Add mechanism to proactively purge old resources in GrResourceCache.

Add mechanism to proactively purge old resources in GrResourceCache.

This change leaves the feature turned off by default.

Committed: https://skia.googlesource.com/skia/+/3f324321cdd8fde7976d958e2888a1ec4e657e35

[bsalomon, 2015-03-30 21:03:30]

bsalomon@google.com changed reviewers:
+ robertphillips@google.com

[robertphillips, 2015-04-06 19:34:51]

First pass. It looks good to me but I would like to take a second pass.

https://codereview.chromium.org/1032873002/diff/120001/src/gpu/GrResourceCach...
File src/gpu/GrResourceCache.cpp (right):

https://codereview.chromium.org/1032873002/diff/120001/src/gpu/GrResourceCach...
src/gpu/GrResourceCache.cpp:395: this->validate();
delete?

https://codereview.chromium.org/1032873002/diff/120001/src/gpu/GrResourceCach...
src/gpu/GrResourceCache.cpp:603: fLastFlushTimestampIndex =
(fLastFlushTimestampIndex + 1) & (fMaxUnusedFlushes - 1);
overlength

https://codereview.chromium.org/1032873002/diff/120001/src/gpu/GrResourceCache.h
File src/gpu/GrResourceCache.h (right):

https://codereview.chromium.org/1032873002/diff/120001/src/gpu/GrResourceCach...
src/gpu/GrResourceCache.h:278: // We keep track of the "timestamps" of the last
n flushes. If a resource hasn't been used in
Delete the parenthetical question? Isn't fMaxUnusedFlushes doing that?

https://codereview.chromium.org/1032873002/diff/120001/src/gpu/GrResourceCach...
src/gpu/GrResourceCache.h:317: /**
detects -> detect ?

[bsalomon, 2015-04-07 20:32:38]

Patchset #8 (id:140001) has been deleted

[bsalomon, 2015-04-07 20:33:46]

https://codereview.chromium.org/1032873002/diff/120001/src/gpu/GrResourceCach...
File src/gpu/GrResourceCache.cpp (right):

https://codereview.chromium.org/1032873002/diff/120001/src/gpu/GrResourceCach...
src/gpu/GrResourceCache.cpp:395: this->validate();
On 2015/04/06 19:34:51, robertphillips wrote:
> delete?

Done.

https://codereview.chromium.org/1032873002/diff/120001/src/gpu/GrResourceCach...
src/gpu/GrResourceCache.cpp:603: fLastFlushTimestampIndex =
(fLastFlushTimestampIndex + 1) & (fMaxUnusedFlushes - 1);
On 2015/04/06 19:34:51, robertphillips wrote:
> overlength

Done.

https://codereview.chromium.org/1032873002/diff/120001/src/gpu/GrResourceCache.h
File src/gpu/GrResourceCache.h (right):

https://codereview.chromium.org/1032873002/diff/120001/src/gpu/GrResourceCach...
src/gpu/GrResourceCache.h:278: // We keep track of the "timestamps" of the last
n flushes. If a resource hasn't been used in
On 2015/04/06 19:34:51, robertphillips wrote:
> Delete the parenthetical question? Isn't fMaxUnusedFlushes doing that?

Done.

https://codereview.chromium.org/1032873002/diff/120001/src/gpu/GrResourceCach...
src/gpu/GrResourceCache.h:317: /**
On 2015/04/06 19:34:51, robertphillips wrote:
> detects -> detect ?

Done.

[robertphillips, 2015-04-08 14:50:26]

lgtm. It seems like we should have some documentation somewhere that summarizes
the entire system:

When proactive purging is enabled, on every flush, the timestamp of that flush
is stored in a n-sized ring buffer. When purging occurs each purgeable
resource's timestamp is compared to the timestamp of the n-th prior purge. If
the resource's timestamp is older than the old flush then the resource is
proactively purged. 'n' in this case is determined by the "maxUnusedFlushes"
parameter to GrResourceCache::setLimits. The ring buffer is reset when the
timestamp wraps around.

[bsalomon, 2015-04-08 17:54:57]

On 2015/04/08 14:50:26, robertphillips wrote:
> lgtm. It seems like we should have some documentation somewhere that
summarizes
> the entire system:
> 
> When proactive purging is enabled, on every flush, the timestamp of that flush
> is stored in a n-sized ring buffer. When purging occurs each purgeable
> resource's timestamp is compared to the timestamp of the n-th prior purge. If
> the resource's timestamp is older than the old flush then the resource is
> proactively purged. 'n' in this case is determined by the "maxUnusedFlushes"
> parameter to GrResourceCache::setLimits. The ring buffer is reset when the
> timestamp wraps around.

Added a comment about GrResourceCache based on this text.

[bsalomon, 2015-04-08 17:54:59]

The CQ bit was checked by bsalomon@google.com

[bsalomon, 2015-04-08 17:55:00]

The patchset sent to the CQ was uploaded after l-g-t-m from
robertphillips@google.com
Link to the patchset: https://codereview.chromium.org/1032873002/#ps180001
(title: "Add comment")

[commit-bot: I haz the power, 2015-04-08 17:55:26]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1032873002/180001

[commit-bot: I haz the power, 2015-04-08 18:02:04]

Committed patchset #9 (id:180001) as
https://skia.googlesource.com/skia/+/3f324321cdd8fde7976d958e2888a1ec4e657e35