8976 - apprunner crash on sidebar startup

Status: VERIFIED FIXED

(Core Graveyard :: RDF, defect, P3)

Description

[tor]

On a clobber Solaris/native build of a morning 6/28/99 CVS pull, apprunner
crashes on startup.  This happens when apprunner is trying to draw the UI
for the first time.  When it gets around to drawing the sidebar it crashes.
This started happening about the time bug 8849 showed up.  Unfortunately
the removal of xpinstall from the build didn't fix the problem on solaris.

The stack trace at the time of the crash is as follows.  The mURLSpec that
it is attempting to free is "0x3".

current thread: t@1
  [1] _free_unlocked(0x3, 0x0, 0xfe031f98, 0x0, 0x0, 0x0), at 0xfdfc60d8
  [2] free(0x3, 0x7ca40, 0x0, 0x0, 0x0, 0x0), at 0xfdfc6088
  [3] PL_strfree(0x3, 0xfd38b57c, 0x9, 0x0, 0x0, 0xfe031f98), at 0xfe781d90
=>[4] RDFXMLDataSourceImpl::Init(this = 0x255614, uri = 0x560c80
"resource:/res/rdf/sidebar-browser.rdf"), line 635 in "nsRDFXMLDataSource.cpp"
  [5] 0xff26b984(0x255614, 0x3, 0x1, 0xffbeddc0, 0x0, 0x0), at 0xff26b983
  [6] nsXPCWrappedNativeClass::CallWrappedMethod(this = 0x54f738, cx = 0x3b3e78,
wrapper = 0x4655d8, desc = 0x46556c, callMode = CALL_METHOD, argc = 1U, argv =
0x4a2e68, vp = 0xffbee050), line 510 in "xpcwrappednativeclass.cpp"
  [7] WrappedNative_CallMethod(cx = 0x3b3e78, obj = 0x49c3a8, argc = 1U, argv =
0x4a2e68, vp = 0xffbee050), line 125 in "xpcwrappednativejsops.cpp"
  [8] js_Invoke(cx = 0x3b3e78, argc = 1U, constructing = 0), line 654 in
"jsinterp.c"
  [9] js_Interpret(cx = 0x3b3e78, result = 0xffbee474), line 2215 in
"jsinterp.c"
  [10] js_Invoke(cx = 0x3b3e78, argc = 2U, constructing = 0), line 670 in
"jsinterp.c"
  [11] js_Interpret(cx = 0x3b3e78, result = 0xffbee8c4), line 2215 in
"jsinterp.c"
  [12] js_Invoke(cx = 0x3b3e78, argc = 0, constructing = 0), line 670 in
"jsinterp.c"
  [13] js_Interpret(cx = 0x3b3e78, result = 0xffbeeeec), line 2215 in
"jsinterp.c"
  [14] js_Execute(cx = 0x3b3e78, chain = 0x2a05b8, script = 0x28b278, fun =
(nil), down = (nil), debugging = 0, result = 0xffbeeeec), line 829 in
"jsinterp.c"
  [15] JS_EvaluateUCScriptForPrincipals(cx = 0x3b3e78, obj = 0x2a05b8,
principals = (nil), chars = 0x2b2660, length = 6U, filename = (nil), lineno = 0,
rval = 0xffbeeeec), line 2584 in "jsapi.c"
  [16] JS_EvaluateUCScript(cx = 0x3b3e78, obj = 0x2a05b8, chars = 0x2b2660,
length = 6U, filename = (nil), lineno = 0, rval = 0xffbeeeec), line 2566 in
"jsapi.c"
  [17] JS_EvaluateScript(cx = 0x3b3e78, obj = 0x2a05b8, bytes = 0x55a660
"Boot()", length = 6U, filename = (nil), lineno = 0, rval = 0xffbeeeec), line
2533 in "jsapi.c"
  [18] GlobalWindowImpl::RunTimeout(this = 0x3b3df8, aTimeout = 0x41a988), line
1516 in "nsGlobalWindow.cpp"
  [19] nsGlobalWindow_RunTimeout(aTimer = 0x4e4c38, aClosure = 0x41a988), line
1433 in "nsGlobalWindow.cpp"
  [20] TimerImpl::FireTimeout(this = 0x4e4c38), line 73 in "nsTimer.cpp"
  [21] nsTimerExpired(aCallData = 0x4e4c38), line 196 in "nsTimer.cpp"
  [22] g_timeout_dispatch(0x286c40, 0xffbef1c8, 0x4e4c38, 0xfe3f1930,
0xfe3f1930, 0xffbef158), at 0xfe3cbaa0
  [23] g_main_dispatch(0xffbef1c8, 0xfe3f3d70, 0x0, 0xfe3f3cb0, 0xfe3f1930,
0x4), at 0xfe3ca7b4
  [24] g_main_iterate(0xffffffff, 0x1, 0xfe3cba4c, 0xfe3f3d78, 0xfe3f3ce0,
0xfe3f3d6c), at 0xfe3caee4
  [25] g_main_run(0x117b88, 0x117b88, 0xfe3f1930, 0xfe3f3d70, 0x91458,
0xfe43faa4), at 0xfe3cb0d8
  [26] gtk_main(0xfe4740a8, 0xfe67d5e0, 0xfe665b1c, 0x117b88, 0x0, 0xfe47105c),
at 0xfe57baa8
  [27] nsAppShell::Run(this = 0x8d140), line 237 in "nsAppShell.cpp"
  [28] nsAppShellService::Run(this = 0x978f0), line 428 in
"nsAppShellService.cpp"
  [29] main(argc = 1, argv = 0xffbef684), line 675 in "nsAppRunner.cpp"

Comment 1

[Chris Waterson]

Suspicious this might be a dup of (i.e., will be fixed by) 8016.

Comment 2

[Chris Waterson]

Per this post:

news://news.mozilla.org/7lv64c%241jt1%40secnews.netscape.com

Added jband and duncan to cc list.

Comment 3

[Chris Waterson]

Okay, after reading more of the followups to the above thread, I am convinced
that this _has_ to be incorrect MI code for Solaris. Reassigning to rogerl. If
I'm wrong, give it back.

Comment 4

[rogerl (gone)]

So I have the MI bug fixed locally, but apprunner won't work for me - I can't
even tell where it fails since gdb seg. faults, too. I can check theMI fix in if
somebody wants to see if that fixes this problem???

Comment 5

[rogerl (gone)]

Moving out to M10 per choffman's request

Comment 6

[rogerl (gone)]

Fixed (see 8016)

Comment 7

[gerardok]

QA Contact massive update.