GitLab

There's no reason to init a DP connector if the encoder just supports
HDMI: we'll just waste hundreds and hundreds of cycles trying to do DP
AUX transactions to detect if there's something there. Same goes for a
DP connector that doesn't support HDMI, but I'm not sure these
actually exist.

v2: - Use bit fields
    - Remove useless identation level
    - Replace DRM_ERROR with DRM_DEBUG_KMS
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@gmail.com> (v1)
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

We currently use the recommended values from BSpec, but the VBT
specifies the correct value to use for the hardware we have, so use
it. We also fall back to the recommended value in case we can't find
the VBT.

In addition, this code also provides some infrastructure to parse more
information about the DDI ports. There's a lot more information we
could extract and use in the future.

v2: - Move some code to init_vbt_defaults.
v3: - Rebase
    - Clarify the "DVO Port" matching code
v4: - Use I915_MAX_PORTS
    - Change the HAS_DDI checks
    - Replace DRM_ERROR with DRM_DEBUG_KMS
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

We currently treat the child_device_config as a simple struct, but
this is not correct: new BDB versions change the meaning of some
offsets, so the struct needs to be adjusted for each version.

Since there are too many changes (today we're in version 170!), making
a big versioned union would be too complicated, so child_device_config
is now a union of 3 things: (i) a "raw" byte array that's safe to use
anywhere; (ii)  an "old" structure that's the one we've been using and
should be safe to keep in the SDVO and TV code; and (iii) a "common"
structure that should contain only fields that are common for all the
known VBT versions.
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

VGA registers/memory live inside the the display power well. Add a power
domain for VGA.
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'd only ever used this define to denote whether or not we have the
dynamic parity feature (DPF) and never to determine whether or not L3
exists. Baytrail is a good example of where L3 exists, and not DPF.

This patch provides clarify in the code for future use cases which might
want to actually query whether or not L3 exists.

v2: Add /* DPF == dynamic parity feature */
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

On both Ivybridge and Haswell, row remapping information is saved and
restored with context. This means, we never actually properly supported
the l3 remapping because our sysfs interface is asynchronous (and not
tied to any context), and the known faulty HW would be reused by the
next context to run.

Not that due to the asynchronous nature of the sysfs entry, there is no
point modifying the registers for the existing context. Instead we set a
flag for all contexts to load the correct remapping information on the
next run. Interested clients can use debugfs to determine whether or not
the row has been remapped.

One could propose at this point that we just do the remapping in the
kernel. I guess since we have to maintain the sysfs interface anyway,
I'm not sure how useful it is, and I do like keeping the policy in
userspace; (it wasn't my original decision to make the
interface the way it is, so I'm not attached).

v2: Force a context switch when we have a remap on the next switch.
(Ville)
Don't let userspace use the interface with disabled contexts.

v3: Don't force a context switch, just let it nop
Improper context slice remap initialization, 1<<1 instead of 1<<i, but I
rewrote it to avoid a second round of confusion.
Error print moved to error path (All Ville)
Added a comment on why the slice remap initialization happens.

CC: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

I have implemented this patch before without creating a separate list
(I'm having trouble finding the links, but the messages ids are:
<1364942743-6041-2-git-send-email-ben@bwidawsk.net>
<1365118914-15753-9-git-send-email-ben@bwidawsk.net>)

However, the code is much simpler to just use a list and it makes the
code from the next patch a lot more pretty.

As you'll see in the next patch, the reason for this is to be able to
specify when a context needs to get L3 remapping. More details there.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Using LRI for setting the remapping registers allows us to stream l3
remapping information. This is necessary to handle per context remaps as
we'll see implemented in an upcoming patch.

Using the ring also means we don't need to frob the DOP clock gating
bits.

v2: Add comment about lack of worry for concurrent register access
(Daniel)
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
[danvet: Bikeshed the comment a bit by doing a s/XXX/Note - there's
nothing to fix.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Certain HSW SKUs have a second bank of L3. This L3 remapping has a
separate register set, and interrupt from the first "slice". A slice is
simply a term to define some subset of the GPU's l3 cache. This patch
implements both the interrupt handler, and ability to communicate with
userspace about this second slice.

v2:  Remove redundant check about non-existent slice.
Change warning about interrupts of unknown slices to WARN_ON_ONCE
Handle the case where we get 2 slice interrupts concurrently, and switch
the tracking of interrupts to be non-destructive (all Ville)
Don't enable/mask the second slice parity interrupt for ivb/vlv (even
though all docs I can find claim it's rsvd) (Ville + Bryan)
Keep BYT excluded from L3 parity

v3: Fix the slice = ffs to be decremented by one (found by Ville). When
I initially did my testing on the series, I was using 1-based slice
counting, so this code was correct. Not sure why my simpler tests that
I've been running since then didn't pick it up sooner.
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Now that adjusted_mode.clock no longer contains the pixel_multiplier, we
can kill the get_clock() callback and instead do the clock readout
in get_pipe_config().

Also i9xx_crtc_clock_get() can now extract the frequency of the PCH
DPLL, so use it to populate port_clock accurately for PCH encoders.
For DP in port A the encoder is still responsible for filling in
port_clock. The FDI adjusted_mode.clock extraction is kept in place
for some extra sanity checking, but we no longer need to pretend it's
also the port_clock.

In the encoder get_config() functions fill out adjusted_mode.clock
based on port_clock and other details such as the DP M/N values,
HDMI 12bpc and SDVO pixel_multiplier. For PCH encoders we will then
do an extra sanity check to make sure the dotclock we derived from
the FDI configuratiuon matches the one we derive from port_clock.

DVO doesn't exist on PCH platforms, so it doesn't need to anything
but assign adjusted_mode.clock=port_clock. And DDI is HSW only, so
none of the changes apply there.

v2: Use hdmi_reg color format to detect 12bpc HDMI case
v3: Set adjusted_mode.clock for LVDS too
v4: Rename ironlake_crtc_clock_get to ironlake_pch_clock_get,
    eliminate the useless link_freq variable.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

When reserving objects during execbuf, it is possible to come across an
object which will not fit given the current fragmentation of the address
space. We do not have any defragment in drm_mm, so the strategy is to
instead evict everything, and reallocate objects.

With the upcoming addition of multiple VMs, there is no point to evict
everything since doing so is overkill for the specific case mentioned
above.
Recommended-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
[danvet: One additional s/evict_everything/evict_vm/ to update a
comment in the code.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Passing the appropriate crtc to intel_update_watermarks() should help
in avoiding needless work in the future.

v2: Avoid clash with internal 'crtc' variable in some wm functions
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Score and action reveals what all the rings were doing
and why hang was declared. Add idle state so that
we can distinguish between waiting and idle ring.

v2: - add idle as a hangcheck action
    - consensed hangcheck status to single line (Chris)
    - mark active explicitly when we are making progress (Chris)
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Now when we have mechanism in place to track which context
was guilty of hanging the gpu, it is possible to punish
for bad behaviour.

If context has recently submitted a faulty batchbuffers guilty of
gpu hang and submits another batch which hangs gpu in quick
succession, ban it permanently. If ctx is banned, no more
batchbuffers will be queued for execution.

There is no need for global wedge machinery anymore and
it would be unwise to wedge the whole gpu if we have multiple
hanging batches queued for execution. Instead just ban
the guilty ones and carry on.

v2: Store guilty ban status bool in gpu_error instead of pointers
    that might become danling before hang is declared.

v3: Use return value for banned status instead of stashing state
    into gpu_error (Chris Wilson)

v4: - rebase on top of fixed hang stats api
    - add define for ban period
    - rename commit and improve commit msg

v5: - rely context banning instead of wedging the gpu
    - beautification and fix for ban calculation (Chris)
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>

The patch doesn't contain functional change, but is to prepare for
future platform which has different DPIO phy.  The additional pipe
parameter will use to select which phy to target for.

v2: Update the commit message and add static for the new function.
(Jani/Ville)
Signed-off-by: Chon Ming Lee <chon.ming.lee@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Notifying the bios lets it enter power saving states.
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

The bios interface seems messy, and it's hard to tell what the bios
really wants. At first, only add support for DDI based machines (hsw+),
and see how it turns out.

The spec says to notify prior to power down and after power up. It is
unclear whether it makes a difference.

v2:
 - squash notification function and callers patches together (Daniel)
 - move callers to haswell_crtc_{enable,disable} (Daniel)
 - rename notification function (Chris)

v3:
 - separate notification function and callers again, as it's not clear
   whether the display power state notification is the right thing to do
   after all

v4: per Paulo's review:
 - drop LVDS
 - WARN on unsupported encoder types
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

SWSCI is a driver to bios call interface.

This checks for SWSCI availability and bios requested callbacks, and
filters out any calls that shouldn't happen. This way the callers don't
need to do the checks all over the place.

v2: silence some checkpatch nagging

v3: set PCI_SWSCI bit 0 to trigger interrupt (Mengdong Lin)

v4: remove an extra #define (Jesse)

v5: spec says s/w is responsible for clearing PCI_SWSCI bit 0 too

v6: per Paulo's review and more:
 - fix sub-function mask
 - add exit parameter
 - add define for set panel details call
 - return more errors from swsci
 - clean up the supported/requested callbacks bit masks mess
 - use DSLP for timeout
 - fix build for CONFIG_ACPI=n

v7: tiny adjustment of requested vs. supported SBCB callbacks handling (Paulo)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

We still maintain code internally that cares about preliminary support.
Leaving the check here doesn't hurt anyone, and should keep things more
in line.

This time around, stick the info in the intel_info structure, and also
change the error from DRM_ERROR->DRM_INFO.

This is a partial revert of:
commit 590e4df8


Author: Jesse Barnes <jbarnes@virtuousgeek.org>
Date:   Wed May 8 10:45:15 2013 -0700

    drm/i915: VLV support is no longer preliminary

Daniel, I'll provide the fix ups for internal too if/when you merge
this (if you want).

Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Batchbuffers constructed by userspace can conditionalise their URB
allocations through the use of the MI_SET_PREDICATE command. This
command can read the MI_PREDICATE_RESULT_2 register to see how many
slices are enabled on GT3, and by virtue of the result, scale their
memory allocations to fit enabled memory.

Of course, this only works if the kernel sets the appropriate bit in the
register first.

v2: Better commit subject and message by Chris Wilson.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Credits-to: Yejun Guo <yejun.guo@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@gmail.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Initial parsing of the VBT MIPI block. For now, just store the panel id
if found.

Note: Again there seems to be no documentation for this piece of lore.
The doc situation for byt+ is just a bad joke :(
Signed-off-by: Shobhit Kumar <shobhit.kumar@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

For GPIO NC, CCK, CCU, and GPS CORE.
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Shobhit Kumar <shobhit.kumar@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

In the execbuf code we don't clean up any vmas which ended up not
getting bound for code simplicity. To make sure that we don't end up
creating multiple vma for the same vm kill the somewhat dangerous
vma_create function and inline it into lookup_or_create.

This is just a safety measure to prevent surprises in the future.

Also update the somewhat confused comment in the execbuf code and
clarify what kind of magic is going on with a new one.

v2: Keep the function separate as requested by Chris. But give it a __
prefix for paranoia and move it tighter together with the other vma
stuff.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ben Widawsky <ben@bwidawsk.net>
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

In order to transition more of our code over to using a VMA instead of
an <OBJ, VM> pair - we must have the vma accessible at execbuf time. Up
until now, we've only had a VMA when actually binding an object.

The previous patch helped handle the distinction on bound vs. unbound.
This patch will help us catch leaks, and other issues before we actually
shuffle a bunch of stuff around.

This attempts to convert all the execbuf code to speak in vmas. Since
the execbuf code is very self contained it was a nice isolated
conversion.

The meat of the code is about turning eb_objects into eb_vma, and then
wiring up the rest of the code to use vmas instead of obj, vm pairs.

Unfortunately, to do this, we must move the exec_list link from the obj
structure. This list is reused in the eviction code, so we must also
modify the eviction code to make this work.

WARNING: This patch makes an already hotly profiled path slower. The cost is
unavoidable. In reply to this mail, I will attach the extra data.

v2: Release table lock early, and two a 2 phase vma lookup to avoid
having to use a GFP_ATOMIC. (Chris)

v3: s/obj_exec_list/obj_exec_link/
Updates to address
commit 6d2b8885


Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Wed Aug 7 18:30:54 2013 +0100

    drm/i915: List objects allocated from stolen memory in debugfs

v4: Use obj = vma->obj for neatness in some places (Chris)
need_reloc_mappable() should return false if ppgtt (Chris)
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
[danvet: Split out prep patches. Also remove a FIXME comment which is
now taken care of.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Historically we've run our own driver hotplug handling in our own
work-queue, which then launched the drm core hotplug handling in the
system workqueue. This is important since we flush our own driver
workqueue in the pageflip code while hodling modeset locks, and only
the drm hotplug code grabbed these locks. But with

commit 69787f7d
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Tue Oct 23 18:23:34 2012 +0000

    drm: run the hpd irq event code directly

this was changed and now we could deadlock in our flip handler if
there's a hotplug work blocking the progress of the crucial unpin
works. So this broke the careful deadlock avoidance implemented in

commit b4a98e57


Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Thu Nov 1 09:26:26 2012 +0000

    drm/i915: Flush outstanding unpin tasks before pageflipping

Since the rule thus far has been that work items on our own workqueue
may never grab modeset locks simply restore that rule again.

v2: Add a comment to the declaration of dev_priv->wq to warn readers
about the tricky implications of using it. Suggested by Chris Wilson.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Stuart Abercrombie <sabercrombie@chromium.org>
Reported-by: Stuart Abercrombie <sabercrombie@chromium.org>
References: http://permalink.gmane.org/gmane.comp.freedesktop.xorg.drivers.intel/26239


Cc: stable@vger.kernel.org
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
[danvet: Squash in a comment at the place where we schedule the work.
Requested after-the-fact by Chris on irc since the hpd work isn't the
only place we botch this.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

This lets drivers see the flags requested by the application

[airlied: fixup for rcar/imx/msm]
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Dave Airlie <airlied@gmail.com>

We currently only enter PC8+ after all its required conditions are
met, there's no rendering, and we stay like that for at least 5
seconds.

I chose "5 seconds" because this value is conservative and won't make
us enter/leave PC8+ thousands of times after the screen is off: some
desktop environments have applications that wake up and do rendering
every 1-3 seconds, even when the screen is off and the machine is
completely idle.

But when I was testing my PC8+ patches I set the default value to
100ms so I could use the bad-behaving desktop environments to
stress-test my patches. I also thought it would be a good idea to ask
our power management team to test different values, but I'm pretty
sure they would ask me for an easy way to change the timeout. So to
help these 2 cases I decided to create an option that would make it
easier to change the default value. I also expect people making
specific products that use our driver could try to find the perfect
timeout for them.

Anyway, fixing the bad-behaving applications will always lead to
better power savings than just changing the timeout value: you need to
stop waking the Kernel, not quickly put it back to sleep again after
you wake it for nothing. Bad sleep leads to bad mood!
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

This patch allows PC8+ states on Haswell. These states can only be
reached when all the display outputs are disabled, and they allow some
more power savings.

The fact that the graphics device is allowing PC8+ doesn't mean that
the machine will actually enter PC8+: all the other devices also need
to allow PC8+.

For now this option is disabled by default. You need i915.allow_pc8=1
if you want it.

This patch adds a big comment inside i915_drv.h explaining how it
works and how it tracks things. Read it.

v2: (this is not really v2, many previous versions were already sent,
     but they had different names)
    - Use the new functions to enable/disable GTIMR and GEN6_PMIMR
    - Rename almost all variables and functions to names suggested by
      Chris
    - More WARNs on the IRQ handling code
    - Also disable PC8 when there's GPU work to do (thanks to Ben for
      the help on this), so apps can run caster
    - Enable PC8 on a delayed work function that is delayed for 5
      seconds. This makes sure we only enable PC8+ if we're really
      idle
    - Make sure we're not in PC8+ when suspending
v3: - WARN if IRQs are disabled on __wait_seqno
    - Replace some DRM_ERRORs with WARNs
    - Fix calls to restore GT and PM interrupts
    - Use intel_mark_busy instead of intel_ring_advance to disable PC8
v4: - Use the force_wake, Luke!
v5: - Remove the "IIR is not zero" WARNs
    - Move the force_wake chunk to its own patch
    - Only restore what's missing from RC6, not everything
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Just like irq_mask and gt_irq_mask, use it to track the status of
GEN6_PMIMR so we don't need to read it again every time we call
snb_update_pm_irq.
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

This define hasn't been used since:

  commit cfdf1fa2


  Author: Kristian Høgsberg <krh@bitplanet.net>
  Date:   Wed Dec 16 15:16:16 2009 -0500

      drm/i915: Implement IS_* macros using static tables
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

The code using this was removed in:

  commit 88f23b8f


  Author: Chris Wilson <chris@chris-wilson.co.uk>
  Date:   Sun Dec 5 15:08:31 2010 +0000

      drm/i915: Avoid using PIPE_CONTROL on Ironlake
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

This define hasn't been used since:

  commit 652c393a


  Author: Jesse Barnes <jbarnes@virtuousgeek.org>
  Date:   Mon Aug 17 13:31:43 2009 -0700

      drm/i915: add dynamic clock frequency control
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

In the new execbuf code we want to track buffers using the vmas even
before they're all properly mapped. Which means that bind_to_vm needs
to deal with buffers which have preallocated vmas which aren't yet
bound.

This patch implements this prep work and adjusts our WARN/BUG checks.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
[danvet: Split out from Ben's big execbuf patch. Also move one BUG
back to its original place to deflate the diff a notch.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

The execbuf wants to do relocations usings vmas, so we need a
vma->exec_list. The eviction code also uses the old obj execbuf list
for it's own book-keeping, but would really prefer to deal in vmas
only. So switch it over to the new list.

Again this is just a prep patch for the big execbuf vma conversion.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
[danvet: Split out from Ben's big execbuf vma patch.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

To convert the execbuf code over to use vmas natively we need to
shuffle the exec_list a bit. This patch here just prepares things with
the debugfs code, which also uses the old exec_list list_head, newly
called obj_exec_link.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
[danvet: Split out from Ben's big patch.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

The machines that fall in this category are the SDVs that have a PCI
ID starting with 0x0C. These are very early pre-production machines
and may not fully work. Other Haswell SDVs have PCI IDs that match the
real Haswell machines and we expect them to work better.

Even though they have problems, they still mostly work so I don't see
a reason to refuse loading our driver. But I do see a reason to reject
bug reports from these machines, so the message should help the bug
triagers.

As far as I know, we don't implement some workarounds that are
specific to these machines and suspend/resume may not work on most of
them, but besides this, they may work.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=61508

Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Haswell GT3e has the unique feature of supporting Write-Through cacheing
of objects within the eLLC/LLC. The purpose of this is to enable the display
plane to remain coherent whilst objects lie resident in the eLLC/LLC - so
that we, in theory, get the best of both worlds, perfect display and fast
access.

However, we still need to be careful as the CPU does not see the WT when
accessing the cache. In particular, this means that we need to flush the
cache lines after writing to an object through the CPU, and on
transitioning from a cached state to WT.

v2: Actually do the clflush on transition to WT, nagging by Ville.
v3: Flush the CPU cache after writes into WT objects.
v4: Rease onto LLC updates and report WT as "uncached" for
get_cache_level_ioctl to remain symmetric with set_cache_level_ioctl.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Now that we skip clflushes more often, return a boolean indicating
whether the clflush was actually performed, and only if it was do the
chipset flush. (Though on most of the architectures where the clflush will
be skipped, the chipset flush is a no-op!)
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
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>

As mentioned in the previous commit, reads and writes from both the CPU
and GPU go through the LLC. This gives us coherency between the CPU and
GPU irrespective of the attribute settings either device sets. We can
use to avoid having to clflush even uncached memory.

Except for the scanout.

The scanout resides within another functional block that does not use
the LLC but reads directly from main memory. So in order to maintain
coherency with the scanout, writes to uncached memory must be flushed.
In order to optimize writes elsewhere, we start tracking whether an
framebuffer is attached to an object.

v2: Use pin_display tracking rather than fb_count (to ensure we flush
cursors as well etc) and only force the clflush along explicit writes to
the scanout paths (i.e. pin_to_display_plane and pwrite into scanout).

v3: Force the flush after hitting the slowpath in pwrite, as after
dropping the lock the object's cache domain may be invalidated. (Ville)

Based on a patch by Ville Syrjälä.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
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>

The display engine has unique coherency rules such that it requires
special handling to ensure that all writes to cursors, scanouts and
sprites are clflushed. This patch introduces the infrastructure to
simply track when an object is being accessed by the display engine.

v2: Explain the is_pin_display() magic as the sources for obj->pin_count
and their individual rules is not obvious. (Ville)
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>