From bug 1332523 comment 0: I have some patches that can land independently of more definite changes on the bootstrap API. I'm building now without this patch to see if it breaks TB.
It breaks without patch. With it it builds and TB works. Jörg, your diff magic is in demand again. :-) I'm away soon, please can you land it, when it's okay?
Assignee: nobody → richard.marti
Status: NEW → ASSIGNED
Attachment #8829200 - Flags: review?(jorgk)
Important: the now checked-in bug 1332827 needs to be applied first.
Comment on attachment 8829200 [details] [diff] [review] bug1332898.patch (In reply to Richard Marti (:Paenglab) from comment #1) > Jörg, your diff magic is in demand again. :-) I know the five differenced between M-C's nsBrowserApp.cpp and C-C's nsMailApp.cpp by heart (German: auswendig) now! > I'm away soon, please can you land it, when it's okay? Landing, now, stand by ;-) Of course, many thanks, like always!
Attachment #8829200 - Flags: review?(jorgk) → review+
https://hg.mozilla.org/comm-central/rev/137c08b2e6e4801d2e884a3e6c8473becd267bc8 This needs porting to IB and SM, like bug 1332827 and bug 1332017.
Status: ASSIGNED → RESOLVED
Closed: 4 years ago
Resolution: --- → FIXED
Target Milestone: --- → Thunderbird 53.0
Thanks. Just found out 15 minutes ago and on it for SeaMonkey. Ewongs Bug 1332173 is still not reviewed and we can probably merge it with it.
(In reply to Richard Marti (:Paenglab) from comment #1) > It breaks without patch. With it it builds and TB works. Hmm, I recompiled and TB doesn't even start :-( I'll try a clobber build.
Hmm, recompiled and opening the application still crashes. I can run xpcshell. I have a 64bit debug build. Stack: mozglue.dll!arena_dalloc(void * ptr, unsigned __int64 offset) Line 4709 C thunderbird.exe!mozilla::GetBootstrap(const char * aXPCOMFile) Line 417 C++ thunderbird.exe!InitXPCOMGlue(const char * argv0) Line 248 C++ thunderbird.exe!NS_internal_main(int argc, char * * argv, char * * envp) Line 297 C++ thunderbird.exe!wmain(int argc, wchar_t * * argv) Line 118 C++ FRG, does you debug build start?
On the console I see: jemalloc.c:4710: Failed assertion: "arena != NULL" Hit MOZ_CRASH() at jemalloc_config.cpp:163
Jorg, I have a non debug build right now only which works fine. I need to clean up the patch a little further and can check later.
Today Daily works and the tests don't look different to yesterday.
I know. I've been looking at this all day, now doing my fourth full build. Basically it just crashes in some memory deallocation called during the program's initialisation in main(). I have all the details. I still had --disable-sandboxing in my mozconfig which isn't used elsewhere. So maybe that's the cause. One hour to go in the compile.
> This needs porting to IB and SM, like bug 1332827 and bug 1332017. ACK.
You need to log in before you can comment on or make changes to this bug.