If the user tries to mmap through the GTT an object that is marked as
snooped, we report an error rather than allow the GPU to hang the
machine. The choice of EINVAL, however, was unfortunate as we turn that
into a WARN rather than a quiet SIGBUS.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Move the MI_ARB_STATE MI_ARB_C3_LP_WRITE_ENABLE setup to
gen3_init_clock_gating() from i915_gem_load() when KMS is enabled. Leave
it in i915_gem_load() for the UMS case, but add an explcit check, just
to make it easier to spot it when we eventually rip out UMS support.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
85x also has a similar AGPBUSY# bit as gen3. Enable it to make
sure vblank interrupts don't get dealyed during C3 state.
There's also another bit which controls whether AGPBUSY# is asserted
based on pending cacheable cycles and interrupts, or just based on
pending commands in the ring and interrupts. Select the cacheable
cycles mode since that seems to be the new way of doing things in
85x, and it does give slightly better C3 residency numbers with
glxgears running.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
My Gen3 Bspec lists the AGPBUSY# bit in INSTPM as an enable bit rather
than a disable bit. Our code has the opposite idea. Make the code match
the spec.
Might fix some gen3 C3 related interrupt delivery problems. Untested
due to lack of hardware.
v2: call it AGPBUSY_INT_EN to make it clearer it has to do with interrupts
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
I don't see why we wouldn't want interrupts to wake up the CPU from C3
always, so just set the AGPBUSY# bit in gen3_init_clock_gating().
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
When doing this, all PLLs should be disabled.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We need to do this anytime we power gate the DPIO common well.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
There may be a dependency here.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This needs to be done before we power back on the CMN_BC well so the PHY
can calibrate properly.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We do this at runtime and later on now.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This is a bit like the CMN reset de-assert we do in DPIO_CTL, except
that it resets the whole common lane section of the PHY. This is
required on machines where the BIOS doesn't do this for us on boot or
resume to properly re-calibrate and get the PHY ready to transmit data.
Without this patch, such machines won't resume correctly much of the time,
with the symptom being a 'port ready' timeout and/or a link training
failure.
Note that simply asserting reset at suspend and de-asserting at resume
is not sufficient, nor is simply de-asserting at boot. Both of these
cases have been tested and have still been found to have failures on
some configurations.
v2: extract simpler set_power_well function for use in reset_dpio (Imre)
move to reset_dpio (Daniel & Ville)
v3: don't reset if DPIO reset is already de-asserted (Imre)
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
If we disable first the port (by disabling DPI) and only then the
display pipe the pipe-off flag will never be set, possibly leading to a
hanged pipe state at the next modeset-enable.
Note that according to the VLV2 display cluster HAS, we should disable
the port before the pipe. This doesn't seem to match reality based on
the above and it's also asymmetric with the enabling sequence, where we
first enable the port and then the pipe.
v2:
- send the panel shutdown command before stopping the pipe, since this
is the recommended sequence (Shobhit)
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Shobhit Kumar <shobhit.kumar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
It seems by default the VBT has MIPI configuration block as well. The
Generic driver will assume always MIPI if MIPI configuration block is found.
This is causing probelm when actually there is eDP. Fix this by looking
into general definition block which will have device configurations. From here
we can figure out what is the LFP type and initialize MIPI only if MIPI
is found.
v2: Addressed review comments by Damien
- Moved PORT definitions to intel_bios.h and renamed as DVO_PORT_MIPIA
- renamed is_mipi to has_mipi and moved definition as suggested
- Check has_mipi inside parse_mipi and intel_dsi_init insted of outside
v3: Make has_mipi as a bitfield as suggested
Signed-off-by: Shobhit Kumar <shobhit.kumar@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
[danvet: fold in conditions to pack everything neatly below 80 chars.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
For disabling L3 clock gating we need to set bit 25 of MMIO
register 940c. Earlier this was being done by just writing 1
into bit 25 and resetting all other bits.
This patch modifies the routine to read-modify-write of the
register, so that the values of other bits are not destroyed.
v2: Modifying the comments and the patch commit message (Chris)
Signed-off-by: Akash Goel <akash.goel@intel.com>
Signed-off-by: Sourab Gupta <sourab.gupta@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
[danvet: Apply checkpatch fixup.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This driver makes use of the generic panel information from the VBT.
Panel information is classified into two - panel configuration and panel
power sequence which is unique to each panel. The generic driver uses the
panel configuration and sequence parsed from VBT block #52 and #53
v2: Address review comments by Jani
- Move all of the things in driver c file from header
- Make all functions static
- Make use of video/mipi_display.c instead of redefining
- Null checks during sequence execution
v3: Address review comments by Damien
- Rename the panel driver file as intel_dsi_panel_vbt.c
- Fix style changes as suggested
- Correct comments for lp->hs and hs->lp count calculations
- General updating comments to have more clarity
- using max() instead of ternary operator
- Fix names (ui_num, ui_den) while using UI in calculations
- compute max of lp_to_hs switch and hs_to_lp switch while computing
hs_lp_switch_count
Signed-off-by: Shobhit Kumar <shobhit.kumar@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Fallout from an intermediate patch revision that I deemed worth saving.
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Currently we do a full re-init of all interrupts after a gpu hang.
Which is pretty bad since we don't restore the interrupts we've
enabled at runtime correctly. Even with that addressed it's rather
horribly race.
But on g4x and later we only reset the gt and not the entire gpu.
Which means we only need to reset the GT interrupt bits. Which has the
nice benefit that vblank waits, pipe CRC interrupts and everything
else display related just keeps on working.
The downside is that gt interrupt handling (i.e. ring->get/put_irq) is
still racy. But as long as the gpu hang reliably wakes all waters and
we have a short time where the refcount drops to 0 we'll recover. So
not that bad really.
v2: Ville noticed that GTIMR and PMIMR don't get cleared, only the
subordinate per-ring registers. So let's rip out all the interrupt dancing.
The FIXME comment is still required though since the ring irq handling
happens at the per-ring interrupt mask registers, too.
Testcase: igt/kms_flip/vblank-vs-hang
Testcase: igt/kms_pipe_crc_basic/hang-*
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Ville figured out that it needs a full display reset since apparently
a lot more goes down than just the GT. Until that's address it's
better to just diable gpu reset.
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
So apparently this is tricky.
We need to consider:
- We start out with all the hw enabling bits disabled, both the
individual fifo underrun interrupts and the shared display error
interrupts masked. Otherwise if the bios config is broken we'll blow
up with a NULL deref in our interrupt handler since the crtc
structures aren't set up yet at driver load time.
- On gmch we need to mask fifo underruns on the sw side, so always
need to set that in sanitize_crtc for those platforms.
- On other platforms we try to set the sw tracking so that it reflects
the real state. But since a few platforms have shared bits we must
_not_ disable fifo underrun reporting. Otherwise we'll never enable
the shared error interrupt.
This is the state before out patch, but unfortunately this is not good
enough. But after a suspend resume operation this is broken:
1. We don't enable the hw interrupts since the same code runs on
resume as on driver load.
2. The fifo underrun state adjustments we do in sanitize_crtc doesn't
fire on resume since (except for hilarious firmware) all pipes are off
at that point. But they also don't hurt since the subsequent crtc
enabling due to force_restore will enable fifo underruns.
Which means when we enable fifo underrun reporting we notice that the
per-crtc state is already correct and short-circuit everthing out. And
the interrupt doesn't get enabled.
A similar problem would happen if the bios doesn't light up anything
when the driver loads. Which is exactly what happens when we reload
the driver since our unload functions disables all outputs.
Now we can't just rip out the short-circuit logic and unconditionally
update the fifo underrun reporting interrupt masking: We have some
checks for shared error interrupts to catch issues that happened when
the shared error interrupt was disabled.
The right fix is to push down this logic so that we can always update
the hardware state, but only check for missed fifo underruns on a real
enabled->disabled transition and ignore them when we're already
disabled.
On platforms with shared error interrupt the pipe CRC interrupts are
grouped together with the fifo underrun reporting this fixes pipe CRC
support after suspend and driver reloads.
Testcase: igt/kms_pipe_crc_basic/suspend-*
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
On platforms with shared interrupt enable bits (which are shared even
with the pipe CRC logic) there's some tricky corner cases. Add
information to make debugging those easier.
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
All drm_fb_helper_restore_fbdev_mode() call sites, save one, do the same
locking. Simplify this into drm_fb_helper_restore_fbdev_mode_unlocked().
Signed-off-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
For atomic, it will be quite necessary to not need to care so much
about locking order. And 'state' gives us a convenient place to stash a
ww_ctx for any sort of update that needs to grab multiple crtc locks.
Because we will want to eventually make locking even more fine grained
(giving locks to planes, connectors, etc), split out drm_modeset_lock
and drm_modeset_acquire_ctx to track acquired locks.
Atomic will use this to keep track of which locks have been acquired
in a transaction.
v1: original
v2: remove a few things not needed until atomic, for now
v3: update for v3 of connection_mutex patch..
v4: squash in docbook
v5: doc tweaks/fixes
Signed-off-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This should avoid races between connector probing and HPD
irqs in the future, currently mode_config.mutex blocks this
possibility.
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
After the split-out of crtc locks from the big mode_config.mutex
there's still two major areas it protects:
- Various connector probe states, like connector->status, EDID
properties, probed mode lists and similar information.
- The links from connector->encoder and encoder->crtc and other
modeset-relevant connector state (e.g. properties which control the
panel fitter).
The later is used by modeset operations. But they don't really care
about the former since it's allowed to e.g. enable a disconnected VGA
output or with a mode not in the probed list.
Thus far this hasn't been a problem, but for the atomic modeset
conversion Rob Clark needs to convert all modeset relevant locks into
w/w locks. This is required because the order of acquisition is
determined by how userspace supplies the atomic modeset data. This has
run into troubles in the detect path since the i915 load detect code
needs _both_ protections offered by the mode_config.mutex: It updates
probe state and it needs to change the modeset configuration to enable
the temporary load detect pipe.
The big deal here is that for the probe/detect users of this lock a
plain mutex fits best, but for atomic modesets we really want a w/w
mutex. To fix this lets split out a new connection_mutex lock for the
modeset relevant parts.
For simplicity I've decided to only add one additional lock for all
connector/encoder links and modeset configuration states. We have
piles of different modeset objects in addition to those (like bridges
or panels), so adding per-object locks would be much more effort.
Also, we're guaranteed (at least for now) to do a full modeset if we
need to acquire this lock. Which means that fine-grained locking is
fairly irrelevant compared to the amount of time the full modeset will
take.
I've done a full audit, and there's just a few things that justify
special focus:
- Locking in drm_sysfs.c is almost completely absent. We should
sprinkle mode_config.connection_mutex over this file a bit, but
since it already lacks mode_config.mutex this patch wont make the
situation any worse. This is material for a follow-up patch.
- omap has a omap_framebuffer_flush function which walks the
connector->encoder->crtc links and is called from many contexts.
Some look like they don't acquire mode_config.mutex, so this is
already racy. Again fixing this is material for a separate patch.
- The radeon hot_plug function to retrain DP links looks at
connector->dpms. Currently this happens without any locking, so is
already racy. I think radeon_hotplug_work_func should gain
mutex_lock/unlock calls for the mode_config.connection_mutex.
- Same applies to i915's intel_dp_hot_plug. But again, this is already
racy.
- i915 load_detect code needs to acquire this lock. Which means the
w/w dance due to Rob's work will be nicely contained to _just_ this
function.
I've added fixme comments everywhere where it looks suspicious but in
the sysfs code. After a quick irc discussion with Dave Airlie it
sounds like the lack of locking in there is due to sysfs cleanup fun
at module unload.
v1: original (only compile tested)
v2: missing mutex_init(), etc (from Rob Clark)
v3: i915 needs more care in the conversion:
- Protect the edp pp logic with the connection_mutex.
- Use connection_mutex in the backlight code due to
get_pipe_from_connector.
- Use drm_modeset_lock_all in suspend/resume paths.
- Update lock checks in the overlay code.
Cc: Alex Deucher <alexdeucher@gmail.com>
Cc: Rob Clark <robdclark@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Rob Clark <robdclark@gmail.com>
This is pure evil. Userspace, I'm looking at you SNA, repacks batch
buffers on the fly after generation as they are being passed to the
kernel for execution. These batches also contain self-referenced
relocations as a single buffer encompasses the state commands, kernels,
vertices and sampler. During generation the buffers are placed at known
offsets within the full batch, and then the relocation deltas (as passed
to the kernel) are tweaked as the batch is repacked into a smaller buffer.
This means that userspace is passing negative relocations deltas, which
subsequently wrap to large values if the batch is at a low address. The
GPU hangs when it then tries to use the large value as a base for its
address offsets, rather than wrapping back to the real value (as one
would hope). As the GPU uses positive offsets from the base, we can
treat the relocation address as the minimum address read by the GPU.
For the upper bound, we trust that userspace will not read beyond the
end of the buffer.
So, how do we fix negative relocations from wrapping? We can either
check that every relocation looks valid when we write it, and then
position each object such that we prevent the offset wraparound, or we
just special-case the self-referential behaviour of SNA and force all
batches to be above 256k. Daniel prefers the latter approach.
This fixes a GPU hang when it tries to use an address (relocation +
offset) greater than the GTT size. The issue would occur quite easily
with full-ppgtt as each fd gets its own VM space, so low offsets would
often be handed out. However, with the rearrangement of the low GTT due
to capturing the BIOS framebuffer, it is already affecting kernels 3.15
onwards. I think only IVB+ is susceptible to this bug, but the workaround
should only kick in rarely, so it seems sensible to always apply it.
v3: Use a bias for batch buffers to prevent small negative delta relocations
from wrapping.
v4 from Daniel:
- s/BIAS/BATCH_OFFSET_BIAS/
- Extract eb_vma_misplaced/i915_vma_misplaced since the conditions
were growing rather cumbersome.
- Add a comment to eb_get_batch explaining why we do this.
- Apply the batch offset bias everywhere but mention that we've only
observed it on gen7 gpus.
- Drop PIN_OFFSET_FIX for now, that slipped in from a feature patch.
v5: Add static to eb_get_batch, spotted by 0-day tester.
Testcase: igt/gem_bad_reloc
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=78533
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> (v3)
Cc: stable@vger.kernel.org
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We only want to modifiy a single field in the userspace view of the
execbuffer command buffer, so explicitly change that rather than copy
everything back again.
This serves two purposes:
1. The single fields are much cheaper to copy (constant size so the
copy uses special case code) and much smaller than the whole array.
2. We modify the array for internal use that need to be masked from
the user.
Note: We need this backported since without it the next bugfix will
blow up when userspace recycles batchbuffers and relocations.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: stable@vger.kernel.org
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
A single object may be referenced by multiple registers fundamentally
breaking the static allotment of ids in the current design. When the
object is used the second time, the physical address of the first
assignment is relinquished and a second one granted. However, the
hardware is still reading (and possibly writing) to the old physical
address now returned to the system. Eventually hilarity will ensue, but
in the short term, it just means that cursors are broken when using more
than one pipe.
v2: Fix up leak of pci handle when handling an error during attachment,
and avoid a double kmap/kunmap. (Ville)
Rebase against -fixes.
v3: And fix the error handling added in v2 (Ville)
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=77351
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: stable@vger.kernel.org
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
It's barely alive now anyway, so give it the "coup de grâce".
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Up until now, contexts had one (and only one) backing object that was
used by the hardware to save/restore render ring contexts (via the
MI_SET_CONTEXT command). Other rings did not have or need this, so
our i915_hw_context struct had a 1:1 relationship with a a real HW
context.
With Logical Ring Contexts and Execlists, this is not possible anymore:
all rings need a backing object, and it cannot be reused. To prepare
for that, rename our contexts to the more generic term intel_context.
No functional changes.
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Manual cleanup after the previous Coccinelle script.
Yes, I could write another Coccinelle script to do this but I
don't want labor-replacing robots making an honest programmer's
work obsolete (also, I'm lazy).
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
As advanced by the previous patch, the ringbuffers and the engine
command streamers belong in different structs. This is so because,
while they used to be tightly coupled together, the new Logical
Ring Contexts (LRC for short) have a ringbuffer each.
In legacy code, we will use the buffer* pointer inside each ring
to get to the pertaining ringbuffer (the actual switch will be
done in the next patch). In the new Execlists code, this pointer
will be NULL and we will use instead the one inside the context
instead.
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
In the upcoming patches we plan to break the correlation between
engine command streamers (a.k.a. rings) and ringbuffers, so it
makes sense to refactor the code and make the change obvious.
No functional changes.
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Atm, we disable GT power saving during the end of the suspend sequence
in i915_save_state(). Doing the disabling at that point seems arbitrary.
One reason to disable it early though is to have a quiescent HW state
before we do anything else (for example save registers). So move the
disabling earlier, which also takes care canceling of the deferred RPS
enabling work done by intel_disable_gt_powersave().
Note that after the move we'll call intel_disable_gt_powersave() only
in case modeset is enabled, but that's anyway the only case where we
have it enabled in the first place.
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Robert Beckett <robert.beckett@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
In
commit c6df39b5ea
Author: Imre Deak <imre.deak@intel.com>
Date: Mon Apr 14 20:24:29 2014 +0300
drm/i915: get a runtime PM ref for the deferred GT powersave enabling
I added an RPM get-ref when enabling RPS from a deferred work, but forgot
to add the corresponding put-ref when canceling the work. This may leave
RPM disabled.
Note that the race is real since we run the rps enabling with a
delayed work item after resume, so leaves enough time (in contrived
examples) to fit a quick autoresum in.
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Robert Beckett <robert.beckett@intel.com>
Testecase: igt/pm_rpm/system-suspend
[danvet: Mention testcase and add note.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Currently user space can access GEM buffers mapped to GTT through
existing mappings concurrently while the platform specific suspend
handlers are running. Since these handlers may change the HW state in a
way that would break such accesses, remove the mappings before calling
the handlers. Spotted by Ville.
Also Chris pointed out that the lists that i915_gem_release_all_mmaps()
walks through need dev->struct_mutex, so take this lock. There is a
potential deadlock against a concurrent RPM resume, resolve this by
aborting and rescheduling the suspend (Daniel).
v2:
- take struct_mutex around i915_gem_release_all_mmaps() (Chris, Daniel)
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Robert Beckett <robert.beckett@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Apparently we need to disable VCP unit clock gating around media reset
on g4x.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
On gen2 the scanline counter behaves a bit differently from the
later generations. Instead of adding one to the raw scanline
counter value, we must subtract one.
On HSW/BDW the scanline counter requires a +2 adjustment on HDMI
outputs. DP outputs on the on the other require the typical +1
adjustment.
As the fixup we must apply to the hardware scanline counter
depends on several factors, compute the desired offset at modeset
time and tuck it away for when it's needed.
v2: Clarify HSW+ situation
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: "Akash Goel <akash.goels@gmail.com>"
Reviewed-by: "Sourab Gupta <sourabgupta@gmail.com>"
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=78997
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The docs are a bit lacking when it comes to describing when certain
timing related events occur in the hardware. Draw a picture which
tries to capture the most important ones.
v2: Clarify a few details (Imre)
v3: Add HSW+ HDMI scanline counter numbers
Acked-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Akash Goel <akash.goels@gmail.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: "Sourab Gupta <sourabgupta@gmail.com>"
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Currently the logic to fix up the frame counter on gen3/4 assumes that
start of vblank occurs at vblank_start*htotal pixels, when in fact
it occurs htotal-hsync_start pixels earlier. Apply the appropriate
adjustment to make the frame counter more accurate.
Also fix the vblank start position for interlaced display modes.
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: "Akash Goel <akash.goels@gmail.com>"
Reviewed-by: "Sourab Gupta <sourabgupta@gmail.com>"
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
In interlaced modes, the pixel counter counts all pixels,
so one field will have htotal more pixels. In order to avoid
the reported position from jumping backwards when the pixel
counter is beyond the length of the shorter field, just
clamp the position the length of the shorter field. This
matches how the scanline counter based position works since
the scanline counter doesn't count the two half lines.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: "Akash Goel <akash.goels@gmail.com>"
Reviewed-by: "Sourab Gupta <sourabgupta@gmail.com>"
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel keeps on ramping up the warning level of the DRM and our display
core to make it complain whenever the locking rules are not followed.
This caught
commit 24576d2397
Author: Jesse Barnes <jbarnes@virtuousgeek.org>
Date: Tue Mar 26 09:25:45 2013 -0700
drm/i915: enable VT switchless resume v3
introducing an unlocked access to the CRTC whilst disabling it for
suspend.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=78114
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Before purging our pages (as opposed to copying back the contents from
the GPU), make sure that there is not an exposed CPU mmapping through
which the user can inspect the results.
Regression from
commit 5537252b6b
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Tue Mar 25 13:23:06 2014 +0000
drm/i915: Invalidate our pages under memory pressure
Testcase: igt/gem_mmap/new-object
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=79005
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: Guo Jinxian <jinxianx.guo@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We have to write to the primary plane base address registrer when we
enable/disable the primary plane in response to sprite coverage. Those
writes will cause the flip counter to increment which could interfere
with the detection of CS flip completion. We could end up completing
CS flips before the CS has even executed the commands from the ring.
To avoid such issues, wait for CS flips to finish before we toggle the
primary plane on/off.
v2: Rebased due to atomic sprite update changes
Testcase: igt/kms_mmio_vs_cs_flip/setplane_vs_cs_flip
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by Rodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
With the current code, we unconditionally touch
HSW_AUD_PIN_ELD_CP_VLD, which means we can touch it when the power
well is off, and that will trigger an "Unclaimed register" message.
Just adding the intel_crtc->config.has_audio should already avoid the
unclaimed register messsages, but since we actually need the power
well to make the Audio code work, it makes sense to also grab the
audio power domain reference, and release it when it's not needed
anymore.
I used IGT's pm_rpm to reproduce this bug, but it can probably be
reproduced on other tests that do modesets. I'm using a machine with
eDP+HDMI connected.
Regression introduced by:
commit acfa75b02e
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date: Thu Apr 24 23:54:51 2014 +0200
drm/i915: Simplify audio handling on DDI ports
Credits to Daniel for suggesting this implementation.
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Because this will trigger "Unclaimed register" messages. All I need to
reproduce this problem is to boot my HSW machine with eDP+HDMI
connected.
Regression introduced by:
commit 9ed109a7b4
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date: Thu Apr 24 23:54:52 2014 +0200
drm/i915: Track has_audio in the pipe config
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Adding stuff at the bottom is really no how this should be done, since
that's the place for ums/dri dungeons.
This was added in
commit a8ebba75b3
Author: Zhao Yakui <yakui.zhao@intel.com>
Date: Thu Apr 17 10:37:40 2014 +0800
drm/i915: Use the coarse ping-pong mechanism based on drm fd to dispatch the BSD command on BDW GT3
Also add a note to prevent this from happening again - people really
should be less lazy and take more time to look for a good home of
their new driver-global state.
Cc: Imre Deak <imre.deak@intel.com>
Cc: Zhao Yakui <yakui.zhao@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Gen2 reports FIFO underruns whenever no planes are enabled on the pipe.
So in order to avoid false positives we must enable the FIFO underrun
reporting only when at least one plane is enabled on the pipe. For
now just move the underrun reporting enable/disable points to the
other side of the plane enable/disable point. That doesn't cover cases
when we turn off all the planes for the pipe but leave the pipe running
on purpose, but it's better than the current situation.
On gen4+ we can actually move the underrun reporting enable/disable to
the opposite ends of the crtc enable/disable hooks. I suppose in theory
we could leave the underrun reporting enabled all the time, except on
VLV where PIPESTAT stops working when the display power well is down.
If we ever get around to unifying the PIPESTAT irq handling for all
gmch platforms, we should still follow the VLV route for other platforms.
It would also micro-optimize the irq handler a bit since we could then
skip the PIPESTAT reads for all disabled pipes.
Gen3 is still a mystery, but for now I'm going to assume it behaves
like gen4+.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Thomas Wood <thomas.wood@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Checking whether the error interrupt was enabled or not isn't really
necessary when we check for uncleared FIFO underruns. If it was enabled
we'll race with the interrupt handler a bit, but that seems OK as we
still claim the interrupt.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Thomas Wood <thomas.wood@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
FIFO underruns don't generate interrupts on gmch platforms, so
if we want to know whether a modeset triggered FIFO underruns we
need to explicitly check for them.
As a modeset on one pipe could cause underruns on other pipes,
check for underruns on all pipes.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Thomas Wood <thomas.wood@intel.com>
[danvet: Fix up merge error, kudos to Ville for noticing it.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
FIFO underruns don't generate an interrupt on gmch platforms, so we
should check whether there were any that we failed to notice when
we're disabling FIFO underrun reporting.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Thomas Wood <thomas.wood@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Document the internal structure of the VLV display PHY a bit to help
people understand how the different register blocks relate to each
other.
v2: Add a bit more text
Make it a DOC: comment, but leave the ascii art out since
it would get mangled
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chon Ming Lee <chon.ming.lee@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
for proper refcounting to take place as we use
i915_add_request() for it.
i915_add_request() also takes the context for the request
from ring->last_context so move the null state batch
submission after the ring context has been set.
v2: we need to check for correct ring now (Ville Syrjälä)
v3: no need to expose i915_gem_move_object_to_active (Chris Wilson)
v4: cargoculted vma/active/inactive error handling removed (Chris Wilson)
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
If a pipe is already active when we init/resume there might not be a
full modeset afterwards so drm_vblank_on() may not get called. In such
a case if someone is holding a vblank reference across a suspend/resume
cycle drm_vblank_get() called after resuming won't re-enable the vblank
interrupts.
So in order to make sure vblank interrupts get re-enabled post-resume,
call drm_vblank_on() in intel_sanitize_crtc() if the crtc is already
active.
v2: Also drm_vblank_off() if the pipe got disabled magically
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Testecase: igt/kms_flip/vblank-vs-suspend
Tested-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Pull in the drm vblank rework from Ville and me. drm core parts acked
by Dave Airlie
Conflicts:
drivers/gpu/drm/i915/intel_display.c
Just a bit of fun around the placement of drm_vblank_on. This merge
resolution has been tested in drm-intel-nightly for a while already.
Acked-by: Dave Airlie <airlied@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We don't have hardware based disable bits on gmch platforms, so need
to block spurious underrun reports in software. Which means that we
_must_ start out with fifo underrun reporting disabled everywhere.
This is in big contrast to ilk/hsw/cpt where there's only _one_
disable bit for all platforms and hence we must allow underrun
reporting on disabled pipes. Otherwise nothing really works,
especially the CRC support since that's key'ed off the same irq
disable bit.
This allows us to ditch the fifo underrun reporting hack from the vlv
runtime pm code and unexport the internal function from i915_irq.c
again. Yay!
v2: Keep the display irq disabling, spotted by Imre.
Cc: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Now that we unconditionally dtrt when disabling/enabling crtcs we
don't need any hacks any longer to keep the vblank logic sane when
all the registers go poof. So let's rip it all out.
This essentially undoes
commit 9dbd8febb4
Author: Paulo Zanoni <paulo.r.zanoni@intel.com>
Date: Tue Jul 23 10:48:11 2013 -0300
drm/i915: update last_vblank when disabling the power well
Apparently igt/kms_flip is already powerful enough to exercise this
properly, yay! See the reference regression report for details.
v2: Update testcase name
References: https://bugs.freedesktop.org/show_bug.cgi?id=66808
Testcase: igt/kms_flip/vblank-vs-*-rpm
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Only the low-level irq handling functions still use integer crtc
indices with this. But fixing that will require a lot more sugery
and some good ideas for backwards compat with old ums userspace.
Both in drivers and in the drm core.
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Originally these functions have been for user modesetting drivers to
ensure vblank processing doesn't fall over completely around modeset
changes. This has been carried over ever since then.
Now that Ville cleaned our vblank handling with an explicit
drm_vblank_off/on braket when disabling/enabling crtcs. So this seems
to be unnecessary now. The most important side effect was that due to
the delayed vblank disabling we have been pretty much guaranteed to
receive a vblank interrupt soonish after a crtc was enabled.
Note that our vblank handling across modeset is still fairly decent
fubar - we don't actually handle vblank counter all to well.
drm_update_vblank_count will make sure that the frame counter always
rolls forward, but userspace isn't really all to ready to cope with
the big jumps this causes.
This isn't a big mostly because the hardware retains the frame
counter. But with runtime pm and also across suspend/resume we fall
over.
Fixing this is a lot more involved and also needs som i-g-ts. So
material for another patch series.
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
All of the .queue_flip() callbacks duplicate the same code to pin the
buffers and calculate the gtt_offset. Move that code to
intel_crtc_page_flip(). In order to do that we must also move the ring
selection logic there.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Now that we've plugged the mmio vs. ring flip race, we shouldn't need
these vblank waits in the modeset codepaths anymore. So get rid of
them.
v2: gen2 needs to wait for planes to turn off before disabling pipe
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Now that the vblank wait is gone from intel_enable_primary_plane(),
hsw_enable_ips() needs to do the vblank wait itself.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Starting from ILK, mmio flips also cause a flip done interrupt to be
signalled. This means if we first do a set_base and follow it
immediately with the CS flip, we might mistake the flip done interrupt
caused by the set_base as the flip done interrupt caused by the CS
flip.
The hardware has a flip counter which increments every time a mmio or
CS flip is issued. It basically counts the number of DSPSURF register
writes. This means we can sample the counter before we put the CS
flip into the ring, and then when we get a flip done interrupt we can
check whether the CS flip has actually performed the surface address
update, or if the interrupt was caused by a previous but yet
unfinished mmio flip.
Even with the flip counter we still have a race condition of the CS flip
base address update happens after the mmio flip done interrupt was
raised but not yet processed by the driver. When the interrupt is
eventually processed, the flip counter will already indicate that the
CS flip has been executed, but it would not actually complete until the
next start of vblank. We can use the DSPSURFLIVE register to check
whether the hardware is actually scanning out of the buffer we expect,
or if we managed hit this race window.
This covers all the cases where the CS flip actually changes the base
address. If the base address remains unchanged, we might still complete
the CS flip before it has actually completed. But since the address
didn't change anyway, the premature flip completion can't result in
userspace overwriting data that's still being scanned out.
CTG already has the flip counter and DSPSURFLIVE registers, and
although the flip done interrupt is still limited to CS flips alone,
the code now also checks the flip counter on CTG as well.
v2: s/dspsurf/gtt_offset/ (Chris)
Testcase: igt/kms_mmio_vs_cs_flip/setcrtc_vs_cs_flip
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=73027
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@gmail.com>
[danvet: Add g4x_ prefix to flip_count_after_eq.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We really just want to go detect displays again and fire off a hotplug
event if things have changed, not go through full hotplug processing.
Requested-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Since
commit 2e82a72031
Author: Imre Deak <imre.deak@intel.com>
Date: Fri Jan 17 15:46:43 2014 +0200
drm/i915: don't disable DP port after a failed link training
and
commit 5d6a1116c6
Author: Imre Deak <imre.deak@intel.com>
Date: Thu Jan 16 18:35:57 2014 +0200
drm/i915: don't disable the DP port if the link is lost
we no longer call intel_dp_link_down from generic DP code, but only
from the !HAS_DDI dp encoder functions. hsw/bdw have their own encoder
disabling callback in intel_ddi.c.
Hence the early return is no longer needed and the big comment just
confusing, so let's rip it out. To ensure what we don't accidentally
use this again on ddi encoders add a WARN_ON instead.
Spotted while reading through intel_dp.c
Cc: Imre Deak <imre.deak@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
drm_vblank_off() will turn off vblank interrupts, but as long as the
refcount is elevated drm_vblank_get() will not re-enable them. This
is a problem is someone is holding a vblank reference while a modeset is
happening, and the driver requires vblank interrupt to work during that
time.
Add drm_vblank_on() as a counterpart to drm_vblank_off() which will
re-enabled vblank interrupts if the refcount is already elevated. This
will allow drivers to choose the specific places in the modeset sequence
at which vblank interrupts get disabled and enabled.
Testcase: igt/kms_flip/*-vs-suspend
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
[danvet: Add Testcase tag for the igt I've written.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Somehow a few functions have been dropped in the middle of backlight
code. Move them around. No functional changes.
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Clear the reset domain after a succesful GPU reset on ilk. We already
do that on gen4, so let's try to be a bit more consistent. And if
ether render or media reset fails, we might use the leftover value
in the register to pinpoint the culprit.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
All the other bits in the GDSR register are read-only, so we don't have
to preserve them when we perform a GPU reset.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
I'm trying to reduce the WARNs during driver reload and this was one of
them. Also while at it remove the redundant condition from before
unregister_shrinker().
v2:
- fix the error path too and move the unregister to its logical place
(Chris)
- remove redundant condition from before unregister_shrinker()
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The comments in i915_reg.h aren't proper kernel-doc comments, so replace
the magic /** with just /*
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The following workarounds should be needed for pre-production hardware
only:
* WaDisablePwrmtrEvent:chv
* WaSetMaskForGfxBusyness:chv
* WaDisableGunitClockGating:chv
* WaDisableFfDopClockGating:chv
* WaDisableDopClockGating:chv
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The spec only tells us to set individual bits here and there. So we use
RMW for most things. Do the same for the swing calc init.
Eventually we should optimize things to just blast the final value in
with group access whenever possible. But to do that someone needs to
take a good look at what's the reset value for each registers, and
possibly if the BIOS manages to frob with some of them. For now
use RMW access always.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Like PCS, TX group reads return 0xffffffff. So we need to target each
lane separately if we want to use RMW cycles to update the registers.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
All PCS groups access reads return 0xffffffff, so we can't use group
access for RMW cycles. Instead target each spline separately.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
[danvet: Fight conflict with misplaced ; .... ARGH!]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The bits we've been setting so far only progagate the reset singal to
the data lanes. To actaully force the reset signal we need to set another
override bit.
v2: Fix mispalced ';' (Mika)
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Seems like we shouldn't leave the data lane resert deasserted when
the port if disabled. So propagate the reset the data lanes in
the encoder .post_disable() hook.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
During the enable sequence we first enable the dclkp output to the
display controller, and then enable the PLL. Do the opposite during
the disable sequence.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We need to pick the correct data lanes based on the port not the
pipe, so move the data lane deassert into the encoder .pre_enable()
hook from the chv_enable_pll().
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Setup the pipe config dpll state correctly for CHV. Also add
a assert_pipe_disabled() to chv_disable_pll(), and program the
DPLL_MD registers in chv_enable_pll().
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Antti Koskipää <antti.koskipaa@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Fix the encoder .get_config hooks to report the correct active pipe for
CHV.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Antti Koskipää <antti.koskipaa@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
CHV has three pipes so let's expose them all.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Antti Koskipää <antti.koskipaa@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Unsurprisingly the cursor C regiters are also at a weird offset on CHV.
Add more pipe offsets to handle them.
This also gets rid of most of the differences between the i9xx vs. ivb
cursor code. We can unify the remaining code as well, but I'll leave
that for another patch.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Antti Koskipää <antti.koskipaa@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
On CHV the GMBUS port for port D is different from other gmch platforms
which have port D. Fix it up.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Antti Koskipää <antti.koskipaa@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
On CHV pipe C can driver only port D, and pipes A and B can drivbe only
ports B and C. Configure the crtc_mask appropriately to reflect that.
v2: Moar braces (Jani)
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Antti Koskipää <antti.koskipaa@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Antti Koskipää <antti.koskipaa@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Antti Koskipää <antti.koskipaa@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Antti Koskipää <antti.koskipaa@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Add support for the third pipe in cherrview
v2: Don't use spaces for indentation (Jani)
Wrap long lines
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Rafael Barbalho <rafael.barbalho@intel.com>
[vsyrjala: slightly massaged the patch]
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Cherryview also needs this WA.
Signed-off-by: Rafael Barbalho <rafael.barbalho@intel.com>
[vsyrjala: Looks like it's for pre-prodution hw only]
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We implement the following workarounds:
* WaDisableAsyncFlipPerfMode:chv
* WaProgramMiArbOnOffAroundMiSetContext:chv
v2: Drop WaDisableSemaphoreAndSyncFlipWait note
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This workaround is listed for CHV, but not for BDW. However BSpec notes
that on BDW CSunit clock gating is always disabled irrespective of the
relevant bit in the GEN6_UGCTL1 registers. For CHV however, such text
is not present in BSpec, so it seems safer to just set the bit.
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
BDW has the same requirement but the w/a database doens't list
this w/a for BDW. Seems to be another one of those "stick a bunch
of known workarounds into this bag and write something on the label"
type of things.
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Besides the fairly useless BUG_ON the logic is completely generic
and cane be used on any platform what wants to reuse the shared
dpll support code.
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This is the last piece of code which write state to the hardware in
the ironalake ->crtc_mode_set callback.
I think we could merge this with the pll->enable hook, but otoh the
ordering requirements with the ldvs port are really tricky. Doing the
FP0/1 writes up-front before we even prepare the lvds port (in the
pre_pll_enable hook) like on i9xx seems safest.
With this ilk+ platforms are now ready for runtime PM with DPMS. Since
hsw/bdw also support runtime pm besides snb we need to first make the
haswell code save before we can touch the core code.
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Instead of every time it isn't active: We only need to do that when
the pll is currently unused, i.e. when pll->refcount == 0. For
paranoia add a warning for the ibx case where plls have a fixed
mapping and hence should always be unused after the call to
intel_put_shared_dpll.
v2: Simplify control flow and use struct assignment instead of memcpy
as suggested by Damien.
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
With this all hw writes are also gone from the ->crtc_mode_set hook on
vlv. I wondered whether we should track more of the pll state in the
pipe config, but otoh as long as we don't have shared plls that's not
really useful - the cross-checking of the port clock should be
sufficient.
While at it also de-magic some of the pipe checks, this has been
irking me since a long time.
Whit this vlv is now ready for runtime PM on dpms. If we'd have
runtime PM support in general ...
Reviewed-by: Shobhit Kumar <shobhit.kumar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
These two writes are the very last hw writes from the
->crtc_modeset_callback on pre-gen5 hardware. As usual vlv is a bit
different, so this here is just warm-up.
Reviewed-by: Shobhit Kumar <shobhit.kumar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Again the same story: This code just transform sw state from the pipe
config into hardware state. And again we can't move the pll code, but
this time around because the state isn't properly tracked in the pipe
config.
Reviewed-by: Shobhit Kumar <shobhit.kumar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Again this code just transforms sw state from the pipe config into
hardware state, so we can just move it around. Unfortunately again a
few forward declarations since intel_display.c is becoming a bit of a
mess.
Note that both for i9xx and ironlake code the only things remaining in
the ->crtc_mode_set hook is now the clock state computation and
sharing code. That needs to be moved into the compute config stage so
that we can catch impossible configurations earlier.
Also note that some of the DPLL hw setup code is still run from within
->crtc_mode_set, namele the pll->mode_set callback. We need to move
that first before we can do fancy things like enable runtime PM for
dpms off.
v2: Make it compile again after the rebase, bisectability issue
reported by Wu Fengguang.
Reviewed-by: Shobhit Kumar <shobhit.kumar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Now this really should be in the pipe config somewhere, but till now
it isn't. We can at least move it up a bit next to all the other pll
code since intel_dp_set_m_n really doesn't depend upon this.
This is just prep work so that moving all the hw frobbing code from
->crtc_mode_set to ->crtc_enable is clean.
v2: Do the same for haswell while at it, not just for ivb.
Reviewed-by: Shobhit Kumar <shobhit.kumar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
All these functions simply convert sw state as encoded in the pipe
config or primary framebuffer into hardware state. So we can move them
all into the crtc enable hook. Unfortunately this means a little bit
of duplication between the i9xx and vlv functions, but since we
already have highly refactored code I think this is acceptable.
Also a pile of forward declarations unfortunately.
Note also that the various <platform>_update_pll functions are still
called from within the ->crtc_mode_set hook. Mostly they compute the
clock state for the pipe config, but unfortunately there are some
random register writes interspersed. Those need to be moved out first
before we can enable runtime PM for DPMS.
Reviewed-by: Shobhit Kumar <shobhit.kumar@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Before the process killer is invoked, oom-notifiers are executed for one
last try at recovering pages. We can hook into this callback to be sure
that everything that can be is purged from our page lists, and to give a
summary of how much memory is still pinned by the GPU in the case of an
oom. This should be really valuable for debugging OOM issues.
Note that the last-ditch effort call to shrink_all we've previously
called from our normal shrinker when we could free as much as the vm
demaned is moved into the oom notifier. Since the shrinker accounting
races against bind/unbind operations we might have called shrink_all
prematurely, which this approach with an oom notifier avoids.
References: https://bugs.freedesktop.org/show_bug.cgi?id=72742
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: lu hua <huax.lu@intel.com>
[danvet: Bikeshed logical | into || and pimp commit message.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We're using the reset domains bits for g4x on ilk. But on ilk those bits
actually shifted by one bit. Fix it up so that we use the correct bits.
We were actually always writing 0x2 to the reset domain bits, which
is a reserved value. In practice it looks like the hardware ignores that
value since nothing happens if I write that value when there's a 3D
workload running. Writing the _correct_ render domain value actually
makes the GPU stop.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We should be waiting for the reset bit to clear, not remain set.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
There are comments in the gen4-5 reset functions stating that we can't
reset render and media without also doing a display reset. But that's
exactly what the code does, ie. we don't perform a display reset. Drop
the bogus comments.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Try to flush out dirty pages into the swapcache (and from there into the
swapfile) when under memory pressure and forced to drop GEM objects from
memory. In effect, this should just allow us to discard unused pages for
memory reclaim and to start writeback earlier.
v2: Hugh Dickins warned that explicitly starting writeback from
shrink_slab was prone to deadlocks within shmemfs.
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Robert Beckett <robert.beckett@intel.com>
Reviewed-by: Rafael Barbalho <rafael.barbalho@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We can share a few lines of tricky lock handling we need to use for both
shrinker routines and in the process fix the return value for count()
when reporting a deadlock.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Robert Beckett <robert.beckett@intel.com>
Reviewed-by: Rafael Barbalho <rafael.barbalho@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
When the machine is under a lot of memory pressure and being stressed by
multiple GPU threads, we quite often report fewer than shrinker->batch
(i.e. SHRINK_BATCH) pages to be freed. This causes the shrink_control to
skip calling into i915.ko to release pages, despite the GPU holding onto
most of the physical pages in its active lists.
References: https://bugs.freedesktop.org/show_bug.cgi?id=72742
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Robert Beckett <robert.beckett@intel.com>
Reviewed-by: Rafael Barbalho <rafael.barbalho@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
shmemfs first checks if there is enough memory to allocate the page
and reports ENOSPC should there be insufficient, along with
the usual ENOMEM for a genuine allocation failure.
We use ENOSPC in our driver to mean that we have run out of aperture
space and so want to translate the error from shmemfs back to
our usual understanding of ENOMEM. None of the the other GEM users
appear to distinguish between ENOMEM and ENOSPC in their error handling,
hence it is easiest to do the fixup in i915.ko
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Robert Beckett <robert.beckett@intel.com>
Reviewed-by: Rafael Barbalho <rafael.barbalho@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This patch adds a mmio base address variable for DSI display,
to make the DSI code generic, so that, if required, the same code
can be re-used for future platforms with different mmio base.
Signed-off-by: Shashank Sharma <shashank.sharma@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
[danvet: Appease checkpatch.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We can apperently miss them, but breaking the entire driver hampers
testing. So bail out after one minute, our customerary "this is a lost
cause" timeout.
References: https://bugs.freedesktop.org/show_bug.cgi?id=78383
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
So far we used the wrong opcodes to access the DSI registers, so the
register writes during DSI programming didn't actually succeed and left
the registers unchanged. This wasn't a problem for the initial modeset,
where the BIOS-programmed values happened to work, but after resuming
from s0ix these registers are reset and failing to program them results
in a blank screen.
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
These opcodes are not specific for an endpoint, but are the same for all
endpoints. So rename them accordingly, using the name the VLV2 sideband
HAS uses. Also move the macros to the .c file, since they aren't used
anywhere else.
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Fixed several switch statements, curly braces, dereference operators
and keywords.
Signed-off-by: Robin Schroer <sulamiification@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Our two ->crtc_mode_set callbacks really don't care whether the fb is
pinned and set up already or not - all the state computation and
handling which originally looked at the framebuffer is already using
the indirection through the pipe configuration.
Eventually we want to move this up a bit more, but as long as the crtc
mode_set callback still exists (and as long as we don't need to pin an
entire pile of planes due to atomic modesets) there's not much point
in it. So I'll let this be for now.
v2: Don't forget about haswell ...
Reviewed-by: Akash Goel <akash.goel@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
A lot of the code in set_base is uncessary when the crtc is off, so we
can get rid of it all. Also, we don't need to call the fbc/psr update
functions since the crtc enable/disable hooks do that already.
The only things we really need are:
- Pin the new framebuffer and potentially unpin the old framebuffer
(if the crtc has been on and we only change the configuration).
- Update the plane registers.
The first step will move out of platform code with the very next
patch.
v2: Don't forget about haswell ...
Reviewed-by: Akash Goel <akash.goel@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
My plan here is to split up set_base into a prepare step, which does
the pinning, and a commit stage, which updates the hw state. Eventually
we should be able to move the prepare step at the beginning of any
atomic update. For now I only want to move the commit step into the
crtc_enable callbacks.
As a prep step sprinkle intel_edp_psr_update all over the place so
that we don't have to concern ourselves with that in the commit step.
v2: Rebase on top of Ville's enable/disable functions for all planes.
v3: Rebase more.
Reviewed-by: Akash Goel <akash.goel@intel.com> (v2)
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Just for consistency, this patch won't fix anything really.
v2: Rebase over all the recent plane enabling shuffling.
Reviewed-by: Akash Goel <akash.goel@intel.com> (v1)
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Way back we've used this to reject framebuffers with unsupported
pixel formats. But since the modesetting reorg with the compute
config stage we reject those much earlier and just BUG() in this
callback. So switch to a void return type.
Reviewed-by: Akash Goel <akash.goel@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
More fallout from
commit c8725f3dc0
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Mon Mar 17 12:21:55 2014 +0000
drm/i915: Do not call retire_requests from wait_for_rendering
is that we can completely fill all of memory using small objects, such
that we exhaust the filp space, and spend all of our time evicting
objects from the aperture. As such, we never fill the ring, and never
trigger the last resort flushing in
commit 1cf0ba1474
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Mon May 5 09:07:33 2014 +0100
drm/i915: Flush request queue when waiting for ring space
and so all the requests are left active and the objects keep that last
active reference. Eventually the system comes to a halt as it runs out
of memory.
The impact is mainly limited to test cases as regular userspace will
trigger retirement by manually checking whether an object is active.
Testcase: igt/gem_lut_handle
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=78724
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: Guo Jinxian <jinxianx.guo@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Since commit 691e6415c8
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Wed Apr 9 09:07:36 2014 +0100
drm/i915: Always use kref tracking for all contexts.
we have contexts everywhere, and so we must be careful to distinguish
fake contexts, which do not have an associated bo, and real ones, which
do. In particular, we now need to be careful not to dereference NULL
pointers.
This is one such example, as the commit highlighted above failed to move
the unpinning of the default ctx object into the real-context-only
branch.
Reported-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=78792
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Some platforms have a shared error interrupt, so if FIFO underrun
reporting gets disabled for one pipe/transcoder it gets disabled
for all pipes/transcoders.
When we disable FIFO underrun reporting we check whether the
interrupt was enabled or not. If it wasn't we might have missed
an underrun and we perform one last check right there. Currently
we print a debug message when an underrun is detect using this
mechanism. Promote the message to DRM_ERROR() to match the other
underrun error messages.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
By exporting the ability to map user address and inserting PTEs
representing their backing pages into the GTT, we can exploit UMA in order
to utilize normal application data as a texture source or even as a
render target (depending upon the capabilities of the chipset). This has
a number of uses, with zero-copy downloads to the GPU and efficient
readback making the intermixed streaming of CPU and GPU operations
fairly efficient. This ability has many widespread implications from
faster rendering of client-side software rasterisers (chromium),
mitigation of stalls due to read back (firefox) and to faster pipelining
of texture data (such as pixel buffer objects in GL or data blobs in CL).
v2: Compile with CONFIG_MMU_NOTIFIER
v3: We can sleep while performing invalidate-range, which we can utilise
to drop our page references prior to the kernel manipulating the vma
(for either discard or cloning) and so protect normal users.
v4: Only run the invalidate notifier if the range intercepts the bo.
v5: Prevent userspace from attempting to GTT mmap non-page aligned buffers
v6: Recheck after reacquire mutex for lost mmu.
v7: Fix implicit padding of ioctl struct by rounding to next 64bit boundary.
v8: Fix rebasing error after forwarding porting the back port.
v9: Limit the userptr to page aligned entries. We now expect userspace
to handle all the offset-in-page adjustments itself.
v10: Prevent vma from being copied across fork to avoid issues with cow.
v11: Drop vma behaviour changes -- locking is nigh on impossible.
Use a worker to load user pages to avoid lock inversions.
v12: Use get_task_mm()/mmput() for correct refcounting of mm.
v13: Use a worker to release the mmu_notifier to avoid lock inversion
v14: Decouple mmu_notifier from struct_mutex using a custom mmu_notifer
with its own locking and tree of objects for each mm/mmu_notifier.
v15: Prevent overlapping userptr objects, and invalidate all objects
within the mmu_notifier range
v16: Fix a typo for iterating over multiple objects in the range and
rearrange error path to destroy the mmu_notifier locklessly.
Also close a race between invalidate_range and the get_pages_worker.
v17: Close a race between get_pages_worker/invalidate_range and fresh
allocations of the same userptr range - and notice that
struct_mutex was presumed to be held when during creation it wasn't.
v18: Sigh. Fix the refactor of st_set_pages() to allocate enough memory
for the struct sg_table and to clear it before reporting an error.
v19: Always error out on read-only userptr requests as we don't have the
hardware infrastructure to support them at the moment.
v20: Refuse to implement read-only support until we have the required
infrastructure - but reserve the bit in flags for future use.
v21: use_mm() is not required for get_user_pages(). It is only meant to
be used to fix up the kernel thread's current->mm for use with
copy_user().
v22: Use sg_alloc_table_from_pages for that chunky feeling
v23: Export a function for sanity checking dma-buf rather than encode
userptr details elsewhere, and clean up comments based on
suggestions by Bradley.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: "Gong, Zhipeng" <zhipeng.gong@intel.com>
Cc: Akash Goel <akash.goel@intel.com>
Cc: "Volkin, Bradley D" <bradley.d.volkin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Reviewed-by: Brad Volkin <bradley.d.volkin@intel.com>
[danvet: Frob ioctl allocation to pick the next one - will cause a bit
of fuss with create2 apparently, but such are the rules.]
[danvet2: oops, forgot to git add after manual patch application]
[danvet3: Appease sparse.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Bit 31 in GEN6_PMINTRMSK is not an interrupt disable bit with gen8.
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Otherwise, we do a NULL pointer dereference.
I've seen this happen while handling an error in
i915_gem_object_pin_to_display_plane():
If i915_gem_object_set_cache_level() fails, we call is_pin_display()
to handle the error. At this point, the object is still not pinned
to GGTT and maybe not even bound, so we have to check before we
dereference its GGTT vma.
The IGT kms_flip/bo-too-big tests for this bug.
v2: Chris Wilson says restoring the old value is easier, but that
is_pin_display is useful as a theory of operation. Take the solomonic
decision: at least this way is_pin_display is a little more robust
(until Chris can kill it off).
v3: Chris suggests the WARN in i915_gem_obj_to_ggtt has outlived its
usefulness: add a reminder to remove it.
Issue: VIZ-3772
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Testcase: igt/kms_flip/bo-too-big
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Looking at our current dsi driver I note that:
- We don't have any slave driver right now.
- There's zero support for the hardware state readout and cross check
code.
- All the modeset state seems to be tracked in the intel_dsi structure
instead of the pipe config.
Given all that I can't properly audit the dsi ->mode_set callback. So
just do it as the first thing in the ->pre_pll_enable hook and hope
for the best.
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
A bit more care required here since there are some very few things
between the call to encoder->mode_set and encoder->pre_enable. But
they're either book-keeping or only matter for the vga port on the
pch. So of no concern.
Note that with the new sequence we write the infoframes after
selecting the clock source, but that shouldn't matter. I've simply
opted for this to have simpler code.
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
All the hard work was already done, only thing left to do is remove
the empty callback. And a now rather misleading comment I've spotted
while reading through code.
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Similar to dp the only thing we do is call intel_write_eld and prepare
a bit of state for the enable hooks. The only difference is that we
write that to the hardware instead of keeping track of it somewhere in
software.
Still we can just move all this to the very first enable hook.
Reviewed-by: Naresh Kumar Kachhi <naresh.kumar.kachhi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Those functions are only used on vlv platforms, so no need to check.
Especially if we're not too consistent about it.
Reviewed-by: Naresh Kumar Kachhi <naresh.kumar.kachhi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
With all the preceding refactoring the dp mode_set callback only
computes a bit of state (all derived from the pipe config) and also
writes the eld. As long as we do that before we enable the audio bit
or depend upon the correct value in intel_dp->DP we'll be fine.
No other hw state is touched.
We therefore only need to check that clearing intel_dp->DP is save.
Which it is since when we re-enable we already mask out all the bits
the link training code sets. And we need to keep on doing that so that
the re-train loop walking over pre-emph/voltage-swing values still
works properly.
Reviewed-by: Naresh Kumar Kachhi <naresh.kumar.kachhi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Only ilk/snb/ivb need the port A pll setup, so move it to the
pre_enable hook for those platforms. We can savely do this since on
those platforms there's nothing that touches the hardware between the
encoder->mode_set and the encoder->pre_enable calls.
Also add a comment that port A is ilk+ only.
Reviewed-by: Naresh Kumar Kachhi <naresh.kumar.kachhi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Including state readout and cross-checking. This allows us to get rid
of crtc->eld_vld on hsw+. It also means that fastboot will be unhappy
if the BIOS hasn't set up the audio routing like we want it too.
Wrt fastboot and external screens I see a few options:
- Don't.
- Try to fix up eld, infoframes and audio settings after the fact. But
that means some pretty extensive reworking of our code which
currently does all this while the pipe/port is still off.
I won't bother with converting SDVO over to this because the audio
support for SDVO is very lacking:
- We don't update the eld.
- We don't update the audio state on the sdvo encoder.
- We don't check whether the platform can even feed audio to the sdvo
encoder.
I've converted hdmi, dp & ddi all in one go since ddi needs both hdmi
and dp converted and so doing it step-by-step would have required a
few intermediate hacks.
Reviewed-by: Naresh Kumar Kachhi <naresh.kumar.kachhi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
There's no need to check whether audio is enabled (which for ddi ports
is done through the crtc->eld_vld flag) since at the cost of a
potentially unecessary register rmw cycle we can unconditionally do
this.
Note that the edp check is just paranoia since we won't ever call the
write_eld function for an edp panel.
Reviewed-by: Naresh Kumar Kachhi <naresh.kumar.kachhi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This way we can rely on the state cross-checker to have a bit
assurance that we'll get it right.
Reviewed-by: Naresh Kumar Kachhi <naresh.kumar.kachhi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
At least on those platforms which have a simple bit and don't rely
on the fully programmable CSC unit to do this.
Note that with the current code this includes CHV, but I guess that
platform will match BYT.
Reviewed-by: Naresh Kumar Kachhi <naresh.kumar.kachhi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We in the pre_enable hook we should only rely on the pipe config and
not on some other state set through properties or detect functions.
Reviewed-by: Naresh Kumar Kachhi <naresh.kumar.kachhi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Also add state readout and cross-check support. The only invasive change
is wiring up the new flag to the ->set_infoframes callbacks.
Reviewed-by: Naresh Kumar Kachhi <naresh.kumar.kachhi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
For compliance we really should be sending NULL infoframes always
when we detect a hdmi capable monitor. Also remove the now redudant
setting for the has_audio case and enforce that audio is only
possible with a hdmi sink.
Reviewed-by: Naresh Kumar Kachhi <naresh.kumar.kachhi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Everything should be in place so enable rc6/rps for bdw.
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Chris Wilson