Skip to contentSkip to conversation

drm/i915: Use ktime on wait_for

PATCH In review Mika Kuoppala 4 participants 6 messages Apr 23, 2018 – Apr 24, 2018

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/2 drm/i915: Use ktime on wait_for

We use jiffies to determine when wait expires. However Imre did find out that jiffies can and will do a >1 increments on certain situations [1]. When this happens in a wait_for loop, we return timeout errorneously much earlier than what the real wallclock would say.

We can't afford our waits to timeout prematurely. Discard jiffies and change to ktime to detect timeouts.

v2: added bugzilla entry (Imre), added stable (Chris)

References: https://lkml.org/lkml/2018/4/18/798 [1]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105771

---
 drivers/gpu/drm/i915/intel_drv.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/intel_drv.h +2 -2
index 44ed248f1fe9..33ff2638c92b 100644
@@ -49,12 +49,12 @@
  * check the condition before the timeout.
  */
 #define __wait_for(OP, COND, US, Wmin, Wmax) ({ \
-	unsigned long timeout__ = jiffies + usecs_to_jiffies(US) + 1;	\
+	const ktime_t end__ = ktime_add_ns(ktime_get_raw(), 1000ll * (US)); \
 	long wait__ = (Wmin); /* recommended min for usleep is 10 us */	\
 	int ret__;							\
 	might_sleep();							\
 	for (;;) {							\
-		bool expired__ = time_after(jiffies, timeout__);	\
+		const bool expired__ = ktime_after(ktime_get_raw(), end__); \
 		OP;							\
 		if (COND) {						\
 			ret__ = 0;					\
Signature · 6 lines
2.14.1

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#1 lore raw

PATCH 2/2 drm/i915: Add compiler barrier to wait_for

We need to be careful to not let compiler evaluate the expiration and the operation on it's terms.

Document and enforce that COND will be evaluated before checking timeout expiration.

---
 drivers/gpu/drm/i915/intel_drv.h | 4 ++++
 1 file changed, 4 insertions(+)
drivers/gpu/drm/i915/intel_drv.h +4 -0
index 33ff2638c92b..58868b93d2a0 100644
@@ -56,6 +56,8 @@
 	for (;;) {							\
 		const bool expired__ = ktime_after(ktime_get_raw(), end__); \
 		OP;							\
+		/* Guarantee COND check prior to timeout */		\
+		barrier();						\
 		if (COND) {						\
 			ret__ = 0;					\
 			break;						\
@@ -96,6 +98,8 @@
 		u64 now = local_clock(); \
 		if (!(ATOMIC)) \
 			preempt_enable(); \
+		/* Guarantee COND check prior to timeout */ \
+		barrier(); \
 		if (COND) { \
 			ret = 0; \
 			break; \
Signature · 6 lines
2.14.1

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#2 lore raw

✓ Fi.CI.BAT: success for series starting with [CI,1/2] drm/i915: Use ktime on wait_for

In reply to Mika Kuoppala
== Series Details ==

== Series Details ==

Series: series starting with [CI,1/2] drm/i915: Use ktime on wait_for
URL : https://patchwork.freedesktop.org/series/42115/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4078 -> Patchwork_8774 =

== Summary - SUCCESS ==

No regressions found.

External URL: https://patchwork.freedesktop.org/api/1.0/series/42115/revisions/1/mbox/

== Known issues ==

Here are the changes found in Patchwork_8774 that come from known issues:

=== IGT changes ===

    ==== Issues hit ====

    igt@kms_pipe_crc_basic@suspend-read-crc-pipe-c:
      fi-cnl-y3:          PASS -> DMESG-WARN (fdo#104951)

    
    ==== Possible fixes ====

    igt@kms_pipe_crc_basic@suspend-read-crc-pipe-c:
      fi-ivb-3520m:       DMESG-WARN (fdo#106084) -> PASS

fdo#104951 https://bugs.freedesktop.org/show_bug.cgi?id=104951
fdo#106084 https://bugs.freedesktop.org/show_bug.cgi?id=106084

== Participating hosts (36 -> 33) ==

Missing (3): fi-ctg-p8600 fi-ilk-m540 fi-skl-6700hq

== Build changes ==

* Linux: CI_DRM_4078 -> Patchwork_8774

CI_DRM_4078: 9391010824b34ec58217f816ba5e314e7312191d @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_4444: dcc44347494231feabc588c2a76998cbc9afdf8c @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
Patchwork_8774: d8fac5f69e93f2197dc4bd591b196a524873a795 @ git://anongit.freedesktop.org/gfx-ci/linux
piglit_4444: a2f486679f467cd6e82578384f56d4aabaa8cf2e @ git://anongit.freedesktop.org/piglit

== Linux commits ==

d8fac5f69e93 drm/i915: Add compiler barrier to wait_for 4dbef50fbcf3 drm/i915: Use ktime on wait_for

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_8774/issues.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#3 lore raw
In reply to Mika Kuoppala
Quoted 7 lines · Mika Kuoppala

We use jiffies to determine when wait expires. However Imre did find out that jiffies can and will do a >1 increments on certain situations [1]. When this happens in a wait_for loop, we return timeout errorneously much earlier than what the real wallclock would say.

We can't afford our waits to timeout prematurely.

Isn't this also the reason why we have *_to_jiffies_timeout() family of functions? Should those
be converted as well?

Lucas De Marchi

Quoted 43 lines

Discard jiffies and change to ktime to detect timeouts.

v2: added bugzilla entry (Imre), added stable (Chris)

References: https://lkml.org/lkml/2018/4/18/798 [1]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105771

---
 drivers/gpu/drm/i915/intel_drv.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/intel_drv.h +2 -3
index 44ed248f1fe9..33ff2638c92b 100644
@@ -49,12 +49,12 @@
  * check the condition before the timeout.
  */
 #define __wait_for(OP, COND, US, Wmin, Wmax) ({ \
-       unsigned long timeout__ = jiffies + usecs_to_jiffies(US) + 1;   \
+       const ktime_t end__ = ktime_add_ns(ktime_get_raw(), 1000ll * (US)); \
        long wait__ = (Wmin); /* recommended min for usleep is 10 us */ \
        int ret__;                                                      \
        might_sleep();                                                  \
        for (;;) {                                                      \
-               bool expired__ = time_after(jiffies, timeout__);        \
+               const bool expired__ = ktime_after(ktime_get_raw(), end__); \
                OP;                                                     \
                if (COND) {                                             \
                        ret__ = 0;                                      \
--

2.14.1

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Signature · 5 lines
Lucas De Marchi
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#4 lore raw
Quoted 8 lines · Lucas De Marchi
Quoted 6 lines · Mika Kuoppala

We use jiffies to determine when wait expires. However Imre did find out that jiffies can and will do a >1 increments on certain situations [1]. When this happens in a wait_for loop, we return timeout errorneously much earlier than what the real wallclock would say.

Quoted 4 lines
Quoted 1 line

We can't afford our waits to timeout prematurely.

Isn't this also the reason why we have *_to_jiffies_timeout() family of functions?

The reason for those is that you can sample jiffies between two clock tick updates and calculate a short expiry time. So a different issue than missed updates of jiffies due to some latency.

Quoted 43 lines

Should those be converted as well?

Lucas De Marchi

Quoted 39 lines

Discard jiffies and change to ktime to detect timeouts.

v2: added bugzilla entry (Imre), added stable (Chris)

References: https://lkml.org/lkml/2018/4/18/798 [1]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105771

---
 drivers/gpu/drm/i915/intel_drv.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/intel_drv.h +2 -3
index 44ed248f1fe9..33ff2638c92b 100644
@@ -49,12 +49,12 @@
  * check the condition before the timeout.
  */
 #define __wait_for(OP, COND, US, Wmin, Wmax) ({ \
-       unsigned long timeout__ = jiffies + usecs_to_jiffies(US) + 1;   \
+       const ktime_t end__ = ktime_add_ns(ktime_get_raw(), 1000ll * (US)); \
        long wait__ = (Wmin); /* recommended min for usleep is 10 us */ \
        int ret__;                                                      \
        might_sleep();                                                  \
        for (;;) {                                                      \
-               bool expired__ = time_after(jiffies, timeout__);        \
+               const bool expired__ = ktime_after(ktime_get_raw(), end__); \
                OP;                                                     \
                if (COND) {                                             \
                        ret__ = 0;                                      \
--

2.14.1

Quoted 13 lines
Quoted 4 lines

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Signature · 5 lines
Lucas De Marchi
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#5 lore raw

✓ Fi.CI.IGT: success for series starting with [CI,1/2] drm/i915: Use ktime on wait_for

In reply to Mika Kuoppala
== Series Details ==

== Series Details ==

Series: series starting with [CI,1/2] drm/i915: Use ktime on wait_for
URL : https://patchwork.freedesktop.org/series/42115/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4078_full -> Patchwork_8774_full =

== Summary - WARNING ==

Minor unknown changes coming with Patchwork_8774_full need to be verified manually.

If you think the reported changes have nothing to do with the changes introduced in Patchwork_8774_full, please notify your bug team to allow them to document this new failure mode, which will reduce false positives in CI.

External URL: https://patchwork.freedesktop.org/api/1.0/series/42115/revisions/1/mbox/

== Possible new issues ==

Here are the unknown changes that may have been introduced in Patchwork_8774_full:

=== IGT changes ===

    ==== Warnings ====

    igt@gem_exec_schedule@deep-bsd2:
      shard-kbl:          SKIP -> PASS

    igt@gem_mocs_settings@mocs-rc6-vebox:
      shard-kbl:          PASS -> SKIP +1

    igt@kms_atomic_transition@plane-all-transition-nonblocking:
      shard-snb:          PASS -> SKIP +2

    igt@kms_vblank@pipe-b-wait-forked-busy-hang:
      shard-glk:          SKIP -> PASS +79

    igt@kms_vblank@pipe-c-wait-forked-busy:
      shard-glk:          PASS -> SKIP +103

== Known issues ==

Here are the changes found in Patchwork_8774_full that come from known issues:

=== IGT changes ===

    ==== Issues hit ====

    igt@kms_flip@2x-flip-vs-wf_vblank-interruptible:
      shard-hsw:          PASS -> FAIL (fdo#100368)

    igt@kms_flip@flip-vs-wf_vblank-interruptible:
      shard-glk:          PASS -> FAIL (fdo#105312)

    igt@kms_flip@plain-flip-ts-check:
      shard-glk:          PASS -> FAIL (fdo#100368)

    igt@kms_frontbuffer_tracking@fbc-1p-offscren-pri-indfb-draw-mmap-wc:
      shard-kbl:          PASS -> INCOMPLETE (fdo#103665)

    igt@kms_vblank@pipe-c-accuracy-idle:
      shard-hsw:          PASS -> FAIL (fdo#102583)

    
    ==== Possible fixes ====

    igt@gem_eio@suspend:
      shard-kbl:          DMESG-WARN -> PASS

    igt@kms_atomic_transition@1x-modeset-transitions-nonblocking-fencing:
      shard-apl:          FAIL (fdo#103207) -> PASS

    igt@kms_flip@2x-flip-vs-absolute-wf_vblank-interruptible:
      shard-hsw:          FAIL (fdo#100368) -> PASS

    igt@kms_flip@blocking-absolute-wf_vblank-interruptible:
      shard-glk:          FAIL (fdo#106134) -> SKIP

    igt@kms_flip@flip-vs-absolute-wf_vblank-interruptible:
      shard-glk:          FAIL (fdo#105312) -> PASS

    igt@kms_flip@modeset-vs-vblank-race:
      shard-hsw:          FAIL (fdo#103060) -> PASS

    igt@kms_flip@wf_vblank-ts-check-interruptible:
      shard-glk:          FAIL (fdo#103933) -> PASS

    igt@kms_setmode@basic:
      shard-glk:          FAIL (fdo#99912) -> PASS

fdo#100368 https://bugs.freedesktop.org/show_bug.cgi?id=100368
fdo#102583 https://bugs.freedesktop.org/show_bug.cgi?id=102583
fdo#103060 https://bugs.freedesktop.org/show_bug.cgi?id=103060
fdo#103207 https://bugs.freedesktop.org/show_bug.cgi?id=103207
fdo#103665 https://bugs.freedesktop.org/show_bug.cgi?id=103665
fdo#103933 https://bugs.freedesktop.org/show_bug.cgi?id=103933
fdo#105312 https://bugs.freedesktop.org/show_bug.cgi?id=105312
fdo#106134 https://bugs.freedesktop.org/show_bug.cgi?id=106134
fdo#99912 https://bugs.freedesktop.org/show_bug.cgi?id=99912

== Participating hosts (6 -> 5) ==

Missing (1): shard-glkb

== Build changes ==

* Linux: CI_DRM_4078 -> Patchwork_8774

CI_DRM_4078: 9391010824b34ec58217f816ba5e314e7312191d @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_4444: dcc44347494231feabc588c2a76998cbc9afdf8c @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
Patchwork_8774: d8fac5f69e93f2197dc4bd591b196a524873a795 @ git://anongit.freedesktop.org/gfx-ci/linux
piglit_4444: a2f486679f467cd6e82578384f56d4aabaa8cf2e @ git://anongit.freedesktop.org/piglit

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_8774/shards.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx