Emil Velikov [Thu, 11 Jul 2013 16:28:03 +0000 (17:28 +0100)]
drm/nv50/gpio: post-nv92 cards have 32 interrupt lines
Since the original merge of nouveau to upstream kernel, we were assuming
that nv90 (and later) cards have 32 lines.
Based on mmio traces of the binary driver, as well as PBUS error messages
during read/write of the e070/e074 registers, we can conclude that nv92
has only 16 lines whereas nv94 (and later) cards have 32.
Reported-and-tested-by: David M. Lloyd <david.lloyd@redhat.com> Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com> Cc: dri-devel@lists.freedesktop.org Cc: Ben Skeggs <bskeggs@redhat.com> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ilia Mirkin [Fri, 19 Jul 2013 10:27:45 +0000 (06:27 -0400)]
drm/nouveau/core: xtensa firmware size needs to be 0x40000 no matter what
The current logic is wrong since we send fw->size >> 8 to the
card. Rounding the size up by 0x100 and 0x1000 didn't seem to help,
the card still hung, so go back to what the blob does -- 0x40000.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
The API allows up to 64-bits allocations, but size is handled as int
inside nouveau almost everywhere. Until this is fixed it's better to
prevent negative sizes.
The 256 kB before INT_MAX is paranoia, because of the large page
aligning below that could flip it above INT_MAX.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 9 Jul 2013 04:20:15 +0000 (14:20 +1000)]
drm/nouveau: use dedicated channel for async moves on GT/GF chipsets.
The moves themselves were generally async to graphics previously, with
the exception that if the "main" channel is used to synchronise a
page flip at the same time, it can end up blocked for a noticable amount
of time for large buffer moves.
Not really critical, and there's better ways of handling this, but they
are all rather invasive, so this is fine for now.
Based on a patch by Maarten Lankhorst addressing the same issue.
Ben Skeggs [Tue, 9 Jul 2013 02:35:55 +0000 (12:35 +1000)]
drm/nv50/kms: fix pin refcnt leaks
Weren't critical previously, the buffers would go away anyway. But with
recent changes to core drm/ttm lockdep will get pissed off now, so let's
fix it.
Ben Skeggs [Mon, 8 Jul 2013 04:15:51 +0000 (14:15 +1000)]
drm/nouveau: fix locking issues in page flipping paths
b580c9e2b7ba5030a795aa2fb73b796523d65a78 introduced additional problems
while trying to solve issues that became apparent while porting to the
new reservation stuff.
The major problem was that the the previously mentioned patch took the
client mutex earlier than previously, but the pinning of new_bo can
can potentially cause a buffer move, which would result in attempting to
acquire the same mutex again.
This commit attempts to fix that "fix".
Thanks to Maarten for the tips on keeping lockdep happy and cooking :)
"drm/nve0-/gr: some new gpc registers can have multiple copies" 5ee86c4190f9e caused a regression for nvc0, because the bit indicating last
transfer has occured was no longer set, resulting in random system lockups.
Reported-by: Ronald Uitermark <ronald645@gmail.com> Tested-by: Ronald Uitermark <ronald645@gmail.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ilia Mirkin [Wed, 3 Jul 2013 07:06:02 +0000 (03:06 -0400)]
drm/nva3/disp: Fix HDMI audio regression
This is the nva3 counterpart to commit beba44b17 (drm/nv84/disp: Fix
HDMI audio regression). The regression happened as a result of
refactoring in commit 8e9e3d2de (drm/nv84/disp: move hdmi control into
core).
Reported-and-tested-by: Max Baldwin <archerseven@gmail.com> Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu> Signed-off-by: Ben Skeggs <bskeggs@redhat.com> Cc: stable@vger.kernel.org
drm/nv50-/disp: initial supervisor support for off-chip encoders
changed the write mask in one of the interrupt functions for on-chip encoders,
causing a regression in certain VGA dual-head setups. This commit reintroduces
the mask thus resolving the regression
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66129 Reported-and-Tested-by: Yves-Alexis <corsac@debian.org> Cc: stable@vger.kernel.org [3.9+] CC: Ben Skeggs <bskeggs@redhat.com> Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com> Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Fri, 5 Jul 2013 00:26:20 +0000 (10:26 +1000)]
drm/nvc0-/gr: factor out yet more unknown magic into versioned functions
NVC1/NVD9 are the only chipsets that should have anything different
happen on them after this. We previously weren't doing these
register modifications, and NVIDIA do.
Dave Airlie [Fri, 5 Jul 2013 00:55:33 +0000 (10:55 +1000)]
Merge branch 'qxl-next' of git://people.freedesktop.org/~airlied/linux into drm-next
Adds 3 features that UMS had to the KMS driver.
dynamic resizing - resizing remote-viewer makes guest resize
multiple crtcs - remote-viewer can access > 1 crtc.
suspend/resume/hibernate: guests can do suspend/resume/hibernate now.
* 'qxl-next' of git://people.freedesktop.org/~airlied/linux:
qxl: use drm helper hotplug support
qxl: add suspend/resume/hibernate support.
qxl: add fb and ttm entry points for use by suspend/resume.
qxl: add ring prep code for s/r
qxl: prepare memslot code for suspend/resume
qxl: split monitors_config object creation out.
drm/qxl: set time on drawables from userspace
drm/qxl: add support for > 1 output
drm/qxl: make dynamic resizing work properly.
Dave Airlie [Mon, 1 Jul 2013 04:14:38 +0000 (14:14 +1000)]
drm/qxl: make dynamic resizing work properly.
qxl has a feature to allow the userspace driver do arbitrary resizes
when the viewer resizes, this fixes it by removing unnecessary code
from the kernel side.
Daniel Vetter [Mon, 1 Jul 2013 20:05:53 +0000 (22:05 +0200)]
drm/mm: kill color_search_free/get_block
drm/i915 is the only user of the color allocation handling and
switched to insert_node a while ago. So we can ditch this.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Dave Airlie <airlied@redhat.com>
Seung-Woo Kim [Mon, 1 Jul 2013 10:44:14 +0000 (19:44 +0900)]
drm: fix print format of sequence in trace point
seq of a trace point is unsigned int but print format was %d. So
it fixes the format as %u.
Signed-off-by: Seung-Woo Kim <sw0312.kim@samsung.com> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Dave Airlie <airlied@redhat.com>
Seung-Woo Kim [Tue, 2 Jul 2013 00:53:28 +0000 (09:53 +0900)]
drm: fix error routines in drm_open_helper
There are missing parts to handle error in drm_open_helper().
The priv->minor, assigned by idr_find() which can return NULL,
should be checked whether it is NULL or not before referencing it.
put_pid(), drm_gem_release(), and drm_prime_destory_file_private()
should be called when error happens after their pair functions are
called. If an error occurs after executing dev->driver->open()
which allocates driver specific per-file private data, then the
private data should be released.
Signed-off-by: YoungJun Cho <yj44.cho@samsung.com> Signed-off-by: Seung-Woo Kim <sw0312.kim@samsung.com> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com> Reviewed-by: Chris Wilson <chris-wilson.co.uk> Signed-off-by: Dave Airlie <airlied@redhat.com>
Seung-Woo Kim [Tue, 2 Jul 2013 08:57:04 +0000 (17:57 +0900)]
drm: add assertion for checking null edid to drm_edid_block_valid
If raw_edid of drm_edid_block_vaild() is null, it will crash, so
checking in bad label is removed and instead assertion is added at
the top of the function.
The type of return for the function is bool, so it fixes to return
true and false instead of 1 and 0.
Signed-off-by: Seung-Woo Kim <sw0312.kim@samsung.com> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Dave Airlie <airlied@redhat.com>
drm: Convert drm class driver from legacy pm ops to dev_pm_ops
Convert drivers/gpu/drm class to use dev_pm_ops for power management and
remove Legacy PM ops hooks. With this change, drm class registers
suspend/resume callbacks via class->pm (dev_pm_ops) instead of Legacy
class->suspend/resume. When __device_suspend() runs call-backs, it will
find class->pm ops for the drm class.
drm_class_suspend() hook calls driver legacy ops with the state information.
e.g: drm_class_suspend() calls into driver suspend routines
via drm_dev->driver->suspend(drm_dev, state).
Once drm_class_suspend() is converted to dev_pm_ops, it will no longer
have access to pm_transition which it has to pass into driver legacy
suspend calls. A new freeze and suspend hooks are added to address the not
having access to the state information. The new freeze and suspend hooks
simply call __drm_class_suspend() with the appropriate pm state information.
__drm_class_suspend() is the original suspend hook with a new name.
Signed-off-by: Shuah Khan <shuah.kh@samsung.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Thu, 4 Jul 2013 00:46:17 +0000 (10:46 +1000)]
Merge tag 'drm-intel-fixes-2013-07-03' of git://people.freedesktop.org/~danvet/drm-intel into drm-next
Pile of fixes for 3.11. A bit large in patch count, but that's simply due
to two fixes being split up into really small parts. Also I've included a
few more vlv patches than I'd have included for other platforms. But since
vlv is officially supported for the first time only in 3.11 that shouldn't
result in unbearable risks.
Highlights:
- ghost eDP fixes for hsw from Paulo
- fix PCH detection in virtualized enviroments (Rui Guo)
- duct-tape dma sg construction when swiotlb is in use (Konrad), dupe with
a patch in your drm-fixes branch
- fix sdvo hotplug on i965g
- tune down a bunch of dmesg ERRORs which can be hit under normal
conditions
- detect invalid pitches for tiled scanout buffers (Chris)
- a pile of vlv fixes from Ville: rps improvements, fixes for the dpll
LPF, fixup the sprite mmio offsets
- fix context size on hsw (Ben)
- locking fixes for the hotplug code, specifically the storm handling
- fix get_config on CPT (Xiong Zhang)
- Fix the domain tracking when an unlocked seqno wait was interrupt
(Chris), this seems to explain tons of little corruption bugs in the
ddx. Chris also added a nice igt to exercise this.
- work around stack-corrupting vnsprintf in our error state dumper
* tag 'drm-intel-fixes-2013-07-03' of git://people.freedesktop.org/~danvet/drm-intel: (39 commits)
drm/i915: Don't try to tear down the stolen drm_mm if it's not there
drm/i915: Break up the large vsnprintf() in print_error_buffers()
drm/i915: Refactor the wait_rendering completion into a common routine
drm/i915: Only clear write-domains after a successful wait-seqno
drm/i915: correct intel_dp_get_config() function for DevCPT
drm/i915: fix hpd interrupt register locking
drm/i915: fold the no-irq check into intel_hpd_irq_handler
drm/i915: fold the queue_work into intel_hpd_irq_handler
drm/i915: fold the hpd_irq_setup call into intel_hpd_irq_handler
drm/i915: s/hotplug_irq_storm_detect/intel_hpd_irq_handler/
drm/i915: close tiny race in the ilk pcu even interrupt setup
drm/i915: fix locking around ironlake_enable|disable_display_irq
drm/i915: Fix context sizes on HSW
drm/i915: Fix VLV sprite register offsets
Revert "drm/i915: Don't use the HDMI port color range bit on Valleyview"
drm/i915: s/LFP/LPF in DPIO PLL register names
drm/i915: Fix VLV PLL LPF coefficients for DAC
drm/i915: Jump to at least RPe on VLV when increasing the GPU frequency
drm/i915: Don't increase the GPU frequency from the delayed VLV rps timer
drm/i915: GEN6_RP_INTERRUPT_LIMITS doesn't seem to exist on VLV
...
Dave Airlie [Thu, 4 Jul 2013 00:44:06 +0000 (10:44 +1000)]
Merge branch 'drm-next-3.11' of git://people.freedesktop.org/~agd5f/linux into drm-next
A few more DPM fixes.
* 'drm-next-3.11' of git://people.freedesktop.org/~agd5f/linux:
drm/radeon/sumo: implement support for disable_gfx_power_gating_in_uvd flag
drm/radeon/tn: disable PG when changing UVD clocks
drm/radeon/sumo: disable PG when changing UVD clocks
drm/radeon/aruba: disable additional rlc features
drm/radeon: fix endian bug in radeon_atom_get_mclk_range_table()
drm/radeon/dpm: fix compilation with certain versions of gcc
drm/radeon/dpm: clarify debugfs warning
Daniel Vetter [Mon, 1 Jul 2013 20:01:04 +0000 (22:01 +0200)]
drm/mm: WARN for unclean mm takedown
The usual drm driver has tons of different drm_mm memory managers so the drm
error message in dmesg is pretty useless. WARN instead so that we have the full
backtrace.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Ben Widawsky <ben@bwidawsk.net> Signed-off-by: Dave Airlie <airlied@redhat.com>
the failure case handling was fixed up. But in the case when we
already had the buffer exported it changed the return value:
Previously we've return 0 on success, now we return the fd.
This ABI change has been caught by i-g-t/prime_self_import/with_one_bo.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66436 Cc: YoungJun Cho <yj44.cho@samsung.com> Cc: Seung-Woo Kim <sw0312.kim@samsung.com> Cc: Kyungmin Park <kyungmin.park@samsung.com> Tested-by: lu hua <huax.lu@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: YoungJun Cho <yj44.cho@samsung.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
Daniel Vetter [Tue, 2 Jul 2013 08:48:31 +0000 (10:48 +0200)]
drm/i915: Don't try to tear down the stolen drm_mm if it's not there
Every other place properly checks whether we've managed to set
up the stolen allocator at boot-up properly, with the exception
of the cleanup code. Which results in an ugly
*ERROR* Memory manager not clean. Delaying takedown
at module unload time since the drm_mm isn't initialized at all.
v2: While at it check whether the stolen drm_mm is initialized instead
of the more obscure stolen_base == 0 check.
v3: Fix up the logic. Also we need to keep the stolen_base check in
i915_gem_object_create_stolen_for_preallocated since that can be
called before stolen memory is fully set up. Spotted by Chris Wilson.
v4: Readd the conversion in i915_gem_object_create_stolen_for_preallocated,
the check is for the dev_priv->mm.gtt_space drm_mm, the stolen
allocatot must already be initialized when calling that function (if
we indeed have stolen memory).
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=65953 Cc: Chris Wilson <chris@chris-wilson.co.uk> Tested-by: lu hua <huax.lu@intel.com> (v3) Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
David Herrmann [Mon, 1 Jul 2013 18:32:58 +0000 (20:32 +0200)]
drm: make drm_mm_init() return void
There is no reason to return "int" as this function never fails.
Furthermore, several drivers (ast, sis) already depend on this.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Tue, 2 Jul 2013 03:31:26 +0000 (13:31 +1000)]
Merge branch 'drm-next-3.11' of git://people.freedesktop.org/~agd5f/linux into drm-next
A few more patches for 3.11:
- add debugfs interface to check current DPM state
- Fix a bug that caused problems with DPM on BTC+ asics.
* 'drm-next-3.11' of git://people.freedesktop.org/~agd5f/linux:
drm/radeon/dpm: add debugfs support for SI
drm/radeon/dpm: add debugfs support for cayman
drm/radeon/dpm: add debugfs support for TN
drm/radeon/dpm: add debugfs support for ON/LN
drm/radeon/dpm: add debugfs support for 7xx/evergreen/btc
drm/radeon/dpm: add debugfs support for rv6xx
drm/radeon/dpm: add infrastructure to support debugfs info
drm/radeon/dpm: re-enable state transitions for Cayman
drm/radeon/dpm: re-enable state transitions for BTC
drm/radeon: fix typo in radeon_atom_init_mc_reg_table()
drm/radeon/atom: fix endian bug in radeon_atom_init_mc_reg_table()
drm/radeon: remove sumo dpm/uvd bringup leftovers
Chris Wilson [Sat, 29 Jun 2013 22:26:50 +0000 (23:26 +0100)]
drm/i915: Break up the large vsnprintf() in print_error_buffers()
So it appears that I have encountered some bogosity when trying to call
i915_error_printf() with many arguments from print_error_buffers(). The
symptom is that the vsnprintf parser tries to interpret an integer arg
as a character string, the resulting OOPS indicating stack corruption.
Replacing the single call with its 13 format specifiers and arguments
with multiple calls to i915_error_printf() worked fine. This patch goes
one step further and introduced i915_error_puts() to pass the strings
simply.
It may not fix the root cause, but it does prevent my box from dying and
I think helps make print_error_buffers() more friendly.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66077 Cc: Mika Kuoppala <mika.kuoppala@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Chris Wilson [Sat, 29 Jun 2013 21:05:26 +0000 (22:05 +0100)]
drm/i915: Refactor the wait_rendering completion into a common routine
Harmonise the completion logic between the non-blocking and normal
wait_rendering paths, and move that logic into a common function.
In the process, we note that the last_write_seqno is by definition the
earlier of the two read/write seqnos and so all successful waits will
have passed the last_write_seqno. Therefore we can unconditionally clear
the write seqno and its domains in the completion logic.
v2: Add the missing ring parameter, because sometimes it is good to have
things compile.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Chris Wilson [Fri, 28 Jun 2013 15:54:08 +0000 (16:54 +0100)]
drm/i915: Only clear write-domains after a successful wait-seqno
In the introduction of the non-blocking wait, I cut'n'pasted the wait
completion code from normal locked path. Unfortunately, this neglected
that the normal path returned early if the wait returned early. The
result is that read-only waits may return whilst the GPU is still
writing to the bo.
Fixes regression from
commit 3236f57a0162391f84b93f39fc1882c49a8998c7 [v3.7]
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Fri Aug 24 09:35:09 2012 +0100
drm/i915: Use a non-blocking wait for set-to-domain ioctl
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66163 Cc: stable@vger.kernel.org Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Xiong Zhang [Fri, 28 Jun 2013 04:59:06 +0000 (12:59 +0800)]
drm/i915: correct intel_dp_get_config() function for DevCPT
On DevCPT, the control register for Transcoder DP Sync Polarity is
TRANS_DP_CTL, not DP_CTL.
Without this patch, Many call trace occur on CPT machine with DP monitor.
The call trace is like: *ERROR* mismatch in adjusted_mode.flags(expected X,found X)
v2: use intel-crtc to simple patch, suggested by Daniel.
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.com>
[danvet: Extend the encoder->get_config comment to specify that we now
also depend upon intel_encoder->base.crtc being correct. Also bikeshed
s/intel_crtc/crtc/.]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=65287 Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Thu, 27 Jun 2013 15:52:15 +0000 (17:52 +0200)]
drm/i915: fix hpd interrupt register locking
Our interrupt handler (in hardirq context) could race with the timer
(in softirq context), hence we need to hold the spinlock around the
call to ->hdp_irq_setup in intel_hpd_irq_handler, too.
But as an optimization (and more so to clarify things) we don't need
to do the irqsave/restore dance in the hardirq context.
Note also that on ilk+ the race isn't just against the hotplug
reenable timer, but also against the fifo underrun reporting. That one
also modifies the SDEIMR register (again protected by the same
dev_priv->irq_lock).
To lock things down again sprinkle a assert_spin_locked. But exclude
the functions touching SDEIMR for now, I want to extract them all into
a new helper function (like we do already for pipestate, display
interrupts and all the various gt interrupts).
v2: Add the missing 't' Egbert spotted in a comment.
v3: Actually fix the right misspelled comment (Paulo).
Cc: Egbert Eich <eich@suse.de> Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Thu, 27 Jun 2013 15:52:14 +0000 (17:52 +0200)]
drm/i915: fold the no-irq check into intel_hpd_irq_handler
The usual pattern for our sub-function irq_handlers is that they check
for the no-irq case themselves. This results in more streamlined code
in the upper irq handlers.
v2: Rebase on top of the i965g/gm sdvo hpd fix.
Cc: Egbert Eich <eich@suse.de> Reviewed-by: Egbert Eich <eich@suse.de> Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Thu, 27 Jun 2013 15:52:13 +0000 (17:52 +0200)]
drm/i915: fold the queue_work into intel_hpd_irq_handler
Everywhere the same.
Note that this patch leaves unnecessary braces behind, but the next
patch will kill those all anyway (including the if itself) so I've
figured I can keep the diff a bit smaller.
v2: Rebase on top of the i965g/gm sdvo hpd fix.
Cc: Egbert Eich <eich@suse.de> Reviewed-by: Egbert Eich <eich@suse.de> Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The combination of Paulo's fifo underrun detection code and Egbert's
hpd storm handling code unfortunately made the hpd storm handling code
racy.
To avoid duplicating tricky interrupt locking code over all platforms
start with a bit of refactoring. This patch is the very first step
since in the end the irq storm handling code will handle all hotplug
logic (and so also encapsulate the locking nicely).
v2: Rebase on top of the i965g/gm sdvo hpd fix.
Cc: Egbert Eich <eich@suse.de> Reviewed-by: Egbert Eich <eich@suse.de> Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Thu, 27 Jun 2013 11:44:59 +0000 (13:44 +0200)]
drm/i915: close tiny race in the ilk pcu even interrupt setup
By the time we write DEIER in the postinstall hook the interrupt
handler could run any time. And it does modify DEIER to handle
interrupts.
Hence the DEIER read-modify-write cycle for enabling the PCU event
source is racy. Close this races the same way we handle vblank
interrupts: Unconditionally enable the interrupt in the IER register,
but conditionally mask it in IMR. The later poses no such race since
the interrupt handler does not touch DEIMR.
Also update the comment, the clearing has already happened
unconditionally above.
v2: Actually shove the updated comment into the right train^W commit,
as spotted by Paulo.
Cc: Paulo Zanoni <przanoni@gmail.com> Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Thu, 27 Jun 2013 11:44:58 +0000 (13:44 +0200)]
drm/i915: fix locking around ironlake_enable|disable_display_irq
The haswell unclaimed register handling code forgot to take the
spinlock. Since this is in the context of the non-rentrant interupt
handler and we only have one interrupt handler it is sufficient to
just grab the spinlock - we do not need to exclude any other
interrupts from running on the same cpu.
To prevent such gaffles in the future sprinkle assert_spin_locked over
these functions. Unfornately this requires us to hold the spinlock in
the ironlake postinstall hook where it is not strictly required:
Currently that is run in single-threaded context and with userspace
exlcuded from running concurrent ioctls. Add a comment explaining
this.
v2: ivb_can_enable_err_int also needs to be protected by the spinlock.
To ensure this won't happen in the future again also sprinkle a
spinlock assert in there.
v3: Kill the 2nd call to ivb_can_enable_err_int I've accidentally left
behind, spotted by Paulo.
Cc: Paulo Zanoni <przanoni@gmail.com> Reviewed-by: Paulo Zanoni <przanoni@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Ben Widawsky [Wed, 26 Jun 2013 04:53:40 +0000 (21:53 -0700)]
drm/i915: Fix context sizes on HSW
With updates to the spec, we can actually see the context layout, and
how many dwords are allocated. That table suggests we need 70720 bytes
per HW context. Rounded up, this is 18 pages. Looking at what lives
after the current 4 pages we use, I can't see too much important (mostly
it's d3d related), but there are a couple of things which look scary. I
am hopeful this can explain some of our odd HSW failures.
v2: Make the context only 17 pages. The power context space isn't used
ever, and execlists aren't used in our driver, making the actual total
66944 bytes.
v3: Add a comment to the code. (Jesse & Paulo)
Reported-by: "Azad, Vinit" <vinit.azad@intel.com> Cc: stable@vger.kernel.org Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Ben Widawsky <ben@bwidawsk.net> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Ville Syrjälä [Tue, 25 Jun 2013 11:16:34 +0000 (14:16 +0300)]
Revert "drm/i915: Don't use the HDMI port color range bit on Valleyview"
The PIPECONF color range bit doesn't appear to be effective, on HDMI
outputs at least. The color range bit in the port register works though,
so let's use it.
I have not yet verified whether the PIPECONF bit works on DP outputs.
Ville Syrjälä [Fri, 14 Jun 2013 11:02:52 +0000 (14:02 +0300)]
drm/i915: Fix VLV PLL LPF coefficients for DAC
The current PLL settings produce a rather unstable picture when
I hook up a VLV to my HP ZR24w display via a VGA cable.
According to VLV2A0_DP_eDP_HDMI_DPIO_driver_vbios_notes_9, we should
use the the same LPF coefficients for DAC as we do for HDMI and RBR DP.
And indeed that seems to cure the shivers.
v2: Add the name of the relevant document to the commit message
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>
Ville Syrjälä [Tue, 25 Jun 2013 18:38:11 +0000 (21:38 +0300)]
drm/i915: Jump to at least RPe on VLV when increasing the GPU frequency
If the current GPU frquency is below RPe, and we're asked to increase
it, just go directly to RPe. This should provide better performance
faster than letting the frequency trickle up in response to the up
threshold interrupts.
For now just do it for VLV, since that matches quite closely how VLV
used to operate when the rps delayed timer kept things at RPe always.
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>
Ville Syrjälä [Tue, 25 Jun 2013 18:38:10 +0000 (21:38 +0300)]
drm/i915: Don't increase the GPU frequency from the delayed VLV rps timer
There's little point in increasing the GPU frequency from the delayed
rps work on VLV. Now when the GPU is idle, the GPU frequency actually
keeps dropping gradually until it hits the minimum, whereas previously
it just ping-ponged constantly between RPe and RPe-1.
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>
Ville Syrjälä [Tue, 25 Jun 2013 16:21:06 +0000 (19:21 +0300)]
drm/i915: GEN6_RP_INTERRUPT_LIMITS doesn't seem to exist on VLV
I can't find GEN6_RP_INTERRUPT_LIMITS (0xA014) anywhere in VLV docs.
Reading it always returns zero from what I can tell, and eliminating
it doesn't seem to make any difference to the behaviour of the system.
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>
Ville Syrjälä [Tue, 25 Jun 2013 16:21:02 +0000 (19:21 +0300)]
drm/i915: Don't wait for Punit after each freq change on VLV
It seems that even though Punit reports the frequency change to have
been completed, it still reports the old frequency in the status
register for some time.
So rather than polling for Punit to complete the frequency change after
each request, poll before. This gets rid of the spurious "Punit overrode
GPU freq" messages.
This also lets us continue working while Punit is performing the actual
frequency change. As a result, openarena demo088-test1 timedemo average
fps is increased by ~5 fps, and the slowest frame duration is reduced
by ~25%.
The sysfs cur_freq file always reads the current frequency from Punit
anyway, so having rps.cur_delay be slightly off at times doesn't matter.
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>
projects / karo-tx-linux.git / log