When we retire a signaled fence, we free the dependency tree. However,
we skip clearing the list so that if we then try to adjust the priority
of the signaled fence, we may walk the list of freed dependencies.

[ 3083.156757] ==================================================================
[ 3083.156806] BUG: KASAN: use-after-free in execlists_schedule+0x199/0x660 [i915]
[ 3083.156810] Read of size 8 at addr ffff8806bf20f400 by task Xorg/831

[ 3083.156815] CPU: 0 PID: 831 Comm: Xorg Not tainted 4.15.0-rc6-no-psn+ #1
[ 3083.156817] Hardware name: Notebook                         N24_25BU/N24_25BU, BIOS 5.12 02/17/2017
[ 3083.156818] Call Trace:
[ 3083.156823]  dump_stack+0x5c/0x7a
[ 3083.156827]  print_address_description+0x6b/0x290
[ 3083.156830]  kasan_report+0x28f/0x380
[ 3083.156872]  ? execlists_schedule+0x199/0x660 [i915]
[ 3083.156914]  execlists_schedule+0x199/0x660 [i915]
[ 3083.156956]  ? intel_crtc_atomic_check+0x146/0x4e0 [i915]
[ 3083.156997]  ? execlists_submit_request+0xe0/0xe0 [i915]
[ 3083.157038]  ? i915_vma_misplaced.part.4+0x25/0xb0 [i915]
[ 3083.157079]  ? __i915_vma_do_pin+0x7c8/0xc80 [i915]
[ 3083.157121]  ? intel_atomic_state_alloc+0x44/0x60 [i915]
[ 3083.157130]  ? drm_atomic_helper_page_flip+0x3e/0xb0 [drm_kms_helper]
[ 3083.157145]  ? drm_mode_page_flip_ioctl+0x7d2/0x850 [drm]
[ 3083.157159]  ? drm_ioctl_kernel+0xa7/0xf0 [drm]
[ 3083.157172]  ? drm_ioctl+0x45b/0x560 [drm]
[ 3083.157211]  i915_gem_object_wait_priority+0x14c/0x2c0 [i915]
[ 3083.157251]  ? i915_gem_get_aperture_ioctl+0x150/0x150 [i915]
[ 3083.157290]  ? i915_vma_pin_fence+0x1d8/0x320 [i915]
[ 3083.157331]  ? intel_pin_and_fence_fb_obj+0x175/0x250 [i915]
[ 3083.157372]  ? intel_rotation_info_size+0x60/0x60 [i915]
[ 3083.157413]  ? intel_link_compute_m_n+0x80/0x80 [i915]
[ 3083.157428]  ? drm_dev_printk+0x1b0/0x1b0 [drm]
[ 3083.157443]  ? drm_dev_printk+0x1b0/0x1b0 [drm]
[ 3083.157485]  intel_prepare_plane_fb+0x2f8/0x5a0 [i915]
[ 3083.157527]  ? intel_crtc_get_vblank_counter+0x80/0x80 [i915]
[ 3083.157536]  drm_atomic_helper_prepare_planes+0xa0/0x1c0 [drm_kms_helper]
[ 3083.157587]  intel_atomic_commit+0x12e/0x4e0 [i915]
[ 3083.157605]  drm_atomic_helper_page_flip+0xa2/0xb0 [drm_kms_helper]
[ 3083.157621]  drm_mode_page_flip_ioctl+0x7d2/0x850 [drm]
[ 3083.157638]  ? drm_mode_cursor2_ioctl+0x10/0x10 [drm]
[ 3083.157652]  ? drm_lease_owner+0x1a/0x30 [drm]
[ 3083.157668]  ? drm_mode_cursor2_ioctl+0x10/0x10 [drm]
[ 3083.157681]  drm_ioctl_kernel+0xa7/0xf0 [drm]
[ 3083.157696]  drm_ioctl+0x45b/0x560 [drm]
[ 3083.157711]  ? drm_mode_cursor2_ioctl+0x10/0x10 [drm]
[ 3083.157725]  ? drm_getstats+0x20/0x20 [drm]
[ 3083.157729]  ? timerqueue_del+0x49/0x80
[ 3083.157732]  ? __remove_hrtimer+0x62/0xb0
[ 3083.157735]  ? hrtimer_try_to_cancel+0x173/0x210
[ 3083.157738]  do_vfs_ioctl+0x13b/0x880
[ 3083.157741]  ? ioctl_preallocate+0x140/0x140
[ 3083.157744]  ? _raw_spin_unlock_irq+0xe/0x30
[ 3083.157746]  ? do_setitimer+0x234/0x370
[ 3083.157750]  ? SyS_setitimer+0x19e/0x1b0
[ 3083.157752]  ? SyS_alarm+0x140/0x140
[ 3083.157755]  ? __rcu_read_unlock+0x66/0x80
[ 3083.157757]  ? __fget+0xc4/0x100
[ 3083.157760]  SyS_ioctl+0x74/0x80
[ 3083.157763]  entry_SYSCALL_64_fastpath+0x1a/0x7d
[ 3083.157765] RIP: 0033:0x7f6135d0c6a7
[ 3083.157767] RSP: 002b:00007fff01451888 EFLAGS: 00003246 ORIG_RAX: 0000000000000010
[ 3083.157769] RAX: ffffffffffffffda RBX: 0000000000000004 RCX: 00007f6135d0c6a7
[ 3083.157771] RDX: 00007fff01451950 RSI: 00000000c01864b0 RDI: 000000000000000c
[ 3083.157772] RBP: 00007f613076f600 R08: 0000000000000001 R09: 0000000000000000
[ 3083.157773] R10: 0000000000000060 R11: 0000000000003246 R12: 0000000000000000
[ 3083.157774] R13: 0000000000000060 R14: 000000000000001b R15: 0000000000000060

[ 3083.157779] Allocated by task 831:
[ 3083.157783]  kmem_cache_alloc+0xc0/0x200
[ 3083.157822]  i915_gem_request_await_dma_fence+0x2c4/0x5d0 [i915]
[ 3083.157861]  i915_gem_request_await_object+0x321/0x370 [i915]
[ 3083.157900]  i915_gem_do_execbuffer+0x1165/0x19c0 [i915]
[ 3083.157937]  i915_gem_execbuffer2+0x1ad/0x550 [i915]
[ 3083.157950]  drm_ioctl_kernel+0xa7/0xf0 [drm]
[ 3083.157962]  drm_ioctl+0x45b/0x560 [drm]
[ 3083.157964]  do_vfs_ioctl+0x13b/0x880
[ 3083.157966]  SyS_ioctl+0x74/0x80
[ 3083.157968]  entry_SYSCALL_64_fastpath+0x1a/0x7d

[ 3083.157971] Freed by task 831:
[ 3083.157973]  kmem_cache_free+0x77/0x220
[ 3083.158012]  i915_gem_request_retire+0x72c/0xa70 [i915]
[ 3083.158051]  i915_gem_request_alloc+0x1e9/0x8b0 [i915]
[ 3083.158089]  i915_gem_do_execbuffer+0xa96/0x19c0 [i915]
[ 3083.158127]  i915_gem_execbuffer2+0x1ad/0x550 [i915]
[ 3083.158140]  drm_ioctl_kernel+0xa7/0xf0 [drm]
[ 3083.158153]  drm_ioctl+0x45b/0x560 [drm]
[ 3083.158155]  do_vfs_ioctl+0x13b/0x880
[ 3083.158156]  SyS_ioctl+0x74/0x80
[ 3083.158158]  entry_SYSCALL_64_fastpath+0x1a/0x7d

[ 3083.158162] The buggy address belongs to the object at ffff8806bf20f400
                which belongs to the cache i915_dependency of size 64
[ 3083.158166] The buggy address is located 0 bytes inside of
                64-byte region [ffff8806bf20f400, ffff8806bf20f440)
[ 3083.158168] The buggy address belongs to the page:
[ 3083.158171] page:00000000d43decc4 count:1 mapcount:0 mapping:          (null) index:0x0
[ 3083.158174] flags: 0x17ffe0000000100(slab)
[ 3083.158179] raw: 017ffe0000000100 0000000000000000 0000000000000000 0000000180200020
[ 3083.158182] raw: ffffea001afc16c0 0000000500000005 ffff880731b881c0 0000000000000000
[ 3083.158184] page dumped because: kasan: bad access detected

[ 3083.158187] Memory state around the buggy address:
[ 3083.158190]  ffff8806bf20f300: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
[ 3083.158192]  ffff8806bf20f380: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
[ 3083.158195] >ffff8806bf20f400: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
[ 3083.158196]                    ^
[ 3083.158199]  ffff8806bf20f480: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
[ 3083.158201]  ffff8806bf20f500: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
[ 3083.158203] ==================================================================
Reported-by: NAlexandru Chirvasitu <achirvasub@gmail.com>
Reported-by: NMike Keehan <mike@keehan.net>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104436
Fixes: 1f181225 ("drm/i915/execlists: Keep request->priority for its lifetime")
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Alexandru Chirvasitu <achirvasub@gmail.com>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Tested-by: NAlexandru Chirvasitu <achirvasub@gmail.com>
Reviewed-by: NMichał Winiarski <michal.winiarski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180106105618.13532-1-chris@chris-wilson.co.uk

After staring at the list_for_each_safe macros for a bit, our current
invocation of list_safe_reset_next in execlists_schedule() simply
reduces to list_for_each.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMichał Winiarski <michal.winiarski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180102151235.3949-11-chris@chris-wilson.co.uk

The dependency chain must be an acyclic graph. This is checked by the
swfence, but for sanity, also do a simple check that we do not corrupt
our list iteration in execlists_schedule() by a shallow dependency
cycle.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMichał Winiarski <michal.winiarski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180102151235.3949-10-chris@chris-wilson.co.uk

Back up our comment that all signalers should have been signaled before
we ourselves were retired with an assert to that effect.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: NMichał Winiarski <michal.winiarski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180102151235.3949-9-chris@chris-wilson.co.uk

Move the register settings for enabling execlists into its own function
for clarity.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMichał Winiarski <michal.winiarski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180102151235.3949-18-chris@chris-wilson.co.uk

Currently, we record the elsp register offset inside init-hw but we only
need to do it once during engine setup (after we know the mmio
iomapping).
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMichał Winiarski <michal.winiarski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180102151235.3949-17-chris@chris-wilson.co.uk

Move the clearing of the CS-interrupt into the engine reset phase,
before the current init-hw phase. This helps clarify that we clear the
pending interrupts prior to any restarting of the execlists.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMichał Winiarski <michal.winiarski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180102151235.3949-16-chris@chris-wilson.co.uk

We already emit a GEM_TRACE for when we start preemption, but we lack
one to show when the preemption is completed and we return to the regular
queue. This is to continue the investigation into the mysterious

<0>[  197.854177]   <idle>-0       1..s1 197837017us : execlists_submission_tasklet: rcs0 cs-irq head=0 [0], tail=0 [0]
<0>[  197.854209] drv_self-6008    2.... 197837390us : reset_common_ring: rcs0 seqno=15515
<0>[  197.854240] drv_self-6008    2.... 197837415us : reset_common_ring: bcs0 seqno=0
<0>[  197.854270] drv_self-6008    2.... 197837443us : reset_common_ring: vcs0 seqno=0
<0>[  197.854300] drv_self-6008    2.... 197837463us : reset_common_ring: vcs1 seqno=0
<0>[  197.854330] drv_self-6008    2.... 197837482us : reset_common_ring: vecs0 seqno=0
<0>[  197.854360] ksoftirq-23      2..s. 197838341us : execlists_submission_tasklet: bcs0 in[0]:  ctx=0.1, seqno=1dce7
<0>[  197.854392]   <idle>-0       1..s1 197838347us : execlists_submission_tasklet: bcs0 cs-irq head=0 [0], tail=0 [0]
<0>[  197.854423] ksoftirq-23      2..s. 197838354us : execlists_submission_tasklet: vcs0 in[0]:  ctx=0.1, seqno=1d027
<0>[  197.854456] ksoftirq-23      2.Ns. 197838361us : execlists_submission_tasklet: vcs1 in[0]:  ctx=0.1, seqno=1e738
<0>[  197.854488] ksoftirq-23      2.Ns. 197838366us : execlists_submission_tasklet: vecs0 in[0]:  ctx=0.1, seqno=235aa
<0>[  197.854520] ksoftirq-23      2.Ns. 197838376us : execlists_submission_tasklet: rcs0 in[0]:  ctx=0.1, seqno=15518
<0>[  197.854552]   <idle>-0       1..s1 197853285us : execlists_submission_tasklet: rcs0 cs-irq head=0 [0], tail=7 [7]
<0>[  197.854584]   <idle>-0       1..s1 197853285us : execlists_submission_tasklet: rcs0 csb[1]: status=0x00000018:0x00000000
<0>[  197.854616]   <idle>-0       1..s1 197853286us : execlists_submission_tasklet: rcs0 out[0]: ctx=0.0, seqno=0
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171222132742.4272-1-chris@chris-wilson.co.ukReviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>

Looking at the coordination of resets with the submission of execlists,
it will be useful to have a GEM_TRACE for when we issue the reset.
Whilst there tidy up the other GEM_TRACE to always include the engine
name, and be careful not to trust any pointers prior to asserts.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171220090626.31643-1-chris@chris-wilson.co.uk

A lesson that has to be relearnt over and over again is that the request
does not keep a reference to the context and so we cannot freely
dereference the context from inside the execlists_submission_tasklet. In
particular, we try to do so in the new GEM_TRACE() so convert those over
to the port->context_id we keep for GEM debugging. This means the
tracing now depends on DRM_I915_GEM_DEBUG.

Fixes: bccd3b83 ("drm/i915: Use trace_printk to provide a death rattle for GEM")
References: https://bugs.freedesktop.org/show_bug.cgi?id=104066
References: https://bugs.freedesktop.org/show_bug.cgi?id=104162
References: https://bugs.freedesktop.org/show_bug.cgi?id=104242
References: https://bugs.freedesktop.org/show_bug.cgi?id=104310Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NMichel Thierry <michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171219220916.30882-1-chris@chris-wilson.co.uk

Currently on every submission, we recalculate the ELSP register offset
for the engine, after chasing the pointers to find the iomem base. Since
this is fixed for the lifetime of the driver, record the offset in the
execlists struct.

In practice the difference is negligible, it just happens to remove 27
bytes of eyesore pointer dancing from next to the hottest instruction
(which is itself due to stalling for a cache miss) in perf profiles of
the execlists_submission_tasklet().

v2: Trim off one more elsp local.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: NMichel Thierry <michel.thierry@intel.com>
Reviewed-by: NRodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171207222434.17686-1-chris@chris-wilson.co.uk

Sagar noticed the check can be consolidated between the engine stats
implementation and the PMU.

My first choice was a static inline helper but that got into include
ordering mess quickly fast so I went with a macro instead. At some point
we should perhaps looking into taking out the non-ringubffer bits from
intel_ringbuffer.h into a new intel_engine.h or something.

v2: Use engine->flags. (Chris Wilson)
v3: Rebase and mark GuC as not yet supported. (Chris Wilson)
v4: Move flag setting to intel_engines_reset_default_submission.
    (Chris Wilson)
v5: Move flag setting to logical_ring_setup.
v6: intel_engines_reset_default_submission is the wrong place to set the
    flag - it needs to be in execlists_set_default_submission. (Sagar)
v7: Flag setting in logical_ring_setup is not required. (Chris)
Signed-off-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Suggested-by: NSagar Arun Kamble <sagar.a.kamble@intel.com>
Cc: Sagar Arun Kamble <sagar.a.kamble@intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Sagar Arun Kamble <sagar.a.kamble@intel.com> (v6)
Link: https://patchwork.freedesktop.org/patch/msgid/20171129102805.22690-1-tvrtko.ursulin@linux.intel.com

Track total time requests have been executing on the hardware.

We add new kernel API to allow software tracking of time GPU
engines are spending executing requests.

Both per-engine and global API is added with the latter also
being exported for use by external users.

v2:
 * Squashed with the internal API.
 * Dropped static key.
 * Made per-engine.
 * Store time in monotonic ktime.

v3: Moved stats clearing to disable.

v4:
 * Comments.
 * Don't export the API just yet.

v5: Whitespace cleanup.

v6:
 * Rename ref to active.
 * Drop engine aggregate stats for now.
 * Account initial busy period after enabling stats.

v7:
 * Rebase.

v8:
 * Move context in notification after the notifier. (Chris Wilson)

v9:

In cases where stats tracking is getting disabled while there is
an active context on an engine, add up the current value to the
total. This also implies we don't clear the total when tracking
is disabled any longer. There is no real need to do so because
we define the stats as relative while enabled, meaning
comparison between two samples while tracking is enabled is the
valid usage. However, when busy stats will later be plugged into
the perf PMU API, it is beneficial to not reset the total, since
the PMU core likes to do some counter disable/enable cycles on
startup, and while doing so during a single long context
executing on an engine we would lose some accuracy and so make
unit testing more difficult than needs to be.

v10:
 * Fix accounting for preemption.

v11:
 * Rebase for i915_modparams.enable_execlists removal.
Signed-off-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20171121181852.16128-5-tvrtko.ursulin@linux.intel.com

No functional change just something which will be handy in the
following patch.
Signed-off-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20171121181852.16128-4-tvrtko.ursulin@linux.intel.com

Execlists and legacy ringbuffer submission are no longer feature
comparable (execlists now offer greater functionality that should
overcome their performance hit) and obsoletes the unsafe module
parameter, i.e. comparing the two modes of execution is no longer
useful, so remove the debug tool.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> #i915_perf.c
Link: https://patchwork.freedesktop.org/patch/msgid/20171120205504.21892-1-chris@chris-wilson.co.uk

The hardware needs some time to process the information received in the
ExecList Submission Port, and expects us to not write anything more until
it has 'acknowledged' this new submission by sending an IDLE_ACTIVE or
PREEMPTED CSB event.

If we do not follow this, the driver could write new data into the ELSP
before HW had finishing fetching the previous one, putting us in
'undefined behaviour' space.

This seems to be the problem causing the spurious PREEMPTED & COMPLETE
events after a COMPLETE like the one below:

[] vcs0: sw rd pointer = 2, hw wr pointer = 0, current 'head' = 3.
[] vcs0:  Execlist CSB[0]: 0x00000018 _ 0x00000007
[] vcs0:  Execlist CSB[1]: 0x00000001 _ 0x00000000
[] vcs0:  Execlist CSB[2]: 0x00000018 _ 0x00000007  <<< COMPLETE
[] vcs0:  Execlist CSB[3]: 0x00000012 _ 0x00000007  <<< PREEMPTED & COMPLETE
[] vcs0:  Execlist CSB[4]: 0x00008002 _ 0x00000006
[] vcs0:  Execlist CSB[5]: 0x00000014 _ 0x00000006

The ELSP writes that lead to this CSB sequence show that the HW hadn't
started executing the previous execlist (the one with only ctx 0x6) by the
time the new one was submitted; this is a bit more clear in the data
show in the EXECLIST_STATUS register at the time of the ELSP write.

[] vcs0: ELSP[0] = 0x0_0        [execlist1] - status_reg = 0x0_302
[] vcs0: ELSP[1] = 0x6_fedb2119 [execlist0] - status_reg = 0x0_8302

[] vcs0: ELSP[2] = 0x7_fedaf119 [execlist1] - status_reg = 0x0_8308
[] vcs0: ELSP[3] = 0x6_fedb2119 [execlist0] - status_reg = 0x7_8308

Note that having to wait for this ack does not disable lite-restores,
although it may reduce their numbers.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=102035Signed-off-by: NMichel Thierry <michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/<20171118003038.7935-1-michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171120123458.23242-4-chris@chris-wilson.co.ukReviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Tested-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

If IDLE_ACTIVE is set, then all other bits are invalid. For us, we can
assert that if we see a COMPLETE | PREEMPTED event, then it should be
impossible for it to also contain an IDLE_ACTIVE flag.
Suggested-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Michal Winiarski <michal.winiarski@intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171120123458.23242-3-chris@chris-wilson.co.ukReviewed-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>

Since we get a COMPLETE event when the context switch occurs on
RING_HEAD == RING_TAIL and a PREEMPTED event when a switch occurs
before that point, COMPLETE | PREEMPTED should cover all possible context
switch completion events. We can move the ELEMENT_SWITCH info message
from the COMPLETED_MASK into an assertion for when we are performing a
switch to port[1].
Suggested-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Michal Winiarski <michal.winiarski@intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171120123458.23242-2-chris@chris-wilson.co.ukReviewed-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>

Since commit e1fee72c
Author: Oscar Mateo <oscar.mateo@intel.com>
Date:   Thu Jul 24 17:04:40 2014 +0100

    drm/i915/bdw: Avoid non-lite-restore preemptions

execlists has listened to (ACTIVE_IDLE | ELEMENT_SWITCH) for detecting
when one context completed and it either continued onto the next (in port
1) or idled. We would always see COMPLETE | ACTIVE_IDLE on the final
context-switch event, but on recent gen it appears that we now get
separate ACTIVE_IDLE and COMPLETE events. In particular, the ACTIVE_IDLE
events may not be coupled to a context (since it is a general state rather
than a specific context completion event).

v2: Update the history, execlists did originally start out by listening
to the COMPLETE event not ACTIVE_IDLE.
v3: Update preempt completion test to also use COMPLETE not ACTIVE_IDLE.

References: bspec/12255
References: https://bugs.freedesktop.org/show_bug.cgi?id=103800Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Oscar Mateo <oscar.mateo@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Michal Winiarski <michal.winiarski@intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
Acked-by: NMichel Thierry <michel.thierry@intel.com>
Reviewed-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171120123458.23242-1-chris@chris-wilson.co.uk

intel_lrc_irq_handler and i915_guc_irq_handler are HW submission related
tasklet functions. Name them with "submission_tasklet" suffix and
remove intel/i915 prefix as they are static. Also rename irq_tasklet
as just tasklet for clarity.

v2: s/_bh/_tasklet (Chris)
Suggested-by: NMichal Wajdeczko <michal.wajdeczko@intel.com>
Signed-off-by: NSagar Arun Kamble <sagar.a.kamble@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Michal Winiarski <michal.winiarski@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: NMichal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/1510839162-25197-2-git-send-email-sagar.a.kamble@intel.comSigned-off-by: NChris Wilson <chris@chris-wilson.co.uk>

At the start of building a request, we would wait for roughly enough
space to fit the average request (to reduce the likelihood of having to
wait and abort partway through request construction). To achieve we
would try to begin a 0-length command packet, this just adds extra
confusion so make the wait-for-space explicit, as in the next patch we
want to move it from the backend to the i915_gem_request_alloc() so it
can ensure that the wait-for-space is the first operation in building a
new request.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171115151204.8105-2-chris@chris-wilson.co.uk

As we now record the default HW state and so only emit the "golden"
renderstate once to prepare the HW, there is no advantage in keeping the
renderstate batch around as it will never be used again.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171110142634.10551-8-chris@chris-wilson.co.uk

Take a copy of the HW state after a reset upon module loading by
executing a context switch from a blank context to the kernel context,
thus saving the default hw state over the blank context image.
We can then use the default hw state to initialise any future context,
ensuring that each starts with the default view of hw state.

v2: Unmap our default state from the GTT after stealing it from the
context. This should stop us from accidentally overwriting it via the
GTT (and frees up some precious GTT space).

Testcase: igt/gem_ctx_isolation
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171110142634.10551-7-chris@chris-wilson.co.uk

In the next few patches, we will want to both copy out of the context
image and write a valid image into a new context. To be completely safe,
we should then couple in our domain tracking to ensure that we don't
have any issues with stale data remaining in unwanted cachelines.

Historically, we omitted the .write=true from the call to set-gtt-domain
in i915_switch_context() in order to avoid a stall between every request
as we would want to wait for the previous context write from the gpu.
Since then, we limit the set-gtt-domain to only occur when we first bind
the vma, so once in use we will never stall, and we are sure to flush
the context following a load from swap.

Equally we never applied the lessons learnt from ringbuffer submission
to execlists; so time to apply the flush of the lrc after load as well.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Acked-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171110142634.10551-6-chris@chris-wilson.co.uk

Trying to enable printk debugging for GEM is fraught with the issue of
spam; interactions with HW are very frequent and often boring. However,
one instance where they are not so boring is just before a BUG; here
ftrace provides a facility to dump its ringbuffer on an oops. So for CI
let's enable trace_printk() to capture the last exchanges with HW as a
death rattle.

For example,
[   79.234110] ------------[ cut here ]------------
[   79.234137] kernel BUG at drivers/gpu/drm/i915/intel_lrc.c:907!
[   79.234145] invalid opcode: 0000 [#1] SMP
[   79.234153] Dumping ftrace buffer:
[   79.234158] ---------------------------------
...
[   79.314044] gem_conc-1059    1..s1 79203443us : intel_lrc_irq_handler: bcs0 out[0]: ctx=5.2, seqno=145
[   79.314089] gem_conc-1059    1..s. 79220800us : intel_lrc_irq_handler: bcs0 csb[1/1]: status=0x00000018:0x00000005
[   79.314133] gem_conc-1059    1..s. 79220803us : intel_lrc_irq_handler: bcs0 out[0]: ctx=5.1, seqno=145
[   79.314177] gem_conc-1062    2..s1 79230458us : intel_lrc_irq_handler: bcs0 in[0]:  ctx=8.1, seqno=146
[   79.314220] gem_conc-1062    2..s1 79230515us : intel_lrc_irq_handler: bcs0 in[0]:  ctx=8.2, seqno=147
[   79.314265] gem_conc-1059    1..s1 79230951us : intel_lrc_irq_handler: bcs0 csb[2/3]: status=0x00000012:0x00000008
[   79.314309] gem_conc-1059    1..s1 79230954us : intel_lrc_irq_handler: bcs0 out[0]: ctx=8.2, seqno=147
[   79.314353] gem_conc-1059    1..s1 79230954us : intel_lrc_irq_handler: bcs0 csb[3/3]: status=0x00008002:0x00000008
[   79.314396] gem_conc-1059    1..s1 79230955us : intel_lrc_irq_handler: bcs0 out[0]: ctx=8.1, seqno=147
[   79.314402] ---------------------------------

v2: Tweak the formatting to be more consistent between in/out.
v3: do {} while (0) stub macro protection
Suggested-by: NMichał Winiarski <michal.winiarski@intel.com>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171109143019.16568-1-chris@chris-wilson.co.uk

Originally we set the priority to max upon inserting the request into
the execlists queue (and removing it from the scheduler lists). We could
then use the prio==INT_MAX as a shortcut within execlists_schedule() to
detect the end of the dependency chain. Since commit 1f181225
("drm/i915/execlists: Keep request->priority for its lifetime") this is
no longer true as we use the request completion as an indicator the
schedule dependency chain is complete instead. (This allows us to then
reschedule requests even when its context is in flight.) However, this
makes the GEM_BUG_ON() inside execlists_schedule() racy as we may change
the rq->prio at the same time. As the assertion is useful, let's keep
the assertion and remove the micro-optimisation.

Fixes: 1f181225 ("drm/i915/execlists: Keep request->priority for its lifetime")
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171024115501.21033-1-chris@chris-wilson.co.ukReviewed-by: NMichał Winiarski <michal.winiarski@intel.com>
(cherry picked from commit 64b80085)
Signed-off-by: NJani Nikula <jani.nikula@intel.com>

Back in commit a4b2b015 ("drm/i915: Don't mark an execlists
context-switch when idle") we noticed the presence of late
context-switch interrupts. We were able to filter those out by looking
at whether the ELSP remained active, but in commit beecec90
("drm/i915/execlists: Preemption!") that became problematic as we now
anticipate receiving a context-switch event for preemption while ELSP
may be empty. To restore the spurious interrupt suppression, add a
counter for the expected number of pending context-switches and skip if
we do not need to handle this interrupt to make forward progress.

v2: Don't forget to switch on for preempt.
v3: Reduce the counter to a on/off boolean tracker. Declare the HW as
active when we first submit, and idle after the final completion event
(with which we confirm the HW says it is idle), and track each source
of activity separately. With a finite number of sources, it should aide
us in debugging which gets stuck.

Fixes: beecec90 ("drm/i915/execlists: Preemption!")
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Michal Winiarski <michal.winiarski@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171023213237.26536-3-chris@chris-wilson.co.ukReviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
(cherry picked from commit 4a118ecb)
Signed-off-by: NJani Nikula <jani.nikula@intel.com>

Pretty similar to what we have on execlists.
We're reusing most of the GEM code, however, due to GuC quirks we need a
couple of extra bits.
Preemption is implemented as GuC action, and actions can be pretty slow.
Because of that, we're using a mutex to serialize them. Since we're
requesting preemption from the tasklet, the task of creating a workitem
and wrapping it in GuC action is delegated to a worker.

To distinguish that preemption has finished, we're using additional
piece of HWSP, and since we're not getting context switch interrupts,
we're also adding a user interrupt.

The fact that our special preempt context has completed unfortunately
doesn't mean that we're ready to submit new work. We also need to wait
for GuC to finish its own processing.

v2: Don't compile out the wait for GuC, handle workqueue flush on reset,
no need for ordered workqueue, put on a reviewer hat when looking at my own
patches (Chris)
Move struct work around in intel_guc, move user interruput outside of
conditional (Michał)
Keep ring around rather than chase though intel_context

v3: Extract WA for flushing ggtt writes to a helper (Chris)
Keep work_struct in intel_guc rather than engine (Michał)
Use ordered workqueue for inject_preempt worker to avoid GuC quirks.

v4: Drop now unused INTEL_GUC_PREEMPT_OPTION_IMMEDIATE (Daniele)
Drop stray newlines, use container_of for intel_guc in worker,
check for presence of workqueue when flushing it, rather than
enable_guc_submission modparam, reorder preempt postprocessing (Chris)

v5: Make wq NULL after destroying it

v6: Swap struct guc_preempt_work members (Michał)
Signed-off-by: NMichał Winiarski <michal.winiarski@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Jeff McGee <jeff.mcgee@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Oscar Mateo <oscar.mateo@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20171026133558.19580-1-michal.winiarski@intel.com

We would also like to make use of execlist_cancel_port_requests and
unwind_incomplete_requests in GuC preemption backend.
Let's rename the functions to use the correct prefixes, so that we can
simply add the declarations in the following patch.
Similar thing for applies for can_preempt, except we're introducing
HAS_LOGICAL_RING_PREEMPTION macro instad, converting other users that
were previously touching device info directly.

v2: s/intel_engine/execlists and pass execlists to unwind (Chris)
v3: use locked version for exporting, drop const qual (Chris)
Signed-off-by: NMichał Winiarski <michal.winiarski@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20171025200020.16636-11-michal.winiarski@intel.com

Let's separate the "emit" part from touching any internal structures,
this way we can have a generic "emit coherent GGTT write" function.
We would like to reuse this functionality for emitting HWSP write, to
confirm that preempt-to-idle has finished.

v2: Reorder args to match emit_pipe_control, s/render/rcs (Chris)
Signed-off-by: NMichał Winiarski <michal.winiarski@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20171025200020.16636-8-michal.winiarski@intel.com

Now that we're handling request resubmission the same way as regular
submission (from the tasklet), we can move GuC initialization earlier,
before restarting the engines. This way, we're no longer being in the
state of flux during engine restart - we're already in user requested
submission mode.
Signed-off-by: NMichał Winiarski <michal.winiarski@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Oscar Mateo <oscar.mateo@intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20171025172519.10670-5-chris@chris-wilson.co.ukSigned-off-by: NChris Wilson <chris@chris-wilson.co.uk>

In the next patch, we will want to install a callback when the engines
(GT as a whole) become idle and similarly when they first become busy.
To enable that callback, first rename intel_engines_mark_idle() to
intel_engines_park() and provide the companion intel_engines_unpark().
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Reviewed-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171025143943.7661-2-chris@chris-wilson.co.uk

Originally we set the priority to max upon inserting the request into
the execlists queue (and removing it from the scheduler lists). We could
then use the prio==INT_MAX as a shortcut within execlists_schedule() to
detect the end of the dependency chain. Since commit 1f181225
("drm/i915/execlists: Keep request->priority for its lifetime") this is
no longer true as we use the request completion as an indicator the
schedule dependency chain is complete instead. (This allows us to then
reschedule requests even when its context is in flight.) However, this
makes the GEM_BUG_ON() inside execlists_schedule() racy as we may change
the rq->prio at the same time. As the assertion is useful, let's keep
the assertion and remove the micro-optimisation.

Fixes: 1f181225 ("drm/i915/execlists: Keep request->priority for its lifetime")
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171024115501.21033-1-chris@chris-wilson.co.ukReviewed-by: NMichał Winiarski <michal.winiarski@intel.com>

Back in commit a4b2b015 ("drm/i915: Don't mark an execlists
context-switch when idle") we noticed the presence of late
context-switch interrupts. We were able to filter those out by looking
at whether the ELSP remained active, but in commit beecec90
("drm/i915/execlists: Preemption!") that became problematic as we now
anticipate receiving a context-switch event for preemption while ELSP
may be empty. To restore the spurious interrupt suppression, add a
counter for the expected number of pending context-switches and skip if
we do not need to handle this interrupt to make forward progress.

v2: Don't forget to switch on for preempt.
v3: Reduce the counter to a on/off boolean tracker. Declare the HW as
active when we first submit, and idle after the final completion event
(with which we confirm the HW says it is idle), and track each source
of activity separately. With a finite number of sources, it should aide
us in debugging which gets stuck.

Fixes: beecec90 ("drm/i915/execlists: Preemption!")
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Michal Winiarski <michal.winiarski@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171023213237.26536-3-chris@chris-wilson.co.ukReviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>

In the next patch, we want to reduce the lock coverage within the
shrinker, and one of the dangerous walks we have is over obj->vma_list.
We are only walking the obj->vma_list in order to check whether it has
been permanently pinned by HW access, typically via use on the scanout.
But we have a couple of other long term pins, the context objects for
which we currently have to check the individual vma pin_count. If we
instead mark these using obj->pin_display, we can forgo the dangerous
and sometimes slow list iteration.

v2: Rearrange code to try and avoid confusion from false associations
due to arrangement of whitespace along with rebasing on obj->pin_global.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171013202621.7276-4-chris@chris-wilson.co.uk

Let GVT-g VM read the CSB and CSB write pointer from virtual HWSP, not all
the host support this feature, need to check the BIT(3) of caps in PVINFO.

v3 : Remove unnecessary comments.
v4 : Separate VM enable patch with GVT-g implementation patch due to code
dependency.
v5 : Use inline for GVT virtual HWSP caps check function.
v6 : Comments refine.
Signed-off-by: NWeinan Li <weinan.z.li@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1508039725-1066-1-git-send-email-weinan.z.li@intel.com

There is function to tell how many ports we have, so use it.
We still have direct relationship with array size and port count,
so no harm was done.

Fixes: 76e70087 ("drm/i915: Make execlist port count variable")
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NMika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171010114857.13108-1-mika.kuoppala@intel.com

Michel Thierry noticed that we were applying WaDisableCtxRestoreArbitration
even to gen9, which does not require the w/a. The rationale is that we
need to enable MI arbitration for execlists to work, and to be safe we
do that before every batch (in addition to every context switch into the
batch). Since this is not clear from the single line comment suggesting
the MI_ARB_ENABLE is solely for the w/a, add a little more detail.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171005191005.13462-1-chris@chris-wilson.co.ukReviewed-by: NMichel Thierry <michel.thierry@intel.com>

When we write to ELSP, it triggers a context preemption at the earliest
arbitration point (3DPRIMITIVE, some PIPECONTROLs, a few other
operations and the explicit MI_ARB_CHECK). If this is to the same
context, it triggers a LITE_RESTORE where the RING_TAIL is merely
updated (used currently to chain requests from the same context
together, avoiding bubbles). However, if it is to a different context, a
full context-switch is performed and it will start to execute the new
context saving the image of the old for later execution.

Previously we avoided preemption by only submitting a new context when
the old was idle. But now we wish embrace it, and if the new request has
a higher priority than the currently executing request, we write to the
ELSP regardless, thus triggering preemption, but we tell the GPU to
switch to our special preemption context (not the target). In the
context-switch interrupt handler, we know that the previous contexts
have finished execution and so can unwind all the incomplete requests
and compute the new highest priority request to execute.

It would be feasible to avoid the switch-to-idle intermediate by
programming the ELSP with the target context. The difficulty is in
tracking which request that should be whilst maintaining the dependency
change, the error comes in with coalesced requests. As we only track the
most recent request and its priority, we may run into the issue of being
tricked in preempting a high priority request that was followed by a
low priority request from the same context (e.g. for PI); worse still
that earlier request may be our own dependency and the order then broken
by preemption. By injecting the switch-to-idle and then recomputing the
priority queue, we avoid the issue with tracking in-flight coalesced
requests. Having tried the preempt-to-busy approach, and failed to find
a way around the coalesced priority issue, Michal's original proposal to
inject an idle context (based on handling GuC preemption) succeeds.

The current heuristic for deciding when to preempt are only if the new
request is of higher priority, and has the privileged priority of
greater than 0. Note that the scheduler remains unfair!

v2: Disable for gen8 (bdw/bsw) as we need additional w/a for GPGPU.
Since, the feature is now conditional and not always available when we
have a scheduler, make it known via the HAS_SCHEDULER GETPARAM (now a
capability mask).
v3: Stylistic tweaks.
v4: Appease Joonas with a snippet of kerneldoc, only to fuel to fire of
the preempt vs preempting debate.
Suggested-by: NMichal Winiarski <michal.winiarski@intel.com>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Michal Winiarski <michal.winiarski@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Cc: Zhenyu Wang <zhenyuw@linux.intel.com>
Cc: Zhi Wang <zhi.a.wang@intel.com>
Reviewed-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171003203453.15692-8-chris@chris-wilson.co.uk

With preemption, we will want to "unsubmit" a request, taking it back
from the hw and returning it to the priority sorted execution list. In
order to know where to insert it into that list, we need to remember
its adjust priority (which may change even as it was being executed).

This also affects reset for execlists as we are now unsubmitting the
requests following the reset (rather than directly writing the ELSP for
the inflight contexts). This turns reset into an accidental preemption
point, as after the reset we may choose a different pair of contexts to
submit to hw.

GuC is not updated as this series doesn't add preemption to the GuC
submission, and so it can keep benefiting from the early pruning of the
DFS inside execlists_schedule() for a little longer. We also need to
find a way of reducing the cost of that DFS...

v2: Include priority in error-state
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Reviewed-by: NMichał Winiarski <michal.winiarski@intel.com>
Reviewed-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171003203453.15692-6-chris@chris-wilson.co.uk