Skip to content

K
Kernel

openeuler / Kernel
1 年多 前同步成功

通知 8
Star 0
Fork 0
K
Kernel
切换分支/标签
  1. 29 6月, 2018 8 次提交
    • C
      drm/i915/execlists: Direct submission of new requests (avoid tasklet/ksoftirqd) · 9512f985
      由 Chris Wilson 提交于 
      Back in commit 27af5eea ("drm/i915: Move execlists irq handler to a
bottom half"), we came to the conclusion that running our CSB processing
and ELSP submission from inside the irq handler was a bad idea. A really
bad idea as we could impose nearly 1s latency on other users of the
system, on average! Deferring our work to a tasklet allowed us to do the
processing with irqs enabled, reducing the impact to an average of about
50us.

We have since eradicated the use of forcewaked mmio from inside the CSB
processing and ELSP submission, bringing the impact down to around 5us
(on Kabylake); an order of magnitude better than our measurements 2
years ago on Broadwell and only about 2x worse on average than the
gem_syslatency on an unladen system.

In this iteration of the tasklet-vs-direct submission debate, we seek a
compromise where by we submit new requests immediately to the HW but
defer processing the CS interrupt onto a tasklet. We gain the advantage
of low-latency and ksoftirqd avoidance when waking up the HW, while
avoiding the system-wide starvation of our CS irq-storms.

Comparing the impact on the maximum latency observed (that is the time
stolen from an RT process) over a 120s interval, repeated several times
(using gem_syslatency, similar to RT's cyclictest) while the system is
fully laden with i915 nops, we see that direct submission an actually
improve the worse case.

Maximum latency in microseconds of a third party RT thread
(gem_syslatency -t 120 -f 2)
  x Always using tasklets (a couple of >1000us outliers removed)
  + Only using tasklets from CS irq, direct submission of requests
+------------------------------------------------------------------------+
|          +                                                             |
|          +                                                             |
|          +                                                             |
|          +       +                                                     |
|          + +     +                                                     |
|       +  + +     +  x     x     x                                      |
|      +++ + +     +  x  x  x  x  x  x                                   |
|      +++ + ++  + +  *x x  x  x  x  x                                   |
|      +++ + ++  + *  *x x  *  x  x  x                                   |
|    + +++ + ++  * * +*xxx  *  x  x  xx                                  |
|    * +++ + ++++* *x+**xx+ *  x  x xxxx x                               |
|   **x++++*++**+*x*x****x+ * +x xx xxxx x          x                    |
|x* ******+***************++*+***xxxxxx* xx*x     xxx +                x+|
|             |__________MA___________|                                  |
|      |______M__A________|                                              |
+------------------------------------------------------------------------+
    N           Min           Max        Median           Avg        Stddev
x 118            91           186           124     125.28814     16.279137
+ 120            92           187           109     112.00833     13.458617
Difference at 95.0% confidence
	-13.2798 +/- 3.79219
	-10.5994% +/- 3.02677%
	(Student's t, pooled s = 14.9237)

However the mean latency is adversely affected:

Mean latency in microseconds of a third party RT thread
(gem_syslatency -t 120 -f 1)
  x Always using tasklets
  + Only using tasklets from CS irq, direct submission of requests
+------------------------------------------------------------------------+
|           xxxxxx                                        +   ++         |
|           xxxxxx                                        +   ++         |
|           xxxxxx                                      + +++ ++         |
|           xxxxxxx                                     +++++ ++         |
|           xxxxxxx                                     +++++ ++         |
|           xxxxxxx                                     +++++ +++        |
|           xxxxxxx                                   + ++++++++++       |
|           xxxxxxxx                                 ++ ++++++++++       |
|           xxxxxxxx                                 ++ ++++++++++       |
|          xxxxxxxxxx                                +++++++++++++++     |
|         xxxxxxxxxxx    x                           +++++++++++++++     |
|x       xxxxxxxxxxxxx   x           +            + ++++++++++++++++++  +|
|           |__A__|                                                      |
|                                                      |____A___|        |
+------------------------------------------------------------------------+
    N           Min           Max        Median           Avg        Stddev
x 120         3.506         3.727         3.631     3.6321417    0.02773109
+ 120         3.834         4.149         4.039     4.0375167   0.041221676
Difference at 95.0% confidence
	0.405375 +/- 0.00888913
	11.1608% +/- 0.244735%
	(Student's t, pooled s = 0.03513)

However, since the mean latency corresponds to the amount of irqsoff
processing we have to do for a CS interrupt, we only need to speed that
up to benefit not just system latency but our own throughput.

v2: Remember to defer submissions when under reset.
v4: Only use direct submission for new requests
v5: Be aware that with mixing direct tasklet evaluation and deferred
tasklets, we may end up idling before running the deferred tasklet.
v6: Remove the redudant likely() from tasklet_is_enabled(), restrict the
annotation to reset_in_progress().
v7: Take the full timeline.lock when enabling perf_pmu stats as the
tasklet is no longer a valid guard. A consequence is that the stats are
now only valid for engines also using the timeline.lock to process
state.

Testcase: igt/gem_exec_latency/*rthog*
References: 27af5eea ("drm/i915: Move execlists irq handler to a bottom half")
Suggested-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180628201211.13837-9-chris@chris-wilson.co.uk
      9512f985
    • C
      drm/i915/execlists: Trust the CSB · fd8526e5
      由 Chris Wilson 提交于 
      Now that we use the CSB stored in the CPU friendly HWSP, we do not need
to track interrupts for when the mmio CSB registers are valid and can
just check where we read up to last from the cached HWSP. This means we
can forgo the atomic bit tracking from interrupt, and in the next patch
it means we can check the CSB at any time.

v2: Change the splitting inside reset_prepare, we only want to lose
testing the interrupt in this patch, the next patch requires the change
in locking
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180628201211.13837-8-chris@chris-wilson.co.uk
      fd8526e5
    • C
      drm/i915/execlists: Stop storing the CSB read pointer in the mmio register · 3800cd19
      由 Chris Wilson 提交于 
      As we now never read back our current head position from the CSB
pointers register, and the HW itself doesn't use it to prevent
overwriting unread CSB entries, we do not need to keep updating the
register. As it turns out this register is not listed as being shadowed,
and so requires forcewake -- but we haven't been taking forcewake around
it so the writes has probably been regularly dropped. Fortuitously, we
only read the value after a reset where it did not matter, and zero was
the right answer (well, close enough).

Mika pointed out that this was how we used to do it (accidentally!)
before he fixed it in commit cc53699b ("drm/i915: Use masked write
for Context Status Buffer Pointer").

References: cc53699b ("drm/i915: Use masked write for Context Status Buffer Pointer")
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180628201211.13837-7-chris@chris-wilson.co.uk
      3800cd19
    • C
      drm/i915/execlists: Reset CSB write pointer after reset · f4b58f04
      由 Chris Wilson 提交于 
      On HW reset, the HW clears the write pointer (to 0). But since it also
writes its first CSB entry to slot 0, we need to reset the write pointer
back to the element before (so the first entry we read is 0).

This is required for the next patch, where we trust the CSB completely!

v2: Use _MASKED_FIELD
v3: Store the reset value, so that we differentiate between mmio/hwsp
transparently and without pretense.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180628201211.13837-6-chris@chris-wilson.co.uk
      f4b58f04
    • C
      drm/i915/execlists: Unify CSB access pointers · bc4237ec
      由 Chris Wilson 提交于 
      Following the removal of the last workarounds, the only CSB mmio access
is for the old vGPU interface. The mmio registers presented by vGPU do
not require forcewake and can be treated as ordinary volatile memory,
i.e. they behave just like the HWSP access just at a different location.
We can reduce the CSB access to a set of read/write/buffer pointers and
treat the various paths identically and not worry about forcewake.
(Forcewake is nightmare for worstcase latency, and we want to process
this all with irqsoff -- no latency allowed!)

v2: Comments, comments, comments. Well, 2 bonus comments.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180628201211.13837-5-chris@chris-wilson.co.uk
      bc4237ec
    • C
      drm/i915/execlists: Process one CSB update at a time · 8ea397fa
      由 Chris Wilson 提交于 
      In the next patch, we will process the CSB events directly from the
submission path, rather than only after a CS interrupt. Hence, we will
no longer have the need for a loop until the has-interrupt bit is clear,
and in the meantime can remove that small optimisation.

v2: Tvrtko pointed out it was safer to unconditionally kick the tasklet
after each irq, when assuming that the tasklet is called for each irq.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180628201211.13837-4-chris@chris-wilson.co.uk
      8ea397fa
    • C
      drm/i915/execlists: Pull CSB reset under the timeline.lock · d8857d54
      由 Chris Wilson 提交于 
      In the following patch, we will process the CSB events under the
timeline.lock and not serialised by the tasklet. This also means that we
will need to protect access to common variables such as
execlists->csb_head with the timeline.lock during reset.

v2: Move sync_irq to avoid deadlocks between taking timeline.lock from
our interrupt handler.
v3: Kill off the synchronize_hardirq as it raises more questions than
answered; now we use the timeline.lock entirely for CSB serialisation
between the irq and elsewhere, we don't need to be so heavy handed with
flushing
v4: Treat request cancellation (wedging after failed reset) similarly
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180628201211.13837-3-chris@chris-wilson.co.uk
      d8857d54
    • C
      drm/i915/execlists: Pull submit after dequeue under timeline lock · 0b02befa
      由 Chris Wilson 提交于 
      In the next patch, we will begin processing the CSB from inside the
submission path (underneath an irqsoff section, and even from inside
interrupt handlers). This means that updating the execlists->port[] will
no longer be serialised by the tasklet but needs to be locked by the
engine->timeline.lock instead. Pull dequeue and submit under the same
lock for protection. (An alternate future plan is to keep the in/out
arrays separate for concurrent processing and reduced lock coverage.)
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180628201211.13837-2-chris@chris-wilson.co.uk
      0b02befa
  3. 25 6月, 2018 2 次提交
  5. 18 6月, 2018 2 次提交
  7. 16 6月, 2018 1 次提交
  9. 15 6月, 2018 1 次提交
  11. 12 6月, 2018 1 次提交
  13. 11 6月, 2018 1 次提交
  15. 06 6月, 2018 1 次提交
  17. 05 6月, 2018 2 次提交
  19. 01 6月, 2018 1 次提交
  21. 25 5月, 2018 2 次提交
    • C
      drm/i915/execlists: Wait for ELSP submission on restart · fe25f304
      由 Chris Wilson 提交于 
      After a reset, we will ensure that there is at least one request
submitted to HW to ensure that a context is loaded for powersaving.
Let's wait for this submission via a tasklet to complete before we drop
our forcewake, ensuring the system is ready for rc6 before we let it
possibly sleep.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180522101937.7738-1-chris@chris-wilson.co.ukReviewed-by: NMika Kuoppala <mika.kuoppala@intel.com>
      fe25f304
    • C
      drm/i915: Flush the ring stop bit after clearing RING_HEAD in reset · 9a4dc803
      由 Chris Wilson 提交于 
      Inside the live_hangcheck (reset) selftests, we occasionally see
failures like

<7>[  239.094840] i915_gem_set_wedged rcs0
<7>[  239.094843] i915_gem_set_wedged 	current seqno 19a98, last 19a9a, hangcheck 0 [5158 ms]
<7>[  239.094846] i915_gem_set_wedged 	Reset count: 6239 (global 1)
<7>[  239.094848] i915_gem_set_wedged 	Requests:
<7>[  239.095052] i915_gem_set_wedged 		first  19a99 [e8c:5f] prio=1024 @ 5159ms: (null)
<7>[  239.095056] i915_gem_set_wedged 		last   19a9a [e81:1a] prio=139 @ 5159ms: igt/rcs0[5977]/1
<7>[  239.095059] i915_gem_set_wedged 		active 19a99 [e8c:5f] prio=1024 @ 5159ms: (null)
<7>[  239.095062] i915_gem_set_wedged 		[head 0220, postfix 0280, tail 02a8, batch 0xffffffff_ffffffff]
<7>[  239.100050] i915_gem_set_wedged 		ring->start:  0x00283000
<7>[  239.100053] i915_gem_set_wedged 		ring->head:   0x000001f8
<7>[  239.100055] i915_gem_set_wedged 		ring->tail:   0x000002a8
<7>[  239.100057] i915_gem_set_wedged 		ring->emit:   0x000002a8
<7>[  239.100059] i915_gem_set_wedged 		ring->space:  0x00000f10
<7>[  239.100085] i915_gem_set_wedged 	RING_START: 0x00283000
<7>[  239.100088] i915_gem_set_wedged 	RING_HEAD:  0x00000260
<7>[  239.100091] i915_gem_set_wedged 	RING_TAIL:  0x000002a8
<7>[  239.100094] i915_gem_set_wedged 	RING_CTL:   0x00000001
<7>[  239.100097] i915_gem_set_wedged 	RING_MODE:  0x00000300 [idle]
<7>[  239.100100] i915_gem_set_wedged 	RING_IMR: fffffefe
<7>[  239.100104] i915_gem_set_wedged 	ACTHD:  0x00000000_0000609c
<7>[  239.100108] i915_gem_set_wedged 	BBADDR: 0x00000000_0000609d
<7>[  239.100111] i915_gem_set_wedged 	DMA_FADDR: 0x00000000_00283260
<7>[  239.100114] i915_gem_set_wedged 	IPEIR: 0x00000000
<7>[  239.100117] i915_gem_set_wedged 	IPEHR: 0x02800000
<7>[  239.100120] i915_gem_set_wedged 	Execlist status: 0x00044052 00000002
<7>[  239.100124] i915_gem_set_wedged 	Execlist CSB read 5 [5 cached], write 5 [5 from hws], interrupt posted? no, tasklet queued? no (enabled)
<7>[  239.100128] i915_gem_set_wedged 		ELSP[0] count=1, ring->start=00283000, rq: 19a99 [e8c:5f] prio=1024 @ 5164ms: (null)
<7>[  239.100132] i915_gem_set_wedged 		ELSP[1] count=1, ring->start=00257000, rq: 19a9a [e81:1a] prio=139 @ 5164ms: igt/rcs0[5977]/1
<7>[  239.100135] i915_gem_set_wedged 		HW active? 0x5
<7>[  239.100250] i915_gem_set_wedged 		E 19a99 [e8c:5f] prio=1024 @ 5164ms: (null)
<7>[  239.100338] i915_gem_set_wedged 		E 19a9a [e81:1a] prio=139 @ 5164ms: igt/rcs0[5977]/1
<7>[  239.100340] i915_gem_set_wedged 		Queue priority: 139
<7>[  239.100343] i915_gem_set_wedged 		Q 0 [e98:19] prio=132 @ 5164ms: igt/rcs0[5977]/8
<7>[  239.100346] i915_gem_set_wedged 		Q 0 [e84:19] prio=121 @ 5165ms: igt/rcs0[5977]/2
<7>[  239.100349] i915_gem_set_wedged 		Q 0 [e87:19] prio=82 @ 5165ms: igt/rcs0[5977]/3
<7>[  239.100352] i915_gem_set_wedged 		Q 0 [e84:1a] prio=44 @ 5164ms: igt/rcs0[5977]/2
<7>[  239.100356] i915_gem_set_wedged 		Q 0 [e8b:19] prio=20 @ 5165ms: igt/rcs0[5977]/4
<7>[  239.100362] i915_gem_set_wedged 	drv_selftest [5894] waiting for 19a99

where the GPU saw an arbitration point and idles; AND HAS NOT BEEN RESET!
The RING_MODE indicates that is idle and has the STOP_RING bit set, so
try clearing it.

v2: Only clear the bit on restarting the ring, as we want to be sure the
STOP_RING bit is kept if reset fails on wedging.
v3: Spot when the ring state doesn't make sense when re-initialising the
engine and dump it to the logs so that we don't have to wait for an
error later and try to guess what happened earlier.
v4: Prepare to print all the unexpected state, not just the first.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180518100933.2239-1-chris@chris-wilson.co.uk
      9a4dc803
  23. 19 5月, 2018 1 次提交
  25. 18 5月, 2018 4 次提交
  27. 17 5月, 2018 6 次提交
  29. 11 5月, 2018 2 次提交
  31. 09 5月, 2018 1 次提交
  33. 08 5月, 2018 3 次提交
  35. 04 5月, 2018 1 次提交