cascardo/linux.git
7 years agodrm/i915/mocs: || vs | typo in get_mocs_settings()
Dan Carpenter [Mon, 13 Jun 2016 06:54:22 +0000 (09:54 +0300)]
drm/i915/mocs: || vs | typo in get_mocs_settings()

It seems pretty clear that bitwise OR was intended here and not logical
OR.

Fixes: 6fc29133eafb ('drm/i915/gen9: Add WaDisableSkipCaching')
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
7 years agodrm/i915: Don't unregister fbdev's fb twice
Lukas Wunner [Wed, 8 Jun 2016 11:15:22 +0000 (13:15 +0200)]
drm/i915: Don't unregister fbdev's fb twice

Calling drm_framebuffer_unregister_private() in intel_fbdev_destroy() is
superfluous because the framebuffer will subsequently be unregistered by
drm_framebuffer_free() when unreferenced in drm_framebuffer_remove().
The call is a leftover, when it was introduced by commit 362063619cf6
("drm: revamp framebuffer cleanup interfaces"), struct intel_framebuffer
was still embedded in struct intel_fbdev rather than being a pointer as
it is today, and drm_framebuffer_remove() wasn't used yet.

As a bonus, the ID of the framebuffer is no longer 0 in the debug log:

Before:
    [   39.680874] [drm:drm_mode_object_unreference] OBJ ID: 0 (3)
    [   39.680878] [drm:drm_mode_object_unreference] OBJ ID: 0 (2)
    [   39.680884] [drm:drm_mode_object_unreference] OBJ ID: 0 (1)

After:
    [  102.504649] [drm:drm_mode_object_unreference] OBJ ID: 45 (3)
    [  102.504651] [drm:drm_mode_object_unreference] OBJ ID: 45 (2)
    [  102.504654] [drm:drm_mode_object_unreference] OBJ ID: 45 (1)

Signed-off-by: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/5031860caad67faa0f1be5965331ef048a311a01.1465383212.git.lukas@wunner.de
7 years agodrm/i915/gen9: implement WaConextSwitchWithConcurrentTLBInvalidate
Tim Gore [Mon, 13 Jun 2016 11:15:01 +0000 (12:15 +0100)]
drm/i915/gen9: implement WaConextSwitchWithConcurrentTLBInvalidate

This patch enables a workaround for a mid thread preemption
issue where a hardware timing problem can prevent the
context restore from happening, leading to a hang.

v2: move to gen9_init_workarounds (Arun)
v3: move to start of gen9_init_workarounds (Arun)

Signed-off-by: Tim Gore <tim.gore@intel.com>
Reviewed-by: Arun Siluvery <arun.siluvery@linux.intel.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465816501-25557-1-git-send-email-tim.gore@intel.com
7 years agodrm/i915: Support for pread/pwrite from/to non shmem backed objects
Ankitprasad Sharma [Fri, 10 Jun 2016 08:53:03 +0000 (14:23 +0530)]
drm/i915: Support for pread/pwrite from/to non shmem backed objects

This patch adds support for extending the pread/pwrite functionality
for objects not backed by shmem. The access will be made through
gtt interface. This will cover objects backed by stolen memory as well
as other non-shmem backed objects.

v2: Drop locks around slow_user_access, prefault the pages before
access (Chris)

v3: Rebased to the latest drm-intel-nightly (Ankit)

v4: Moved page base & offset calculations outside the copy loop,
corrected data types for size and offset variables, corrected if-else
braces format (Tvrtko/kerneldocs)

v5: Enabled pread/pwrite for all non-shmem backed objects including
without tiling restrictions (Ankit)

v6: Using pwrite_fast for non-shmem backed objects as well (Chris)

v7: Updated commit message, Renamed i915_gem_gtt_read to i915_gem_gtt_copy,
added pwrite slow path for non-shmem backed objects (Chris/Tvrtko)

v8: Updated v7 commit message, mutex unlock around pwrite slow path for
non-shmem backed objects (Tvrtko)

v9: Corrected check during pread_ioctl, to avoid shmem_pread being
called for non-shmem backed objects (Tvrtko)

v10: Moved the write_domain check to needs_clflush and tiling mode check
to pwrite_fast (Chris)

v11: Use pwrite_fast fallback for all objects (shmem and non-shmem backed),
call fast_user_write regardless of pagefault in previous iteration

v12: Use page-by-page copy for slow user access too (Chris)

v13: Handled EFAULT, Avoid use of WARN_ON, put_fence only if whole obj
pinned (Chris)

v14: Corrected datatypes/initializations (Tvrtko)

Testcase: igt/gem_stolen, igt/gem_pread, igt/gem_pwrite

Signed-off-by: Ankitprasad Sharma <ankitprasad.r.sharma@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465548783-19712-1-git-send-email-ankitprasad.r.sharma@intel.com
7 years agodrm/i915: Use insert_page for pwrite_fast
Ankitprasad Sharma [Fri, 10 Jun 2016 08:53:01 +0000 (14:23 +0530)]
drm/i915: Use insert_page for pwrite_fast

In pwrite_fast, map an object page by page if obj_ggtt_pin fails. First,
we try a nonblocking pin for the whole object (since that is fastest if
reused), then failing that we try to grab one page in the mappable
aperture. It also allows us to handle objects larger than the mappable
aperture (e.g. if we need to pwrite with vGPU restricting the aperture
to a measely 8MiB or something like that).

v2: Pin pages before starting pwrite, Combined duplicate loops (Chris)

v3: Combined loops based on local patch by Chris (Chris)

v4: Added i915 wrapper function for drm_mm_insert_node_in_range (Chris)

v5: Renamed wrapper function for drm_mm_insert_node_in_range (Chris)

v5: Added wrapper for drm_mm_remove_node() (Chris)

v6: Added get_pages call before pinning the pages (Tvrtko)
Added remove_mappable_node() wrapper for drm_mm_remove_node() (Chris)

v7: Added size argument for insert_mappable_node (Tvrtko)

v8: Do not put_pages after pwrite, do memset of node in the wrapper
function (insert_mappable_node) (Chris)

v9: Rebase (Ankit)

Signed-off-by: Ankitprasad Sharma <ankitprasad.r.sharma@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
7 years agodrm/i915: Introduce i915_gem_object_get_dma_address()
Chris Wilson [Fri, 10 Jun 2016 08:53:00 +0000 (14:23 +0530)]
drm/i915: Introduce i915_gem_object_get_dma_address()

This utility function is a companion to i915_gem_object_get_page() that
uses the same cached iterator for the scatterlist to perform fast
sequential lookup of the dma address associated with any page within the
object.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Ankitprasad Sharma <ankitprasad.r.sharma@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
7 years agodrm/i915: Add support for mapping an object page by page
Chris Wilson [Fri, 10 Jun 2016 08:52:59 +0000 (14:22 +0530)]
drm/i915: Add support for mapping an object page by page

Introduced a new vm specfic callback insert_page() to program a single pte in
ggtt or ppgtt. This allows us to map a single page in to the mappable aperture
space. This can be iterated over to access the whole object by using space as
meagre as page size.

v2: Added low level rpm assertions to insert_page routines (Chris)

v3: Added POSTING_READ post register write (Tvrtko)

v4: Rebase (Ankit)

v5: Removed wmb() and FLUSH_CTL from insert_page, caller to take care
of it (Chris)

v6: insert_page not working correctly without FLSH_CNTL write, added the
write again.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Ankitprasad Sharma <ankitprasad.r.sharma@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
7 years agodrm/i915/guc: suppress GuC-related message on non-GuC platforms
Dave Gordon [Fri, 10 Jun 2016 16:21:25 +0000 (17:21 +0100)]
drm/i915/guc: suppress GuC-related message on non-GuC platforms

If the user doesn't override the default values of the GuC-related
kernel parameters, then on a non-GuC-based platform we shouldn't
mention that we haven't loaded the GuC firmware.

The various messages have been reordered into a least->most severe
cascade (none/INFO/INFO/ERROR) for ease of comprehension.

Signed-off-by: Dave Gordon <david.s.gordon@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465575685-34169-1-git-send-email-david.s.gordon@intel.com
7 years agoRevert "drm/i915/ilk: Don't disable SSC source if it's in use"
Daniel Vetter [Thu, 9 Jun 2016 16:39:07 +0000 (18:39 +0200)]
Revert "drm/i915/ilk: Don't disable SSC source if it's in use"

This reverts commit f165d2834ceb3d5c29bebadadc27629bebf402ac.

It breaks one of our CI systems. Quoting from Ville:

[   13.100979] [drm:ironlake_init_pch_refclk] has_panel 1 has_lvds 1 has_ck505 0 using_ssc_source 1
[   13.101413] ------------[ cut here ]------------
[   13.101429] kernel BUG at drivers/gpu/drm/i915/intel_display.c:8528!

"which is the 'BUG_ON(val != final)' at the end of ironlake_init_pch_refclk()."

Cc: stable@vger.kernel.org
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Lyude <cpaul@redhat.com>
Cc: marius.c.vlad@intel.com
References: https://www.spinics.net/lists/dri-devel/msg109557.html
Acked-by: Lyude <cpaul@redhat.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
7 years agodrm/i915/gen9: Add WaFbcHighMemBwCorruptionAvoidance
Mika Kuoppala [Tue, 7 Jun 2016 14:19:19 +0000 (17:19 +0300)]
drm/i915/gen9: Add WaFbcHighMemBwCorruptionAvoidance

Add this fbc related workaround for all gen9

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-28-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i195/fbc: Add WaFbcNukeOnHostModify
Mika Kuoppala [Tue, 7 Jun 2016 14:19:18 +0000 (17:19 +0300)]
drm/i195/fbc: Add WaFbcNukeOnHostModify

Bspec states that we need to set nuke on modify all to prevent
screen corruption with fbc on skl and kbl.

v2: proper workaround name

References: HSD#2227109, HSDES#1404569388
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-27-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/gen9: Add WaFbcWakeMemOn
Mika Kuoppala [Tue, 7 Jun 2016 14:19:17 +0000 (17:19 +0300)]
drm/i915/gen9: Add WaFbcWakeMemOn

Set bit 8 in 0x43224 to prevent screen corruption and system
hangs on high memory bandwidth conditions. The same wa also suggest
setting bit 31 on ARB_CTL. According to another workaround we gain
better idle power savings when FBC is enabled.

v2: use correct workaround name
v3: split out overlapping wa for corruption avoidance (Ville)

References: HSD#2137218, HSD#2227171, HSD#2136579, BSID#883
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-26-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/gen9: Add WaFbcTurnOffFbcWatermark
Mika Kuoppala [Tue, 7 Jun 2016 14:19:16 +0000 (17:19 +0300)]
drm/i915/gen9: Add WaFbcTurnOffFbcWatermark

According to bspec this prevents screen corruption when fbc is
used.

v2: This workaround has a name, use it (Ville)
v3: remove bogus gen check on ilk/vlv wm path (Ville)

References: HSD#2135555, HSD#2137270, BSID#562
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-25-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/kbl: Add WaClearSlmSpaceAtContextSwitch
Mika Kuoppala [Tue, 7 Jun 2016 14:19:15 +0000 (17:19 +0300)]
drm/i915/kbl: Add WaClearSlmSpaceAtContextSwitch

This workaround for bdw and chv, is also needed for kbl A0.

References: HSD#1911519, BSID#569
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-24-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/skl: Extend WaDisableChickenBitTSGBarrierAckForFFSliceCS
Mika Kuoppala [Tue, 7 Jun 2016 14:19:14 +0000 (17:19 +0300)]
drm/i915/skl: Extend WaDisableChickenBitTSGBarrierAckForFFSliceCS

There is ambiguity in the documentation between D0 and E0.
Extend this workaround to E0.

References: BSID#779
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-23-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/gen9: Add WaEnableChickenDCPR
Mika Kuoppala [Tue, 7 Jun 2016 14:19:13 +0000 (17:19 +0300)]
drm/i915/gen9: Add WaEnableChickenDCPR

Workaround for display underrun issues with Y & Yf Tiling.
Set this on all gen9 as stated by bspec.

v2: proper workaround name

References: HSD#2136383, BSID#857
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-22-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/kbl: Add WaDisableSbeCacheDispatchPortSharing
Mika Kuoppala [Tue, 7 Jun 2016 14:19:12 +0000 (17:19 +0300)]
drm/i915/kbl: Add WaDisableSbeCacheDispatchPortSharing

This is needed for all kbl revision.

v2: Don't add revid checks to generic gen9 init (Arun)

References: HSD#2135593
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-21-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/kbl: Add WaDisableGafsUnitClkGating
Mika Kuoppala [Tue, 7 Jun 2016 14:19:11 +0000 (17:19 +0300)]
drm/i915/kbl: Add WaDisableGafsUnitClkGating

We need to disable clock gating in this unit to work around
hardware issue causing possible corruption/hang.

v2: name the bit (Ville)
v3: leave the fix enabled for 2227050 and set correct bit (Matthew)
v4: Split out the skl part in separate commit for easier backport

References: HSD#2227156, HSD#2227050
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-20-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/kbl: Add WaForGAMHang
Mika Kuoppala [Tue, 7 Jun 2016 14:19:10 +0000 (17:19 +0300)]
drm/i915/kbl: Add WaForGAMHang

Add this workaround for A0 and B0 revisions

References: HSD#2226935
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-19-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/skl: Add WAC6entrylatency
Mika Kuoppala [Tue, 7 Jun 2016 14:19:09 +0000 (17:19 +0300)]
drm/i915/skl: Add WAC6entrylatency

This workaround is for fbc working with rc6 on skylake. Bspec
states that setting this bit needs to be coordinated with uncore
but offers no further details.

v2: rebase

References: HSD#4712857
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-18-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/gen9: Add WaDisableSkipCaching
Mika Kuoppala [Tue, 7 Jun 2016 14:19:08 +0000 (17:19 +0300)]
drm/i915/gen9: Add WaDisableSkipCaching

Make sure that we never enable skip caching on gen9 by
accident.

References: HSD#2134698
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-17-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915: Add WaInsertDummyPushConstP for bxt and kbl
Mika Kuoppala [Tue, 7 Jun 2016 14:19:07 +0000 (17:19 +0300)]
drm/i915: Add WaInsertDummyPushConstP for bxt and kbl

Add this workaround for both bxt and kbl up to until
rev B0.

References: HSD#2136703
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-16-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/kbl: Add WaDisableDynamicCreditSharing
Mika Kuoppala [Tue, 7 Jun 2016 14:19:06 +0000 (17:19 +0300)]
drm/i915/kbl: Add WaDisableDynamicCreditSharing

Bspec states that we need to turn off dynamic credit
sharing on kbl revid a0 and b0. This happens by writing bit 28
on 0x4ab8.

References: HSD#2225601, HSD#2226938, HSD#2225763
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-15-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/kbl: Add WaDisableGamClockGating
Mika Kuoppala [Tue, 7 Jun 2016 14:19:05 +0000 (17:19 +0300)]
drm/i915/kbl: Add WaDisableGamClockGating

According to bspec we need to disable gam unit clock gating on
on kbl revids A0 and B0.

References: HSD#2226858, HSD#1944358
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-14-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/gen9: Enable must set chicken bits in config0 reg
Mika Kuoppala [Tue, 7 Jun 2016 14:19:04 +0000 (17:19 +0300)]
drm/i915/gen9: Enable must set chicken bits in config0 reg

The bspec states that these must be set in CONFIG0 for all gen9.

v2: rebase
v3: fix spacing (Matthew)

References: HSD#2134995
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-13-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/kbl: Add WaDisableLSQCROPERFforOCL
Mika Kuoppala [Tue, 7 Jun 2016 14:19:03 +0000 (17:19 +0300)]
drm/i915/kbl: Add WaDisableLSQCROPERFforOCL

Extend the scope of this workaround, already used in skl,
to also take effect in kbl.

v2: Fix KBL_REVID_E0 (Matthew)

References: HSD#2132677
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-12-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/edp: Add WaKVMNotificationOnConfigChange:bdw
Mika Kuoppala [Tue, 7 Jun 2016 14:19:02 +0000 (17:19 +0300)]
drm/i915/edp: Add WaKVMNotificationOnConfigChange:bdw

According to bspec this workaround helps to reduce lag and improve
performance on edp.

Documentation suggests this for bdw and all gen9. However evidence
shows that this register is missing on gen9 and causing unclaimed mmio
access if we access it. So apply to bdw only where the reg
exists and can hold its value.

v2: drop skl

References: HSD#2134579
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-11-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/kbl: Add WaDisableSDEUnitClockGating
Mika Kuoppala [Tue, 7 Jun 2016 14:19:01 +0000 (17:19 +0300)]
drm/i915/kbl: Add WaDisableSDEUnitClockGating

Add this workaround until upto kbl revid B0.

References: HSD#1802092
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-10-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/kbl: Add WaDisableFenceDestinationToSLM for A0
Mika Kuoppala [Tue, 7 Jun 2016 14:19:00 +0000 (17:19 +0300)]
drm/i915/kbl: Add WaDisableFenceDestinationToSLM for A0

Add this workaround for kbl revid A0 only.

v2: rebase
v3: carve out a non related workaround (Chris)

References: HSD#1911714
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-9-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/kbl: Add WaEnableGapsTsvCreditFix
Mika Kuoppala [Tue, 7 Jun 2016 14:18:59 +0000 (17:18 +0300)]
drm/i915/kbl: Add WaEnableGapsTsvCreditFix

We need this crucial workaround from skl also to all kbl revisions.
Lack of it was causing system hangs on skl enabling so this is
a must have.

v2: Don't add revid checks to gen9 init workarounds (Arun)

References: HSD#2126660
Cc: Arun Siluvery <arun.siluvery@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-8-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915: Mimic skl with WaForceEnableNonCoherent
Mika Kuoppala [Tue, 7 Jun 2016 14:18:58 +0000 (17:18 +0300)]
drm/i915: Mimic skl with WaForceEnableNonCoherent

Past evidence with system hangs and hsds tie
WaForceEnableNonCoherent and WaDisableHDCInvalidation to
WaForceContextSaveRestoreNonCoherent. Documentation
states that WaForceContextSaveRestoreNonCoherent would
not be needed on skl past E0 but evidence proved otherwise. See
commit <510650e8b2ab> ("drm/i915/skl: Fix spurious gpu hang with gt3/gt4
revs"). In this scope consider kbl to be skl with a bigger revision than
E0 so play it safe and bind these two workarounds to the
WaForceContextSaveRestoreNonCoherent, and apply to all gen9.

v2: fix comment (Matthew)

References: HSD#2134449, HSD#2131413
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-7-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/gen9: Always apply WaForceContextSaveRestoreNonCoherent
Mika Kuoppala [Tue, 7 Jun 2016 14:18:57 +0000 (17:18 +0300)]
drm/i915/gen9: Always apply WaForceContextSaveRestoreNonCoherent

The revision id range for this workaround has changed. So apply
it to all revids on all gen9.

References: HSD#2134449
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-6-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/kbl: Add WaSkipStolenMemoryFirstPage for A0
Mika Kuoppala [Tue, 7 Jun 2016 14:18:56 +0000 (17:18 +0300)]
drm/i915/kbl: Add WaSkipStolenMemoryFirstPage for A0

We need this for kbl a0 boards. Note that this should be also
for bxt A0 but we omit that on purpose as bxt A0's are
out of fashion already.

References: HSD#1912158, HSD#4393097
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-5-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/kbl: Add REVID macro
Mika Kuoppala [Tue, 7 Jun 2016 14:18:55 +0000 (17:18 +0300)]
drm/i915/kbl: Add REVID macro

Add REVID macro for kbl to limit wa applicability to particular
revision range.

Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-4-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/kbl: Init gen9 workarounds
Mika Kuoppala [Tue, 7 Jun 2016 14:18:54 +0000 (17:18 +0300)]
drm/i915/kbl: Init gen9 workarounds

Kabylake is part of gen9 family so init the generic gen9
workarounds for it.

v2: rebase

Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-3-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915/skl: Add WaDisableGafsUnitClkGating
Mika Kuoppala [Tue, 7 Jun 2016 14:18:53 +0000 (17:18 +0300)]
drm/i915/skl: Add WaDisableGafsUnitClkGating

We need to disable clock gating in this unit to work around
hardware issue causing possible corruption/hang.

v2: name the bit (Ville)
v3: leave the fix enabled for 2227050 and set correct bit (Matthew)

References: HSD#2227156, HSD#2227050
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465309159-30531-2-git-send-email-mika.kuoppala@intel.com
7 years agodrm/i915: only disable memory self-refresh on GMCH
David Weinehall [Thu, 19 May 2016 12:50:36 +0000 (15:50 +0300)]
drm/i915: only disable memory self-refresh on GMCH

The atomic version of intel_pre_plane_update did not check
for HAS_GMCH_DISPLAY before calling intel_set_memory_cxsr().
While this doesn't cause any issues on its own (it will
return without doing anything if the hardware doesn't
have the required feature), the drm_wait_one_vblank() that
is needed if memory self-refresh is disabled introduces
an unnecessary delay in the suspend path.

In cases where i915 is on the critical path it means that
we slow down suspend by 16.8ms on platforms that don't
need to disable memory self-refresh.

Signed-off-by: David Weinehall <david.weinehall@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1463662236-18192-1-git-send-email-david.weinehall@linux.intel.com
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
7 years agodrm/i915: Check VBT for port presence in addition to the strap on VLV/CHV
Ville Syrjälä [Fri, 3 Jun 2016 09:17:43 +0000 (12:17 +0300)]
drm/i915: Check VBT for port presence in addition to the strap on VLV/CHV

Apparently some CHV boards failed to hook up the port presence straps
for HDMI ports as well (earlier we assumed this problem only affected
eDP ports). So let's check the VBT in addition to the strap, and if
either one claims that the port is present go ahead and register the
relevant connector.

While at it, change port D to register DP before HDMI as we do for ports
B and C since
commit 457c52d87e5d ("drm/i915: Only ignore eDP ports that are connected")

Also print a debug message when we register a HDMI connector to aid
in diagnosing missing/incorrect ports. We already had such a print for
DP/eDP.

v2: Improve the comment in the code a bit, note the port D change in
    the commit message

Cc: Radoslav Duda <radosd@radosd.com>
Tested-by: Radoslav Duda <radosd@radosd.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=96321
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464945463-14364-1-git-send-email-ville.syrjala@linux.intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
7 years agodrm/i915/guc: enable GuC loading & submission by default
Dave Gordon [Tue, 7 Jun 2016 08:14:51 +0000 (09:14 +0100)]
drm/i915/guc: enable GuC loading & submission by default

The recent patch
fce91f2 drm/i915/guc: add enable_guc_loading parameter
enabled GuC loading and submission by default, but as issues
were found with warnings being issued during suspend-resume
cycles, GuC loading was disabled by default, by patch
2335986 drm/i915/guc: Disable automatic GuC firmware loading

Those warnings have been resolved, so this patch re-enables GuC
loading and submission by default.

Signed-off-by: Dave Gordon <david.s.gordon@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465287291-2187-1-git-send-email-david.s.gordon@intel.com
7 years agodrm/i915/guc: disable GuC submission earlier during GuC (re)load
Dave Gordon [Tue, 7 Jun 2016 08:14:50 +0000 (09:14 +0100)]
drm/i915/guc: disable GuC submission earlier during GuC (re)load

When resetting and reloading the GuC, the GuC submission management code
also needs to destroy and recreate the GuC client(s). Currently this is
done by a separate call from the GuC loader, but really, it's just an
internal detail of the submission code. So here we remove the call from
the loader (which is too late, really, because the GuC has already been
reloaded at this point) and put it into guc_submission_init() instead.
This means that any preexisting client is destroyed *before* the GuC
(re)load and then recreated after, iff the firmware was successfully
loaded. If the GuC reload fails, we don't recreate the client, so
fallback to execlists mode (if active) won't leak the client object
(previously, the now-unusable client would have been left allocated,
and leaked if the driver were unloaded).

Signed-off-by: Dave Gordon <david.s.gordon@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
7 years agodrm/i915/guc: fix GuC loading/submission check
Dave Gordon [Tue, 7 Jun 2016 08:14:49 +0000 (09:14 +0100)]
drm/i915/guc: fix GuC loading/submission check

The last stage of the GuC loader also sanitises the GuC submission
settings, so should be called unconditionally (even on platforms
without a GuC) to ensure consistent settings; in particular, this
prevents any attempt to use GuC submission on GuCless platforms!

Also fix error path handling and clarify DRM_INFO fallback message.

Signed-off-by: Dave Gordon <david.s.gordon@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
7 years agodrm/i915/gen9: Add WaVFEStateAfterPipeControlwithMediaStateClear
arun.siluvery@linux.intel.com [Mon, 6 Jun 2016 08:52:49 +0000 (09:52 +0100)]
drm/i915/gen9: Add WaVFEStateAfterPipeControlwithMediaStateClear

Kernel only need to add a register to HW whitelist, required for a
preemption related issue.

Reference: HSD#2131039
Reviewed-by: Jeff McGee <jeff.mcgee@intel.com>
Signed-off-by: Arun Siluvery <arun.siluvery@linux.intel.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1465203169-16591-1-git-send-email-arun.siluvery@linux.intel.com
7 years agodrm/i915: Fix a buch of kerneldoc warnings
Tvrtko Ursulin [Fri, 3 Jun 2016 13:02:17 +0000 (14:02 +0100)]
drm/i915: Fix a buch of kerneldoc warnings

Just a bunch of stale kerneldocs generating warnings when
building the docs. Mostly function parameters so not very
useful but still.

v2: Tidy.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1464958937-23344-1-git-send-email-tvrtko.ursulin@linux.intel.com
7 years agodrm/i915/dsi: fix bxt split screen and color issue
Jani Nikula [Fri, 3 Jun 2016 14:57:05 +0000 (17:57 +0300)]
drm/i915/dsi: fix bxt split screen and color issue

Fix the failure mode where the display appears split, or shifted about
2/3 of the screen, and the color components are cycled. Turns out we
were missing the crucial BXT_DEFEATURE_DPI_FIFO_CTR bit in the
EOT_DISABLE register.

Per bspec, with the bit set, the "mipi_dpf_vblank_start" signal is
asserted only when the complete frame is transferred in the DPHY line
and also the DPI FIFO is flushed out at the end of each frame.

The problem was mitigated by keeping the panel fitter enabled, but that
only limited the issue to a shift of about 0..10 pixels. With the fix
here, the panel fitter workaround does not seem to be needed at all.

While at it, set BXT_DPHY_DEFEATURE_EN in EOT_DISABLE register which is
also needed per the BXT DSI mode set sequence.

Issue: VIZ-7610
Cc: Mika Kahola <mika.kahola@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Ramalingam C <ramalingam.c@intel.com>
Cc: Uma Shankar <uma.shankar@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464965825-31035-1-git-send-email-jani.nikula@intel.com
7 years agodrm/i915: Update DRIVER_DATE to 20160606
Daniel Vetter [Sun, 5 Jun 2016 22:29:53 +0000 (00:29 +0200)]
drm/i915: Update DRIVER_DATE to 20160606

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
7 years agodrm/i915: Extract physical display dimensions from VBT
Ville Syrjälä [Tue, 31 May 2016 09:08:34 +0000 (12:08 +0300)]
drm/i915: Extract physical display dimensions from VBT

The VBT has these mysterious H/V image sizes as part of the display
timings. Looking at some dumps those appear to be the physical
dimensions in mm. Which makes sense since the timing descriptor matches
the format used by EDID detailed timing descriptor, which defines these
as "H/V Addressable Video Image Size in mm".

So let's use that information from the panel fixed mode to get the
physical dimensions for LVDS/eDP/DSI displays. And with that we can
fill out the display_info so that userspace can get at it via
GetConnector.

v2: Use (hi<<8)|lo instead of broken (hi<<4)+lo
    Handle LVDS and eDP too

Cc: Stephen Just <stephenjust@gmail.com>
Tested-by: Stephen Just <stephenjust@gmail.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=96255
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464685714-30507-1-git-send-email-ville.syrjala@linux.intel.com
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
7 years agodrm/i915: Silence "unexpected child device config size" for VBT on 845g
Chris Wilson [Wed, 1 Jun 2016 17:08:43 +0000 (18:08 +0100)]
drm/i915: Silence "unexpected child device config size" for VBT on 845g

My old 845g complains that the child_device_size inside its VBT,
version 110, is incorrect. Let's fiddle with the version matching such
that it works with this VBT (i.e. treat BIOS v110 as having the same size
as v108).

Fixes [drm:intel_bios_init] *ERROR* Unexpected child device config
size 27 (expected 33 for VBT version 110)

Whether this is correct, no one knows - but it works for this particular
machine.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Acked-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464800923-6054-1-git-send-email-chris@chris-wilson.co.uk
7 years agoMerge remote-tracking branch 'airlied/drm-next' into drm-intel-next-queued
Daniel Vetter [Thu, 2 Jun 2016 07:54:12 +0000 (09:54 +0200)]
Merge remote-tracking branch 'airlied/drm-next' into drm-intel-next-queued

Git got absolutely destroyed with all our cherry-picking from
drm-intel-next-queued to various branches. It ended up inserting
intel_crtc_page_flip 2x even in intel_display.c.

Backmerge to get back to sanity.

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
7 years agoMerge branch 'drm-intel-next' of git://anongit.freedesktop.org/drm-intel into drm...
Dave Airlie [Wed, 1 Jun 2016 21:58:36 +0000 (07:58 +1000)]
Merge branch 'drm-intel-next' of git://anongit.freedesktop.org/drm-intel into drm-next

drm-intel-next-2016-05-22:
- cmd-parser support for direct reg->reg loads (Ken Graunke)
- better handle DP++ smart dongles (Ville)
- bxt guc fw loading support (Nick Hoathe)
- remove a bunch of struct typedefs from dpll code (Ander)
- tons of small work all over to avoid casting between drm_device and the i915
  dev struct (Tvrtko&Chris)
- untangle request retiring from other operations, also fixes reset stat corner
  cases (Chris)
- skl atomic watermark support from Matt Roper, yay!
- various wm handling bugfixes from Ville
- big pile of cdclck rework for bxt/skl (Ville)
- CABC (Content Adaptive Brigthness Control) for dsi panels (Jani&Deepak M)
- nonblocking atomic commits for plane-only updates (Maarten Lankhorst)
- bunch of PSR fixes&improvements
- untangle our map/pin/sg_iter code a bit (Dave Gordon)
drm-intel-next-2016-05-08:
- refactor stolen quirks to share code between early quirks and i915 (Joonas)
- refactor gem BO/vma funcstion (Tvrtko&Dave)
- backlight over DPCD support (Yetunde Abedisi)
- more dsi panel sequence support (Jani)
- lots of refactoring around handling iomaps, vma, ring access and related
  topics culmulating in removing the duplicated request tracking in the execlist
  code (Chris & Tvrtko) includes a small patch for core iomapping code
- hw state readout for bxt dsi (Ramalingam C)
- cdclk cleanups (Ville)
- dedupe chv pll code a bit (Ander)
- enable semaphores on gen8+ for legacy submission, to be able to have a direct
  comparison against execlist on the same platform (Chris) Not meant to be used
  for anything else but performance tuning
- lvds border bit hw state checker fix (Jani)
- rpm vs. shrinker/oom-notifier fixes (Praveen Paneri)
- l3 tuning (Imre)
- revert mst dp audio, it's totally non-functional and crash-y (Lyude)
- first official dmc for kbl (Rodrigo)
- and tons of small things all over as usual

* 'drm-intel-next' of git://anongit.freedesktop.org/drm-intel: (194 commits)
  drm/i915: Revert async unpin and nonblocking atomic commit
  drm/i915: Update DRIVER_DATE to 20160522
  drm/i915: Inline sg_next() for the optimised SGL iterator
  drm/i915: Introduce & use new lightweight SGL iterators
  drm/i915: optimise i915_gem_object_map() for small objects
  drm/i915: refactor i915_gem_object_pin_map()
  drm/i915/psr: Implement PSR2 w/a for gen9
  drm/i915/psr: Use ->get_aux_send_ctl functions
  drm/i915/psr: Order DP aux transactions correctly
  drm/i915/psr: Make idle_frames sensible again
  drm/i915/psr: Try to program link training times correctly
  drm/i915/userptr: Convert to drm_i915_private
  drm/i915: Allow nonblocking update of pageflips.
  drm/i915: Check for unpin correctness.
  Reapply "drm/i915: Avoid stalling on pending flips for legacy cursor updates"
  drm/i915: Make unpin async.
  drm/i915: Prepare connectors for nonblocking checks.
  drm/i915: Pass atomic states to fbc update functions.
  drm/i915: Remove reset_counter from intel_crtc.
  drm/i915: Remove queue_flip pointer.
  ...

7 years agoMerge tag 'topic/drm-misc-2016-06-01' of git://anongit.freedesktop.org/drm-intel...
Dave Airlie [Wed, 1 Jun 2016 21:50:23 +0000 (07:50 +1000)]
Merge tag 'topic/drm-misc-2016-06-01' of git://anongit.freedesktop.org/drm-intel into drm-next

Frist -misc pull for 4.8, with pretty much just random all over plus a few
more lockless gem BO patches acked/reviewed by driver maintainers.

I'm starting a bit earlier this time around because there's a few invasive
patch series to land (nonblocking atomic prep work, fence prep work,
rst/sphinx kerneldoc finally happening) and I need a baseline with all the
branches merged.

* tag 'topic/drm-misc-2016-06-01' of git://anongit.freedesktop.org/drm-intel: (21 commits)
  drm/vc4: Use lockless gem BO free callback
  drm/vc4: Use drm_gem_object_unreference_unlocked
  drm: Initialize a linear gamma table by default
  drm/vgem: Use lockless gem BO free callback
  drm/qxl: Don't set a gamma table size
  drm/msm: Nuke dummy gamma_set/get functions
  drm/cirrus: Drop redundnant gamma size check
  drm/fb-helper: Remove dead code in setcolreg
  drm/mediatek: Use lockless gem BO free callback
  drm/hisilicon: Use lockless gem BO free callback
  drm/hlcd: Use lockless gem BO free callback
  vga_switcheroo: Support deferred probing of audio clients
  vga_switcheroo: Add helper for deferred probing
  virtio-gpu: fix output lookup
  drm/doc: Unify KMS Locking docs
  drm/atomic-helper: Do not call ->mode_fixup for CRTC which will be disabled
  Fix annoyingly awkward typo in drm_edid_load.c
  drm/doc: Drop vblank_disable_allow wording
  drm: use seqlock for vblank time/count
  drm/mm: avoid possible null pointer dereference
  ...

7 years agodrm/i915/skl+: Use scaling amount for plane data rate calculation (v4)
Kumar, Mahesh [Thu, 19 May 2016 22:03:01 +0000 (15:03 -0700)]
drm/i915/skl+: Use scaling amount for plane data rate calculation (v4)

if downscaling is enabled plane data rate increases according to scaling
amount. take scaling amount under consideration while calculating plane
data rate

v2: Address Matt's comments, where data rate was overridden because of
missing else.

v3 (by Matt):
 - Add braces to 'else' branch to match kernel coding style
 - Adjust final calculation now that skl_plane_downscale_amount()
   returns 16.16 fixed point value instead of a decimal fixed point

v4 (by Matt):
 - Avoid integer overflow by making sure final multiplication is
   treated as 64-bit.

Cc: matthew.d.roper@intel.com
Signed-off-by: Kumar, Mahesh <mahesh1.kumar@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Kumar Mahesh <mahesh1.kumar@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1463695381-21368-1-git-send-email-matthew.d.roper@intel.com
7 years agodrm/i915/skl+: calculate plane pixel rate (v4)
Kumar, Mahesh [Mon, 16 May 2016 22:52:00 +0000 (15:52 -0700)]
drm/i915/skl+: calculate plane pixel rate (v4)

Don't use pipe pixel rate for plane pixel rate. Calculate plane pixel according
to formula

adjusted plane_pixel_rate = adjusted pipe_pixel_rate * downscale ammount

downscale amount = max[1, src_h/dst_h] * max[1, src_w/dst_w]
if 90/270 rotation use rotated width & height

v2: use intel_plane_state->visible instead of (fb == NULL) as per Matt's
    comment.

v3 (by Matt):
 - Keep downscale amount in 16.16 fixed point rather than converting to
   decimal fixed point.
 - Store adjusted plane pixel rate in plane state instead of the plane
   parameters structure that we no longer use.

v4 (by Matt):
 - Significant rebasing onto latest atomic watermark work
 - Don't bother storing plane pixel rate in state; just calculate it
   right before the calls that make use of it.
 - Fix downscale calculations to actually use width values when
   computing downscale_w rather than copy/pasted height values.

Cc: matthew.d.roper@intel.com
Signed-off-by: Kumar, Mahesh <mahesh1.kumar@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Kumar Mahesh <mahesh1.kumar@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1463439121-28974-4-git-send-email-matthew.d.roper@intel.com
7 years agodrm/i915/skl+: calculate ddb minimum allocation (v6)
Kumar, Mahesh [Tue, 31 May 2016 16:58:59 +0000 (09:58 -0700)]
drm/i915/skl+: calculate ddb minimum allocation (v6)

don't always use 8 ddb as minimum, instead calculate using proper
algorithm.

v2: optimizations as per Matt's comments.

v3 (by Matt):
 - Fix boolean logic for !fb test in skl_ddb_min_alloc()
 - Adjust negative tiling format comparisons in skl_ddb_min_alloc() to
   improve readability.

v4 (by Matt):
 - Rebase onto recent atomic watermark changes
 - Slight tweaks to code flow to make the logic more closely match the
   description in the bspec.

v5 (by Matt):
 - Handle minimum scanline calculation properly for 4 & 8 bpp formats.
   8bpp isn't actually possible right now, but it's listed in the bspec
   so I've included it here for forward compatibility (similar to how
   we have logic for NV12).

v6 (by Matt):
 - Calculate plane_bpp correctly for non-NV12 formats. (Mahesh)

Cc: matthew.d.roper@intel.com
Signed-off-by: Kumar, Mahesh <mahesh1.kumar@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Kumar Mahesh <mahesh1.kumar@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464713939-10440-1-git-send-email-matthew.d.roper@intel.com
7 years agodrm/i915: Don't try to calculate relative data rates during hw readout
Matt Roper [Mon, 16 May 2016 22:51:58 +0000 (15:51 -0700)]
drm/i915: Don't try to calculate relative data rates during hw readout

We don't actually read out full plane state during driver startup (only
whether the primary plane is enabled/disabled), so all of the src/dest
rectangles are invalid at this point.  However this calculation was
needless anyway since we re-calculate them from scratch on the very
first atomic transaction after boot anyway.

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Kumar Mahesh <mahesh1.kumar@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1463439121-28974-2-git-send-email-matthew.d.roper@intel.com
7 years agodrm/i915: Only ignore eDP ports that are connected
Chris Wilson [Wed, 1 Jun 2016 07:27:50 +0000 (08:27 +0100)]
drm/i915: Only ignore eDP ports that are connected

If the VBT says that a certain port should be eDP (and hence fused off
from HDMI), but in reality it isn't, we need to try and acquire the HDMI
connection instead. So only trust the VBT edp setting if we can connect
to an eDP device on that port.

Fixes: d2182a6608 (drm/i915: Don't register HDMI connectors for eDP ports on VLV/CHV)
References: https://bugs.freedesktop.org/show_bug.cgi?id=96288
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: Phidias Chiang <phidias.chiang@canonical.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464766070-31623-1-git-send-email-chris@chris-wilson.co.uk
7 years agodrm/i915: Revert async unpin and nonblocking atomic commit
Daniel Vetter [Tue, 24 May 2016 15:13:53 +0000 (17:13 +0200)]
drm/i915: Revert async unpin and nonblocking atomic commit

This reverts the following patches:

d55dbd06bb5e1399aba9ab5227465339d1bbefff drm/i915: Allow nonblocking update of pageflips.
15c86bdb760185e871c7a0f559978328aa500971 drm/i915: Check for unpin correctness.
95c2ccdc82d520f59ae3b6fdc097b63c9b7082bb Reapply "drm/i915: Avoid stalling on pending flips for legacy cursor updates"
a6747b7304a9d66758a196d885dab8bbfa5e7d1f drm/i915: Make unpin async.
03f476e1fcb42fca88fc50b94b0d3adbdbe887f0 drm/i915: Prepare connectors for nonblocking checks.
2099deffef4404f949ba1b68d2b17e0608190bc2 drm/i915: Pass atomic states to fbc update functions.
ee7171af72c39c18b7d7571419a4ac6ca30aea66 drm/i915: Remove reset_counter from intel_crtc.
2ee004f7c59b2e642f0bb2834f847d756f2dd7b7 drm/i915: Remove queue_flip pointer.
b8d2afae557dbb9b9c7bc6f6ec4f5278f3c4c34e drm/i915: Remove use_mmio_flip kernel parameter.
8dd634d922615ec3a9af7976029110ec037f8b50 drm/i915: Remove cs based page flip support.
143f73b3bf48c089b40f58462dd7f7c199fd4f0f drm/i915: Rework intel_crtc_page_flip to be almost atomic, v3.
84fc494b64e8c591be446a966b7447a9db519c88 drm/i915: Add the exclusive fence to plane_state.
6885843ae164e11f6c802209d06921e678a3f3f3 drm/i915: Convert flip_work to a list.
aa420ddd8eeaa5df579894a412289e4d07c2fee9 drm/i915: Allow mmio updates on all platforms, v2.
afee4d8707ab1f21b7668de995be3a5961e83582 Revert "drm/i915: Avoid stalling on pending flips for legacy cursor updates"

"drm/i915: Allow nonblocking update of pageflips" should have been
split up, misses a proper commit message and seems to cause issues in
the legacy page_flip path as demonstrated by kms_flip.

"drm/i915: Make unpin async" doesn't handle the unthrottled cursor
updates correctly, leading to an apparent pin count leak. This is
caught by the WARN_ON in i915_gem_object_do_pin which screams if we
have more than DRM_I915_GEM_OBJECT_MAX_PIN_COUNT pins.

Unfortuantely we can't just revert these two because this patch series
came with a built-in bisect breakage in the form of temporarily
removing the unthrottled cursor update hack for legacy cursor ioctl.
Therefore there's no other option than to revert the entire pile :(

There's one tiny conflict in intel_drv.h due to other patches, nothing
serious.

Normally I'd wait a bit longer with doing a maintainer revert, but
since the minimal set of patches we need to revert (due to the bisect
breakage) is so big, time is running out fast. And very soon
(especially after a few attempts at fixing issues) it'll be really
hard to revert things cleanly.

Lessons learned:
- Not a good idea to rush the review (done by someone fairly new to
  the area) and not make sure domain experts had a chance to read it.

- Patches should be properly split up. I only looked at the two
  patches that should be reverted in detail, but both look like the
  mix up different things in one patch.

- Patches really should have proper commit messages. Especially when
  doing more than one thing, and especially when touching critical and
  tricky core code.

- Building a patch series and r-b stamping it when it has a built-in
  bisect breakage is not a good idea.

- I also think we need to stop building up technical debt by
  postponing atomic igt testcases even longer. I think it's clear that
  there's enough corner cases in this beast that we really need to
  have the testcases _before_ the next step lands.

(cherry picked from commit 5a21b6650a239ebc020912968a44047701104159
from drm-intel-next-queeud)

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Patrik Jakobsson <patrik.jakobsson@linux.intel.com>
Cc: John Harrison <John.C.Harrison@Intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Acked-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Acked-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Dave Airlie <airlied@redhat.com>
Acked-by: Jani Nikula <jani.nikula@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
7 years agodrm/i915: Update GEN6_PMINTRMSK setup with GuC enabled
Sagar Arun Kamble [Tue, 31 May 2016 08:28:27 +0000 (13:58 +0530)]
drm/i915: Update GEN6_PMINTRMSK setup with GuC enabled

On Loading, GuC sets PM interrupts routing (bit 31) and clears ARAT
expired interrupt (bit 9). Host turbo also updates this register
in RPS flows. This patch ensures bit 31 and bit 9 setup by GuC persists.
ARAT timer interrupt is needed in GuC for various features. It also
facilitates halting GuC and hence achieving RC6. PM interrupt routing
will not impact RPS interrupt reception by host as GuC will redirect
them.
This patch fixes igt test pm_rc6_residency that was failing with guc
load/submission enabled. Tested with SKL GuC v6.1 and BXT GuC v5.1 and v8.7.

v2: i915_irq/i915_pm decoupling from intel_guc. (ChrisW)

v3: restructuring the mask update and rebase w.r.t Ville's patch. (ChrisW)

v4: Updating the pm_intr_keep during direct_interrupts_to_guc. (Sagar)

Cc: Chris Harris <chris.harris@intel.com>
Cc: Zhe Wang <zhe1.wang@intel.com>
Cc: Deepak S <deepak.s@intel.com>
Cc: Satyanantha, Rama Gopal M <rama.gopal.m.satyanantha@intel.com>
Cc: Akash Goel <akash.goel@intel.com>
Testcase: igt/pm_rc6_residency
Signed-off-by: Sagar Arun Kamble <sagar.a.kamble@intel.com>
Tested-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464683307-19475-1-git-send-email-sagar.a.kamble@intel.com
7 years agodrm/vc4: Use lockless gem BO free callback
Daniel Vetter [Mon, 30 May 2016 17:53:07 +0000 (19:53 +0200)]
drm/vc4: Use lockless gem BO free callback

No dev->struct_mutex anywhere to be seen.

Cc: Eric Anholt <eric@anholt.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464630800-30786-16-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm/vc4: Use drm_gem_object_unreference_unlocked
Daniel Vetter [Mon, 30 May 2016 17:53:06 +0000 (19:53 +0200)]
drm/vc4: Use drm_gem_object_unreference_unlocked

Since my last struct_mutex crusade someone escaped!

This already has the advantage that for the common case when someone
else holds a ref the unref won't even acquire dev->struct_mutex. And
I'm working on code to allow drivers to completely opt-out of any and
all dev->struct_mutex usage, but that only works if they use the
_unlocked variants everywhere.

v2: Drop comment too.

v3: Drop the other comment too.

Cc: Eric Anholt <eric@anholt.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464630800-30786-15-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm: Initialize a linear gamma table by default
Daniel Vetter [Wed, 30 Mar 2016 09:51:16 +0000 (11:51 +0200)]
drm: Initialize a linear gamma table by default

Code stolen from gma500.

This is just a minor bit of safety code that I spotted and figured it
might be useful if we put it into the core. This is to make the
get_gamma ioctl reflect likely reality even before the first set_gamma
ioctl call.

v2 on irc: Extend commit message per Maarten's suggestions.

Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1459331485-28376-2-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm/vgem: Use lockless gem BO free callback
Daniel Vetter [Mon, 30 May 2016 17:53:08 +0000 (19:53 +0200)]
drm/vgem: Use lockless gem BO free callback

No dev->struct_mutex anywhere to be seen.

Cc: seanpaul@chromium.org
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1464630800-30786-17-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm/qxl: Don't set a gamma table size
Daniel Vetter [Wed, 30 Mar 2016 09:51:24 +0000 (11:51 +0200)]
drm/qxl: Don't set a gamma table size

qxl doesn't have any functions for setting the gamma table, so this is
completely defunct.

Not nice to lie to userspace, so let's stop!

Cc: Dave Airlie <airlied@redhat.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1459331485-28376-10-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm/msm: Nuke dummy gamma_set/get functions
Daniel Vetter [Wed, 30 Mar 2016 09:51:21 +0000 (11:51 +0200)]
drm/msm: Nuke dummy gamma_set/get functions

Again the fbdev emulation gamma_set/get functions are only needed for
drivers that try to also use 8bpp paletted mode. Which msm doesn't, so
this is dead code. Let's rip it out.

Cc: Rob Clark <robdclark@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1459331485-28376-7-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm/cirrus: Drop redundnant gamma size check
Daniel Vetter [Wed, 30 Mar 2016 09:51:20 +0000 (11:51 +0200)]
drm/cirrus: Drop redundnant gamma size check

The core does this for us already.

Cc: Dave Airlie <airlied@redhat.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1459331485-28376-6-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm/fb-helper: Remove dead code in setcolreg
Daniel Vetter [Wed, 30 Mar 2016 09:51:17 +0000 (11:51 +0200)]
drm/fb-helper: Remove dead code in setcolreg

DRM fbdev emulation only supports pallete_color with depth == 8, and
truecolor with depth > 8. Handling depth == 16 for palettes is hence
dead code, let's remove it.

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1459331485-28376-3-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm/mediatek: Use lockless gem BO free callback
Daniel Vetter [Mon, 30 May 2016 17:53:19 +0000 (19:53 +0200)]
drm/mediatek: Use lockless gem BO free callback

No dev->struct_mutex anywhere to be seen.

Cc: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Acked-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1464630800-30786-28-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm/hisilicon: Use lockless gem BO free callback
Daniel Vetter [Mon, 30 May 2016 17:53:18 +0000 (19:53 +0200)]
drm/hisilicon: Use lockless gem BO free callback

No dev->struct_mutex anywhere to be seen.

Cc: Xinliang Liu <xinliang.liu@linaro.org>
Cc: Xinwei Kong <kong.kongxinwei@hisilicon.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Acked-by: Xinwei Kong <kong.kongxinwei@hisilicon.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1464630800-30786-27-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm/hlcd: Use lockless gem BO free callback
Daniel Vetter [Mon, 30 May 2016 17:53:17 +0000 (19:53 +0200)]
drm/hlcd: Use lockless gem BO free callback

No dev->struct_mutex anywhere to be seen.

Cc: Liviu Dudau <Liviu.Dudau@arm.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Acked-by: Liviu Dudau <Liviu.Dudau@arm.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1464630800-30786-26-git-send-email-daniel.vetter@ffwll.ch
7 years agovga_switcheroo: Support deferred probing of audio clients
Lukas Wunner [Tue, 31 May 2016 09:13:27 +0000 (11:13 +0200)]
vga_switcheroo: Support deferred probing of audio clients

Daniel Vetter pointed out that vga_switcheroo_client_probe_defer() could
be needed by audio clients as well. To avoid mistakes when someone adds
conditions for these in the future, constrain the single existing
condition to VGA clients by checking for PCI_BASE_CLASS_DISPLAY. This
encompasses both PCI_CLASS_DISPLAY_VGA as well as PCI_CLASS_DISPLAY_3D,
which is used by some Nvidia Optimus GPUs.

Any future checks for audio clients should then be constrained to
PCI_BASE_CLASS_MULTIMEDIA.

v6: Spun out from commit introducing vga_switcheroo_client_probe_defer()
    to keep it a pure refactoring change. (Emil Velikov, Jani Nikula)

Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Emil Velikov <emil.l.velikov@gmail.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/358d58490eb9dda5f270d844b0dce511a2a20828.1464685538.git.lukas@wunner.de
7 years agovga_switcheroo: Add helper for deferred probing
Lukas Wunner [Tue, 31 May 2016 09:13:27 +0000 (11:13 +0200)]
vga_switcheroo: Add helper for deferred probing

So far we've got one condition when DRM drivers need to defer probing
on a dual GPU system and it's coded separately into each of the relevant
drivers. As suggested by Daniel Vetter, deduplicate that code in the
drivers and move it to a new vga_switcheroo helper. This yields better
encapsulation of concepts and lets us add further checks in a central
place. (The existing check pertains to pre-retina MacBook Pros and an
additional check is expected to be needed for retinas.)

One might be tempted to check deferred probing conditions in
vga_switcheroo_register_client(), but this is usually called fairly late
during driver load. The GPU is fully brought up and ready for switching
at that point. On boot the ->probe hook is potentially called dozens of
times until it finally succeeds, and each time we'd repeat bringup and
teardown of the GPU, lengthening boot time considerably and cluttering
logfiles. A separate helper is therefore needed which can be called
right at the beginning of the ->probe hook.

Note that amdgpu currently does not call this helper as the AMD GPUs
built into MacBook Pros are only supported by radeon so far.

v2: This helper could eventually be used by audio clients as well,
    so rephrase kerneldoc to refer to "client" instead of "GPU"
    and move the single existing check in an if block specific
    to PCI_CLASS_DISPLAY_VGA devices. Move documentation on
    that check from kerneldoc to a comment. (Daniel Vetter)

v3: Mandate in kerneldoc that registration of client shall only
    happen after calling this helper. (Daniel Vetter)

v4: Rebase on 412c8f7de011 ("drm/radeon: Return -EPROBE_DEFER when
    amdkfd not loaded")

v5: Some Optimus GPUs use PCI_CLASS_DISPLAY_3D, make sure those are
    matched as well. (Emil Velikov)

v6: The if-condition referring to PCI_BASE_CLASS_DISPLAY may be
    considered a functional change. Move to a separate commit to
    keep this a pure refactoring change. (Emil Velikov, Jani Nikula)

Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/575885fd440c2b13c3f19ddf44360cfbbff35f50.1464685538.git.lukas@wunner.de
7 years agovirtio-gpu: fix output lookup
Gerd Hoffmann [Mon, 30 May 2016 12:03:26 +0000 (14:03 +0200)]
virtio-gpu: fix output lookup

Needed for multihead setups where we can have disabled
outputs and therefore plane->crtc can be NULL.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1464609806-22013-1-git-send-email-kraxel@redhat.com
7 years agodrm/i915: kill STANDARD/CURSOR plane screams
Ville Syrjälä [Fri, 27 May 2016 17:59:25 +0000 (20:59 +0300)]
drm/i915: kill STANDARD/CURSOR plane screams

Stop yelling the plane type. "STANDARD" doesn't mean anything anyway.
Let's just use the plane name here.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464371966-15190-8-git-send-email-ville.syrjala@linux.intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
7 years agodrm/i915: Give encoders useful names
Ville Syrjälä [Fri, 27 May 2016 17:59:24 +0000 (20:59 +0300)]
drm/i915: Give encoders useful names

Rather than let the core generate usless encoder names, let's pass in
something that actually identifies the piece of hardware we're dealing
with.

v2: Use 'DSI %c' instead of 'MIPI %c' for DSI encoders (Jani)
v3: Use port_name() in DSI code since we have it

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464371966-15190-7-git-send-email-ville.syrjala@linux.intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
7 years agodrm/i915: Give meaningful names to all the planes
Ville Syrjälä [Fri, 27 May 2016 17:59:23 +0000 (20:59 +0300)]
drm/i915: Give meaningful names to all the planes

Let's name our planes in a way that makes sense wrt. the spec:
- skl+ -> "plane 1A", "plane 2A", "plane 1C", "cursor A" etc.
- g4x+ -> "primary A", "primary B", "sprite A", "cursor C" etc.
- pre-g4x -> "plane A", "cursor B" etc.

v2: Rebase on top of the fixed/cleaned error paths
    Use a local 'name' variable to make things easier
v3: Pass the name as a function argument to drm_universal_plane_init() (Jani)
v3: Pass the printf style string to drm_universal_plane_init()

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464371966-15190-6-git-send-email-ville.syrjala@linux.intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
7 years agodrm/i915: Don't leak primary/cursor planes on crtc init failure
Ville Syrjälä [Fri, 27 May 2016 17:59:22 +0000 (20:59 +0300)]
drm/i915: Don't leak primary/cursor planes on crtc init failure

Call intel_plane_destroy() instead of drm_plane_cleanup() so that we
also free the plane struct itself when bailing out of the crtc init.

And make intel_plane_destroy() NULL tolerant to avoid having to check
for it in the caller.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464371966-15190-5-git-send-email-ville.syrjala@linux.intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
7 years agodrm/i915: Set crtc->name to "pipe A", "pipe B", etc.
Ville Syrjälä [Fri, 27 May 2016 17:59:21 +0000 (20:59 +0300)]
drm/i915: Set crtc->name to "pipe A", "pipe B", etc.

v2: Fix intel_crtc leak on failure to allocate the name
    Use a local 'name' variable to make things easier
v3: Pass the name as a function arguemnt to drm_crtc_init_with_planes() (Jani)
v4: Pass the printf style format string to drm_crtc_init_with_planes()
v5: Drop spurious code changes

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464371966-15190-4-git-send-email-ville.syrjala@linux.intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
7 years agodrm/i915: Use plane->name in debug prints
Ville Syrjälä [Fri, 27 May 2016 17:59:20 +0000 (20:59 +0300)]
drm/i915: Use plane->name in debug prints

We have plane->name, so let's use that in debug messages instead
of just printing the more or less useless object ID.

v2: slap on a commit message

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464371966-15190-3-git-send-email-ville.syrjala@linux.intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
7 years agodrm/i915: Use crtc->name in debug messages
Ville Syrjälä [Fri, 27 May 2016 17:59:19 +0000 (20:59 +0300)]
drm/i915: Use crtc->name in debug messages

We have crtc->name, so let's use that in debug messages instead
of just printing the more or less useless object ID.

v2: Rebased due to intel_dpll_mgr.c, slap on a commit message

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1464371966-15190-2-git-send-email-ville.syrjala@linux.intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
7 years agodrm/doc: Unify KMS Locking docs
Daniel Vetter [Mon, 30 May 2016 09:10:49 +0000 (11:10 +0200)]
drm/doc: Unify KMS Locking docs

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Acked-by: Jani Nikula <jani.nikula@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1464599449-12509-1-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm/atomic-helper: Do not call ->mode_fixup for CRTC which will be disabled
Liu Ying [Fri, 27 May 2016 09:35:54 +0000 (17:35 +0800)]
drm/atomic-helper: Do not call ->mode_fixup for CRTC which will be disabled

When a CRTC is going to be disabled, it's state may contain a display mode
with zeroed content.  This could be reproduced by HDMI cable hotplug out
operation with legacy fbdev support in dual display cases.  It would confuse
driver's CRTC callback ->mode_fixup and make the total state be rejected.
So, let's don't call the callback for the CRTC.

Signed-off-by: Liu Ying <gnuiyl@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1464341754-7087-1-git-send-email-gnuiyl@gmail.com
7 years agoFix annoyingly awkward typo in drm_edid_load.c
Valdis Kletnieks [Mon, 30 May 2016 06:26:38 +0000 (02:26 -0400)]
Fix annoyingly awkward typo in drm_edid_load.c

Fix egregious typo in comment.

Signed-off-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/10576.1464589598@turing-police.cc.vt.edu
7 years agoLinux 4.7-rc1 v4.7-rc1
Linus Torvalds [Sun, 29 May 2016 16:29:24 +0000 (09:29 -0700)]
Linux 4.7-rc1

7 years agohash_string: Fix zero-length case for !DCACHE_WORD_ACCESS
George Spelvin [Sun, 29 May 2016 12:05:56 +0000 (08:05 -0400)]
hash_string: Fix zero-length case for !DCACHE_WORD_ACCESS

The self-test was updated to cover zero-length strings; the function
needs to be updated, too.

Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Fixes: fcfd2fbf22d2 ("fs/namei.c: Add hashlen_string() function")
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoRename other copy of hash_string to hashlen_string
George Spelvin [Sun, 29 May 2016 05:26:41 +0000 (01:26 -0400)]
Rename other copy of hash_string to hashlen_string

The original name was simply hash_string(), but that conflicted with a
function with that name in drivers/base/power/trace.c, and I decided
that calling it "hashlen_" was better anyway.

But you have to do it in two places.

[ This caused build errors for architectures that don't define
  CONFIG_DCACHE_WORD_ACCESS   - Linus ]

Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Reported-by: Guenter Roeck <linux@roeck-us.net>
Fixes: fcfd2fbf22d2 ("fs/namei.c: Add hashlen_string() function")
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agohpfs: implement the show_options method
Mikulas Patocka [Tue, 24 May 2016 20:49:18 +0000 (22:49 +0200)]
hpfs: implement the show_options method

The HPFS filesystem used generic_show_options to produce string that is
displayed in /proc/mounts.  However, there is a problem that the options
may disappear after remount.  If we mount the filesystem with option1
and then remount it with option2, /proc/mounts should show both option1
and option2, however it only shows option2 because the whole option
string is replaced with replace_mount_options in hpfs_remount_fs.

To fix this bug, implement the hpfs_show_options function that prints
options that are currently selected.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoaffs: fix remount failure when there are no options changed
Mikulas Patocka [Tue, 24 May 2016 20:48:33 +0000 (22:48 +0200)]
affs: fix remount failure when there are no options changed

Commit c8f33d0bec99 ("affs: kstrdup() memory handling") checks if the
kstrdup function returns NULL due to out-of-memory condition.

However, if we are remounting a filesystem with no change to
filesystem-specific options, the parameter data is NULL.  In this case,
kstrdup returns NULL (because it was passed NULL parameter), although no
out of memory condition exists.  The mount syscall then fails with
ENOMEM.

This patch fixes the bug.  We fail with ENOMEM only if data is non-NULL.

The patch also changes the call to replace_mount_options - if we didn't
pass any filesystem-specific options, we don't call
replace_mount_options (thus we don't erase existing reported options).

Fixes: c8f33d0bec99 ("affs: kstrdup() memory handling")
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org # v4.1+
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agohpfs: fix remount failure when there are no options changed
Mikulas Patocka [Tue, 24 May 2016 20:47:00 +0000 (22:47 +0200)]
hpfs: fix remount failure when there are no options changed

Commit ce657611baf9 ("hpfs: kstrdup() out of memory handling") checks if
the kstrdup function returns NULL due to out-of-memory condition.

However, if we are remounting a filesystem with no change to
filesystem-specific options, the parameter data is NULL.  In this case,
kstrdup returns NULL (because it was passed NULL parameter), although no
out of memory condition exists.  The mount syscall then fails with
ENOMEM.

This patch fixes the bug.  We fail with ENOMEM only if data is non-NULL.

The patch also changes the call to replace_mount_options - if we didn't
pass any filesystem-specific options, we don't call
replace_mount_options (thus we don't erase existing reported options).

Fixes: ce657611baf9 ("hpfs: kstrdup() out of memory handling")
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoMerge branch 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus
Linus Torvalds [Sat, 28 May 2016 23:41:39 +0000 (16:41 -0700)]
Merge branch 'upstream' of git://git.linux-mips.org/ralf/upstream-linus

Pull more MIPS updates from Ralf Baechle:
 "This is the secondnd batch of MIPS patches for 4.7. Summary:

  CPS:
   - Copy EVA configuration when starting secondary VPs.

  EIC:
   - Clear Status IPL.

  Lasat:
   - Fix a few off by one bugs.

  lib:
   - Mark intrinsics notrace.  Not only are the intrinsics
     uninteresting, it would cause infinite recursion.

  MAINTAINERS:
   - Add file patterns for MIPS BRCM device tree bindings.
   - Add file patterns for mips device tree bindings.

  MT7628:
   - Fix MT7628 pinmux typos.
   - wled_an pinmux gpio.
   - EPHY LEDs pinmux support.

  Pistachio:
   - Enable KASLR

  VDSO:
   - Build microMIPS VDSO for microMIPS kernels.
   - Fix aliasing warning by building with `-fno-strict-aliasing' for
     debugging but also tracing them might result in recursion.

  Misc:
   - Add missing FROZEN hotplug notifier transitions.
   - Fix clk binding example for varioius PIC32 devices.
   - Fix cpu interrupt controller node-names in the DT files.
   - Fix XPA CPU feature separation.
   - Fix write_gc0_* macros when writing zero.
   - Add inline asm encoding helpers.
   - Add missing VZ accessor microMIPS encodings.
   - Fix little endian microMIPS MSA encodings.
   - Add 64-bit HTW fields and fix its configuration.
   - Fix sigreturn via VDSO on microMIPS kernel.
   - Lots of typo fixes.
   - Add definitions of SegCtl registers and use them"

* 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus: (49 commits)
  MIPS: Add missing FROZEN hotplug notifier transitions
  MIPS: Build microMIPS VDSO for microMIPS kernels
  MIPS: Fix sigreturn via VDSO on microMIPS kernel
  MIPS: devicetree: fix cpu interrupt controller node-names
  MIPS: VDSO: Build with `-fno-strict-aliasing'
  MIPS: Pistachio: Enable KASLR
  MIPS: lib: Mark intrinsics notrace
  MIPS: Fix 64-bit HTW configuration
  MIPS: Add 64-bit HTW fields
  MAINTAINERS: Add file patterns for mips device tree bindings
  MAINTAINERS: Add file patterns for mips brcm device tree bindings
  MIPS: Simplify DSP instruction encoding macros
  MIPS: Add missing tlbinvf/XPA microMIPS encodings
  MIPS: Fix little endian microMIPS MSA encodings
  MIPS: Add missing VZ accessor microMIPS encodings
  MIPS: Add inline asm encoding helpers
  MIPS: Spelling fix lets -> let's
  MIPS: VR41xx: Fix typo
  MIPS: oprofile: Fix typo
  MIPS: math-emu: Fix typo
  ...

7 years agofs: fix binfmt_aout.c build error
Guenter Roeck [Sat, 28 May 2016 22:26:02 +0000 (15:26 -0700)]
fs: fix binfmt_aout.c build error

Various builds (such as i386:allmodconfig) fail with

  fs/binfmt_aout.c:133:2: error: expected identifier or '(' before 'return'
  fs/binfmt_aout.c:134:1: error: expected identifier or '(' before '}' token

[ Oops. My bad, I had stupidly thought that "allmodconfig" covered this
  on x86-64 too, but it obviously doesn't.  Egg on my face.  - Linus ]

Fixes: 5d22fc25d4fc ("mm: remove more IS_ERR_VALUE abuses")
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoMerge branch 'hash' of git://ftp.sciencehorizons.net/linux
Linus Torvalds [Sat, 28 May 2016 23:15:25 +0000 (16:15 -0700)]
Merge branch 'hash' of git://ftp.sciencehorizons.net/linux

Pull string hash improvements from George Spelvin:
 "This series does several related things:

   - Makes the dcache hash (fs/namei.c) useful for general kernel use.

     (Thanks to Bruce for noticing the zero-length corner case)

   - Converts the string hashes in <linux/sunrpc/svcauth.h> to use the
     above.

   - Avoids 64-bit multiplies in hash_64() on 32-bit platforms.  Two
     32-bit multiplies will do well enough.

   - Rids the world of the bad hash multipliers in hash_32.

     This finishes the job started in commit 689de1d6ca95 ("Minimal
     fix-up of bad hashing behavior of hash_64()")

     The vast majority of Linux architectures have hardware support for
     32x32-bit multiply and so derive no benefit from "simplified"
     multipliers.

     The few processors that do not (68000, h8/300 and some models of
     Microblaze) have arch-specific implementations added.  Those
     patches are last in the series.

   - Overhauls the dcache hash mixing.

     The patch in commit 0fed3ac866ea ("namei: Improve hash mixing if
     CONFIG_DCACHE_WORD_ACCESS") was an off-the-cuff suggestion.
     Replaced with a much more careful design that's simultaneously
     faster and better.  (My own invention, as there was noting suitable
     in the literature I could find.  Comments welcome!)

   - Modify the hash_name() loop to skip the initial HASH_MIX().  This
     would let us salt the hash if we ever wanted to.

   - Sort out partial_name_hash().

     The hash function is declared as using a long state, even though
     it's truncated to 32 bits at the end and the extra internal state
     contributes nothing to the result.  And some callers do odd things:

      - fs/hfs/string.c only allocates 32 bits of state
      - fs/hfsplus/unicode.c uses it to hash 16-bit unicode symbols not bytes

   - Modify bytemask_from_count to handle inputs of 1..sizeof(long)
     rather than 0..sizeof(long)-1.  This would simplify users other
     than full_name_hash"

  Special thanks to Bruce Fields for testing and finding bugs in v1.  (I
  learned some humbling lessons about "obviously correct" code.)

  On the arch-specific front, the m68k assembly has been tested in a
  standalone test harness, I've been in contact with the Microblaze
  maintainers who mostly don't care, as the hardware multiplier is never
  omitted in real-world applications, and I haven't heard anything from
  the H8/300 world"

* 'hash' of git://ftp.sciencehorizons.net/linux:
  h8300: Add <asm/hash.h>
  microblaze: Add <asm/hash.h>
  m68k: Add <asm/hash.h>
  <linux/hash.h>: Add support for architecture-specific functions
  fs/namei.c: Improve dcache hash function
  Eliminate bad hash multipliers from hash_32() and  hash_64()
  Change hash_64() return value to 32 bits
  <linux/sunrpc/svcauth.h>: Define hash_str() in terms of hashlen_string()
  fs/namei.c: Add hashlen_string() function
  Pull out string hash to <linux/stringhash.h>

7 years agoh8300: Add <asm/hash.h>
George Spelvin [Wed, 25 May 2016 18:19:49 +0000 (14:19 -0400)]
h8300: Add <asm/hash.h>

This will improve the performance of hash_32() and hash_64(), but due
to complete lack of multi-bit shift instructions on H8, performance will
still be bad in surrounding code.

Designing H8-specific hash algorithms to work around that is a separate
project.  (But if the maintainers would like to get in touch...)

Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: uclinux-h8-devel@lists.sourceforge.jp
7 years agomicroblaze: Add <asm/hash.h>
George Spelvin [Wed, 25 May 2016 15:06:09 +0000 (11:06 -0400)]
microblaze: Add <asm/hash.h>

Microblaze is an FPGA soft core that can be configured various ways.

If it is configured without a multiplier, the standard __hash_32()
will require a call to __mulsi3, which is a slow software loop.

Instead, use a shift-and-add sequence for the constant multiply.
GCC knows how to do this, but it's not as clever as some.

Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Cc: Alistair Francis <alistair.francis@xilinx.com>
Cc: Michal Simek <michal.simek@xilinx.com>
7 years agom68k: Add <asm/hash.h>
George Spelvin [Thu, 26 May 2016 15:36:19 +0000 (11:36 -0400)]
m68k: Add <asm/hash.h>

This provides a multiply by constant GOLDEN_RATIO_32 = 0x61C88647
for the original mc68000, which lacks a 32x32-bit multiply instruction.

Yes, the amount of optimization effort put in is excessive. :-)

Shift-add chain found by Yevgen Voronenko's Hcub algorithm at
http://spiral.ece.cmu.edu/mcm/gen.html

Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: Andreas Schwab <schwab@linux-m68k.org>
Cc: Philippe De Muyter <phdm@macq.eu>
Cc: linux-m68k@lists.linux-m68k.org
7 years ago<linux/hash.h>: Add support for architecture-specific functions
George Spelvin [Fri, 27 May 2016 02:11:51 +0000 (22:11 -0400)]
<linux/hash.h>: Add support for architecture-specific functions

This is just the infrastructure; there are no users yet.

This is modelled on CONFIG_ARCH_RANDOM; a CONFIG_ symbol declares
the existence of <asm/hash.h>.

That file may define its own versions of various functions, and define
HAVE_* symbols (no CONFIG_ prefix!) to suppress the generic ones.

Included is a self-test (in lib/test_hash.c) that verifies the basics.
It is NOT in general required that the arch-specific functions compute
the same thing as the generic, but if a HAVE_* symbol is defined with
the value 1, then equality is tested.

Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: Andreas Schwab <schwab@linux-m68k.org>
Cc: Philippe De Muyter <phdm@macq.eu>
Cc: linux-m68k@lists.linux-m68k.org
Cc: Alistair Francis <alistai@xilinx.com>
Cc: Michal Simek <michal.simek@xilinx.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: uclinux-h8-devel@lists.sourceforge.jp
7 years agofs/namei.c: Improve dcache hash function
George Spelvin [Mon, 23 May 2016 11:43:58 +0000 (07:43 -0400)]
fs/namei.c: Improve dcache hash function

Patch 0fed3ac866 improved the hash mixing, but the function is slower
than necessary; there's a 7-instruction dependency chain (10 on x86)
each loop iteration.

Word-at-a-time access is a very tight loop (which is good, because
link_path_walk() is one of the hottest code paths in the entire kernel),
and the hash mixing function must not have a longer latency to avoid
slowing it down.

There do not appear to be any published fast hash functions that:
1) Operate on the input a word at a time, and
2) Don't need to know the length of the input beforehand, and
3) Have a single iterated mixing function, not needing conditional
   branches or unrolling to distinguish different loop iterations.

One of the algorithms which comes closest is Yann Collet's xxHash, but
that's two dependent multiplies per word, which is too much.

The key insights in this design are:

1) Barring expensive ops like multiplies, to diffuse one input bit
   across 64 bits of hash state takes at least log2(64) = 6 sequentially
   dependent instructions.  That is more cycles than we'd like.
2) An operation like "hash ^= hash << 13" requires a second temporary
   register anyway, and on a 2-operand machine like x86, it's three
   instructions.
3) A better use of a second register is to hold a two-word hash state.
   With careful design, no temporaries are needed at all, so it doesn't
   increase register pressure.  And this gets rid of register copying
   on 2-operand machines, so the code is smaller and faster.
4) Using two words of state weakens the requirement for one-round mixing;
   we now have two rounds of mixing before cancellation is possible.
5) A two-word hash state also allows operations on both halves to be
   done in parallel, so on a superscalar processor we get more mixing
   in fewer cycles.

I ended up using a mixing function inspired by the ChaCha and Speck
round functions.  It is 6 simple instructions and 3 cycles per iteration
(assuming multiply by 9 can be done by an "lea" instruction):

x ^= *input++;
y ^= x; x = ROL(x, K1);
x += y; y = ROL(y, K2);
y *= 9;

Not only is this reversible, two consecutive rounds are reversible:
if you are given the initial and final states, but not the intermediate
state, it is possible to compute both input words.  This means that at
least 3 words of input are required to create a collision.

(It also has the property, used by hash_name() to avoid a branch, that
it hashes all-zero to all-zero.)

The rotate constants K1 and K2 were found by experiment.  The search took
a sample of random initial states (I used 1023) and considered the effect
of flipping each of the 64 input bits on each of the 128 output bits two
rounds later.  Each of the 8192 pairs can be considered a biased coin, and
adding up the Shannon entropy of all of them produces a score.

The best-scoring shifts also did well in other tests (flipping bits in y,
trying 3 or 4 rounds of mixing, flipping all 64*63/2 pairs of input bits),
so the choice was made with the additional constraint that the sum of the
shifts is odd and not too close to the word size.

The final state is then folded into a 32-bit hash value by a less carefully
optimized multiply-based scheme.  This also has to be fast, as pathname
components tend to be short (the most common case is one iteration!), but
there's some room for latency, as there is a fair bit of intervening logic
before the hash value is used for anything.

(Performance verified with "bonnie++ -s 0 -n 1536:-2" on tmpfs.  I need
a better benchmark; the numbers seem to show a slight dip in performance
between 4.6.0 and this patch, but they're too noisy to quote.)

Special thanks to Bruce fields for diligent testing which uncovered a
nasty fencepost error in an earlier version of this patch.

[checkpatch.pl formatting complaints noted and respectfully disagreed with.]

Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Tested-by: J. Bruce Fields <bfields@redhat.com>
7 years agoEliminate bad hash multipliers from hash_32() and hash_64()
George Spelvin [Fri, 27 May 2016 03:00:23 +0000 (23:00 -0400)]
Eliminate bad hash multipliers from hash_32() and  hash_64()

The "simplified" prime multipliers made very bad hash functions, so get rid
of them.  This completes the work of 689de1d6ca.

To avoid the inefficiency which was the motivation for the "simplified"
multipliers, hash_64() on 32-bit systems is changed to use a different
algorithm.  It makes two calls to hash_32() instead.

drivers/media/usb/dvb-usb-v2/af9015.c uses the old GOLDEN_RATIO_PRIME_32
for some horrible reason, so it inherits a copy of the old definition.

Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Cc: Antti Palosaari <crope@iki.fi>
Cc: Mauro Carvalho Chehab <m.chehab@samsung.com>
7 years agoChange hash_64() return value to 32 bits
George Spelvin [Fri, 27 May 2016 02:22:01 +0000 (22:22 -0400)]
Change hash_64() return value to 32 bits

That's all that's ever asked for, and it makes the return
type of hash_long() consistent.

It also allows (upcoming patch) an optimized implementation
of hash_64 on 32-bit machines.

I tried adding a BUILD_BUG_ON to ensure the number of bits requested
was never more than 32 (most callers use a compile-time constant), but
adding <linux/bug.h> to <linux/hash.h> breaks the tools/perf compiler
unless tools/perf/MANIFEST is updated, and understanding that code base
well enough to update it is too much trouble.  I did the rest of an
allyesconfig build with such a check, and nothing tripped.

Signed-off-by: George Spelvin <linux@sciencehorizons.net>
7 years ago<linux/sunrpc/svcauth.h>: Define hash_str() in terms of hashlen_string()
George Spelvin [Fri, 20 May 2016 17:31:33 +0000 (13:31 -0400)]
<linux/sunrpc/svcauth.h>: Define hash_str() in terms of hashlen_string()

Finally, the first use of previous two patches: eliminate the
separate ad-hoc string hash functions in the sunrpc code.

Now hash_str() is a wrapper around hash_string(), and hash_mem() is
likewise a wrapper around full_name_hash().

Note that sunrpc code *does* call hash_mem() with a zero length, which
is why the previous patch needed to handle that in full_name_hash().
(Thanks, Bruce, for finding that!)

This also eliminates the only caller of hash_long which asks for
more than 32 bits of output.

The comment about the quality of hashlen_string() and full_name_hash()
is jumping the gun by a few patches; they aren't very impressive now,
but will be improved greatly later in the series.

Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Tested-by: J. Bruce Fields <bfields@redhat.com>
Acked-by: J. Bruce Fields <bfields@redhat.com>
Cc: Jeff Layton <jlayton@poochiereds.net>
Cc: linux-nfs@vger.kernel.org
7 years agofs/namei.c: Add hashlen_string() function
George Spelvin [Fri, 20 May 2016 12:41:37 +0000 (08:41 -0400)]
fs/namei.c: Add hashlen_string() function

We'd like to make more use of the highly-optimized dcache hash functions
throughout the kernel, rather than have every subsystem create its own,
and a function that hashes basic null-terminated strings is required
for that.

(The name is to emphasize that it returns both hash and length.)

It's actually useful in the dcache itself, specifically d_alloc_name().
Other uses in the next patch.

full_name_hash() is also tweaked to make it more generally useful:
1) Take a "char *" rather than "unsigned char *" argument, to
   be consistent with hash_name().
2) Handle zero-length inputs.  If we want more callers, we don't want
   to make them worry about corner cases.

Signed-off-by: George Spelvin <linux@sciencehorizons.net>
7 years agoPull out string hash to <linux/stringhash.h>
George Spelvin [Fri, 20 May 2016 11:26:00 +0000 (07:26 -0400)]
Pull out string hash to <linux/stringhash.h>

... so they can be used without the rest of <linux/dcache.h>

The hashlen_* macros will make sense next patch.

Signed-off-by: George Spelvin <linux@sciencehorizons.net>