[Why]
HDMI 2.0 adds aspect ratio attribute to distinguish different
4k modes. According to Appendix E of HDMI 2.0 spec, source should
use VSIF to indicate video mode only when the mode is one defined
in HDMI 1.4b 4K modes. Otherwise, use AVI infoframes to convey VIC.
Current code doesn't take aspect ratio into consideration while
constructing avi infoframe. Should modify that.
[How]
Inherit Ville Syrjälä's work
"drm/edid: Prep for HDMI VIC aspect ratio" at
https://patchwork.kernel.org/patch/11174639/
Add picture_aspect_ratio attributes to edid_4k_modes[] and
construct VIC and HDMI_VIC by taking aspect ratio into
consideration.
v2: Correct missing initializer error at adding aspect ratio of
SMPTE mode.
Signed-off-by: Wayne Lin <Wayne.Lin@amd.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191118101832.15487-1-Wayne.Lin@amd.com
A coccicheck run provided information like the following.
drivers/gpu/drm/qxl/qxl_kms.c:295:1-7: ERROR: missing iounmap;
ioremap on line 178 and execution via conditional on line 185
Generated by: scripts/coccinelle/free/iounmap.cocci
A jump target was specified in an if branch. The corresponding function
call did not release the desired system resource then.
Thus use the label “rom_unmap” instead to fix the exception handling
for this function implementation.
Fixes: 5043348a49 ("drm: qxl: Fix error handling at qxl_device_init")
Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Link: http://patchwork.freedesktop.org/patch/msgid/5e5ef9c4-4d85-3c93-cf28-42cfcb5b0649@web.de
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
If rockchip would switch over to the generic fbdev setup we could
grabage collect even more of all this code (all of the remaining fb
handling code really).
v2: Actually use _with_dirty like the patch subject promised (Andrzej)
Cc: Andrzej Pietrasiewicz <andrzej.p@collabora.com>
Reviewed-by: Andrzej Pietrasiewicz <andrzej.p@collabora.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Sandy Huang <hjc@rock-chips.com>
Cc: "Heiko Stübner" <heiko@sntech.de>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-rockchip@lists.infradead.org
Link: https://patchwork.freedesktop.org/patch/msgid/20191127180035.416209-1-daniel.vetter@ffwll.ch
This was forgotten in f96bdf564f ("drm/rect: Handle rounding errors
in drm_rect_clip_scaled, v3.")
Spotted while reviewing patches from Ville touching this area.
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Fixes: f96bdf564f ("drm/rect: Handle rounding errors in drm_rect_clip_scaled, v3.")
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Benjamin Gaignard <benjamin.gaignard@st.com>
Cc: Ville Syrjala <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191126145213.380079-1-daniel.vetter@ffwll.ch
Add selftests for drm_rect. A few basic ones for clipped and unclipped
cases, and a few special ones for specific bugs we had in the code.
I'm too lazy to think of more corner cases to check at this time.
Maybe later.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191122175623.13565-5-ville.syrjala@linux.intel.com
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Benjamin Gaignard <benjamin.gaignard@st.com>
Now that we've constrained the clipped source rectangle such
that it can't have negative dimensions doing the same for the
dst rectangle seems appropriate. Should at least result in
the clipped src and dst rectangles being a bit more consistent
with each other.
Cc: Benjamin Gaignard <benjamin.gaignard@st.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191122175623.13565-4-ville.syrjala@linux.intel.com
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Benjamin Gaignard <benjamin.gaignard@st.com>
Limit the scaled clip to only clip at most dst_w/h pixels.
This avoids the problem with clip_scaled() not being able
to return negative values. Since new_src_w/h is now properly
bounded we can remove the clamp()s.
Cc: Benjamin Gaignard <benjamin.gaignard@st.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Testcase: igt/kms_selftest/drm_rect_clip_scaled_signed_vs_unsigned
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191122175623.13565-3-ville.syrjala@linux.intel.com
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Benjamin Gaignard <benjamin.gaignard@st.com>
Check for zero width/height destination rectangle in
drm_rect_clip_scaled() to avoid a division by zero.
Cc: stable@vger.kernel.org
Fixes: f96bdf564f ("drm/rect: Handle rounding errors in drm_rect_clip_scaled, v3.")
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Benjamin Gaignard <benjamin.gaignard@st.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Testcase: igt/kms_selftest/drm_rect_clip_scaled_div_by_zero
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191122175623.13565-2-ville.syrjala@linux.intel.com
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Benjamin Gaignard <benjamin.gaignard@st.com>
0 means 1 as the default, but it's mighty confusing if the block size
for the first plane is spelled out explicitly, but not for the 2nd
plane.
No cc: stable because this is just confusion, but 0 functional issue.
Acked-by: Liviu Dudau <liviu.dudau@arm.com>
Fixes: 7ba0fee247 ("drm/fourcc: Add AFBC yuv fourccs for Mali")
Cc: Brian Starkey <brian.starkey@arm.com>
Cc: Ayan Kumar Halder <ayan.halder@arm.com>
Cc: Liviu Dudau <liviu.dudau@arm.com>
Cc: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <mripard@kernel.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191126091414.226070-2-daniel.vetter@ffwll.ch
0 means 1 as the default, but it's mighty confusing if the block size
for the first plane is spelled out explicitly, but not for the 2nd
plane.
No cc: stable because this is just confusion, but 0 functional issue.
Acked-by: Liviu Dudau <liviu.dudau@arm.com>
Fixes: 05f8bc82fc ("drm/fourcc: Add new P010, P016 video format")
Cc: Daniel Stone <daniel@fooishbar.org>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Randy Li <ayaka@soulik.info>
Cc: Clint Taylor <clinton.a.taylor@intel.com>
Cc: Ayan Kumar Halder <ayan.halder@arm.com>
Cc: Neil Armstrong <narmstrong@baylibre.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <mripard@kernel.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191126091414.226070-1-daniel.vetter@ffwll.ch
Problem:
Due to a race between drm_sched_cleanup_jobs in sched thread and
drm_sched_job_timedout in timeout work there is a possiblity that
bad job was already freed while still being accessed from the
timeout thread.
Fix:
Instead of just peeking at the bad job in the mirror list
remove it from the list under lock and then put it back later when
we are garanteed no race with main sched thread is possible which
is after the thread is parked.
v2: Lock around processing ring_mirror_list in drm_sched_cleanup_jobs.
v3: Rebase on top of drm-misc-next. v2 is not needed anymore as
drm_sched_get_cleanup_job already has a lock there.
v4: Fix comments to relfect latest code in drm-misc.
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Emily Deng <Emily.Deng@amd.com>
Tested-by: Emily Deng <Emily.Deng@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/342356
The udl driver's struct udl_framebuffer stores a DRM framebuffer
with an associated GEM object. This functionality is also provided by
generic code. Switch udl over.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20191114141025.32198-6-tzimmermann@suse.de
The framebuffer's 'active_16' flag signals which framebuffer to flush
to device memory. Moving the 'active_16' state from struct udl_framebuffer
into struct udl_device prepares for using the generic GEM framebuffer.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20191114141025.32198-4-tzimmermann@suse.de
Udl's custom implementation for struct drm_gem_object_funcs.free_object
unmaps perma-mapped memory buffer before freeing the buffer object.
After switching to generic fbdev emulation and fixing the damage
handler, no perma-mapped buffers have to be released. Switch to SHMEM's
implementation of free_object.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20191114141025.32198-3-tzimmermann@suse.de
Udl keeps a BO mapped for its entire lifetime if it has been used in a
damage update at least once. The BO's free callback release the mapping
before it frees the BO.
Change this behaviour to unmap immediately after the damage update, so
SHMEM's implementation of free can be used.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20191114141025.32198-2-tzimmermann@suse.de
There's no in-kernel users for the k(un)map stuff. And the mmap one is
actively harmful - return 0 and then _not_ actually mmaping can't end
well.
Acked-by: Sumit Semwal <sumit.semwal@linaro.org>
Reviewed-by: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
Reviewed-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: Juergen Gross <jgross@suse.com>
Cc: Stefano Stabellini <sstabellini@kernel.org>
Cc: xen-devel@lists.xenproject.org
Link: https://patchwork.freedesktop.org/patch/msgid/20191118103536.17675-14-daniel.vetter@ffwll.ch
No need for stubs, dma-buf.c takes care of that.
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Acked-by: Sumit Semwal <sumit.semwal@linaro.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: VMware Graphics <linux-graphics-maintainer@vmware.com>
Cc: Thomas Hellstrom <thellstrom@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191118103536.17675-11-daniel.vetter@ffwll.ch
It's unused.
10 years ago, back when 32bit was still fairly common and trying to
not exhaust vmalloc space sounded like a worthwhile goal, adding these
to dma_buf made sense.
Reality is that they simply never caught on, and nowadays everyone who
needs plenty of buffers will run in 64bit mode anyway.
Also update the docs in this area to adjust them to reality.
The actual hooks in dma_buf_ops will be removed once all the
implementations are gone.
Acked-by: Sumit Semwal <sumit.semwal@linaro.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: linux-media@vger.kernel.org
Cc: linaro-mm-sig@lists.linaro.org
Link: https://patchwork.freedesktop.org/patch/msgid/20191118103536.17675-10-daniel.vetter@ffwll.ch
No in-tree users left.
Note that this is one of the few (if only) implementations of dma-buf
that provided a kmap, but not a vmap implemenation. Given that the
only real user (in-tree at least) of kmap was tegra, and it's
impossible to buy a chip with tegra host1x and ompadrm on the same
SoC, there's no problem here.
Acked-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Sumit Semwal <sumit.semwal@linaro.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191118103536.17675-8-daniel.vetter@ffwll.ch
No in-tree users left.
Aside, I think mock_dmabuf would be a nice addition to drm
mock/selftest helpers (we have some already), with an
EXPORT_SYMBOL_FOR_TESTS_ONLY.
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Acked-by: Sumit Semwal <sumit.semwal@linaro.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: "Christian König" <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191118103536.17675-7-daniel.vetter@ffwll.ch
There's no callers in-tree anymore.
For merging probably best to stuff this into drm-misc, since that's
where the dma-buf heaps will land too. And the resulting conflict
hopefully ensures that dma-buf heaps wont have a new ->kmap/unmap
implemenation.
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Laura Abbott <labbott@redhat.com>
Acked-by: Sumit Semwal <sumit.semwal@linaro.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: devel@driverdev.osuosl.org
Cc: linaro-mm-sig@lists.linaro.org
Link: https://patchwork.freedesktop.org/patch/msgid/20191118103536.17675-5-daniel.vetter@ffwll.ch
It's the only user left in the entire kernel for dma_buf_kmap/_kunmap.
Delete it, before we start garbage-collecting the various
implementations.
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Acked-by: Sumit Semwal <sumit.semwal@linaro.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Dave Airlie <airlied@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191118103536.17675-4-daniel.vetter@ffwll.ch
It doesn't have any callers anymore.
Aside: The ->mmap/munmap hooks have a bit a confusing name, they don't
do userspace mmaps, but a kernel vmap. I think most places use vmap
for this, except ttm, which uses kmap for vmap for added confusion.
mmap seems entirely for userspace mappings set up through mmap(2)
syscall.
Reviewed-by: Thierry Reding <treding@nvidia.com>
Tested-by: Thierry Reding <treding@nvidia.com>
Acked-by: Sumit Semwal <sumit.semwal@linaro.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Thierry Reding <thierry.reding@gmail.com>
Cc: Jonathan Hunter <jonathanh@nvidia.com>
Cc: linux-tegra@vger.kernel.org
Link: https://patchwork.freedesktop.org/patch/msgid/20191118103536.17675-3-daniel.vetter@ffwll.ch
A few reasons to drop kmap:
- For native objects all we do is look at obj->vaddr anyway, so might
as well not call functions for every page.
- Reloc-processing on dma-buf is ... questionable.
- Plus most dma-buf that bother kernel cpu mmaps give you at least
vmap, much less kmaps. And all the ones relevant for arm-soc are
again doing a obj->vaddr game anyway, there's no real kmap going on
on arm it seems.
Plus this seems to be the only real in-tree user of dma_buf_kmap, and
I'd like to get rid of that.
Reviewed-by: Thierry Reding <treding@nvidia.com>
Tested-by: Thierry Reding <treding@nvidia.com>
Acked-by: Sumit Semwal <sumit.semwal@linaro.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Thierry Reding <thierry.reding@gmail.com>
Cc: linux-tegra@vger.kernel.org
Link: https://patchwork.freedesktop.org/patch/msgid/20191118103536.17675-2-daniel.vetter@ffwll.ch
This reverts commit f25c7a006c ("drm/fbdev: Fallback to non tiled mode
if all tiles not present"). The commit causes flip done timeouts in CI.
Below are the sample errors thrown in logs:
[IGT] core_getversion: executing
[IGT] core_getversion: exiting, ret=0
Setting dangerous option reset - tainting kernel
drm:drm_atomic_helper_wait_for_dependencies] ERROR [CRTC:152:pipe B] flip_done timed out
drm:drm_atomic_helper_wait_for_dependencies] ERROR [CONNECTOR:299:DP-2] flip_done timed out
drm:drm_atomic_helper_wait_for_dependencies] ERROR [PLANE:92:plane 1B] flip_done timed out
[drm:drm_atomic_helper_wait_for_flip_done] ERROR [CRTC:152:pipe B] flip_done timed out
[drm:drm_atomic_helper_wait_for_dependencies] ERROR [CRTC:152:pipe B] flip_done timed out
[drm:drm_atomic_helper_wait_for_dependencies] ERROR [CONNECTOR:299:DP-2] flip_done timed out
[drm:drm_atomic_helper_wait_for_dependencies] ERROR [PLANE:92:plane 1B] flip_done timed out
[drm:drm_atomic_helper_wait_for_flip_done] ERROR [CRTC:152:pipe B] flip_done timed out
Console: switching to colour frame buffer device 480x135
[drm:drm_atomic_helper_wait_for_dependencies] ERROR [CRTC:152:pipe B] flip_done timed out
[drm:drm_atomic_helper_wait_for_dependencies] ERROR [CONNECTOR:299:DP-2] flip_done timed out
Reverting the change for now to unblock CI execution.
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Manasi Navare <manasi.d.navare@intel.com>
Signed-off-by: Uma Shankar <uma.shankar@intel.com>
Fixes: f25c7a006c ("drm/fbdev: Fallback to non tiled mode if all tiles not present")
Closes: https://gitlab.freedesktop.org/drm/intel/issues/6
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191123091840.32382-1-uma.shankar@intel.com
The gma500 driver stores the console framebuffer in struct psb_fbdev.
Moving it into struct drm_fb_helper will allow for removal of struct
psb_fbdev.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191122100545.16812-6-tzimmermann@suse.de
Several framebuffer functions take a pointer to an object of type
struct gtt_range when they actually need the GEM base object. Passing
the GEM object removes some type casting and clutter.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191122100545.16812-5-tzimmermann@suse.de
After removing all unnecessary fields, struct psb_framebuffer is just a
wrapper around struct drm_framebuffer. So we can replace the former with
the latter.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191122100545.16812-4-tzimmermann@suse.de
The i index i always 0..3 in these statements so there
is no need to tag "& 3" to clamp it to 3 here. Make
the operator precedence explicit even if it's correct
as it is, the paranthesis creates less cognitive stress
for humans.
Reviewed-by: Stephan Gerhold <stephan@gerhold.net>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20191122072508.25677-1-linus.walleij@linaro.org
Aside: There's a few other fb_create implementations which
simply check for valid buffer format (or an approximation thereof),
and then call drm_gem_fb_create. For atomic drivers at least we could
walk all planes and make sure the format/modifier combo is valid,
and remove even more code.
For non-atomic drivers that's not possible, since the format list for
the primary buffer might be garbage (and most likely it is).
Also delete mtk_drm_fb.[hc] since it would now only contain one
function.
Acked-by: CK Hu <ck.hu@mediatek.com>
Cc: CK Hu <ck.hu@mediatek.com>
Cc: Philipp Zabel <p.zabel@pengutronix.de>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-mediatek@lists.infradead.org
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20191115092120.4445-4-daniel.vetter@ffwll.ch
For locking semantics it really doesn't matter when we grab the
ticket. But for lockdep validation it does: the acquire ctx is a fake
lockdep. Since other drivers might want to do a full multi-lock dance
in their fault-handler, not just lock a single dma_resv. Therefore we
must init the acquire_ctx only after we've done all the copy_*_user or
anything else that might trigger a pagefault. For msm this means we
need to move it past submit_lookup_objects.
Aside: Why is msm still using struct_mutex, it seems to be using
dma_resv_lock for general buffer state protection?
v2:
- Add comment to explain why the ww ticket setup is separate (Rob)
- Fix up error handling, we need to make sure we don't call
ww_acquire_fini without _init.
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-and-tested-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Rob Clark <robdclark@gmail.com>
Cc: Sean Paul <sean@poorly.run>
Cc: linux-arm-msm@vger.kernel.org
Cc: freedreno@lists.freedesktop.org
Link: https://patchwork.freedesktop.org/patch/msgid/20191120105607.3023-1-daniel.vetter@ffwll.ch
Semnatically it really doesn't matter where we grab the ticket. But
since the ticket is a fake lockdep lock, it matters for lockdep
validation purposes.
This means stuff like grabbing a ticket and then doing
copy_from/to_user isn't allowed anymore. This is a changed compared to
the current ttm fault handler, which doesn't bother with having a full
reservation. Since I'm looking into fixing the TODO entry in
ttm_mem_evict_wait_busy() I think that'll have to change sooner or
later anyway, better get started. A bit more context on why I'm
looking into this: For backwards compat with existing i915 gem code I
think we'll have to do full slowpath locking in the i915 equivalent of
the eviction code. And with dynamic dma-buf that will leak across
drivers, so another thing we need to standardize and make sure it's
done the same way everyway.
Unfortunately this means another full audit of all drivers:
- gem helpers: acquire_init is done right before taking locks, so no
problem. Same for acquire_fini and unlocking, which means nothing
that's not already covered by the dma_resv_lock rules will be caught
with this extension here to the acquire_ctx.
- etnaviv: An absolute massive amount of code is run between the
acquire_init and the first lock acquisition in submit_lock_objects.
But nothing that would touch user memory and could cause a fault.
Furthermore nothing that uses the ticket, so even if I missed
something, it would be easy to fix by pushing the acquire_init right
before the first use. Similar on the unlock/acquire_fini side.
- i915: Right now (and this will likely change a lot rsn) the acquire
ctx and actual locks are right next to each another. No problem.
- msm has a problem: submit_create calls acquire_init, but then
submit_lookup_objects() has a bunch of copy_from_user to do the
object lookups. That's the only thing before submit_lock_objects
call dma_resv_lock(). Despite all the copypasta to etnaviv, etnaviv
does not have this issue since it copies all the userspace structs
earlier. submit_cleanup does not have any such issues.
With the prep patch to pull out the acquire_ctx and reorder it msm
is going to be safe too.
- nouveau: acquire_init is right next to ttm_bo_reserve, so all good.
Similar on the acquire_fini/ttm_bo_unreserve side.
- ttm execbuf utils: acquire context and locking are even in the same
functions here (one function to reserve everything, the other to
unreserve), so all good.
- vc4: Another case where acquire context and locking are handled in
the same functions (one function to lock everything, the other to
unlock).
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Christian König <christian.koenig@amd.com>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: linux-media@vger.kernel.org
Cc: linaro-mm-sig@lists.linaro.org
Cc: Huang Rui <ray.huang@amd.com>
Cc: Eric Anholt <eric@anholt.net>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Rob Herring <robh@kernel.org>
Cc: Lucas Stach <l.stach@pengutronix.de>
Cc: Russell King <linux+etnaviv@armlinux.org.uk>
Cc: Christian Gmeiner <christian.gmeiner@gmail.com>
Cc: Rob Clark <robdclark@gmail.com>
Cc: Sean Paul <sean@poorly.run>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191119210844.16947-3-daniel.vetter@ffwll.ch
It's kinda really hard to get this wrong on a driver with both display
and dma_resv locking. But who ever knows, so better to make sure that
really all drivers nest these the same way.
For actual lock semantics the acquire context nesting doesn't matter.
But to teach lockdep what's going on with ww_mutex the acquire ctx is
a fake lockdep lock, hence from a lockdep pov it does matter. That's
why I figured better to include it.
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Christian König <christian.koenig@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191119210844.16947-2-daniel.vetter@ffwll.ch
Cc: Dave Airlie <airlied@gmail.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <mripard@kernel.org>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Sean Paul <sean@poorly.run>
Link: https://patchwork.freedesktop.org/patch/msgid/20191115205302.246625-1-sean@poorly.run
