firefly-linux-kernel-4.4.55.git
8 years agoMerge branch 'topic/atomic-conversion' into drm-intel-next-queued
Daniel Vetter [Mon, 15 Jun 2015 15:43:48 +0000 (17:43 +0200)]
Merge branch 'topic/atomic-conversion' into drm-intel-next-queued

The i915 atomic conversion is a real beast and it's not getting easier
wrangling in a separate branch. I'm might be regretting this, but
right after vacation nothing can burst my little bubble here!

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
8 years agodrm/i915: Report to userspace if we have a (presumed) working GPU reset
Chris Wilson [Mon, 15 Jun 2015 11:23:48 +0000 (12:23 +0100)]
drm/i915: Report to userspace if we have a (presumed) working GPU reset

In igt, we want to test handling of GPU hangs, both for recovery
purposes and for reporting. However, we don't want to inject a genuine
GPU hang onto a machine that cannot recover and so be permenantly
wedged. Rather than embed heuristics into igt, have the kernel report
exactly when it expects the GPU reset to work.

This can also be usefully extended in future to indicate different
levels of fine-grained resets.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Tim Gore <tim.gore@intel.com>
Cc: Tomas Elf <tomas.elf@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Fix build without CONFIG_PM
Chris Wilson [Mon, 15 Jun 2015 11:52:28 +0000 (12:52 +0100)]
drm/i915: Fix build without CONFIG_PM

drivers/gpu/drm/i915/i915_debugfs.c: In function ‘i915_runtime_pm_status’:
drivers/gpu/drm/i915/i915_debugfs.c:2528:34: error: ‘struct dev_pm_info’ has no member named ‘usage_count’
      atomic_read(&dev->dev->power.usage_count));

Regression from commit a6aaec8be22652a808d6e316d4a92e58cb75e986
Author: Damien Lespiau <damien.lespiau@intel.com>
Date:   Thu Jun 4 18:23:58 2015 +0100

    drm/i915: Add runtime PM's usage_count in i915_runtime_pm_status

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Damien Lespiau <damien.lespiau@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Make sure our labels start at column 0
Damien Lespiau [Thu, 4 Jun 2015 15:56:18 +0000 (16:56 +0100)]
drm/i915: Make sure our labels start at column 0

I noticed one of those and it turned out we have a few lingering around.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915/skl: Add debug messages at the start/end of DMC firmware loading
Damien Lespiau [Thu, 4 Jun 2015 15:42:16 +0000 (16:42 +0100)]
drm/i915/skl: Add debug messages at the start/end of DMC firmware loading

It's handy to have debug message for the "big" events and this one
qualifies IMHO. Also helpful to see what's happening while we're loading
the firwmare and how much time it takes.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Remove unnecessary () used with WARN()
Damien Lespiau [Thu, 4 Jun 2015 15:42:15 +0000 (16:42 +0100)]
drm/i915: Remove unnecessary () used with WARN()

In Linux, macros are usually well done and protect their arguments
properly, even avoiding multiple evaluations of the parameters. Extra ()
are really not needed.

Cc: Suketu Shah <suketu.j.shah@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Add SCRATCH1 and ROW_CHICKEN3 to the register whitelist.
Francisco Jerez [Fri, 29 May 2015 13:44:15 +0000 (16:44 +0300)]
drm/i915: Add SCRATCH1 and ROW_CHICKEN3 to the register whitelist.

Only bit 27 of SCRATCH1 and bit 6 of ROW_CHICKEN3 are allowed to be
set because of security-sensitive bits we don't want userspace to mess
with.  On HSW hardware the whitelisted bits control whether atomic
read-modify-write operations are performed on L3 or on GTI, and when
set to L3 (which can be 10x-30x better performing than on GTI,
depending on the application) require great care to avoid a system
hang, so we currently program them to be handled on GTI by default.

Beignet can immediately start taking advantage of this change to
enable L3 atomics.  Mesa should eventually switch to L3 atomics too,
but a number of non-trivial changes are still required so it will
continue using GTI atomics for now.

Signed-off-by: Francisco Jerez <currojerez@riseup.net>
Reviewed-by: Zhigang Gong <zhigang.gong@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Extend the parser to check register writes against a mask/value pair.
Francisco Jerez [Fri, 29 May 2015 13:44:14 +0000 (16:44 +0300)]
drm/i915: Extend the parser to check register writes against a mask/value pair.

In some cases it might be unnecessary or dangerous to give userspace
the right to write arbitrary values to some register, even though it
might be desirable to give it control of some of its bits.  This patch
extends the register whitelist entries to contain a mask/value pair in
addition to the register offset.  For registers with non-zero mask,
any LRM writes and LRI writes where the bits of the immediate given by
the mask don't match the specified value will be rejected.

This will be used in my next patch to grant userspace partial write
access to some sensitive registers.

Signed-off-by: Francisco Jerez <currojerez@riseup.net>
Reviewed-by: Zhigang Gong <zhigang.gong@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Fix command parser to validate multiple register access with the same command.
Francisco Jerez [Fri, 29 May 2015 13:44:13 +0000 (16:44 +0300)]
drm/i915: Fix command parser to validate multiple register access with the same command.

Until now the software command checker assumed that commands could
read or write at most a single register per packet.  This is not
necessarily the case, MI_LOAD_REGISTER_IMM expects a variable-length
list of offset/value pairs and writes them in sequence.  The previous
code would only check whether the first entry was valid, effectively
allowing userspace to write unrestricted registers of the MMIO space
by sending a multi-register write with a legal first register, with
potential security implications on Gen6 and 7 hardware.

Fix it by extending the drm_i915_cmd_descriptor table to represent
multi-register access and making validate_cmd() iterate for all
register offsets present in the command packet.

Signed-off-by: Francisco Jerez <currojerez@riseup.net>
Reviewed-by: Zhigang Gong <zhigang.gong@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Bump CHV PFI credits to 63 when cdclk>=czclk
Ville Syrjälä [Tue, 26 May 2015 17:22:40 +0000 (20:22 +0300)]
drm/i915: Bump CHV PFI credits to 63 when cdclk>=czclk

Switch from using 31 PFI credits to 63 PFI credits when cdclk>=czclk on
CHV. The spec lists both 31 and 63 as "suggested" values, but based on
feedback from hardware folks we should actually be using 63. Originally
I picked the 31 basically by flipping a coin.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Clint Taylor <Clinton.A.Taylor@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agoFix resume from suspend on IBM X30
Thomas Richter [Sat, 30 May 2015 18:25:53 +0000 (20:25 +0200)]
Fix resume from suspend on IBM X30

This patch fixes the resume from suspend-to-ram on the IBM X30
laptop. The problem is caused by the Bios missing to re-initialize
the iVCH registers, especially the PLL registers.

This patch records the iVCH registers during initialization, and
re-installs this register set when resuming.

Signed-off-by: Thomas Richter <thor@math.tu-berlin.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915/gtt: Remove _single from page table allocator
Mika Kuoppala [Fri, 22 May 2015 17:04:59 +0000 (20:04 +0300)]
drm/i915/gtt: Remove _single from page table allocator

We are always allocating a single page. No need to be verbose so
remove the suffix.

Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915/gtt: Don't leak scratch page on mapping error
Mika Kuoppala [Fri, 22 May 2015 17:04:58 +0000 (20:04 +0300)]
drm/i915/gtt: Don't leak scratch page on mapping error

Free the scratch page if dma mapping fails.

Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915/vlv: fix RC6 residency time calculation
Imre Deak [Mon, 1 Jun 2015 07:32:01 +0000 (10:32 +0300)]
drm/i915/vlv: fix RC6 residency time calculation

The divider value to convert from CZ clock rate to ms needs a +1
adjustment on VLV just like on CHV. This matches both the spec and
the accuracy test by pm_rc6_residency.

v2:
- simplify logic checking for the CHV 320MHz special case (Rodrigo)

Testcase: igt/pm_rc6_residency
Signed-off-by: Imre Deak <imre.deak@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76877
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Double the port clock when using double clocked modes with 12bpc
Ville Syrjälä [Tue, 5 May 2015 14:06:27 +0000 (17:06 +0300)]
drm/i915: Double the port clock when using double clocked modes with 12bpc

Currently we're forgetting to double the port clock when using double
clocked modes with 12bpc on HDMI. We're only accounting for the 1.5x
factor due to the 12bpc. So further double the 1.5x port clock when we
have a double clocked mode.

Unfortunately I don't have any displays that support both 12bpc and
double clocked modes, so I was unable to test this.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chandra Konduru <Chandra.konduru@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Fix hdmi clock readout with pixel repeat
Ville Syrjälä [Tue, 5 May 2015 14:06:26 +0000 (17:06 +0300)]
drm/i915: Fix hdmi clock readout with pixel repeat

Account for the pixel multiplier when reading out the HDMI
mode dotclock. Makes the state checked happier on my ILK when using
double clocked modes.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chandra Konduru <Chandra.konduru@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Check infoframe state more diligently.
Ville Syrjälä [Tue, 5 May 2015 14:06:25 +0000 (17:06 +0300)]
drm/i915: Check infoframe state more diligently.

Check that the DIP is enabled on the right port on IBX and VLV/CHV as
we're doing on g4x, and also check for all the infoframe enable bits on
all platforms.

Eventually we should track each infoframe type independently, and also
their contents. This is a small step in that direction as .infoframe_enabled()
return value could be easily turned into a bitmask.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chandra Konduru <Chandra.konduru@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Disable all infoframes when turning off the HDMI port
Ville Syrjälä [Tue, 5 May 2015 14:06:24 +0000 (17:06 +0300)]
drm/i915: Disable all infoframes when turning off the HDMI port

Currently we just disable the GCP infoframe when turning off the port.
That means if the same transcoder is used on a DP port next, we might
end up pushing infoframes over DP, which isn't intended. Just disable
all the infoframes when turning off the port.

Also protect against two ports stomping on each other on g4x due to
the single video DIP instance. Now only the first port to enable
gets to send infoframes.

v2: Rebase

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chandra Konduru <Chandra.konduru@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Fix 12bpc HDMI enable for IBX
Ville Syrjälä [Tue, 5 May 2015 14:06:23 +0000 (17:06 +0300)]
drm/i915: Fix 12bpc HDMI enable for IBX

Follow the procedure listed in Bspec to toggle the port enable bit off
and on when enabling HDMI with 12bpc and pixel repeat on IBX. The old
code didn't actually enable the port before "toggling" the bit back off,
so the whole workaround was essentially a nop.

Also take the opportunity to clarify the code by splitting the gmch
platforms to a separate (much more straightforward) function.

v2: Rebased due to crtc->config changes

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chandra Konduru <Chandra.konduru@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Fix HDMI 12bpc TRANSCONF bpc value
Ville Syrjälä [Tue, 5 May 2015 14:06:22 +0000 (17:06 +0300)]
drm/i915: Fix HDMI 12bpc TRANSCONF bpc value

IBX BSpec says we must specify 8bpc in TRANSCONF for both 8bpc
and 12bpc HDMI output. Do so.

v2: Pass intel_crtc to intel_pipe_has_type()

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chandra Konduru <Chandra.konduru@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Enable default_phase in GCP when possible
Ville Syrjälä [Tue, 5 May 2015 14:06:21 +0000 (17:06 +0300)]
drm/i915: Enable default_phase in GCP when possible

When the video timings are suitably aligned so that all different
periods start at phase 0 (ie. none of the periods start mid-pixel)
we can inform the sink about this. Supposedly the sink can then
optimize certain things. Obviously this is only relevant when
outputting >8bpc data since otherwise there are no mid-pixel phases.

v2: Rebased due to crtc->config changes

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chandra Konduru <Chandra.konduru@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Send GCP infoframes for deep color HDMI sinks
Ville Syrjälä [Tue, 5 May 2015 14:06:20 +0000 (17:06 +0300)]
drm/i915: Send GCP infoframes for deep color HDMI sinks

GCP infoframes are required to inform the HDMI sink about the color
depth.

Send the GCP infoframe whenever the sink supports any deep color modes
since such sinks must anyway be capable of receiving them. For sinks
that don't support deep color let's skip the GCP in case it might
confuse the sink, although HDMI 1.4 spec does say all sinks must be
capable of reciving them. In theory we could skip the GCP infoframe
for deep color sinks in 8bpc mode as well since sinks must fall back to
8bpc whenever GCP isn't received for some time.

BSpec says we should disable GCP after disabling the port, so do that as
well.

v2: s/intel_set_gcp_infoframe/intel_hdmi_set_gcp_infoframe/
    Rebased due to crtc->config changes

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
[danvet: Resolve conflict with lack of chv phy patches and fixup typo
Chandra spotted.]
Reviewed-by: Chandra Konduru <Chandra.konduru@intel.com>
Reviewed-by: Ander Conselvan de Oliveira <conselvan2@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 years agodrm/i915: Implement WaEnableHDMI8bpcBefore12bpc:snb, ivb
Ville Syrjälä [Tue, 5 May 2015 14:06:19 +0000 (17:06 +0300)]
drm/i915: Implement WaEnableHDMI8bpcBefore12bpc:snb, ivb

CPT/PPT require a specific procedure for enabling 12bpc HDMI. Implement
it, and to keep things neat pull the code into a function.

v2: Rebased due to crtc->config changes
    s/HDMI_GC/HDMIUNIT_GC/ to match spec better
    Factor out intel_enable_hdmi_audio()

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ander Conselvan de Oliveira <conselvan2@gmail.com>
Reviewed-By: Chandra Konduru <Chandra.konduru@intel.com>
Testecase: igt/kms_render/*
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agoMerge remote-tracking branch 'origin/drm-intel-next-queued' into drm-intel-next-queued
Daniel Vetter [Mon, 15 Jun 2015 05:26:47 +0000 (07:26 +0200)]
Merge remote-tracking branch 'origin/drm-intel-next-queued' into drm-intel-next-queued

Pull in patches Jani applied while I was on vacation.

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
9 years agodrm/i915: Only enable cursor if it can be enabled.
Maarten Lankhorst [Fri, 12 Jun 2015 09:15:42 +0000 (11:15 +0200)]
drm/i915: Only enable cursor if it can be enabled.

The cursor should only be enabled if it's visible. This fixes
igt/kms_cursor_crc, which may otherwise produce the following
warning:

------------[ cut here ]------------
WARNING: CPU: 0 PID: 3425 at drivers/gpu/drm/i915/intel_display.c:9995 intel_crtc_update_cursor+0x14c/0x4d0 [i915]()
Missing switch case (0) in i9xx_update_cursor
Modules linked in: i915
CPU: 0 PID: 3425 Comm: kms_cursor_crc Tainted: G        W       4.1.0-rc7-patser+ #4079
Hardware name: LENOVO 2349AV8/2349AV8, BIOS G1ETA5WW (2.65 ) 04/15/2014
 ffffffffc01aad10 ffff8800b083faa8 ffffffff817f7827 0000000080000001
 ffff8800b083faf8 ffff8800b083fae8 ffffffff81084955 ffff8800b083fad8
 ffff8800c4931148 0000000001200000 ffff8800c48b0000 0000000000000000
Call Trace:
 [<ffffffff817f7827>] dump_stack+0x4f/0x7b
 [<ffffffff81084955>] warn_slowpath_common+0x85/0xc0
 [<ffffffff810849d1>] warn_slowpath_fmt+0x41/0x50
 [<ffffffffc0139f2c>] intel_crtc_update_cursor+0x14c/0x4d0 [i915]
 [<ffffffffc01497f4>] __intel_set_mode+0x6c4/0x750 [i915]
 [<ffffffffc0150143>] intel_crtc_set_config+0x473/0x5c0 [i915]
 [<ffffffff81467da9>] drm_mode_set_config_internal+0x69/0x120
 [<ffffffff8146c1b9>] drm_mode_setcrtc+0x189/0x540
 [<ffffffff8145c7e0>] drm_ioctl+0x1a0/0x6a0
 [<ffffffff810b3b41>] ? get_parent_ip+0x11/0x50
 [<ffffffff811e9c28>] do_vfs_ioctl+0x2f8/0x530
 [<ffffffff810d0f7d>] ? trace_hardirqs_on+0xd/0x10
 [<ffffffff812e7746>] ? selinux_file_ioctl+0x56/0x100
 [<ffffffff811e9ee1>] SyS_ioctl+0x81/0xa0
 [<ffffffff81801617>] system_call_fastpath+0x12/0x6f
---[ end trace abf0f71163290a96 ]---

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Set hwmode during readout.
Maarten Lankhorst [Fri, 12 Jun 2015 09:15:41 +0000 (11:15 +0200)]
drm/i915: Set hwmode during readout.

This was introduced after converting hw readout to atomic,
so it should have been part of the revert too.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90929
Reported-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: get rid of intel_plane_restore in intel_crtc_page_flip
Maarten Lankhorst [Fri, 12 Jun 2015 09:15:40 +0000 (11:15 +0200)]
drm/i915: get rid of intel_plane_restore in intel_crtc_page_flip

Use a full atomic call instead. intel_crtc_page_flip will still
have to live until async updates are allowed.

This doesn't seem to be a regression from the convert to atomic,
part 3 patch. During GPU reset it fixes the following warning:

 ------------[ cut here ]------------
WARNING: CPU: 0 PID: 752 at drivers/gpu/drm/drm_crtc.c:5337 drm_mode_page_flip_ioctl+0x27b/0x360()
Modules linked in: i915
CPU: 0 PID: 752 Comm: Xorg Not tainted 4.1.0-rc7-patser+ #4090
Hardware name: NUC5i7RYB, BIOS RYBDWi35.86A.0246.2015.0309.1355 03/09/2015
 ffffffff81c90866 ffff8800d87c3ca8 ffffffff817f7d87 0000000080000001
 0000000000000000 ffff8800d87c3ce8 ffffffff81084955 ffff880000000000
 ffff8800d87c3dc0 ffff8800d93d1208 0000000000000000 ffff8800b7d1f3e0
Call Trace:
 [<ffffffff817f7d87>] dump_stack+0x4f/0x7b
 [<ffffffff81084955>] warn_slowpath_common+0x85/0xc0
 [<ffffffff81084a35>] warn_slowpath_null+0x15/0x20
 [<ffffffff8146dffb>] drm_mode_page_flip_ioctl+0x27b/0x360
 [<ffffffff8145ccb0>] drm_ioctl+0x1a0/0x6a0
 [<ffffffff810b3b41>] ? get_parent_ip+0x11/0x50
 [<ffffffff812e5540>] ? avc_has_perm+0x20/0x280
 [<ffffffff810b3b41>] ? get_parent_ip+0x11/0x50
 [<ffffffff811ea0f8>] do_vfs_ioctl+0x2f8/0x530
 [<ffffffff811f6001>] ? expand_files+0x261/0x270
 [<ffffffff812e7c16>] ? selinux_file_ioctl+0x56/0x100
 [<ffffffff811ea3b1>] SyS_ioctl+0x81/0xa0
 [<ffffffff81801b97>] system_call_fastpath+0x12/0x6f
---[ end trace 9ce834560085bd64 ]---

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Do not use atomic modesets in hw readout.
Maarten Lankhorst [Fri, 12 Jun 2015 09:15:39 +0000 (11:15 +0200)]
drm/i915: Do not use atomic modesets in hw readout.

This should fix fallout caused by making intel_crtc_control
and update_dpms atomic, which became a problem after reverting the
atomic hw readout patch.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90929
Reported-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agoRevert "drm/i915: Read hw state into an atomic state struct, v2."
Maarten Lankhorst [Wed, 10 Jun 2015 08:24:20 +0000 (10:24 +0200)]
Revert "drm/i915: Read hw state into an atomic state struct, v2."

This reverts commit 3bae26eb2991c00670df377cf6c3bc2b0577e82a.

Seems it introduces regressions for 3 different reasons, oh boy..

In bug #90868 as I can see the atomic state will be restored on
resume without the planes being set up properly. Because plane
setup here requires the atomic state, we'll have to settle
for committing atomic planes first.

In bug #90861 the failure appears to affect mostly DP devices,
and happens because reading out the atomic state prevents a modeset
on boot, which would require better hw state readout.

In bug #90874 it's shown that cdclk should be part of the atomic
state, so only performing a single modeset during resume excarbated
the issue.

It's better to fix those issues first, and then commit this patch,
so do that temporarily.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90868
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90861
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90874
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Acked-by: Ander Conselvan de Oliveira <conselvan2@gmail.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agoRevert "drm/i915: Make intel_display_suspend atomic, v2."
Maarten Lankhorst [Wed, 10 Jun 2015 08:24:19 +0000 (10:24 +0200)]
Revert "drm/i915: Make intel_display_suspend atomic, v2."

This reverts commit 490f400db5d886fc28566af69b02f6497f31be4b.

We're not ready yet to make it atomic, we calculate some state in
advance, but without atomic plane support atomic the hw readout will
fail.

It's required to revert this commit to revert the atomic hw
state readout patch.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90868
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90861
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Acked-by: Ander Conselvan de Oliveira <conselvan2@gmail.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: use calculated state for vblank evasion
Maarten Lankhorst [Mon, 1 Jun 2015 10:50:11 +0000 (12:50 +0200)]
drm/i915: use calculated state for vblank evasion

crtc->active will be gone eventually, and this check should be just as good.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Use atomic state for calculating DVO_2X_MODE on i830.
Maarten Lankhorst [Mon, 1 Jun 2015 10:50:10 +0000 (12:50 +0200)]
drm/i915: Use atomic state for calculating DVO_2X_MODE on i830.

This is a small behavioral change because it leaves DVO_2X_MODE
set between crtc_disable and crtc_enable. This is probably harmless
though and if not should be fixed by calculating 2x mode before
enable/disable pll.

This is needed because intel_crtc->active will be removed eventually.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Calculate haswell plane workaround, v5.
Maarten Lankhorst [Mon, 1 Jun 2015 10:50:09 +0000 (12:50 +0200)]
drm/i915: Calculate haswell plane workaround, v5.

This needs to be done last after all modesets have been calculated.

A modeset first disables all crtc's, so any crtc that undergoes a
modeset counts as inactive.

If no modeset's done, or > 1 crtc's stay w/a doesn't apply.
Apply workaround on the first crtc if 1 crtc stays active.
Apply workaround on the second crtc if no crtc was active.

Changes since v1:
 - Use intel_crtc->atomic as a place to put hsw_workaround_pipe.
 - Make sure quirk only applies to haswell.
 - Use first loop to iterate over newly enabled crtc's only.
   This increases readability.
Changes since v2:
 - Move hsw_workaround_pipe back to crtc_state.
Changes since v3:
 - Return errors from haswell_mode_set_planes_workaround.
Changes since v4:
- Clean up commit message.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Remove use of crtc->config from i915_debugfs.c
Maarten Lankhorst [Mon, 1 Jun 2015 10:50:08 +0000 (12:50 +0200)]
drm/i915: Remove use of crtc->config from i915_debugfs.c

crtc->config is updated to always contain to the active crtc_state
and only differs from crtc_state during crtc_disable. It will
eventually be removed, so start with some low hanging fruit.

For crtc->active the situation is the same; it will be removed
eventually. Instead use crtc->state->active.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Use crtc->hwmode for vblanks, v2.
Maarten Lankhorst [Mon, 1 Jun 2015 10:50:07 +0000 (12:50 +0200)]
drm/i915: Use crtc->hwmode for vblanks, v2.

intel_crtc->config will be removed eventually, so use crtc->hwmode.
drm_atomic_helper_update_legacy_modeset_state updates hwmode,
but crtc->active will eventually be gone too. Set dotclock to zero
to indicate the crtc is inactive.

Changes since v1:
- With the hwmode update in drm*update_legacy_modeset_state removed,
  intel_modeset_update_state has to assign it instead.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: move swap state to the right place
Maarten Lankhorst [Mon, 1 Jun 2015 10:50:06 +0000 (12:50 +0200)]
drm/i915: move swap state to the right place

This is a preparation for passing crtc state to the helpers.
When converting all users of crtc->config to use the old or
new state it's easier to find regressions when swap_state is
done first.

If crtc->config is swapped at the same place as swap_state
bugs will never be found.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Make intel_display_suspend atomic, v2.
Maarten Lankhorst [Mon, 1 Jun 2015 10:50:05 +0000 (12:50 +0200)]
drm/i915: Make intel_display_suspend atomic, v2.

Calculate all state using a normal transition, but afterwards fudge
crtc->state->active back to its old value. This should still allow
state restore in setup_hw_state to work properly.

Calling intel_set_mode will cause intel_display_set_init_power to be
called, make sure init_power gets set again afterwards.

Changes since v1:
- Fix to compile with v2 of the patch that adds intel_display_suspend.
- Add intel_display_set_init_power.
- Set return value to int to allow error checking.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Implement intel_crtc_control using atomic state, v4
Maarten Lankhorst [Mon, 1 Jun 2015 10:50:04 +0000 (12:50 +0200)]
drm/i915: Implement intel_crtc_control using atomic state, v4

Assume the callers lock everything with drm_modeset_lock_all.

This change had to be done after converting suspend/resume to
use atomic_state so the atomic state is preserved, otherwise
all transitional state is erased.

Now all callers of .crtc_enable and .crtc_disable go through
atomic modeset! :-D

Changes since v1:
- Only check for crtc_state->active in valleyview_modeset_global_pipes.
- Only check for crtc_state->active in modeset_update_crtc_power_domains.
Changes since v2:
- Rework on top of the changed patch order.
Changes since v3:
- Rename intel_crtc_toggle in description to *_control
- Change return value to int.
- Do not add plane state, should be done implicitly already.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Read hw state into an atomic state struct, v2.
Ander Conselvan de Oliveira [Mon, 1 Jun 2015 10:50:03 +0000 (12:50 +0200)]
drm/i915: Read hw state into an atomic state struct, v2.

To make this work we load the new hardware state into the
atomic_state, then swap it with the sw state.

This lets us change the force restore path in setup_hw_state()
to use a single call to intel_mode_set() to restore all the
previous state.

As a nice bonus this kills off encoder->new_encoder,
connector->new_enabled and crtc->new_enabled. They were used only
to restore the state after a modeset.

Changes since v1:
- Make sure all possible planes are added with their crtc set,
  so they will be turned off on first modeset.

Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Move cdclk and pll setup to intel_modeset_compute_config(), v2.
Ander Conselvan de Oliveira [Mon, 1 Jun 2015 10:50:02 +0000 (12:50 +0200)]
drm/i915: Move cdclk and pll setup to intel_modeset_compute_config(), v2.

It makes more sense there, since these are computation steps that can
fail.

Changes since v1:
- Rename __intel_set_mode_checks to intel_modeset_checks (Matt Roper)
- Move intel_modeset_checks to before check_planes, so it won't
  have to be moved later.

Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Swap planes on each crtc separately, v2.
Maarten Lankhorst [Mon, 1 Jun 2015 10:50:01 +0000 (12:50 +0200)]
drm/i915: Swap planes on each crtc separately, v2.

Repeated calls to begin_crtc_commit can cause warnings like this:
[  169.127746] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:616
[  169.127835] in_atomic(): 0, irqs_disabled(): 1, pid: 1947, name: kms_flip
[  169.127840] 3 locks held by kms_flip/1947:
[  169.127843]  #0:  (&dev->mode_config.mutex){+.+.+.}, at: [<ffffffff814774bc>] __drm_modeset_lock_all+0x9c/0x130
[  169.127860]  #1:  (crtc_ww_class_acquire){+.+.+.}, at: [<ffffffff814774cd>] __drm_modeset_lock_all+0xad/0x130
[  169.127870]  #2:  (crtc_ww_class_mutex){+.+.+.}, at: [<ffffffff81477178>] drm_modeset_lock+0x38/0x110
[  169.127879] irq event stamp: 665690
[  169.127882] hardirqs last  enabled at (665689): [<ffffffff817ffdb5>] _raw_spin_unlock_irqrestore+0x55/0x70
[  169.127889] hardirqs last disabled at (665690): [<ffffffffc0197a23>] intel_pipe_update_start+0x113/0x5c0 [i915]
[  169.127936] softirqs last  enabled at (665470): [<ffffffff8108a766>] __do_softirq+0x236/0x650
[  169.127942] softirqs last disabled at (665465): [<ffffffff8108ae75>] irq_exit+0xc5/0xd0
[  169.127951] CPU: 1 PID: 1947 Comm: kms_flip Not tainted 4.1.0-rc4-patser+ #4039
[  169.127954] Hardware name: LENOVO 2349AV8/2349AV8, BIOS G1ETA5WW (2.65 ) 04/15/2014
[  169.127957]  ffff8800c49036f0 ffff8800cde5fa28 ffffffff817f6907 0000000080000001
[  169.127964]  0000000000000000 ffff8800cde5fa58 ffffffff810aebed 0000000000000046
[  169.127970]  ffffffff81c5d518 0000000000000268 0000000000000000 ffff8800cde5fa88
[  169.127981] Call Trace:
[  169.127992]  [<ffffffff817f6907>] dump_stack+0x4f/0x7b
[  169.128001]  [<ffffffff810aebed>] ___might_sleep+0x16d/0x270
[  169.128008]  [<ffffffff810aed38>] __might_sleep+0x48/0x90
[  169.128017]  [<ffffffff817fc359>] mutex_lock_nested+0x29/0x410
[  169.128073]  [<ffffffffc01635f0>] ? vgpu_write64+0x220/0x220 [i915]
[  169.128138]  [<ffffffffc017fddf>] ? ironlake_update_primary_plane+0x2ff/0x410 [i915]
[  169.128198]  [<ffffffffc0190e75>] intel_frontbuffer_flush+0x25/0x70 [i915]
[  169.128253]  [<ffffffffc01831ac>] intel_finish_crtc_commit+0x4c/0x180 [i915]
[  169.128279]  [<ffffffffc00784ac>] drm_atomic_helper_commit_planes+0x12c/0x240 [drm_kms_helper]
[  169.128338]  [<ffffffffc0184264>] __intel_set_mode+0x684/0x830 [i915]
[  169.128378]  [<ffffffffc018a84a>] intel_crtc_set_config+0x49a/0x620 [i915]
[  169.128385]  [<ffffffff817fdd39>] ? mutex_unlock+0x9/0x10
[  169.128391]  [<ffffffff81467b69>] drm_mode_set_config_internal+0x69/0x120
[  169.128398]  [<ffffffff8119b547>] ? might_fault+0x57/0xb0
[  169.128403]  [<ffffffff8146bf93>] drm_mode_setcrtc+0x253/0x620
[  169.128409]  [<ffffffff8145c600>] drm_ioctl+0x1a0/0x6a0
[  169.128415]  [<ffffffff810b3b41>] ? get_parent_ip+0x11/0x50
[  169.128424]  [<ffffffff811e9ab8>] do_vfs_ioctl+0x2f8/0x530
[  169.128429]  [<ffffffff810d0fcd>] ? trace_hardirqs_on+0xd/0x10
[  169.128435]  [<ffffffff812e7676>] ? selinux_file_ioctl+0x56/0x100
[  169.128439]  [<ffffffff811e9d71>] SyS_ioctl+0x81/0xa0
[  169.128445]  [<ffffffff81800697>] system_call_fastpath+0x12/0x6f

Solve it by using the newly introduced drm_atomic_helper_commit_planes_on_crtc.

The problem here was that the drm_atomic_helper_commit_planes() helper
we were using was basically designed to do

    begin_crtc_commit(crtc #1)
    begin_crtc_commit(crtc #2)
    ...
    commit all planes
    finish_crtc_commit(crtc #1)
    finish_crtc_commit(crtc #2)

The problem here is that since our hardware relies on vblank evasion,
our CRTC 'begin' function waits until we're out of the danger zone in
which register writes might wind up straddling the vblank, then disables
interrupts; our 'finish' function re-enables interrupts after the
registers have been written.  The expectation is that the operations between
'begin' and 'end' must be performed without sleeping (since interrupts
are disabled) and should happen as quickly as possible.  By clumping all
of the 'begin' calls together, we introducing a couple problems:
 * Subsequent 'begin' invocations might sleep (which is illegal)
 * The first 'begin' ensured that we were far enough from the vblank that
   we could write our registers safely and ensure they all fell within
   the same frame.  Adding extra delay waiting for subsequent CRTC's
   wasn't accounted for and could put us back into the 'danger zone' for
   CRTC #1.

This commit solves the problem by using a new helper that allows an
order of operations like:

   for each crtc {
        begin_crtc_commit(crtc)  // sleep (maybe), then disable interrupts
        commit planes for this specific CRTC
        end_crtc_commit(crtc)    // reenable interrupts
   }

so that sleeps will only be performed while interrupts are enabled and
we can be sure that registers for a CRTC will be written immediately
once we know we're in the safe zone.

The crtc->config->base.crtc update may seem unrelated, but the helper
will use it to obtain the crtc for the state. Without the update it
will dereference NULL and crash.

Changes since v1:
- Use Matt Roper's commit message.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Use drm_atomic_helper_swap_state in intel_atomic_commit.
Maarten Lankhorst [Mon, 1 Jun 2015 10:50:00 +0000 (12:50 +0200)]
drm/i915: Use drm_atomic_helper_swap_state in intel_atomic_commit.

And update crtc->config to point to the new state. There is no point
in swapping only part of the state when the rest of the state
should be untouched.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Use global atomic state for staged pll, config, v3.
Maarten Lankhorst [Thu, 4 Jun 2015 08:21:28 +0000 (10:21 +0200)]
drm/i915: Use global atomic state for staged pll, config, v3.

Now that we can subclass drm_atomic_state we can also use it to keep
track of all the pll settings. atomic_state is a better place to hold
all shared state than keeping pll->new_config everywhere.

Changes since v1:
- Assert connection_mutex is held.
Changes since v2:
- Fix swapped arguments to kzalloc for intel_atomic_state_alloc. (Jani Nikula)

Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Zap call to drm_plane_helper_disable, v2.
Maarten Lankhorst [Mon, 1 Jun 2015 10:49:58 +0000 (12:49 +0200)]
drm/i915: Zap call to drm_plane_helper_disable, v2.

The primary plane can still be configured when crtc is off,
furthermore this is also a noop now that affected planes are
added on modesets.

Changes since v1:
- Move commit so no frontbuffer_bits warnings are generated.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Support modeset across multiple pipes
Ander Conselvan de Oliveira [Mon, 1 Jun 2015 10:49:57 +0000 (12:49 +0200)]
drm/i915: Support modeset across multiple pipes

Compute new pipe_configs for all crtcs in the atomic state. The commit
part of the mode set (__intel_set_mode()) is already enabled to support
multiple pipes, the only thing missing was calculating a new pipe_config
for every crtc.

Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: calculate primary visibility changes instead of calling from set_config
Maarten Lankhorst [Mon, 1 Jun 2015 10:49:56 +0000 (12:49 +0200)]
drm/i915: calculate primary visibility changes instead of calling from set_config

This should be much cleaner, with the same effects.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: do not wait for vblank when crtc is off
Maarten Lankhorst [Mon, 1 Jun 2015 10:49:55 +0000 (12:49 +0200)]
drm/i915: do not wait for vblank when crtc is off

This can happen when turning off a sprite plane. Because the crtc state
is not yet always swapped correctly and transitional helpers are used
the crtc state cannot be relied on.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: update plane state during init
Maarten Lankhorst [Mon, 1 Jun 2015 10:49:54 +0000 (12:49 +0200)]
drm/i915: update plane state during init

Atomic planes updates rely on having a accurate plane_mask.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Make sure all planes and connectors are added on modeset.
Maarten Lankhorst [Mon, 1 Jun 2015 10:49:53 +0000 (12:49 +0200)]
drm/i915: Make sure all planes and connectors are added on modeset.

Add missing calls to drm_atomic_add_affected_*. This is needed
to convert to atomic planes. When converting to atomic all planes
are needed on modeset. For good measure make sure all connectors
are added too.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Use crtc_state->active instead of crtc_state->enable
Maarten Lankhorst [Mon, 1 Jun 2015 10:49:52 +0000 (12:49 +0200)]
drm/i915: Use crtc_state->active instead of crtc_state->enable

crtc_state->enable means a crtc is configured, but it may be turned
off for dpms. Until the commit "use intel_crtc_control everywhere"
crtc_state->active was not updated on crtc off, but now
crtc_state->active should be used for tracking whether a crtc is
scanning out or not.

A few commits from now dpms will be handled by calling
intel_set_mode with a different value for crtc_state->active,
which causes a crtc to turn on or off.

At this point crtc->active should mirror crtc_state->active,
so some paranoia from the crtc_disable functions can be removed.

intel_set_mode_setup_plls still checks for ->enable, because all
resources that are needed have to be calculated, else
dpms changes may not succeed.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Set mode_changed for audio in intel_modeset_pipe_config()
Ander Conselvan de Oliveira [Mon, 1 Jun 2015 10:49:51 +0000 (12:49 +0200)]
drm/i915: Set mode_changed for audio in intel_modeset_pipe_config()

A follow up patch will make intel_modeset_compute_config() deal with
multiple crtcs, so move crtc specific stuff into the lower level crtc
specific function.

Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Make __intel_set_mode() take only atomic state as argument
Ander Conselvan de Oliveira [Mon, 1 Jun 2015 10:49:50 +0000 (12:49 +0200)]
drm/i915: Make __intel_set_mode() take only atomic state as argument

With the use of drm_atomic_helper_update_legacy_modeset_state the
last user of modeset_crtc is removed from this function.

Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Use drm_atomic_helper_update_legacy_modeset_state, v2.
Maarten Lankhorst [Mon, 1 Jun 2015 10:49:49 +0000 (12:49 +0200)]
drm/i915: Use drm_atomic_helper_update_legacy_modeset_state, v2.

Now that the helper is exported there's no need to duplicate
this code any more.

Changes since v1:
- move intel_modeset_update_staged_output_state call to the right place.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: use intel_crtc_control everywhere, v3.
Maarten Lankhorst [Mon, 1 Jun 2015 10:49:48 +0000 (12:49 +0200)]
drm/i915: use intel_crtc_control everywhere, v3.

Having a single path for everything makes it a lot easier to keep
crtc_state->active in sync with intel_crtc->active.

A crtc cannot be changed to active when not enabled, because it means
no mode is set and no connectors are connected.

This should also make intel_crtc->active match crtc_state->active.

Changes since v1:
- Reworded commit message, there's no intel_crtc_toggle.
Changes since v2:
- Change some callers of intel_crtc_control to intel_display_suspend.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: add intel_display_suspend, v2
Maarten Lankhorst [Mon, 1 Jun 2015 10:49:47 +0000 (12:49 +0200)]
drm/i915: add intel_display_suspend, v2

This is a function used to disable all crtc's. This makes it clearer
to distinguish between when mode needs to be preserved and when
it can be trashed.

Changes since v1:
- Copy power changes from intel_crtc_control.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: get rid of intel_crtc_disable and related code, v3
Maarten Lankhorst [Mon, 1 Jun 2015 10:49:46 +0000 (12:49 +0200)]
drm/i915: get rid of intel_crtc_disable and related code, v3

Now that the dpll updates are (mostly) atomic, the .off() code is a noop,
and intel_crtc_disable does mostly the same as intel_modeset_update_state.

Move all logic for connectors_active and setting dpms to that function.

Changes since v1:
- Move drm_atomic_helper_swap_state up.
Changes since v2:
- Split out intel_put_shared_dpll removal.
Changes since v3:
- Rebase on top of latest drm-intel.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: get rid of put_shared_dpll
Maarten Lankhorst [Mon, 1 Jun 2015 10:49:45 +0000 (12:49 +0200)]
drm/i915: get rid of put_shared_dpll

Now that the pll updates are staged the put_shared_dpll function
consists only of checks that are done in check_shared_dpll_state
after a modeset too.

The changes to pll->config are overwritten by
intel_shared_dpll_commit, so this entire function is a noop.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915/bxt: fix DDI PHY vswing scale value setting
Imre Deak [Thu, 4 Jun 2015 15:01:35 +0000 (18:01 +0300)]
drm/i915/bxt: fix DDI PHY vswing scale value setting

According to bspec the DDI PHY vswing scale value is "don't care" in
case the scale enable bit [27] is clear. But this doesn't seem to be
correct. The scale value seems to also matter if the scale mode bit
[26] is set. So both bit 26 and 27 depend on the value. Setting the
scale value to 0 while either bit is set results in a failed modeset on
HDMI (sink reports no signal).

After reset the scale value is 0x98, but according to the spec we have
to program it to 0x9a. So for consistency program it always to 0x9a
regardless of the scale enable bit.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Tested-by: Matt Roper <matthew.d.roper@intel.com>
Acked-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Add runtime PM's usage_count in i915_runtime_pm_status
Damien Lespiau [Thu, 4 Jun 2015 17:23:58 +0000 (18:23 +0100)]
drm/i915: Add runtime PM's usage_count in i915_runtime_pm_status

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Make pc8_status report status for all runtime PM platforms
Damien Lespiau [Thu, 4 Jun 2015 17:23:57 +0000 (18:23 +0100)]
drm/i915: Make pc8_status report status for all runtime PM platforms

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915/bxt: Use intel_update_cdclk() to update dev_priv->cdclk_freq
Damien Lespiau [Thu, 4 Jun 2015 17:21:34 +0000 (18:21 +0100)]
drm/i915/bxt: Use intel_update_cdclk() to update dev_priv->cdclk_freq

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915/skl: Update the cached CDCLK at the end of set_cdclk()
Damien Lespiau [Thu, 4 Jun 2015 17:21:33 +0000 (18:21 +0100)]
drm/i915/skl: Update the cached CDCLK at the end of set_cdclk()

Ville's and Mika's cdclk series was in flight at the same time as the
SKL S3 patches so we were missing that update.

intel_update_max_cdclk() and intel_update_cdclk() had to be moved up a
bit to avoid forward declarations.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Don't display the boot CDCLK twice
Damien Lespiau [Thu, 4 Jun 2015 17:21:32 +0000 (18:21 +0100)]
drm/i915: Don't display the boot CDCLK twice

intel_update_cdclk() will already display the boot CDCLK for DDI
platforms, no need to repeat there.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915/skl: Don't warn if reading back DPLL0 is disabled
Damien Lespiau [Thu, 4 Jun 2015 17:21:31 +0000 (18:21 +0100)]
drm/i915/skl: Don't warn if reading back DPLL0 is disabled

We can operate with DPLL0 off with CDCLK backed by the 24Mhz reference
clock, and that's a supported configuration. Don't warn when notice
DPLL0 is off then.

We still have a separate warn at boot if cdclk is disabled (because we
don't currently try to handle the case (that shouldn't happen on SKL as
far as I know) where we boot with display not initialized.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915/skl: Derive the max CDCLK from DFSM
Damien Lespiau [Thu, 4 Jun 2015 17:21:30 +0000 (18:21 +0100)]
drm/i915/skl: Derive the max CDCLK from DFSM

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Make broxton_set_cdclk() static
Damien Lespiau [Thu, 4 Jun 2015 17:21:29 +0000 (18:21 +0100)]
drm/i915: Make broxton_set_cdclk() static

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: BDW clock change support
Ville Syrjälä [Wed, 3 Jun 2015 12:45:13 +0000 (15:45 +0300)]
drm/i915: BDW clock change support

Add support for changing cdclk frequency during runtime on BDW.

Also with IPS enabled the actual pixel rate mustn't exceed 95% of cdclk,
so take that into account when computing the max pixel rate.

v2: Grab rps.hw_lock around sandybridge_pcode_write()
v3: Rebase due to power well vs. .global_resources() reordering
v4: Rebased to the latest
v5: Rebased to the latest
v6: Patch order shuffle so that Broadwell CD clock change is
    applied before the patch for Haswell CD clock change
v7: Fix for patch style problems

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Mika Kahola <mika.kahola@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Add IS_BDW_ULX
Ville Syrjälä [Wed, 3 Jun 2015 12:45:12 +0000 (15:45 +0300)]
drm/i915: Add IS_BDW_ULX

We need to tell BDW ULT and ULX apart.

v2: Rebased to the latest
v3: Rebased to the latest
v4: Fix for patch style problems

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Mika Kahola <mika.kahola@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Don't enable IPS when pixel rate exceeds 95%
Ville Syrjälä [Wed, 3 Jun 2015 12:45:11 +0000 (15:45 +0300)]
drm/i915: Don't enable IPS when pixel rate exceeds 95%

Bspec says we shouldn't enable IPS on BDW when the pipe pixel rate
exceeds 95% of the core display clock. Apparently this can cause
underruns.

There's no similar restriction listed for HSW, so leave that one alone
for now.

v2: Add pipe_config_supports_ips() (Chris)
v3: Compare against the max cdclk insted of the current cdclk
v4: Rebased to the latest
v5: Rebased to the latest
v6: Fix for patch style problems

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=83497
Tested-by: Timo Aaltonen <tjaalton@ubuntu.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Mika Kahola <mika.kahola@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Store max cdclk value in dev_priv
Ville Syrjälä [Wed, 3 Jun 2015 12:45:10 +0000 (15:45 +0300)]
drm/i915: Store max cdclk value in dev_priv

Keep the cdclk maximum supported frequency around in dev_priv so that we
can verify certain things against it before actually changing the cdclk
frequency.

For now only VLV/CHV have support changing cdclk frequency, so other
plarforms get to assume cdclk is fixed.

v2: Rebased to the latest
v3: Rebased to the latest
v4: Fix for patch style problems

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Mika Kahola <mika.kahola@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Use cached cdclk value
Ville Syrjälä [Wed, 3 Jun 2015 12:45:08 +0000 (15:45 +0300)]
drm/i915: Use cached cdclk value

Rather than reading out the current cdclk value use the cached value we
have tucked away in dev_priv.

v2: Rebased to the latest
v3: Rebased to the latest
v4: Fix for patch style problems

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Mika Kahola <mika.kahola@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Cache current cdclk frequency in dev_priv
Ville Syrjälä [Wed, 3 Jun 2015 12:45:07 +0000 (15:45 +0300)]
drm/i915: Cache current cdclk frequency in dev_priv

Rather that extracting the current cdclk freuqncy every time someone
wants to know it, cache the current value and use that. VLV/CHV already
stored a cached value there so just expand that to cover all platforms.

v2: Rebased to the latest
v3: Rebased to the latest
v4: Rebased to the latest
v5: Removed spurious call to 'intel_update_cdclk(dev)' based on
    Damien Lespiau's comment

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Mika Kahola <mika.kahola@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agoMerge remote-tracking branch 'drm-upstream/drm-next' into drm-intel-next-queued
Jani Nikula [Thu, 4 Jun 2015 07:26:17 +0000 (10:26 +0300)]
Merge remote-tracking branch 'drm-upstream/drm-next' into drm-intel-next-queued

Backmerge drm-next so I can apply Maarten's drm/i915 atomic conversion
patches.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agoMerge branch 'virtio-gpu-drm-next' of git://git.kraxel.org/linux into drm-next
Dave Airlie [Wed, 3 Jun 2015 23:36:39 +0000 (09:36 +1000)]
Merge branch 'virtio-gpu-drm-next' of git://git.kraxel.org/linux into drm-next

Yay, thanks to Gerd for pull this together.

* 'virtio-gpu-drm-next' of git://git.kraxel.org/linux:
  Add MAINTAINERS entry for virtio-gpu.
  Add virtio gpu driver.
  drm_vblank_get: don't WARN_ON in case vblanks are not initialized
  break kconfig dependency loop

9 years agoMerge branch 'linux-4.1.0-rc5-tilcdc-refactor' of https://github.com/jsarha/linux...
Dave Airlie [Wed, 3 Jun 2015 23:24:28 +0000 (09:24 +1000)]
Merge branch 'linux-4.1.0-rc5-tilcdc-refactor' of https://github.com/jsarha/linux into drm-next

Please pull the contents of "Use DRM component API in tilcdc to
connect to tda998x" patch series.

* 'linux-4.1.0-rc5-tilcdc-refactor' of https://github.com/jsarha/linux:
  drm/tilcdc: Force building of DRM_TILCDC_SLAVE_COMPAT
  drm/tilcdc: Add DRM_TILCDC_SLAVE_COMPAT for ti,tilcdc,slave binding support
  drm/tilcdc: use pm_runtime_irq_safe()
  drm/tilcdc: Add support for external tda998x encoder
  drm/tilcdc: Remove tilcdc slave support for tda998x driver
  drm/tilcdc: Fix module unloading

9 years agoMerge tag 'v4.1-rc6' into drm-next
Dave Airlie [Wed, 3 Jun 2015 23:23:51 +0000 (09:23 +1000)]
Merge tag 'v4.1-rc6' into drm-next

Linux 4.1-rc6

backmerge 4.1-rc6 as some of the later pull reqs are based on newer bases
and I'd prefer to do the fixup myself.

9 years agodrm/atomic: Clear crtc_state->active in drm_atomic_helper_set_config.
Maarten Lankhorst [Mon, 1 Jun 2015 06:59:53 +0000 (08:59 +0200)]
drm/atomic: Clear crtc_state->active in drm_atomic_helper_set_config.

This fixes some regressions in i915 when converting to atomic.
set_config failed with -EINVAL, and I received the following warning
in dmesg:

[drm:drm_atomic_crtc_check] [CRTC:20] active without enabled

Solve this by clearing active when a crtc is disabled.

Because crtc_state->enable implies that connectors are active the
change from disabled->enabled can only happen for the crtc that's
being set_config'd, and checking for !crtc_state->enable is sufficient
here.

Cc: dri-devel@lists.freedesktop.org
Reviewed-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
9 years agoMerge tag 'topic/drm-misc-2015-05-27' of git://anongit.freedesktop.org/drm-intel...
Dave Airlie [Wed, 3 Jun 2015 23:17:45 +0000 (09:17 +1000)]
Merge tag 'topic/drm-misc-2015-05-27' of git://anongit.freedesktop.org/drm-intel into drm-next

One more round of drm-misc, again mostly atomic. Big thing is the
userspace blob code from Daniel Stone, with support for the mode_id blob
now added to the atomic ioctl. Finally we can do atomic modesets!

Note that the atomic ioctl is still behind the module knob since the
weston patches aren't quite ready yet imo - they lack TEST_ONLY support,
which is a fairly crucial bit of the atomic api. But besides that I think
it's all good to go. That's also why we didn't bother to hide the new blob
ioctls behind the knob, that part won't need to change. And if weston
patches get in shape in time we could throw the "atomic by default patch"
on top for 4.2.

* tag 'topic/drm-misc-2015-05-27' of git://anongit.freedesktop.org/drm-intel:
  drm: Fix off-by-one in vblank hardware counter wraparound handling
  drm/atomic: fix out of bounds read in for_each_*_in_state helpers
  drm/atomic: Add MODE_ID property
  drm/atomic: Add current-mode blob to CRTC state
  drm: Add drm_atomic_set_mode_for_crtc
  drm: check for garbage in unused addfb2 fields
  drm: Retain reference to blob properties in lookup
  drm/mode: Add user blob-creation ioctl
  drm: Return error value from blob creation
  drm: Allow creating blob properties without copy
  drm/mode: Unstatic kernel-userspace mode conversion
  drm/mode: Validate modes inside drm_crtc_convert_umode
  drm/crtc_helper: Replace open-coded CRTC state helpers
  drm: kerneldoc fixes for blob properties
  drm/DocBook: Add more drm_bridge documentation
  drm: bridge: Allow daisy chaining of bridges
  drm/atomic: add all affected planes in drm_atomic_helper_check_modeset
  drm/atomic: add drm_atomic_add_affected_planes
  drm/atomic: add commit_planes_on_crtc helper

9 years agoMerge tag 'drm-amdkfd-next-2015-06-03' of git://people.freedesktop.org/~gabbayo/linux...
Dave Airlie [Wed, 3 Jun 2015 23:15:39 +0000 (09:15 +1000)]
Merge tag 'drm-amdkfd-next-2015-06-03' of git://people.freedesktop.org/~gabbayo/linux into drm-next

drm-amdkfd-next-2015-06-03:

- Add the H/W debugger support module, including new IOCTLs to:
  - register/unregister a process as a debugged process
  - Set address watch-point in the debugged process's GPU kernel
  - Do a wave control operation in the debugged process's waves
  See the commit messages for more details on the available operations.

  The debugged process can only perform debug operation on itself. It is
  blocked by the amdkfd+H/W from performing operations on other processes's
  waves or GPU kernels. The blocking is done by setting the VMID and PASID of
  the debugged process in the packets that are sent to the CP with the debug
  instructions.

- Add support for static user-mode queues. These queues are regular queues,
  but because they belong to the debugged process, we need to make sure the CP
  doesn't preempt them during a debug operation. Therefore, we mark them as
  static for the CP ignore them during preemption.

- Support killing all the waves when a process is terminated. This is needed
  in case a process is terminated but we can't UNMAP its queues (can occur due
  to several reasons). In that case, the CP could be stuck unless we kill all
  its waves. This function is *very* important as it provides the kernel a high
  level of control over the GPU. The reason we didn't upstream this function
  so far, is because it is implemented using the H/W debugger module functions,
  so we had to wait until we can upstream the H/W debugger module.

- Replace declaration of bitmap from unsigned long to standard DECLARE_BITMAP

* tag 'drm-amdkfd-next-2015-06-03' of git://people.freedesktop.org/~gabbayo/linux:
  drm/amdkfd: Enforce kill all waves on process termination
  drm/radeon: Add ATC VMID<-->PASID functions to kfd->kgd
  drm/amdkfd: Implement address watch debugger IOCTL
  drm/amdkfd: Implement wave control debugger IOCTL
  drm/amdkfd: Implement (un)register debugger IOCTLs
  drm/amdkfd: Add address watch operation to debugger
  drm/amdkfd: Add wave control operation to debugger
  drm/amdkfd: Add skeleton H/W debugger module support
  drm/amdkfd: Add static user-mode queues support
  drm/amdkfd: add H/W debugger IOCTL set definitions
  drm/radeon: Add H/W debugger kfd->kgd functions
  drm/amdkfd: Use DECLARE_BITMAP

9 years agoAdd MAINTAINERS entry for virtio-gpu.
Gerd Hoffmann [Tue, 2 Jun 2015 08:49:03 +0000 (10:49 +0200)]
Add MAINTAINERS entry for virtio-gpu.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
9 years agoAdd virtio gpu driver.
Dave Airlie [Mon, 9 Sep 2013 00:02:56 +0000 (10:02 +1000)]
Add virtio gpu driver.

This patch adds a kms driver for the virtio gpu.  The xorg modesetting
driver can handle the device just fine, the framebuffer for fbcon is
there too.

Qemu patches for the host side are under review currently.

The pci version of the device comes in two variants: with and without
vga compatibility.  The former has a extra memory bar for the vga
framebuffer, the later is a pure virtio device.  The only concern for
this driver is that in the virtio-vga case we have to kick out the
firmware framebuffer.

Initial revision has only 2d support, 3d (virgl) support requires
some more work on the qemu side and will be added later.

Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
9 years agodrm/i915: Apply WaDisableAsyncFlipPerfMode via LRIs on gen8
Ville Syrjälä [Tue, 2 Jun 2015 12:37:37 +0000 (15:37 +0300)]
drm/i915: Apply WaDisableAsyncFlipPerfMode via LRIs on gen8

MI_MODE is saved in the logical context so WaDisableAsyncFlipPerfMode
must be applied using LRIs on gen8.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Set INSTPM_FORCE_ORDERING via LRI on gen8, drop it on gen9+
Ville Syrjälä [Tue, 2 Jun 2015 12:37:36 +0000 (15:37 +0300)]
drm/i915: Set INSTPM_FORCE_ORDERING via LRI on gen8, drop it on gen9+

INSTPM is saved in the logical context so we should initialize it using
LRIs on gen8. It actually defaults to 1 starting from HSW, but let's
keep the write around anyway.

Also drop the INSTPM_FORCE_ORDERING setup entirely on gen9+ since it's
now a reserved bit.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/amdkfd: Enforce kill all waves on process termination
Ben Goz [Wed, 20 May 2015 15:05:44 +0000 (18:05 +0300)]
drm/amdkfd: Enforce kill all waves on process termination

This commit makes sure that on process termination, after
we're destroying all the active queues, we're killing all the
existing wave front of the current process.

By doing this we're making sure that if any of the CUs were blocked
by infinite loop we're enforcing it to end the shader explicitly.

Signed-off-by: Ben Goz <ben.goz@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/radeon: Add ATC VMID<-->PASID functions to kfd->kgd
Alexey Skidanov [Tue, 19 May 2015 16:25:01 +0000 (19:25 +0300)]
drm/radeon: Add ATC VMID<-->PASID functions to kfd->kgd

This patch adds three new interfaces to kfd2kgd interface file of radeon.

The interfaces are:

- Check if a specific VMID has a valid PASID mapping
- Retrieve the PASID which is mapped to a specific VMID
- Issue a VMID invalidation request to the ATC

Signed-off-by: Alexey Skidanov <Alexey.Skidanov@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: Implement address watch debugger IOCTL
Yair Shachar [Wed, 20 May 2015 11:09:39 +0000 (14:09 +0300)]
drm/amdkfd: Implement address watch debugger IOCTL

v2:

- rename get_dbgmgr_mutex to kfd_get_dbgmgr_mutex to namespace it
- change void* to uint64_t inside ioctl arguments
- use kmalloc instead of kzalloc because we use copy_from_user
  immediately after it

Signed-off-by: Yair Shachar <yair.shachar@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: Implement wave control debugger IOCTL
Yair Shachar [Wed, 20 May 2015 11:09:24 +0000 (14:09 +0300)]
drm/amdkfd: Implement wave control debugger IOCTL

v2:

- rename get_dbgmgr_mutex to kfd_get_dbgmgr_mutex to namespace it
- change void* to uint64_t inside ioctl arguments
- use kmalloc instead of kzalloc because we use copy_from_user
  immediately after it

Signed-off-by: Yair Shachar <yair.shachar@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: Implement (un)register debugger IOCTLs
Yair Shachar [Wed, 20 May 2015 11:08:55 +0000 (14:08 +0300)]
drm/amdkfd: Implement (un)register debugger IOCTLs

v2: rename get_dbgmgr_mutex to kfd_get_dbgmgr_mutex to namespace it

Signed-off-by: Yair Shachar <yair.shachar@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: Add address watch operation to debugger
Yair Shachar [Wed, 20 May 2015 10:59:17 +0000 (13:59 +0300)]
drm/amdkfd: Add address watch operation to debugger

The address watch operation gives the ability to specify watch points
which will generate a shader breakpoint, based on a specified single
address or range of addresses.

There is support for read/write/any access modes.

Signed-off-by: Yair Shachar <yair.shachar@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: Add wave control operation to debugger
Yair Shachar [Wed, 20 May 2015 10:58:12 +0000 (13:58 +0300)]
drm/amdkfd: Add wave control operation to debugger

The wave control operation supports several command types executed upon
existing wave fronts that belong to the currently debugged process.

The available commands are:

HALT   - Freeze wave front(s) execution
RESUME - Resume freezed wave front(s) execution
KILL   - Kill existing wave front(s)

Signed-off-by: Yair Shachar <yair.shachar@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: Add skeleton H/W debugger module support
Yair Shachar [Wed, 20 May 2015 10:48:26 +0000 (13:48 +0300)]
drm/amdkfd: Add skeleton H/W debugger module support

This patch adds the skeleton H/W debugger module support. This code
enables registration and unregistration of a single HSA process at a
time.

The module saves the process's pasid and use it to verify that only the
registered process is allowed to execute debugger operations through the
kernel driver.

v2: rename get_dbgmgr_mutex to kfd_get_dbgmgr_mutex to namespace it

Signed-off-by: Yair Shachar <yair.shachar@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: Add static user-mode queues support
Yair Shachar [Wed, 20 May 2015 10:43:04 +0000 (13:43 +0300)]
drm/amdkfd: Add static user-mode queues support

This patch adds support for static user-mode queues in QCM.
Queues which are designated as static can NOT be preempted by
the CP microcode when it is executing its scheduling algorithm.

This is needed for supporting the debugger feature, because we
can't allow the CP to preempt queues which are currently being debugged.

The number of queues that can be designated as static is limited by the
number of HQDs (Hardware Queue Descriptors).

Signed-off-by: Yair Shachar <yair.shachar@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: add H/W debugger IOCTL set definitions
Yair Shachar [Sun, 7 Dec 2014 15:05:22 +0000 (17:05 +0200)]
drm/amdkfd: add H/W debugger IOCTL set definitions

This patch adds four new IOCTLs to amdkfd. These IOCTLs expose a H/W
debugger functionality to the userspace.

The IOCTLs are:

- AMDKFD_IOC_DBG_REGISTER:

The purpose of this IOCTL is to notify amdkfd that a process wants to use
GPU debugging facilities on itself only.
It is expected that this IOCTL would be called before any other H/W
debugger requests are sent to amdkfd and for each GPU where the H/W
debugging needs to be enabled. The use of this IOCTL ensures that only
one instance of a debugger is active in the system.

- AMDKFD_IOC_DBG_UNREGISTER:

This IOCTL detaches the debugger/debugged process from the H/W
Debug which was established by the AMDKFD_IOC_DBG_REGISTER IOCTL.

- AMDKFD_IOC_DBG_ADDRESS_WATCH:

This IOCTL allows to set different watchpoints with various conditions as
indicated by the IOCTL's arguments. The available number of watchpoints
is retrieved from topology. This operation is confined to the current
debugged process, which was registered through AMDKFD_IOC_DBG_REGISTER.

- AMDKFD_IOC_DBG_WAVE_CONTROL:

This IOCTL allows to control a wavefront as indicated by the IOCTL's
arguments. For example, you can halt/resume or kill either a
single wavefront or a set of wavefronts. This operation is confined to
the current debugged process, which was registered through
AMDKFD_IOC_DBG_REGISTER.

Because the arguments for the address watch IOCTL and wave control IOCTL
are dynamic, meaning that they could vary in size, the userspace passes a
pointer to a structure (in userspace) that contains the value of the
arguments. The kernel driver is responsible to parse this structure and
validate its contents.

v2: change void* to uint64_t inside ioctl arguments

Signed-off-by: Yair Shachar <yair.shachar@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/radeon: Add H/W debugger kfd->kgd functions
Yair Shachar [Sun, 28 Sep 2014 08:51:15 +0000 (11:51 +0300)]
drm/radeon: Add H/W debugger kfd->kgd functions

This patch adds new interface functions to the kfd2kgd interface file. The
new functions allow to perform H/W debugger operations by writing to GPU
registers.

Signed-off-by: Yair Shachar <yair.shachar@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: Use DECLARE_BITMAP
Joe Perches [Wed, 20 May 2015 01:37:51 +0000 (18:37 -0700)]
drm/amdkfd: Use DECLARE_BITMAP

Use the generic mechanism to declare a bitmap instead of unsigned long.

It seems that "struct kfd_process.allocated_queue_bitmap" is unused.
Maybe it could be deleted instead.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/i915/bxt: edp1.4 Intermediate Freq support
Sonika Jindal [Tue, 26 May 2015 12:20:13 +0000 (17:50 +0530)]
drm/i915/bxt: edp1.4 Intermediate Freq support

BXT supports following intermediate link rates for edp:
2.16GHz, 2.43GHz, 3.24GHz, 4.32GHz.
Adding support for programming the intermediate rates.

v2: Adding clock in bxt_clk_div struct and then look for the entry with
required rate (Ville)
v3: 'clock' has the selected value, no need to use link_bw or rate_select
for selecting pll(Ville)
v4: Make bxt_dp_clk_val const and remove size (Ville)
v5: Rebased
v6: Removed setting of vco while rebasing in v5, adding it back

Signed-off-by: Sonika Jindal <sonika.jindal@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> (v4)
Reviewed-by: Vandana Kannan <vandana.kannan@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Move WaBarrierPerformanceFixDisable:skl to skl code from chv code
Ville Syrjälä [Tue, 2 Jun 2015 12:37:35 +0000 (15:37 +0300)]
drm/i915: Move WaBarrierPerformanceFixDisable:skl to skl code from chv code

 commit 65ca7514e21adbee25b8175fc909759c735d00ff
 Author: Damien Lespiau <damien.lespiau@intel.com>
 Date:   Mon Feb 9 19:33:22 2015 +0000

    drm/i915/skl: Implement WaBarrierPerformanceFixDisable

got misapplied and the code landed in chv_init_workarounds() instead of
the intended skl_init_workarounds(). Move it over to the right place.

Cc: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm_vblank_get: don't WARN_ON in case vblanks are not initialized
Gerd Hoffmann [Mon, 30 Mar 2015 10:50:50 +0000 (12:50 +0200)]
drm_vblank_get: don't WARN_ON in case vblanks are not initialized

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
9 years agobreak kconfig dependency loop
Gerd Hoffmann [Tue, 11 Nov 2014 08:17:00 +0000 (09:17 +0100)]
break kconfig dependency loop

After adding virtio-gpu I get this funky kconfig dependency loop.

scripts/kconfig/conf --oldconfig Kconfig
drivers/video/fbdev/Kconfig:5:error: recursive dependency detected!
drivers/video/fbdev/Kconfig:5:  symbol FB is selected by DRM_KMS_FB_HELPER
drivers/gpu/drm/Kconfig:34:     symbol DRM_KMS_FB_HELPER is selected by DRM_VIRTIO_GPU
drivers/gpu/drm/virtio/Kconfig:1:       symbol DRM_VIRTIO_GPU depends on VIRTIO
drivers/virtio/Kconfig:1:       symbol VIRTIO is selected by REMOTEPROC
drivers/remoteproc/Kconfig:4:   symbol REMOTEPROC is selected by OMAP_REMOTEPROC
drivers/remoteproc/Kconfig:12:  symbol OMAP_REMOTEPROC depends on OMAP_IOMMU
drivers/iommu/Kconfig:141:      symbol OMAP_IOMMU is selected by VIDEO_OMAP3
drivers/media/platform/Kconfig:96:      symbol VIDEO_OMAP3 depends on VIDEO_V4L2
drivers/media/v4l2-core/Kconfig:6:      symbol VIDEO_V4L2 depends on I2C
drivers/i2c/Kconfig:7:  symbol I2C is selected by FB_DDC
drivers/video/fbdev/Kconfig:59: symbol FB_DDC is selected by FB_CYBER2000_DDC
drivers/video/fbdev/Kconfig:374:        symbol FB_CYBER2000_DDC depends on FB_CYBER2000
drivers/video/fbdev/Kconfig:362:        symbol FB_CYBER2000 depends on FB

Making VIDEO_OMAP3 depend on OMAP_IOMMU instead of selecting it breaks the
loop, which looks like the best way to handle it to me.  Updated OMAP_IOMMU
help text accordingly.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
9 years agoMerge tag 'drm-intel-next-fixes-2015-05-29' of git://anongit.freedesktop.org/drm...
Dave Airlie [Tue, 2 Jun 2015 08:10:50 +0000 (18:10 +1000)]
Merge tag 'drm-intel-next-fixes-2015-05-29' of git://anongit.freedesktop.org/drm-intel into drm-next

Fixes for 4.2. Nothing too serious (given that it's still pre merge
window). With that it's off for 2 weeks of vacation for me and taking care
of 4.2 fixes for Jani.

* tag 'drm-intel-next-fixes-2015-05-29' of git://anongit.freedesktop.org/drm-intel:
  drm/i915: limit PPGTT size to 2GB in 32-bit platforms
  drm/i915: Another fbdev hack to avoid PSR on fbcon.
  drm/i915: Return the frontbuffer flip to enable intel_crtc_enable_planes.
  drm/i915: disable IPS while getting the sink CRCs
  drm/i915: Disable 12bpc hdmi for now
  drm/i915: Adjust sideband locking a bit for CHV/VLV
  drm/i915: s/dpio_lock/sb_lock/
  drm/i915: Kill intel_flush_primary_plane()
  drm/i915: Throw out WIP CHV power well definitions
  drm/i915: Use the default 600ns LDO programming sequence delay
  drm/i915: Remove unnecessary null check in execlists_context_unqueue
  drm/i915: Use spinlocks for checking when to waitboost
  drm/i915: Fix the confusing comment about the ioctl limits
  Revert "drm/i915: Force clean compilation with -Werror"