We got a few conflicts in drm_atomic.c after merging the DRM writeback support,
now we need a backmerge to unlock develop development on drm-misc-next.
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.com>
Commit b9f19259b8 ("drm/vc4: Add the DRM_IOCTL_VC4_GEM_MADVISE ioctl")
introduced a mechanism to mark some BOs as purgeable to allow the driver
to drop them under memory pressure. In order to implement this feature
we had to add a mechanism to mark BOs as currently used by a piece of
hardware which materialized through the ->usecnt counter.
Plane code is supposed to increment usecnt when it attaches a BO to a
plane and decrement it when it's done with this BO, which was done in
the ->prepare_fb() and ->cleanup_fb() hooks. The problem is, async page
flip logic does not go through the regular atomic update path, and
->prepare_fb() and ->cleanup_fb() are not called in this case.
Fix that by manually calling vc4_bo_{inc,dec}_usecnt() in the
async-page-flip path.
Note that all this should go away as soon as we get generic async page
flip support in the core, in the meantime, this fix should do the
trick.
Fixes: b9f19259b8 ("drm/vc4: Add the DRM_IOCTL_VC4_GEM_MADVISE ioctl")
Reported-by: Peter Robinson <pbrobinson@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20180430133232.32457-1-boris.brezillon@bootlin.com
Link: https://patchwork.freedesktop.org/patch/msgid/20180430133232.32457-1-boris.brezillon@bootlin.com
The hardware has a single block for applying a CTM prior to gamma lut.
It can be fed with pixels from one of our CRTC at a time and uses a
matrix with S0.9 scalars. Use private atomic state to reject attempts
from userland to apply CTM for more than one CRTC at a time and reject
matrices with scalars that we can't approximate without integer bits.
Signed-off-by: Stefan Schake <stschake@gmail.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
Link: https://patchwork.freedesktop.org/patch/218067/
If we restrict this helper to only kms drivers (which is the case) we
can look up the correct mode easily ourselves. But it's a bit tricky:
- All legacy drivers look at crtc->hwmode. But that is updated already
at the beginning of the modeset helper, which means when we disable
a pipe. Hence the final timestamps might be a bit off. But since
this is an existing bug I'm not going to change it, but just try to
be bug-for-bug compatible with the current code. This only applies
to radeon&amdgpu.
- i915 tries to get it perfect by updating crtc->hwmode when the pipe
is off (i.e. vblank->enabled = false).
- All other atomic drivers look at crtc->state->adjusted_mode. Those
that look at state->requested_mode simply don't adjust their mode,
so it's the same. That has two problems: Accessing crtc->state from
interrupt handling code is unsafe, and it's updated before we shut
down the pipe. For nonblocking modesets it's even worse.
For atomic drivers try to implement what i915 does. To do that we add
a new hwmode field to the vblank structure, and update it from
drm_calc_timestamping_constants(). For atomic drivers that's called
from the right spot by the helper library already, so all fine. But
for safety let's enforce that.
For legacy driver this function is only called at the end (oh the
fun), which is broken, so again let's not bother and just stay
bug-for-bug compatible.
The benefit is that we can use drm_calc_vbltimestamp_from_scanoutpos
directly to implement ->get_vblank_timestamp in every driver, deleting
a lot of code.
v2: Completely new approach, trying to mimick the i915 solution.
v3: Fixup kerneldoc.
v4: Drop the WARN_ON to check that the vblank is off, atomic helpers
currently unconditionally call this. Recomputing the same stuff should
be harmless.
v5: Fix typos and move misplaced hunks to the right patches (Neil).
v6: Undo hunk movement (kbuild).
Cc: Mario Kleiner <mario.kleiner@tuebingen.mpg.de>
Cc: Eric Anholt <eric@anholt.net>
Cc: Rob Clark <robdclark@gmail.com>
Cc: linux-arm-msm@vger.kernel.org
Cc: freedreno@lists.freedesktop.org
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Reviewed-by: Neil Armstrong <narmstrong@baylibre.com>
Acked-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170509140329.24114-4-daniel.vetter@ffwll.ch
I had written most of my comments as if I was describing the
individual code files the way I used to for doxygen, while for RST we
want to describe things in a more chapter/section way where there's no
obvious relation to .c files.
Additionally, several of the files had stub descriptions that I've
taken this opportunity to extend.
Signed-off-by: Eric Anholt <eric@anholt.net>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227201144.10970-4-eric@anholt.net
Backmerge the main pull request to sync up with all the newly landed
drivers. Otherwise we'll have chaos even before 4.12 started in
earnest.
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
The underscores variant frees the pointers inside, while the
no-underscores variant calls underscores and then frees the struct.
Signed-off-by: Eric Anholt <eric@anholt.net>
Fixes: d8dbf44f13 ("drm/vc4: Make the CRTCs cooperate on allocating display lists.")
Cc: stable@vger.kernel.org
PV_CONTROL_CLK_SELECT_VEC is actually 2 and not 0. Fix the definition and
rework the vc4_set_crtc_possible_masks() to cover the full range of the
PV_CONTROL_CLK_SELECT field.
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
There was a small window where a userspace program could submit
a pageflip after receiving a pageflip completion event yet still
receive EBUSY.
Signed-off-by: Derek Foreman <derekf@osg.samsung.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Now that we have infoframes to report the pixel repeat flag, we can
start using it. Fixes locking the 720x480i and 720x576i modes on my
Dell 2408WFP. Like the 1920x1080i case, they don't fit properly on
the screen, though.
Signed-off-by: Eric Anholt <eric@anholt.net>
We really do need to be using the halved V fields. I had been
confused by the code I was using as a reference because it stored
halved vsync fields but not halved vdisplay, so it looked like I only
needed to divide vdisplay by 2.
This reverts part of Mario's timestamping fixes that prevented
CRTC_HALVE_V from applying, and instead adjusts the timestamping code
to not use the crtc field in that case.
Fixes locking of 1920x1080x60i on my Dell 2408WFP. There are black
bars on the top and bottom, but I suspect that might be an
under/overscan flags problem as opposed to video timings.
Signed-off-by: Eric Anholt <eric@anholt.net>
Add missing drm_crtc_vblank_on/off() calls so vblank irq
handling/updating/timestamping never runs with a crtc shut down
or during its shutdown/startup, as that causes large jumps in
vblank count and trouble for compositors.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
On top of the interlaced video mode fix and with some additional
adjustments, this now works well. It has almost the same accuracy
as on regular progressive scan modes.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
We can't handle doublescan modes at the moment, so if
userspace tries to set one, reject the mode set.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
We must not apply CRTC_INTERLACE_HALVE_V to interlaced modes during
mode enumeration, as drm_helper_probe_single_connector_modes
does, so wrap it and reset the effect of CRTC_INTERLACE_HALVE_V
on affected interlaced modes.
Also mode_fixup interlaced modes passed in from user space.
This fixes the vblank timestamping constants and entries in
the mode->crtc_xxx fields needed for precise vblank timestamping.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
This pull request brings in new vc4 plane formats for Android, precise
vblank timestamping, and a couple of small cleanups.
* tag 'drm-vc4-next-2016-07-12' of https://github.com/anholt/linux:
drm/vc4: remove redundant ret status check
drm/vc4: Implement precise vblank timestamping.
drm/vc4: Bind the HVS before we bind the individual CRTCs.
gpu: drm: vc4_hdmi: add missing of_node_put after calling of_parse_phandle
drm: vc4: enable XBGR8888 and ABGR8888 pixel formats
drm/vc4: clean up error exit path on failed dpi_connector allocation
Precise vblank timestamping is implemented via the
usual scanout position based method. On VC4 the
pixelvalves PV do not have a scanout position
register. Only the hardware video scaler HVS has a
similar register which describes which scanline for
the output is currently composited and stored in the
HVS fifo for later consumption by the PV.
This causes a problem in that the HVS runs at a much
faster clock (system clock / audio gate) than the PV
which runs at video mode dot clock, so the unless the
fifo between HVS and PV is full, the HVS will progress
faster in its observable read line position than video
scan rate, so the HVS position reading can't be directly
translated into a scanout position for timestamp correction.
Additionally when the PV is in vblank, it doesn't consume
from the fifo, so the fifo gets full very quickly and then
the HVS stops compositing until the PV enters active scanout
and starts consuming scanlines from the fifo again, making
new space for the HVS to composite.
Therefore a simple translation of HVS read position into
elapsed time since (or to) start of active scanout does
not work, but for the most interesting cases we can still
get useful and sufficiently accurate results:
1. The PV enters active scanout of a new frame with the
fifo of the HVS completely full, and the HVS can refill
any fifo line which gets consumed and thereby freed up by
the PV during active scanout very quickly. Therefore the
PV and HVS work effectively in lock-step during active
scanout with the fifo never having more than 1 scanline
freed up by the PV before it gets refilled. The PV's
real scanout position is therefore trailing the HVS
compositing position as scanoutpos = hvspos - fifosize
and we can get the true scanoutpos as HVS readpos minus
fifo size, so precise timestamping works while in active
scanout, except for the last few scanlines of the frame,
when the HVS reaches end of frame, stops compositing and
the PV catches up and drains the fifo. This special case
would only introduce minor errors though.
2. If we are in vblank, then we can only guess something
reasonable. If called from vblank irq, we assume the irq is
usually dispatched with minimum delay, so we can take a
timestamp taken at entry into the vblank irq handler as a
baseline and then add a full vblank duration until the
guessed start of active scanout. As irq dispatch is usually
pretty low latency this works with relatively low jitter and
good results.
If we aren't called from vblank then we could be anywhere
within the vblank interval, so we return a neutral result,
simply the current system timestamp, and hope for the best.
Measurement shows the generated timestamps to be rather precise,
and at least never off more than 1 vblank duration worst-case.
Limitations: Doesn't work well yet for interlaced video modes,
therefore disabled in interlaced mode for now.
v2: Use the DISPBASE registers to determine the FIFO size (changes
by anholt)
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-and-tested-by: Mario Kleiner <mario.kleiner.de@gmail.com> (v2)
Protect both the setup of the pageflip event and the
latching of the new requested displaylist head pointer
by the event lock, so we can't get into a situation
where vc4_atomic_flush latches the new display list via
HVS_WRITE, then immediately gets preempted before queueing
the pageflip event, then the page-flip completes in hw and
the vc4_crtc_handle_page_flip() runs and no-ops due to
lack of a pending pageflip event, then vc4_atomic_flush
continues and only then queues the pageflip event - after
the page flip handling already no-oped. This would cause
flip completion handling only at the next vblank - one
frame too late.
In vc4_crtc_handle_page_flip() check the actual DL head
pointer in SCALER_DISPLACTX against the requested pointer
for page flip to make sure that the flip actually really
completed in the current vblank and doesn't get deferred
to the next one because the DL head pointer was written
a bit too late into SCALER_DISPLISTX, after start of
vblank, and missed the boat. This avoids handling a
pageflip completion too early - one frame too early.
According to Eric, DL head pointer updates which were
written into the HVS DISPLISTX reg get committed to hardware
at the last pixel of active scanout. Our vblank interrupt
handler, as triggered by PV_INT_VFP_START irq, gets to run
earliest at the first pixel of HBLANK at the end of the
last scanline of active scanout, ie. vblank irq handling
runs at least 1 pixel duration after a potential pageflip
completion happened in hardware.
This ordering of events in the hardware, together with the
lock protection and SCALER_DISPLACTX sampling of this patch,
guarantees that pageflip completion handling only runs at
exactly the vblank irq of actual pageflip completion in all
cases.
Background info from Eric about the relative timing of
HVS, PV's and trigger points for interrupts, DL updates:
https://lists.freedesktop.org/archives/dri-devel/2016-May/107510.html
Tested on RPi 2B with hardware timing measurement equipment
and shown to no longer complete flips too early or too late.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
