Closed
Bug 1380214
Opened 8 years ago
Closed 8 years ago
Crash in IPCError-browser | PBrowserParent::RecvPDocAccessibleConstructor Constructing a top-level PDocAccessible with null COM
Categories
(Core :: Disability Access APIs, defect)
Tracking
()
RESOLVED
FIXED
mozilla56
Tracking | Status | |
---|---|---|
firefox-esr52 | --- | unaffected |
firefox54 | --- | unaffected |
firefox55 | --- | unaffected |
firefox56 | --- | fixed |
People
(Reporter: ting, Assigned: bugzilla)
References
Details
(Keywords: crash, Whiteboard: aes+)
Crash Data
Attachments
(1 file, 2 obsolete files)
1.60 KB,
patch
|
eeejay
:
review+
|
Details | Diff | Splinter Review |
This bug was filed from the Socorro interface and is
report bp-fcb2988f-32ca-4188-ac71-362590170710.
=============================================================
Top #17 of Nightly 20170709030204 on Windows.
Reporter | ||
Updated•8 years ago
|
![]() |
||
Updated•8 years ago
|
Whiteboard: aes+
Assignee | ||
Comment 1•8 years ago
|
||
One cause of this is that if we fail to serialize the handler payload, we fail marshaling entirely.
This is entirely possible if for example there is a sandboxing violation during serialization.
Since failing to serialize the handler payload is non-fatal, I think we can modify our code so that we successfully marshal with an empty payload.
Assignee | ||
Updated•8 years ago
|
Assignee: nobody → aklotz
Status: NEW → ASSIGNED
Assignee | ||
Comment 2•8 years ago
|
||
This patch adds a check that ends up calling
https://dxr.mozilla.org/mozilla-central/rev/aff336ac161daa3ea350e59a288963edbd58ed39/ipc/mscom/StructStream.h#92
to check for a successful error code. It is non-fatal for payload serialization to fail, so we just resort to writing an empty struct in that case.
This allows us to successfully marshal an accessible even if the payload fails to serialize for some reason, thus preventing some (hopefully most or all) instances of the crash in this bug.
Attachment #8887256 -
Flags: review?(eitan)
Assignee | ||
Comment 3•8 years ago
|
||
I realized that we should also be checking this when computing the payload size.
Attachment #8887256 -
Attachment is obsolete: true
Attachment #8887256 -
Flags: review?(eitan)
Attachment #8887281 -
Flags: review?(eitan)
Assignee | ||
Comment 4•8 years ago
|
||
For real this time.
Attachment #8887281 -
Attachment is obsolete: true
Attachment #8887281 -
Flags: review?(eitan)
Attachment #8887289 -
Flags: review?(eitan)
Updated•8 years ago
|
Attachment #8887289 -
Flags: review?(eitan) → review+
Assignee | ||
Comment 5•8 years ago
|
||
https://hg.mozilla.org/integration/mozilla-inbound/rev/9fa7419d9080260365669174e4ed13c65280492d
Bug 1380214: Ensure that if the handler payload fails to serialize that we may still marshal the accessible itself; r=eeejay
Comment 6•8 years ago
|
||
bugherder |
Status: ASSIGNED → RESOLVED
Closed: 8 years ago
status-firefox56:
--- → fixed
Resolution: --- → FIXED
Target Milestone: --- → mozilla56
Updated•8 years ago
|
status-firefox54:
--- → unaffected
status-firefox55:
--- → unaffected
status-firefox-esr52:
--- → unaffected
Comment 7•8 years ago
|
||
This signature is still showing up in the Windows nightly of 20170720030203,
as topcrash #3.
Comment 8•8 years ago
|
||
Aaron, do we need to file a new bug for this stack signature (comment 7)?
Flags: needinfo?(aklotz)
You need to log in
before you can comment on or make changes to this bug.
Description
•