|
|
DescriptionSpeculative fix for crash in blink::WebFrameWidgetBase::dragTargetDragEnterOrOver.
BUG=671504
Review-Url: https://codereview.chromium.org/2695043002
Cr-Commit-Position: refs/heads/master@{#451149}
Committed: https://chromium.googlesource.com/chromium/src/+/1b7963b7210d4d1a45049841cf83eb30274ba82d
Patch Set 1 #Patch Set 2 : Added TODO. #Messages
Total messages: 14 (5 generated)
Description was changed from ========== # Enter a description of the change. Speculative fix for crash in blink::WebFrameWidgetBase::dragTargetDragEnterOrOver. BUG=671504 ========== to ========== Speculative fix for crash in blink::WebFrameWidgetBase::dragTargetDragEnterOrOver. BUG=671504 ==========
paulmeyer@chromium.org changed reviewers: + pdr@chromium.org
On 2017/02/14 at 15:53:17, paulmeyer wrote: > I think this is the bug, but we might be able to reason about this crash and make a real fix. It looks like we're somehow getting into the dragTargetDragOver state without first going through dragTargetDragEnter. Should that be possible? If so, can you refactor the code to make that more clear?
On 2017/02/15 18:11:35, pdr. wrote: > On 2017/02/14 at 15:53:17, paulmeyer wrote: > > > > I think this is the bug, but we might be able to reason about this crash and > make a real fix. It looks like we're somehow getting into the dragTargetDragOver > state without first going through dragTargetDragEnter. Should that be possible? > If so, can you refactor the code to make that more clear? It is not meant to be possible for drag over to occur before drag enter, and I'm not sure yet how this is happening. I haven't actually been able to repro this crash locally yet. What I was hoping to do is land this patch which will at least prevent the crash and unblock stable. I can then continue looking into figuring out exactly how this is happening and try to fix the underlying cause.
On 2017/02/16 19:30:30, paulmeyer_ wrote: > On 2017/02/15 18:11:35, pdr. wrote: > > On 2017/02/14 at 15:53:17, paulmeyer wrote: > > > > > > > I think this is the bug, but we might be able to reason about this crash and > > make a real fix. It looks like we're somehow getting into the > dragTargetDragOver > > state without first going through dragTargetDragEnter. Should that be > possible? > > If so, can you refactor the code to make that more clear? > > It is not meant to be possible for drag over to occur before drag enter, and I'm > not sure yet how this is happening. I haven't actually been able to repro this > crash locally yet. What I was hoping to do is land this patch which will at > least prevent the crash and unblock stable. I can then continue looking into > figuring out exactly how this is happening and try to fix the underlying cause. Please add a TODO to figure out the real fix. Drag and drop is already pretty messy, so it's important to have a tracking followup to understand why the invariants the embedder (Aura/content) should be providing aren't holding.
On 2017/02/16 19:47:20, dcheng wrote: > On 2017/02/16 19:30:30, paulmeyer_ wrote: > > On 2017/02/15 18:11:35, pdr. wrote: > > > On 2017/02/14 at 15:53:17, paulmeyer wrote: > > > > > > > > > > I think this is the bug, but we might be able to reason about this crash and > > > make a real fix. It looks like we're somehow getting into the > > dragTargetDragOver > > > state without first going through dragTargetDragEnter. Should that be > > possible? > > > If so, can you refactor the code to make that more clear? > > > > It is not meant to be possible for drag over to occur before drag enter, and > I'm > > not sure yet how this is happening. I haven't actually been able to repro this > > crash locally yet. What I was hoping to do is land this patch which will at > > least prevent the crash and unblock stable. I can then continue looking into > > figuring out exactly how this is happening and try to fix the underlying > cause. > > Please add a TODO to figure out the real fix. Drag and drop is already pretty > messy, so it's important to have a tracking followup to understand why the > invariants the embedder (Aura/content) should be providing aren't holding. Added.
On 2017/02/16 at 20:36:15, paulmeyer wrote: > On 2017/02/16 19:47:20, dcheng wrote: > > On 2017/02/16 19:30:30, paulmeyer_ wrote: > > > On 2017/02/15 18:11:35, pdr. wrote: > > > > On 2017/02/14 at 15:53:17, paulmeyer wrote: > > > > > > > > > > > > > I think this is the bug, but we might be able to reason about this crash and > > > > make a real fix. It looks like we're somehow getting into the > > > dragTargetDragOver > > > > state without first going through dragTargetDragEnter. Should that be > > > possible? > > > > If so, can you refactor the code to make that more clear? > > > > > > It is not meant to be possible for drag over to occur before drag enter, and > > I'm > > > not sure yet how this is happening. I haven't actually been able to repro this > > > crash locally yet. What I was hoping to do is land this patch which will at > > > least prevent the crash and unblock stable. I can then continue looking into > > > figuring out exactly how this is happening and try to fix the underlying > > cause. > > > > Please add a TODO to figure out the real fix. Drag and drop is already pretty > > messy, so it's important to have a tracking followup to understand why the > > invariants the embedder (Aura/content) should be providing aren't holding. > > Added. Thanks, LGTM
The CQ bit was checked by paulmeyer@chromium.org
CQ is trying da patch. Follow status at https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...
CQ is committing da patch. Bot data: {"patchset_id": 20001, "attempt_start_ts": 1487277795999530, "parent_rev": "0a8d26e24fbf3e0a262f2e0fe72100c979a77428", "commit_rev": "1b7963b7210d4d1a45049841cf83eb30274ba82d"}
Message was sent while issue was closed.
Description was changed from ========== Speculative fix for crash in blink::WebFrameWidgetBase::dragTargetDragEnterOrOver. BUG=671504 ========== to ========== Speculative fix for crash in blink::WebFrameWidgetBase::dragTargetDragEnterOrOver. BUG=671504 Review-Url: https://codereview.chromium.org/2695043002 Cr-Commit-Position: refs/heads/master@{#451149} Committed: https://chromium.googlesource.com/chromium/src/+/1b7963b7210d4d1a45049841cf83... ==========
Message was sent while issue was closed.
Committed patchset #2 (id:20001) as https://chromium.googlesource.com/chromium/src/+/1b7963b7210d4d1a45049841cf83... |