The new GEM object function drm_gem_cma_mmap() sets the VMA flags
and offset as in the old implementation and immediately maps in the
buffer's memory pages.
Changing CMA helpers to use the GEM object function allows for the
removal of the special implementations for mmap and gem_prime_mmap
callbacks. The regular functions drm_gem_mmap() and drm_gem_prime_mmap()
are now used.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201123115646.11004-3-tzimmermann@suse.de
Add support for the BOE NV110WTM-N61 panel. The EDID lists two modes
(one for 60 Hz refresh rate and one for 40 Hz), so we'll list both of
them here.
Note that the panel datasheet requires 80 ms between HPD asserting and
the backlight power being turned on. We'll use the new timing
constraints structure to do this cleanly. This assumes that the
backlight will be enabled _after_ the panel enable finishes. This is
how it works today and seems a sane assumption.
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201109170018.v4.4.I71b2118dfc00fd7b43b02d28e7b890081c2acfa2@changeid
On the panel I'm looking at, there's an 80 ms minimum time between HPD
being asserted by the panel and setting the backlight enable GPIO.
While we could just add an 80 ms "enable" delay, this is not ideal.
Link training is allowed to happen in parallel with this delay so the
fixed 80 ms delay over-delays.
We'll support this by logging the time at the end of prepare and then
delaying in enable if enough time hasn't passed.
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201109170018.v4.3.Ib9ce3c6482f464bf594161581521ced46bbd54ed@changeid
It is believed that all of the current users of the "unprepare" delay
don't actually need to wait the amount of time specified directly in
the unprepare phase. The purpose of the delay that's specified is to
allow the panel to fully power off so that we don't try to power it
back on before it's managed to full power down.
Let's use this observation to avoid the fixed delay that we currently
have. Instead of delaying, we'll note the current time when the
unprepare happens. If someone then tries to prepare the panel later
and not enough time has passed, we'll do the delay before starting the
prepare phase.
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201109170018.v4.2.I06a95d83e7fa1bd919c8edd63dacacb5436e495a@changeid
When I run:
scripts/kernel-doc -rst drivers/gpu/drm/panel/panel-simple.c
I see that several of the kernel-doc entries aren't showing up because
they don't specify the full path down the hierarchy. Let's fix that
and also move to inline kernel docs.
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201109170018.v4.1.Icaa86f0a4ca45a9a7184da4bc63386b29792d613@changeid
The NULL checking isn't done consistently in this function and it leads
to a static checker warning:
drivers/gpu/drm/kmb/kmb_drv.c:561 kmb_pm_suspend()
error: we previously assumed 'drm' could be null (see line 559)
Fortunately "drm" cannot be NULL at this point so the check can just be
removed.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Anitha Chrisanthus <anitha.chrisanthus@intel.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201117072137.GB1111239@mwanda
Fix following W=1 warnings:
- Fix set but not used variables that were used only for logging.
Fixed by introducing no_printk() to trick compiler to think variables
are used
- Fix kernel-doc warning by deleting an empty comment line
v3:
- Fix grammar in commit message (Thomas)
v2:
- Subject updated (Lee)
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Kristoffer Ericson <kristoffer.ericson@gmail.com>
Cc: Lee Jones <lee.jones@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201128224114.1033617-29-sam@ravnborg.org
Replacing DPRINTK() statements with pr_debug fixes set but not used
warnings. And moves to a more standard logging setup at the same time.
v2:
- Fix indent (Joe)
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Joe Perches <joe@perches.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Jiri Slaby <jirislaby@kernel.org>
Cc: Peilin Ye <yepeilin.cs@gmail.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: George Kennedy <george.kennedy@oracle.com>
Cc: Nathan Chancellor <natechancellor@gmail.com>
Cc: Peter Rosin <peda@axentia.se>
Link: https://patchwork.freedesktop.org/patch/msgid/20201128224114.1033617-3-sam@ravnborg.org
Call drm_mode_config_helper_suspend() and
drm_mode_config_helper_resume() on suspend and resume, respectively.
This makes sure that the display stack is properly disabled when the
hardware is put to sleep.
Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201128171606.132830-1-paul@crapouillou.net
Not technically a problem for ttm, but very likely a driver bug and
pretty big time confusing for reviewing code.
So warn about it, both at cleanup time (so we catch these for sure)
and at pin/unpin time (so we know who's the culprit).
Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Christian Koenig <christian.koenig@amd.com>
Cc: Huang Rui <ray.huang@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201028113120.3641237-1-daniel.vetter@ffwll.ch
It looks like SPARC64 requires FB_ATY_CT to build without errors,
so have FB_ATY select FB_ATY_CT if both SPARC64 and PCI are enabled
instead of using "default y if SPARC64 && PCI", which is not strong
enough to prevent build errors.
As it currently is, FB_ATY_CT can be disabled, resulting in build
errors:
ERROR: modpost: "aty_postdividers" [drivers/video/fbdev/aty/atyfb.ko] undefined!
ERROR: modpost: "aty_ld_pll_ct" [drivers/video/fbdev/aty/atyfb.ko] undefined!
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Fixes: f7018c2135 ("video: move fbdev to drivers/video/fbdev")
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: sparclinux@vger.kernel.org
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: dri-devel@lists.freedesktop.org
Cc: linux-fbdev@vger.kernel.org
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: David Airlie <airlied@linux.ie>
Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20201127031752.10371-1-rdunlap@infradead.org
This change also fix checkpatch.pl warning:
WARNING: Prefer using '"%s...", __func__' to using
'via_driver_irq_postinstall', this function's name, in a string
+ DRM_DEBUG("via_driver_irq_postinstall\n");
Signed-off-by: Bernard Zhao <bernard@vivo.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20201119072957.108941-1-bernard@vivo.com
Reorder the code to fix checking if blitting is available.
Signed-off-by: Christian König <christian.koenig@amd.com>
Fixes: 28a68f8282 ("drm/radeon/ttm: use multihop")
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/403847/
This patch adds support for using NN interpolation scaling by setting the
SCALING_FILTER plane property to 1. Otherwise, the default method is used.
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@oss.nxp.com>
Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20201105145018.27255-1-laurentiu.palcu@oss.nxp.com
This small patch fixes a warning that I got while running coccinelle:
CHECK drivers/gpu/drm/imx/dcss/dcss-plane.c
drivers/gpu/drm/imx/dcss/dcss-plane.c:107:21-23: WARNING !A || A && B is equivalent to !A || B
Fixes: 9021c317b7 ("drm/imx: Add initial support for DCSS on iMX8MQ")
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@oss.nxp.com>
Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20201105140127.25249-3-laurentiu.palcu@oss.nxp.com
DCSS supports 90/180/270 degree rotations for Vivante tiled and super-tiled
formats. Unfortunately, with the current code, they didn't work properly.
This simple patch makes the rotations work by fixing the way the scaler is set
up for 90/270 degree rotations. In this particular case, the source width and
height need to be swapped since DPR is sending the buffer to scaler already
rotated.
Also, make sure to allow full rotations for DRM_FORMAT_MOD_VIVANTE_SUPER_TILED.
Fixes: 9021c317b7 ("drm/imx: Add initial support for DCSS on iMX8MQ")
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@oss.nxp.com>
Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20201105140127.25249-2-laurentiu.palcu@oss.nxp.com
The masking of val with ~MCDE_CRX1_CLKSEL_MASK is currently being
ignored because there seems to be a missing bitwise-or of val in the
following statement. Fix this by replacing the assignment of val
with a bitwise-or.
Fixes: d795fd3220 ("drm/mcde: Support DPI output")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Addresses-Coverity: ("Unused valued")
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201124121528.395681-1-colin.king@canonical.com
We hardcode the maximum number of shared fences to 4, instead of
respecting num_fences. Use a minimum of 4, but more if num_fences
is higher.
This seems to have been an oversight when first implementing the
api.
Fixes: 04a5faa8cb ("reservation: update api and add some helpers")
Cc: <stable@vger.kernel.org> # v3.17+
Reported-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201124115707.406917-1-maarten.lankhorst@linux.intel.com
The Innolux N125HCE-GN1 display is used in the MNT Reform 2.0 laptop,
attached via eDP to a SN65DSI86 MIPI-DSI to eDP bridge. This patch
contains the DT binding for "innolux,n125hce-gn1".
Signed-off-by: Lukas F. Hartmann <lukas@mntre.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
[reorder so comments comes before the compatible]
Link: https://patchwork.freedesktop.org/patch/msgid/20201124172604.981746-2-lukas@mntre.com
Flushing the fbdev's shadow buffer requires vmap'ing the BO memory, which
in turn requires pinning the BO. While being pinned, the BO cannot be moved
into VRAM for scanout. Consequently, a concurrent modeset operation that
involves the fbdev framebuffer would likely fail.
Resolve this problem be acquiring the modeset lock of the planes that use
the fbdev framebuffer. On non-atomic drivers, also acquire the mode-config
lock. This serializes the flushing of the framebuffer with concurrent
modeset operations.
v2:
* only acquire struct drm_fb_helper.lock in damage blitter (Daniel,
Christian)
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Maxime Ripard <mripard@kernel.org>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201120102545.4047-11-tzimmermann@suse.de
Copy the vmap()'ed instance of struct dma_buf_map before modifying it,
in case the implementation of vunmap() depends on the exact address.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Maxime Ripard <mripard@kernel.org>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201120102545.4047-10-tzimmermann@suse.de
If the damage handling fails, restore the damage area. The next invocation
of the damage worker will then perform the update.
v3:
* Use drm_WARN_ONCE() with an error message to print warning
v2:
* print a single warning if dirty callback fails (Daniel, Sebastian)
* update comment
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Maxime Ripard <mripard@kernel.org>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201120102545.4047-9-tzimmermann@suse.de
Introduce a separate function for the blit code and its vmap setup. Done
in preparation of additional changes. No functional changes are made.
v3:
* Use drm_WARN_ONCE() with an error message to print warning
v2:
* print a single warning if damage blitter fails
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201120102545.4047-8-tzimmermann@suse.de
Flushing the shadow framebuffer and invoking the dirty callback are two
separate operations, so do them separately. The flush operation is paired
with calls to vmap and vunmap. They are not needed for the dirty callback,
which performs its own invocations if necessary.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201120102545.4047-7-tzimmermann@suse.de
Returning early in the damage worker if no update is required. Makes the
code more readable. No functional changes are being made.
v3:
* s/dirty/damage in commit message (Sam)
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201120102545.4047-6-tzimmermann@suse.de
The dirty worker handles all damage updates, instead of just calling
the framebuffer's dirty callback. Rename it to damage worker. Also
rename related variables accordingly. No functional changes are made.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201120102545.4047-5-tzimmermann@suse.de
DRM client's vmap/vunmap functions don't allow for multiple vmap
operations. Calling drm_client_buffer_vmap() twice returns the same
mapping, then calling drm_client_buffer_vunmap() twice already unmaps
on the first call. This leads to unbalanced vmap refcounts. Fix this
by calling drm_gem_vmap() unconditionally in drm_client_buffer_vmap().
All drivers that support DRM clients have to implement correct ref-
counting for their vmap operations, or not vunmap at all. This is the
case for drivers that use CMA, SHMEM and VRAM helpers, and QXL. Other
drivers are not affected.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201120102545.4047-4-tzimmermann@suse.de
The fbdev helper's generic probe function establishes a mapping for
framebuffers without shadow buffer. The clean-up function did not unmap
the buffer object. Add the unmap operation.
As fbdev devices are usally released during system shutdown, this has
not been a problem in practice.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201120102545.4047-3-tzimmermann@suse.de
If fbdev uses a shadow framebuffer, call the damage handler. Otherwise
the update might not make it to the screen.
v2:
* mark virtual screen as dirty (Ville)
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Fixes: 222ec45f4c ("drm/fb_helper: Support framebuffers in I/O memory")
Acked-by: Maxime Ripard <mripard@kernel.org>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: dri-devel@lists.freedesktop.org
Cc: virtualization@lists.linux-foundation.org
Link: https://patchwork.freedesktop.org/patch/msgid/20201120102545.4047-2-tzimmermann@suse.de
SHMEM-buffer backing storage is allocated from system memory; which is
typically cachable. The default mode for SHMEM objects is writecombine
though.
Unify SHMEM semantics by defaulting to cached mappings. The exception
is pages imported via dma-buf. DMA memory is usually not cached.
DRM drivers that require write-combined mappings set the map_wc flag
in struct drm_gem_shmem_object to true. This currently affects lima,
panfrost and v3d.
The drivers mgag200, udl, virtio and vkms continue to use default
shmem mappings.
The drivers cirrus and gm12u320 change caching flags. Both used
writecombine and now switch over to shmem defaults. Both drivers use
SHMEM objects as shadow buffers for internal video memory, so cached
mappings will not affect them negatively.
v3:
* set value of shmem pointer before dereferencing it in
__drm_gem_shmem_create() (Dan, kernel test robot)
v2:
* recreate patch on top of latest SHMEM helpers
* update lima, panfrost, v3d to select writecombine (Daniel, Rob)
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201117133156.26822-2-tzimmermann@suse.de
"val" isn't initialized on the default: errorpath.
Just return from the function if this happens.
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201119140707.1008407-1-linus.walleij@linaro.org
This implements support for DPI output using the port node
in the device tree to connect a DPI LCD display to the
MCDE. The block also supports TV-out but we leave that
for another day when we have a hardware using it.
We implement parsing and handling of the "port" node,
and follow that to the DPI endpoint.
The clock divider used by the MCDE to divide down the
"lcdclk" (this has been designed for TV-like frequencies)
is represented by an ordinary clock provider internally
in the MCDE. This idea was inspired by the PL111 solution
by Eric Anholt: the divider also works very similar to
the Pl111 clock divider.
We take care to clear up some errors regarding the number
of available formatters and their type. We have 6 DSI
formatters and 2 DPI formatters.
Tested on the Samsung GT-I9070 Janice mobile phone.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Stephan Gerhold <stephan@gerhold.net>
Cc: phone-devel@vger.kernel.org
Cc: upstreaming@lists.sr.ht
Link: https://patchwork.freedesktop.org/patch/msgid/20201112142925.2571179-2-linus.walleij@linaro.org
I was confused when the graphics came out with blue
penguins on the DPI panel.
It turns out that the so-called "packed RGB666" mode
on the DSI formatter is incorrect: this mode is the
actual RGB888 mode, and the mode called RGB888 is
BGR888.
The claims that the MCDE had inverse RGB/BGR buffer
formats was wrong, so correct this and the buggy
register and everything is much more consistent, and
graphics look good on all targets, both DPI and
DSI.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Cc: phone-devel@vger.kernel.org
Cc: Stephan Gerhold <stephan@gerhold.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20201117175413.869871-1-linus.walleij@linaro.org
While the system heap can return non-contiguous pages,
try to allocate larger order pages if possible.
This will allow slight performance gains and make implementing
page pooling easier.
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: Liam Mark <lmark@codeaurora.org>
Cc: Laura Abbott <labbott@kernel.org>
Cc: Brian Starkey <Brian.Starkey@arm.com>
Cc: Hridya Valsaraju <hridya@google.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Sandeep Patil <sspatil@google.com>
Cc: Daniel Mentz <danielmentz@google.com>
Cc: Chris Goldsworthy <cgoldswo@codeaurora.org>
Cc: Ørjan Eide <orjan.eide@arm.com>
Cc: Robin Murphy <robin.murphy@arm.com>
Cc: Ezequiel Garcia <ezequiel@collabora.com>
Cc: Simon Ser <contact@emersion.fr>
Cc: James Jones <jajones@nvidia.com>
Cc: linux-media@vger.kernel.org
Cc: dri-devel@lists.freedesktop.org
Reviewed-by: Brian Starkey <brian.starkey@arm.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201121235002.69945-6-john.stultz@linaro.org
This patch is basically a port of Ørjan Eide's similar patch for ION
https://lore.kernel.org/lkml/20200414134629.54567-1-orjan.eide@arm.com/
Only sync the sg-list of dma-buf heap attachment when the attachment
is actually mapped on the device.
dma-bufs may be synced at any time. It can be reached from user space
via DMA_BUF_IOCTL_SYNC, so there are no guarantees from callers on when
syncs may be attempted, and dma_buf_end_cpu_access() and
dma_buf_begin_cpu_access() may not be paired.
Since the sg_list's dma_address isn't set up until the buffer is used
on the device, and dma_map_sg() is called on it, the dma_address will be
NULL if sync is attempted on the dma-buf before it's mapped on a device.
Before v5.0 (commit 55897af630 ("dma-direct: merge swiotlb_dma_ops
into the dma_direct code")) this was a problem as the dma-api (at least
the swiotlb_dma_ops on arm64) would use the potentially invalid
dma_address. How that failed depended on how the device handled physical
address 0. If 0 was a valid address to physical ram, that page would get
flushed a lot, while the actual pages in the buffer would not get synced
correctly. While if 0 is an invalid physical address it may cause a
fault and trigger a crash.
In v5.0 this was incidentally fixed by commit 55897af630 ("dma-direct:
merge swiotlb_dma_ops into the dma_direct code"), as this moved the
dma-api to use the page pointer in the sg_list, and (for Ion buffers at
least) this will always be valid if the sg_list exists at all.
But, this issue is re-introduced in v5.3 with
commit 449fa54d68 ("dma-direct: correct the physical addr in
dma_direct_sync_sg_for_cpu/device") moves the dma-api back to the old
behaviour and picks the dma_address that may be invalid.
dma-buf core doesn't ensure that the buffer is mapped on the device, and
thus have a valid sg_list, before calling the exporter's
begin_cpu_access.
Logic and commit message originally by: Ørjan Eide <orjan.eide@arm.com>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: Liam Mark <lmark@codeaurora.org>
Cc: Laura Abbott <labbott@kernel.org>
Cc: Brian Starkey <Brian.Starkey@arm.com>
Cc: Hridya Valsaraju <hridya@google.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Sandeep Patil <sspatil@google.com>
Cc: Daniel Mentz <danielmentz@google.com>
Cc: Chris Goldsworthy <cgoldswo@codeaurora.org>
Cc: Ørjan Eide <orjan.eide@arm.com>
Cc: Robin Murphy <robin.murphy@arm.com>
Cc: Ezequiel Garcia <ezequiel@collabora.com>
Cc: Simon Ser <contact@emersion.fr>
Cc: James Jones <jajones@nvidia.com>
Cc: linux-media@vger.kernel.org
Cc: dri-devel@lists.freedesktop.org
Reviewed-by: Brian Starkey <brian.starkey@arm.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20201121235002.69945-5-john.stultz@linaro.org
