Mobile/4.0b5 crashes

Don't bother looking at crashes newer than March 4, and even ones from then can be hit or miss - some have been reprocessed, most haven't. The numbers on the main crash listing page mean nothing and are frequently incorrect, unfortunately.

tl;dr: The generic OOM (libc.so, 0xdeadbaad), nsJNIString leak (libdvm.so, 0xdeadd00d), and CloseSuperfluousFds (libc.so, non-0xdeadbaad) triumvirate take the cake for probably >90% of all crashes. The JNIString leak is fixed, and the CloseSuperfluousFds crash only happens in the content process, so we're in decent shape.

Now, drilling down:

Most named stacks are invalid recent ones or references to old reprocessed ones - the vast majority of crashes are hidden behind libc.so, libdvm.so and libxul.so. Continue to see libz.so stacks, not necessarily happening on cyanogenmod devices. OOM stacks are still pretty screwed up and don't give much useful info (libc.so, crash address 0xdeadbaad) Child process crashes don't display hardware type - this is a known bug.

OOM stacks - why are they so wacky? Either they have a single common nonsense frame showing after mozalloc_abort (nsXULTemplateQueryProcessorRDF::ComputeContainmentProperties), or they have a bunch of unrelated stacks interleaved,like https://crash-stats.mozilla.com/report/index/697fd5b5-fca4-4bb3-a748-72fd12110227

Other common crashes:

gfxFont::AddRef

nsThebesFontMetrics::GetMetrics

nsXPConnect::GetPrincipal and friends, but that's fixed in recent builds (yay!)

js::Interpret coming out of nsFrameMessageManager

Large numbers of libc-2.5.so crashes with non-useful stack, quite mysterious

Interesting crashes encountered:

A couple js_EmitTree crashes at startup

Strange PR_AtomicDecrement stacks, ipc-related: https://crash-stats.mozilla.com/report/list?range_value=2&range_unit=weeks&date=2011-03-05%2005%3A00%3A00&signature=PR_AtomicDecrement%20%7C%20PR_DestroyCondVar&version=Fennec%3A4.0b5

Seems like any stack that begins with "@0x0 |" contains Java_org_mozilla_gecko_GeckoAppShell_callObserver

The regular nsTHashtable<mozilla::FrameLayerBuilder::ThebesLayerItemsEntry>::s_ClearEntry> crash is still around: https://crash-stats.mozilla.com/report/list?range_value=2&range_unit=weeks&date=2011-03-05%2005%3A00%3A00&signature=nsTArray_base%3CnsTArrayDefaultAllocator%3E%3A%3AShiftData%20%7C%20nsTHashtable%3Cmozilla%3A%3AFrameLayerBuilder%3A%3AThebesLayerItemsEntry%3E%3A%3As_ClearEntry&version=Fennec%3A4.0b5

libGLES* crashes: https://crash-stats.mozilla.com/report/list?range_value=2&range_unit=weeks&date=2011-03-05%2005%3A00%3A00&signature=libGLESv2_adreno200.so%400x7afd6&version=Fennec%3A4.0b5 https://crash-stats.mozilla.com/report/index/4bff1f97-1180-4247-acee-501002110301