Skip to contentSkip to conversation

drm/i915/perf: Do not clear pollin for small user read buffers

PATCH In review Ashutosh Dixit 3 participants 6 messages Mar 31, 2020 – Apr 1, 2020

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

PATCH 1/1 drm/i915/perf: Do not clear pollin for small user read buffers

It is wrong to block the user thread in the next poll when OA data is already available which could not fit in the user buffer provided in the previous read. In several cases the exact user buffer size is not known. Blocking user space in poll can lead to data loss when the buffer size used is smaller than the available data.

This change fixes this issue and allows user space to read all OA data even when using a buffer size smaller than the available data using multiple non-blocking reads rather than staying blocked in poll till the next timer interrupt.

v2: Fix ret value for blocking reads (Umesh)
v3: Mistake during patch send (Ashutosh)
v4: Remove -EAGAIN from comment (Umesh)
v5: Improve condition for clearing pollin and return (Lionel)
v6: Improve blocking read loop and other cleanups (Lionel)

Cc Umesh Nerlige Ramappa <[email protected]>
Cc Lionel Landwerlin <[email protected]>
---
 drivers/gpu/drm/i915/i915_perf.c | 61 ++++++--------------------------
 1 file changed, 11 insertions(+), 50 deletions(-)
drivers/gpu/drm/i915/i915_perf.c +11 -50
index 28e3d76fa2e6..2f78b147bb2d 100644
@@ -2963,49 +2963,6 @@ void i915_oa_init_reg_state(const struct intel_context *ce,
 		gen8_update_reg_state_unlocked(ce, stream);
 }
 
-/**
- * i915_perf_read_locked - &i915_perf_stream_ops->read with error normalisation
- * @stream: An i915 perf stream
- * @file: An i915 perf stream file
- * @buf: destination buffer given by userspace
- * @count: the number of bytes userspace wants to read
- * @ppos: (inout) file seek position (unused)
- *
- * Besides wrapping &i915_perf_stream_ops->read this provides a common place to
- * ensure that if we've successfully copied any data then reporting that takes
- * precedence over any internal error status, so the data isn't lost.
- *
- * For example ret will be -ENOSPC whenever there is more buffered data than
- * can be copied to userspace, but that's only interesting if we weren't able
- * to copy some data because it implies the userspace buffer is too small to
- * receive a single record (and we never split records).
- *
- * Another case with ret == -EFAULT is more of a grey area since it would seem
- * like bad form for userspace to ask us to overrun its buffer, but the user
- * knows best:
- *
- *   http://yarchive.net/comp/linux/partial_reads_writes.html
- *
- * Returns: The number of bytes copied or a negative error code on failure.
- */
-static ssize_t i915_perf_read_locked(struct i915_perf_stream *stream,
-				     struct file *file,
-				     char __user *buf,
-				     size_t count,
-				     loff_t *ppos)
-{
-	/* Note we keep the offset (aka bytes read) separate from any
-	 * error status so that the final check for whether we return
-	 * the bytes read with a higher precedence than any error (see
-	 * comment below) doesn't need to be handled/duplicated in
-	 * stream->ops->read() implementations.
-	 */
-	size_t offset = 0;
-	int ret = stream->ops->read(stream, buf, count, &offset);
-
-	return offset ?: (ret ?: -EAGAIN);
-}
-
 /**
  * i915_perf_read - handles read() FOP for i915 perf stream FDs
  * @file: An i915 perf stream file
@@ -3031,7 +2988,8 @@ static ssize_t i915_perf_read(struct file *file,
 {
 	struct i915_perf_stream *stream = file->private_data;
 	struct i915_perf *perf = stream->perf;
-	ssize_t ret;
+	size_t offset = 0;
+	int ret;
 
 	/* To ensure it's handled consistently we simply treat all reads of a
 	 * disabled stream as an error. In particular it might otherwise lead
@@ -3054,13 +3012,12 @@ static ssize_t i915_perf_read(struct file *file,
 				return ret;
 
 			mutex_lock(&perf->lock);
-			ret = i915_perf_read_locked(stream, file,
-						    buf, count, ppos);
+			ret = stream->ops->read(stream, buf, count, &offset);
 			mutex_unlock(&perf->lock);
-		} while (ret == -EAGAIN);
+		} while (!offset && !ret);
 	} else {
 		mutex_lock(&perf->lock);
-		ret = i915_perf_read_locked(stream, file, buf, count, ppos);
+		ret = stream->ops->read(stream, buf, count, &offset);
 		mutex_unlock(&perf->lock);
 	}
 
@@ -3071,11 +3028,15 @@ static ssize_t i915_perf_read(struct file *file,
 	 * and read() returning -EAGAIN. Clearing the oa.pollin state here
 	 * effectively ensures we back off until the next hrtimer callback
 	 * before reporting another EPOLLIN event.
+	 * The exception to this is if ops->read() returned -ENOSPC which means
+	 * that more OA data is available than could fit in the user provided
+	 * buffer. In this case we want the next poll() call to not block.
 	 */
-	if (ret >= 0 || ret == -EAGAIN)
+	if (ret != -ENOSPC)
 		stream->pollin = false;
 
-	return ret;
+	/* Possible values for ret are 0, -EFAULT, -ENOSPC, -EIO, ... */
+	return offset ?: (ret ?: -EAGAIN);
 }
 
 static enum hrtimer_restart oa_poll_check_timer_cb(struct hrtimer *hrtimer)
Signature · 6 lines
2.25.2

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

[Intel-gfx] ✓ Fi.CI.BAT: success for drm/i915/perf: Do not clear pollin for small user read buffers (rev6)

== Series Details ==

== Series Details ==

Series: drm/i915/perf: Do not clear pollin for small user read buffers (rev6)
URL : https://patchwork.freedesktop.org/series/75085/
State : success

== Summary ==

CI Bug Log - changes from CI_DRM_8230 -> Patchwork_17166
====================================================

Summary
-------

**SUCCESS**

No regressions found.

External URL: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/index.html

Known issues
------------

Here are the changes found in Patchwork_17166 that come from known issues:

### IGT changes ###

#### Issues hit ####

* igt@i915_selftest@live@late_gt_pm:
- fi-bsw-n3050: [PASS][1] -> [INCOMPLETE][2] ([i915#1382])
[1]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/fi-bsw-n3050/igt@i915_selftest@live@late_gt_pm.html
[2]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/fi-bsw-n3050/igt@i915_selftest@live@late_gt_pm.html

{name}: This element is suppressed. This means it is ignored when computing
the status of the difference (SUCCESS, WARNING, or FAILURE).

[i915#1382]: https://gitlab.freedesktop.org/drm/intel/issues/1382
[i915#529]: https://gitlab.freedesktop.org/drm/intel/issues/529

Participating hosts (40 -> 36)
------------------------------

Additional (8): fi-byt-j1900 fi-skl-guc fi-ilk-650 fi-kbl-7500u fi-hsw-4770 fi-bsw-kefka fi-bsw-nick fi-skl-6700k2
Missing (12): fi-kbl-soraka fi-hsw-4200u fi-byt-squawks fi-bsw-cyan fi-bwr-2160 fi-ctg-p8600 fi-ivb-3770 fi-cfl-8109u fi-byt-n2820 fi-byt-clapper fi-bdw-samus fi-snb-2600

Build changes
-------------

* CI: CI-20190529 -> None
* Linux: CI_DRM_8230 -> Patchwork_17166

CI-20190529: 20190529
CI_DRM_8230: fa9f8453ffb88a4fc4e36d68b84a7ff9bf90f769 @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_5550: 98927dfde17aecaecfe67bb9853ceca326ca2b23 @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
Patchwork_17166: db60a6358595b602a7eaae13fb3c579af95f8da6 @ git://anongit.freedesktop.org/gfx-ci/linux

== Linux commits ==

db60a6358595 drm/i915/perf: Do not clear pollin for small user read buffers

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/index.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#2 lore raw
In reply to Ashutosh Dixit
Quoted 89 lines · Ashutosh Dixit

It is wrong to block the user thread in the next poll when OA data is already available which could not fit in the user buffer provided in the previous read. In several cases the exact user buffer size is not known. Blocking user space in poll can lead to data loss when the buffer size used is smaller than the available data.

This change fixes this issue and allows user space to read all OA data even when using a buffer size smaller than the available data using multiple non-blocking reads rather than staying blocked in poll till the next timer interrupt.

v2: Fix ret value for blocking reads (Umesh)
v3: Mistake during patch send (Ashutosh)
v4: Remove -EAGAIN from comment (Umesh)
v5: Improve condition for clearing pollin and return (Lionel)
v6: Improve blocking read loop and other cleanups (Lionel)

Cc Umesh Nerlige Ramappa <[email protected]>
Cc Lionel Landwerlin <[email protected]>
---
  drivers/gpu/drm/i915/i915_perf.c | 61 ++++++--------------------------
  1 file changed, 11 insertions(+), 50 deletions(-)
drivers/gpu/drm/i915/i915_perf.c +2 -44
index 28e3d76fa2e6..2f78b147bb2d 100644
@@ -2963,49 +2963,6 @@ void i915_oa_init_reg_state(const struct intel_context *ce,
  		gen8_update_reg_state_unlocked(ce, stream);
  }
 
-/**
- * i915_perf_read_locked - &i915_perf_stream_ops->read with error normalisation
- * @stream: An i915 perf stream
- * @file: An i915 perf stream file
- * @buf: destination buffer given by userspace
- * @count: the number of bytes userspace wants to read
- * @ppos: (inout) file seek position (unused)
- *
- * Besides wrapping &i915_perf_stream_ops->read this provides a common place to
- * ensure that if we've successfully copied any data then reporting that takes
- * precedence over any internal error status, so the data isn't lost.
- *
- * For example ret will be -ENOSPC whenever there is more buffered data than
- * can be copied to userspace, but that's only interesting if we weren't able
- * to copy some data because it implies the userspace buffer is too small to
- * receive a single record (and we never split records).
- *
- * Another case with ret == -EFAULT is more of a grey area since it would seem
- * like bad form for userspace to ask us to overrun its buffer, but the user
- * knows best:
- *
- *   http://yarchive.net/comp/linux/partial_reads_writes.html
- *
- * Returns: The number of bytes copied or a negative error code on failure.
- */
-static ssize_t i915_perf_read_locked(struct i915_perf_stream *stream,
-				     struct file *file,
-				     char __user *buf,
-				     size_t count,
-				     loff_t *ppos)
-{
-	/* Note we keep the offset (aka bytes read) separate from any
-	 * error status so that the final check for whether we return
-	 * the bytes read with a higher precedence than any error (see
-	 * comment below) doesn't need to be handled/duplicated in
-	 * stream->ops->read() implementations.
-	 */
-	size_t offset = 0;
-	int ret = stream->ops->read(stream, buf, count, &offset);
-
-	return offset ?: (ret ?: -EAGAIN);
-}
-
  /**
   * i915_perf_read - handles read() FOP for i915 perf stream FDs
   * @file: An i915 perf stream file
@@ -3031,7 +2988,8 @@ static ssize_t i915_perf_read(struct file *file,
  {
  	struct i915_perf_stream *stream = file->private_data;
  	struct i915_perf *perf = stream->perf;
-	ssize_t ret;
+	size_t offset = 0;
+	int ret;
 
  	/* To ensure it's handled consistently we simply treat all reads of a
  	 * disabled stream as an error. In particular it might otherwise lead

drivers/gpu/drm/i915/i915_perf.c static ssize_t i915_perf_read(struct file *file, · line 3017

Quoted 9 lines
drivers/gpu/drm/i915/i915_perf.c +2 -3
@@ -3054,13 +3012,12 @@ static ssize_t i915_perf_read(struct file *file,
  				return ret;
 
  			mutex_lock(&perf->lock);
-			ret = i915_perf_read_locked(stream, file,
-						    buf, count, ppos);
+			ret = stream->ops->read(stream, buf, count, &offset);
  			mutex_unlock(&perf->lock);
-		} while (ret == -EAGAIN);
+		} while (!offset && !ret);

This doesn't sound right, !offset means it will stop as soon as some data was written.

But for the blocking read we want to fill the buffer up to -ENOSPC.

while (ret >= 0) doesn't work?

-Lionel

Quoted 15 lines
drivers/gpu/drm/i915/i915_perf.c +4 -1
  	} else {
  		mutex_lock(&perf->lock);
-		ret = i915_perf_read_locked(stream, file, buf, count, ppos);
+		ret = stream->ops->read(stream, buf, count, &offset);
  		mutex_unlock(&perf->lock);
  	}
 
@@ -3071,11 +3028,15 @@ static ssize_t i915_perf_read(struct file *file,
  	 * and read() returning -EAGAIN. Clearing the oa.pollin state here
  	 * effectively ensures we back off until the next hrtimer callback
  	 * before reporting another EPOLLIN event.
+	 * The exception to this is if ops->read() returned -ENOSPC which means
+	 * that more OA data is available than could fit in the user provided
+	 * buffer. In this case we want the next poll() call to not block.
  	 */

drivers/gpu/drm/i915/i915_perf.c static ssize_t i915_perf_read(struct file *file, · line 3042

Quoted 10 lines
drivers/gpu/drm/i915/i915_perf.c +3 -2
-	if (ret >= 0 || ret == -EAGAIN)
+	if (ret != -ENOSPC)
  		stream->pollin = false;
 
-	return ret;
+	/* Possible values for ret are 0, -EFAULT, -ENOSPC, -EIO, ... */
+	return offset ?: (ret ?: -EAGAIN);
  }
 
  static enum hrtimer_restart oa_poll_check_timer_cb(struct hrtimer *hrtimer)

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

#3 lore raw
Quoted 102 lines · Lionel Landwerlin
Quoted 88 lines · Ashutosh Dixit

It is wrong to block the user thread in the next poll when OA data is already available which could not fit in the user buffer provided in the previous read. In several cases the exact user buffer size is not known. Blocking user space in poll can lead to data loss when the buffer size used is smaller than the available data.

This change fixes this issue and allows user space to read all OA data even when using a buffer size smaller than the available data using multiple non-blocking reads rather than staying blocked in poll till the next timer interrupt.

v2: Fix ret value for blocking reads (Umesh)
v3: Mistake during patch send (Ashutosh)
v4: Remove -EAGAIN from comment (Umesh)
v5: Improve condition for clearing pollin and return (Lionel)
v6: Improve blocking read loop and other cleanups (Lionel)

Cc Umesh Nerlige Ramappa <[email protected]>
Cc Lionel Landwerlin <[email protected]>
---
  drivers/gpu/drm/i915/i915_perf.c | 61 ++++++--------------------------
  1 file changed, 11 insertions(+), 50 deletions(-)
drivers/gpu/drm/i915/i915_perf.c +2 -43
index 28e3d76fa2e6..2f78b147bb2d 100644
@@ -2963,49 +2963,6 @@ void i915_oa_init_reg_state(const struct intel_context *ce,
 		gen8_update_reg_state_unlocked(ce, stream);
  }
  -/**
- * i915_perf_read_locked - &i915_perf_stream_ops->read with error normalisation
- * @stream: An i915 perf stream
- * @file: An i915 perf stream file
- * @buf: destination buffer given by userspace
- * @count: the number of bytes userspace wants to read
- * @ppos: (inout) file seek position (unused)
- *
- * Besides wrapping &i915_perf_stream_ops->read this provides a common place to
- * ensure that if we've successfully copied any data then reporting that takes
- * precedence over any internal error status, so the data isn't lost.
- *
- * For example ret will be -ENOSPC whenever there is more buffered data than
- * can be copied to userspace, but that's only interesting if we weren't able
- * to copy some data because it implies the userspace buffer is too small to
- * receive a single record (and we never split records).
- *
- * Another case with ret == -EFAULT is more of a grey area since it would seem
- * like bad form for userspace to ask us to overrun its buffer, but the user
- * knows best:
- *
- *   http://yarchive.net/comp/linux/partial_reads_writes.html
- *
- * Returns: The number of bytes copied or a negative error code on failure.
- */
-static ssize_t i915_perf_read_locked(struct i915_perf_stream *stream,
-				     struct file *file,
-				     char __user *buf,
-				     size_t count,
-				     loff_t *ppos)
-{
-	/* Note we keep the offset (aka bytes read) separate from any
-	 * error status so that the final check for whether we return
-	 * the bytes read with a higher precedence than any error (see
-	 * comment below) doesn't need to be handled/duplicated in
-	 * stream->ops->read() implementations.
-	 */
-	size_t offset = 0;
-	int ret = stream->ops->read(stream, buf, count, &offset);
-
-	return offset ?: (ret ?: -EAGAIN);
-}
-
  /**
   * i915_perf_read - handles read() FOP for i915 perf stream FDs
   * @file: An i915 perf stream file
@@ -3031,7 +2988,8 @@ static ssize_t i915_perf_read(struct file *file,
  {
 	struct i915_perf_stream *stream = file->private_data;
 	struct i915_perf *perf = stream->perf;
-	ssize_t ret;
+	size_t offset = 0;
+	int ret;
 		/* To ensure it's handled consistently we simply treat all reads of

a
* disabled stream as an error. In particular it might otherwise lead

drivers/gpu/drm/i915/i915_perf.c static ssize_t i915_perf_read(struct file *file, · line 3016

Quoted 8 lines
drivers/gpu/drm/i915/i915_perf.c +2 -3
@@ -3054,13 +3012,12 @@ static ssize_t i915_perf_read(struct file *file,
 				return ret;
 				mutex_lock(&perf->lock);
-			ret = i915_perf_read_locked(stream, file,
-						    buf, count, ppos);
+			ret = stream->ops->read(stream, buf, count, &offset);
 			mutex_unlock(&perf->lock);
-		} while (ret == -EAGAIN);
+		} while (!offset && !ret);

This doesn't sound right, !offset means it will stop as soon as some data was written.

Quoted 1 line

But for the blocking read we want to fill the buffer up to -ENOSPC.

I don't think that's true. Here's 'man 2 read': "read() attempts to read /up to/ count bytes" and "It is not an error if this number is smaller than the number of bytes requested".

The condition (!offset && !ret) is exactly equivalent to the condition (ret == -EAGAIN) in the original code (currently on drm-tip). The driver is free to unblock the blocking read whenever it determines "there is data". Our determination of "there is data" is "we are woken up by the OA timer and call ops->read() and offset > 0". (Offset will be equal to min(amount of data available, space in the user buffer)). The only constraint seems to be that the blocking read cannot return -EAGAIN (0 bytes) and the loop in the code guards against that.

Quoted 1 line

while (ret >= 0) doesn't work?

Because this is not the logic in the original code and I see no reason to change that logic. It will also change the blocking read behavior which according to some people is a breakage of the uAPI. The purpose of the patch is to fix the non blocking read path (poll + non-blocking read). It should not affect blocking read imo.

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

Reviewed-by
#4 lore raw
Quoted 122 lines · Ashutosh
Quoted 100 lines · Lionel Landwerlin
Quoted 88 lines · Ashutosh Dixit

It is wrong to block the user thread in the next poll when OA data is already available which could not fit in the user buffer provided in the previous read. In several cases the exact user buffer size is not known. Blocking user space in poll can lead to data loss when the buffer size used is smaller than the available data.

This change fixes this issue and allows user space to read all OA data even when using a buffer size smaller than the available data using multiple non-blocking reads rather than staying blocked in poll till the next timer interrupt.

v2: Fix ret value for blocking reads (Umesh)
v3: Mistake during patch send (Ashutosh)
v4: Remove -EAGAIN from comment (Umesh)
v5: Improve condition for clearing pollin and return (Lionel)
v6: Improve blocking read loop and other cleanups (Lionel)

Cc Umesh Nerlige Ramappa <[email protected]>
Cc Lionel Landwerlin <[email protected]>
---
   drivers/gpu/drm/i915/i915_perf.c | 61 ++++++--------------------------
   1 file changed, 11 insertions(+), 50 deletions(-)
drivers/gpu/drm/i915/i915_perf.c +2 -43
index 28e3d76fa2e6..2f78b147bb2d 100644
@@ -2963,49 +2963,6 @@ void i915_oa_init_reg_state(const struct intel_context *ce,
 		gen8_update_reg_state_unlocked(ce, stream);
   }
   -/**
- * i915_perf_read_locked - &i915_perf_stream_ops->read with error normalisation
- * @stream: An i915 perf stream
- * @file: An i915 perf stream file
- * @buf: destination buffer given by userspace
- * @count: the number of bytes userspace wants to read
- * @ppos: (inout) file seek position (unused)
- *
- * Besides wrapping &i915_perf_stream_ops->read this provides a common place to
- * ensure that if we've successfully copied any data then reporting that takes
- * precedence over any internal error status, so the data isn't lost.
- *
- * For example ret will be -ENOSPC whenever there is more buffered data than
- * can be copied to userspace, but that's only interesting if we weren't able
- * to copy some data because it implies the userspace buffer is too small to
- * receive a single record (and we never split records).
- *
- * Another case with ret == -EFAULT is more of a grey area since it would seem
- * like bad form for userspace to ask us to overrun its buffer, but the user
- * knows best:
- *
- *   http://yarchive.net/comp/linux/partial_reads_writes.html
- *
- * Returns: The number of bytes copied or a negative error code on failure.
- */
-static ssize_t i915_perf_read_locked(struct i915_perf_stream *stream,
-				     struct file *file,
-				     char __user *buf,
-				     size_t count,
-				     loff_t *ppos)
-{
-	/* Note we keep the offset (aka bytes read) separate from any
-	 * error status so that the final check for whether we return
-	 * the bytes read with a higher precedence than any error (see
-	 * comment below) doesn't need to be handled/duplicated in
-	 * stream->ops->read() implementations.
-	 */
-	size_t offset = 0;
-	int ret = stream->ops->read(stream, buf, count, &offset);
-
-	return offset ?: (ret ?: -EAGAIN);
-}
-
   /**
    * i915_perf_read - handles read() FOP for i915 perf stream FDs
    * @file: An i915 perf stream file
@@ -3031,7 +2988,8 @@ static ssize_t i915_perf_read(struct file *file,
   {
 	struct i915_perf_stream *stream = file->private_data;
 	struct i915_perf *perf = stream->perf;
-	ssize_t ret;
+	size_t offset = 0;
+	int ret;
 		/* To ensure it's handled consistently we simply treat all reads of

a
* disabled stream as an error. In particular it might otherwise lead

drivers/gpu/drm/i915/i915_perf.c static ssize_t i915_perf_read(struct file *file, · line 3016

Quoted 8 lines
drivers/gpu/drm/i915/i915_perf.c +2 -3
@@ -3054,13 +3012,12 @@ static ssize_t i915_perf_read(struct file *file,
 				return ret;
 				mutex_lock(&perf->lock);
-			ret = i915_perf_read_locked(stream, file,
-						    buf, count, ppos);
+			ret = stream->ops->read(stream, buf, count, &offset);
 			mutex_unlock(&perf->lock);
-		} while (ret == -EAGAIN);
+		} while (!offset && !ret);

This doesn't sound right, !offset means it will stop as soon as some data was written.

Quoted 1 line

But for the blocking read we want to fill the buffer up to -ENOSPC.

I don't think that's true. Here's 'man 2 read': "read() attempts to read /up to/ count bytes" and "It is not an error if this number is smaller than the number of bytes requested".

The condition (!offset && !ret) is exactly equivalent to the condition (ret == -EAGAIN) in the original code (currently on drm-tip). The driver is free to unblock the blocking read whenever it determines "there is data". Our determination of "there is data" is "we are woken up by the OA timer and call ops->read() and offset > 0". (Offset will be equal to min(amount of data available, space in the user buffer)). The only constraint seems to be that the blocking read cannot return -EAGAIN (0 bytes) and the loop in the code guards against that.

Quoted 1 line

while (ret >= 0) doesn't work?

Because this is not the logic in the original code and I see no reason to change that logic. It will also change the blocking read behavior which according to some people is a breakage of the uAPI. The purpose of the patch is to fix the non blocking read path (poll + non-blocking read). It should not affect blocking read imo.

Quoted 3 lines

Thanks!
--
Ashutosh

Ah sorry, you're right :)

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

#5 lore raw

[Intel-gfx] ✓ Fi.CI.IGT: success for drm/i915/perf: Do not clear pollin for small user read buffers (rev6)

In reply to Ashutosh Dixit
== Series Details ==

== Series Details ==

Series: drm/i915/perf: Do not clear pollin for small user read buffers (rev6)
URL : https://patchwork.freedesktop.org/series/75085/
State : success

== Summary ==

CI Bug Log - changes from CI_DRM_8230_full -> Patchwork_17166_full
====================================================

Summary
-------

**SUCCESS**

No regressions found.

New tests
---------

New tests have been introduced between CI_DRM_8230_full and Patchwork_17166_full:

### New IGT tests (1) ###

* igt@gem_exec_capture@capture:
- Statuses :
- Exec time: [None] s

Known issues
------------

Here are the changes found in Patchwork_17166_full that come from known issues:

### IGT changes ###

#### Issues hit ####

* igt@gem_exec_schedule@implicit-read-write-bsd2:
- shard-iclb: [PASS][1] -> [SKIP][2] ([fdo#109276] / [i915#677])
[1]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-iclb2/igt@[email protected]
[2]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-iclb5/igt@[email protected]

* igt@gem_exec_schedule@out-order-bsd2:
- shard-iclb: [PASS][3] -> [SKIP][4] ([fdo#109276]) +7 similar issues
[3]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-iclb4/igt@[email protected]
[4]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-iclb3/igt@[email protected]

* igt@gem_exec_schedule@pi-common-bsd:
- shard-iclb: [PASS][5] -> [SKIP][6] ([i915#677])
[5]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-iclb6/igt@[email protected]
[6]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-iclb2/igt@[email protected]

* igt@gem_exec_schedule@preempt-other-chain-bsd:
- shard-iclb: [PASS][7] -> [SKIP][8] ([fdo#112146]) +6 similar issues
[7]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-iclb7/igt@[email protected]
[8]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-iclb1/igt@[email protected]

* igt@gem_exec_suspend@basic-s3:
- shard-kbl: [PASS][9] -> [DMESG-WARN][10] ([i915#180]) +1 similar issue
[9]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-kbl6/igt@[email protected]
[10]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-kbl6/igt@[email protected]

* igt@i915_pm_dc@dc6-psr:
- shard-iclb: [PASS][11] -> [FAIL][12] ([i915#454])
[11]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-iclb3/igt@[email protected]
[12]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-iclb4/igt@[email protected]

* igt@kms_cursor_crc@pipe-a-cursor-64x64-random:
- shard-apl: [PASS][13] -> [FAIL][14] ([i915#54] / [i915#95])
[13]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-apl3/igt@[email protected]
[14]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-apl8/igt@[email protected]

* igt@kms_cursor_crc@pipe-b-cursor-suspend:
- shard-skl: [PASS][15] -> [INCOMPLETE][16] ([i915#300])
[15]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-skl8/igt@[email protected]
[16]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-skl8/igt@[email protected]

* igt@kms_cursor_legacy@flip-vs-cursor-atomic-transitions:
- shard-skl: [PASS][17] -> [FAIL][18] ([IGT#5] / [i915#697])
[17]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-skl10/igt@[email protected]
[18]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-skl9/igt@[email protected]

* igt@kms_flip@plain-flip-ts-check-interruptible:
- shard-skl: [PASS][19] -> [FAIL][20] ([i915#34])
[19]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-skl3/igt@[email protected]
[20]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-skl5/igt@[email protected]

* igt@kms_plane_alpha_blend@pipe-a-constant-alpha-min:
- shard-skl: [PASS][21] -> [FAIL][22] ([fdo#108145]) +1 similar issue
[21]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-skl10/igt@[email protected]
[22]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-skl10/igt@[email protected]

* igt@kms_plane_alpha_blend@pipe-b-coverage-7efc:
- shard-skl: [PASS][23] -> [FAIL][24] ([fdo#108145] / [i915#265])
[23]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-skl3/igt@[email protected]
[24]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-skl5/igt@[email protected]

* igt@kms_psr@psr2_cursor_mmap_cpu:
- shard-iclb: [PASS][25] -> [SKIP][26] ([fdo#109441]) +2 similar issues
[25]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-iclb2/igt@kms_psr@psr2_cursor_mmap_cpu.html
[26]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-iclb4/igt@kms_psr@psr2_cursor_mmap_cpu.html

* igt@kms_setmode@basic:
- shard-apl: [PASS][27] -> [FAIL][28] ([i915#31])
[27]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-apl2/igt@[email protected]
[28]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-apl1/igt@[email protected]

* igt@perf_pmu@busy-accuracy-2-vcs1:
- shard-iclb: [PASS][29] -> [SKIP][30] ([fdo#112080]) +7 similar issues
[29]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-iclb2/igt@[email protected]
[30]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-iclb7/igt@[email protected]

#### Possible fixes ####

* igt@gem_exec_schedule@implicit-read-write-bsd:
- shard-iclb: [SKIP][31] ([i915#677]) -> [PASS][32]
[31]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-iclb1/igt@[email protected]
[32]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-iclb7/igt@[email protected]

* igt@gem_exec_schedule@implicit-read-write-bsd1:
- shard-iclb: [SKIP][33] ([fdo#109276] / [i915#677]) -> [PASS][34] +2 similar issues
[33]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-iclb3/igt@[email protected]
[34]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-iclb2/igt@[email protected]

* igt@gem_exec_schedule@preemptive-hang-bsd:
- shard-iclb: [SKIP][35] ([fdo#112146]) -> [PASS][36] +3 similar issues
[35]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-iclb4/igt@[email protected]
[36]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-iclb3/igt@[email protected]

* igt@i915_pm_rc6_residency@rc6-idle:
- shard-apl: [FAIL][37] -> [PASS][38]
[37]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-apl1/igt@[email protected]
[38]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-apl6/igt@[email protected]

* igt@i915_suspend@fence-restore-tiled2untiled:
- shard-kbl: [DMESG-WARN][39] ([i915#180]) -> [PASS][40]
[39]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-kbl3/igt@[email protected]
[40]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-kbl1/igt@[email protected]

* igt@kms_flip@flip-vs-expired-vblank:
- shard-glk: [FAIL][41] ([i915#79]) -> [PASS][42]
[41]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-glk9/igt@[email protected]
[42]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-glk3/igt@[email protected]

* igt@kms_flip@plain-flip-fb-recreate:
- shard-glk: [FAIL][43] -> [PASS][44]
[43]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-glk4/igt@[email protected]
[44]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-glk9/igt@[email protected]
- shard-skl: [FAIL][45] ([i915#34]) -> [PASS][46]
[45]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-skl7/igt@[email protected]
[46]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-skl7/igt@[email protected]

* igt@kms_flip_tiling@flip-changes-tiling:
- shard-kbl: [FAIL][47] ([i915#699] / [i915#93] / [i915#95]) -> [PASS][48]
[47]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-kbl2/igt@[email protected]
[48]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-kbl4/igt@[email protected]

* igt@kms_frontbuffer_tracking@fbc-suspend:
- shard-apl: [DMESG-WARN][49] ([i915#180] / [i915#95]) -> [PASS][50]
[49]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-apl4/igt@[email protected]
[50]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-apl4/igt@[email protected]

* igt@kms_plane@plane-panning-bottom-right-suspend-pipe-c-planes:
- shard-apl: [DMESG-WARN][51] ([i915#180]) -> [PASS][52] +1 similar issue
[51]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-apl1/igt@[email protected]
[52]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-apl6/igt@[email protected]

* igt@kms_psr@psr2_cursor_plane_move:
- shard-iclb: [SKIP][53] ([fdo#109441]) -> [PASS][54]
[53]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-iclb3/igt@kms_psr@psr2_cursor_plane_move.html
[54]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-iclb2/igt@kms_psr@psr2_cursor_plane_move.html

* igt@kms_setmode@basic:
- shard-skl: [FAIL][55] ([i915#31]) -> [PASS][56]
[55]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-skl2/igt@[email protected]
[56]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-skl3/igt@[email protected]

* igt@perf_pmu@busy-no-semaphores-vcs1:
- shard-iclb: [SKIP][57] ([fdo#112080]) -> [PASS][58] +6 similar issues
[57]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-iclb7/igt@[email protected]
[58]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-iclb1/igt@[email protected]

* igt@prime_vgem@fence-wait-bsd2:
- shard-iclb: [SKIP][59] ([fdo#109276]) -> [PASS][60] +16 similar issues
[59]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-iclb7/igt@[email protected]
[60]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-iclb1/igt@[email protected]

#### Warnings ####

* igt@kms_flip@flip-vs-suspend-interruptible:
- shard-kbl: [DMESG-WARN][61] ([i915#180]) -> [INCOMPLETE][62] ([i915#155] / [i915#600])
[61]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8230/shard-kbl7/igt@[email protected]
[62]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/shard-kbl7/igt@[email protected]

{name}: This element is suppressed. This means it is ignored when computing
the status of the difference (SUCCESS, WARNING, or FAILURE).

[IGT#5]: https://gitlab.freedesktop.org/drm/igt-gpu-tools/issues/5
[fdo#108145]: https://bugs.freedesktop.org/show_bug.cgi?id=108145
[fdo#109276]: https://bugs.freedesktop.org/show_bug.cgi?id=109276
[fdo#109441]: https://bugs.freedesktop.org/show_bug.cgi?id=109441
[fdo#112080]: https://bugs.freedesktop.org/show_bug.cgi?id=112080
[fdo#112146]: https://bugs.freedesktop.org/show_bug.cgi?id=112146
[i915#155]: https://gitlab.freedesktop.org/drm/intel/issues/155
[i915#180]: https://gitlab.freedesktop.org/drm/intel/issues/180
[i915#265]: https://gitlab.freedesktop.org/drm/intel/issues/265
[i915#300]: https://gitlab.freedesktop.org/drm/intel/issues/300
[i915#31]: https://gitlab.freedesktop.org/drm/intel/issues/31
[i915#34]: https://gitlab.freedesktop.org/drm/intel/issues/34
[i915#454]: https://gitlab.freedesktop.org/drm/intel/issues/454
[i915#54]: https://gitlab.freedesktop.org/drm/intel/issues/54
[i915#600]: https://gitlab.freedesktop.org/drm/intel/issues/600
[i915#677]: https://gitlab.freedesktop.org/drm/intel/issues/677
[i915#697]: https://gitlab.freedesktop.org/drm/intel/issues/697
[i915#699]: https://gitlab.freedesktop.org/drm/intel/issues/699
[i915#79]: https://gitlab.freedesktop.org/drm/intel/issues/79
[i915#93]: https://gitlab.freedesktop.org/drm/intel/issues/93
[i915#95]: https://gitlab.freedesktop.org/drm/intel/issues/95

Participating hosts (10 -> 10)
------------------------------

No changes in participating hosts

Build changes
-------------

* CI: CI-20190529 -> None
* Linux: CI_DRM_8230 -> Patchwork_17166

CI-20190529: 20190529
CI_DRM_8230: fa9f8453ffb88a4fc4e36d68b84a7ff9bf90f769 @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_5550: 98927dfde17aecaecfe67bb9853ceca326ca2b23 @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
Patchwork_17166: db60a6358595b602a7eaae13fb3c579af95f8da6 @ git://anongit.freedesktop.org/gfx-ci/linux
piglit_4509: fdc5a4ca11124ab8413c7988896eec4c97336694 @ git://anongit.freedesktop.org/piglit

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17166/index.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx