs: add a full barrier in erofs_workgroup_unfreeze
Just like other generic locks, insert a full barrier
in case of memory reorder.

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:53:08 +01:00
Gao Xiang
73f5c66df3 staging: erofs: fix `erofs_workgroup_{try_to_freeze, unfreeze}'
There are two minor issues in the current freeze interface:

   1) Freeze interfaces have not related with CONFIG_DEBUG_SPINLOCK,
      therefore fix the incorrect conditions;

   2) For SMP platforms, it should also disable preemption before
      doing atomic_cmpxchg in case that some high priority tasks
      preempt between atomic_cmpxchg and disable_preempt, then spin
      on the locked refcount later.

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:53:07 +01:00
Gao Xiang
df134b8d17 staging: erofs: atomic_cond_read_relaxed on ref-locked workgroup
It's better to use atomic_cond_read_relaxed, which is implemented
in hardware instructions to monitor a variable changes currently
for ARM64, instead of open-coded busy waiting.

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:53:07 +01:00
Gao Xiang
51232df5e4 staging: erofs: fix race when the managed cache is enabled
When the managed cache is enabled, the last reference count
of a workgroup must be used for its workstation.

Otherwise, it could lead to incorrect (un)freezes in
the reclaim path, and it would be harmful.

A typical race as follows:

Thread 1 (In the reclaim path)  Thread 2
workgroup_freeze(grp, 1)                                refcnt = 1
...
workgroup_unfreeze(grp, 1)                              refcnt = 1
                                workgroup_get(grp)      refcnt = 2 (x)
workgroup_put(grp)                                      refcnt = 1 (x)
                                ...unexpected behaviors

* grp is detached but still used, which violates cache-managed
  freeze constraint.

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:53:07 +01:00
Gao Xiang
eef1687898 staging: erofs: fix the definition of DBG_BUGON
It's better not to positively BUG_ON the kernel, however developers
need a way to locate issues as soon as possible.

DBG_BUGON is introduced and it could only crash when EROFS_FS_DEBUG
(EROFS developping feature) is on. It is helpful for developers
to find and solve bugs quickly by eng builds.

Previously, DBG_BUGON is defined as ((void)0) if EROFS_FS_DEBUG is off,
but some unused variable warnings as follows could occur:

drivers/staging/erofs/unzip_vle.c: In function `init_alway:':
drivers/staging/erofs/unzip_vle.c:61:33: warning: unused variable `work' [-Wunused-variable]
  struct z_erofs_vle_work *const work =
                                 ^~~~

Fix it to #define DBG_BUGON(x) ((void)(x)).

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:53:07 +01:00
Gao Xiang
ba9ce771b0 staging: erofs: fix `trace_erofs_readpage' position
`trace_erofs_readpage' was designed for .readpage() interface hook
in order to trace the detailed synchronized read rather than
the internal implementation `z_erofs_do_read_page' which both
.readpage() and .readpages() use.

It seems the tracepoint was placed to a wrong place by mistake and
it should be moved to the right place.

Fixes: 284db12cfd ("staging: erofs: add trace points for reading zipped data")
Reviewed-by: Chen Gong <gongchen4@huawei.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-11-23 10:53:07 +01:00