Seen while trying to load https://www.youtube.com/watch?v=V1IBH1bz5-c (while trying to reproduce bug 870654). STR: 0. BRANCH=v1.0.1 ./config.sh hamachi && ./build.sh && ./flash.sh 1. complete FTU, connecting to available WiFi 2. open the Browser app 3. go to https://www.youtube.com/watch?v=V1IBH1bz5-c Expected results: page should open Observed results: "Well, this is embarassing. We tried to display this Web page, but it's not responding." logcat shows: F( 620:0x26c) Assertion failure: NS_SUCCEEDED(rv), at /home/mikeh/dev/mozilla/m-c/b2g18/dom/ipc/TabChild.cpp:505 F( 620:0x26c) Fatal signal 11 (SIGSEGV) at 0x00000000 (code=1) Relevant lines in the code: 502 // by AsyncPanZoomController and causes a blurry flash. 503 bool isFirstPaint; 504 nsresult rv = utils->GetIsFirstPaint(&isFirstPaint); 505 MOZ_ASSERT(NS_SUCCEEDED(rv)); http://hg.mozilla.org/releases/mozilla-b2g18/file/b87559cef261/dom/ipc/TabChild.cpp#l505 Observed with: - gecko: b2g18:b87559cef261 - gaia: 7d0ca6a07fc3d4817f80f4ebc24314b6cbcf049a
Note: this is with a debug build, so step 0 should include |B2G_DEBUG=1 ./build.sh|.
ajones, this assertion was introduced in changeset 117519; any idea what's going on here? 1. http://hg.mozilla.org/releases/mozilla-b2g18/diff/3dd36851bd7c/dom/ipc/TabChild.cpp
If I comment out the above MOZ_ASSERT(), the URL https://www.youtube.com/watch?v=V1IBH1bz5-c loads correctly.
I'd expect to see a regression in bug 799401. Just need to figure out if that could possibly happen in this context. Is GetPresShell() returning nullptr?
Needsinfo on Mike to see if he can with comment #4 ?
Triage - Mike, can you describe the user impact here and if there is any occurrence rate? Unable to make triage decision base on the bug as is.
(In reply to Wayne Chang [:wchang] from comment #6) > > Triage - Mike, can you describe the user impact here and if there is any > occurrence rate? Unable to make triage decision base on the bug as is. It happens only on DEBUG builds, so no one is going to see it in the field; but presumably the assertion exists for a reason, and it's failure is an error, so we should fix it. It also makes debugging impossible unless you manually comment out that line of code, like I have done.
It should be fixed, but we can't hold a release for it, so it will have to land to v1-train and bubble up from there to v1.next releases.