Correction, in the prototype [3], I just use normal blur filter.

Just to make sure I'm understanding, is the issue that MapRectInternal (https://cs.chromium.org/chromium/src/cc/base/filter_operation.cc?rcl=0238418b07cbe9ea31f40cb3804e7ebea141ab43&l=325) maps empty rects into non-empty 
rects for blur filters, and so the surface a blur filter draws into is considered damaged every frame even when nothing changes? That would certainly be a bug.

Would you be interested in fixing MapRectInternal?

I think we need to separate the concepts of "what is damaged/what should i swap" from "what should i draw". Today they are one rect, but we want to draw more than what changed for background filters, in order for them to read, and if they are in the cache then we don't need to draw.

#4 sounds good. The change could also resolve the partial swap for blur in this  bug 737255 . I am adding a second rect to track the real damaged area. But I could make it more general to fit here.

#3, right, an empty rect becomes non-empty rects due to result.Inset(-spread_x, -spread_y, -spread_x, -spread_y).

We might have multiple places to check if this rect is empty. One place could be here [1], before we expand, we check is_rect_valid. But is_rect_valid could be empty. I wonder whether we can define an empty rect is not valid.

[1] https://cs.chromium.org/chromium/src/cc/trees/damage_tracker.cc?rcl=88a57b64c852326ba2a5d84c395882c01692ea32&l=461

From https://bugs.chromium.org/p/chromium/issues/detail?id=738185#c29 it sounds like this bug actually isn't what I thought, as there is no background filter?

With a normal filter, whatever damage there is needs to expand, because the pixels move and changing a green box to red would change pixels outside of that box too.

So that made me confused why the damage expansion is a problem. Is it going from empty to non-empty?

> #3, right, an empty rect becomes non-empty rects due to result.Inset(-spread_x, -spread_y, -spread_x, -spread_y).

OH I should read this bug's comment. It's already here, thanks for explaining wutao@. We should def not expand an empty rect.

It seems to me that FilterOperations::MapRect should not expand an empty rect.

I will upload a cl for this soon.

Is there any exception that when the input rect is empty, we still want to expand/shrink the rect?

cl can be found at here:

https://chromium-review.googlesource.com/c/587090/

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/92b48551c98e87d75510ee552b1ef96de7641893

commit 92b48551c98e87d75510ee552b1ef96de7641893
Author: wutao <wutao@chromium.org>
Date: Sat Jul 29 02:04:17 2017

Do not map empty rect in FilterOperation.

If the input rect is empty, we do not want to map the rect to a
non-empty rect. Because it could cause problem, for example, in
damage_tracker to have a damaged region due to having pixel-moving
filters.

DamageTrackerTest.VerifyDamageForBackgroundBlurredChild

Bug:  745101 
Test: DamageTrackerTest.VerifyDamageForImageFilter and
Cq-Include-Trybots: master.tryserver.blink:linux_trusty_blink_rel
Change-Id: I8719f5ad2e6da08b3a95df45c5194532c3bc5bd4
Reviewed-on: https://chromium-review.googlesource.com/587090
Commit-Queue: Tao Wu <wutao@chromium.org>
Reviewed-by: Ali Juma <ajuma@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
Cr-Commit-Position: refs/heads/master@{#490580}
[modify] https://crrev.com/92b48551c98e87d75510ee552b1ef96de7641893/cc/trees/damage_tracker.cc
[modify] https://crrev.com/92b48551c98e87d75510ee552b1ef96de7641893/cc/trees/damage_tracker_unittest.cc

This code need to be merged back to M61 to make Render surface cache work correctly.
[1]  issue 708513 
Render surface cache is required to make the signin/lock screen blur work efficiently.
[2]  issue 738185 

Your change meets the bar and is auto-approved for M61. Please go ahead and merge the CL to branch 3163 manually. Please contact milestone owner if you have questions.
Owners: amineer@(Android), cmasso@(iOS), ketakid @(ChromeOS), govind@(Desktop)

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/e4d934b8951f910c65a4276d4d30478f587fbe9b

commit e4d934b8951f910c65a4276d4d30478f587fbe9b
Author: Qiang Xu <warx@chromium.org>
Date: Mon Jul 31 18:35:48 2017

[merge to m61] Do not map empty rect in FilterOperation.

merge to m61 on behalf of wutao@

If the input rect is empty, we do not want to map the rect to a
non-empty rect. Because it could cause problem, for example, in
damage_tracker to have a damaged region due to having pixel-moving
filters.

DamageTrackerTest.VerifyDamageForBackgroundBlurredChild

(cherry picked from commit 92b48551c98e87d75510ee552b1ef96de7641893)

TBR: ajuma@chromium.org, senorblanco@chromium.org
Bug:  745101 
Test: DamageTrackerTest.VerifyDamageForImageFilter and
Cq-Include-Trybots: master.tryserver.blink:linux_trusty_blink_rel
Change-Id: I8719f5ad2e6da08b3a95df45c5194532c3bc5bd4
Reviewed-on: https://chromium-review.googlesource.com/587090
Commit-Queue: Tao Wu <wutao@chromium.org>
Reviewed-by: Ali Juma <ajuma@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#490580}
Reviewed-on: https://chromium-review.googlesource.com/594669
Reviewed-by: Qiang(Joe) Xu <warx@chromium.org>
Cr-Commit-Position: refs/branch-heads/3163@{#170}
Cr-Branched-From: ff259bab28b35d242e10186cd63af7ed404fae0d-refs/heads/master@{#488528}
[modify] https://crrev.com/e4d934b8951f910c65a4276d4d30478f587fbe9b/cc/trees/damage_tracker.cc
[modify] https://crrev.com/e4d934b8951f910c65a4276d4d30478f587fbe9b/cc/trees/damage_tracker_unittest.cc