Got logs or bugreport?

What is the best way of fetching the necessary logs, without adding private data? 

Actually screenshots is probably enough here.

G930F, so it's the internaltional version of S7. Crash is clearly in v8, which we know we have a known bug in. I'll just dup this into that one, although it's private.

> What is the best way of fetching the necessary logs, without adding private data? 

Manually audit it I guess. For crashes, usually including the breakpad microdump, or the java exception stacks are enough.

I see the same kind of crash in Google Search as well. Attached screenshot of stack trace. 

Deleted: Screenshot_20160318-120357.png 298 KB

	Screenshot_20160318-120357.png 298 KB View Download

Can you attach the whole crash stack from the logs. That screenshot didn't include everything. Also if you see "BREAKPAD MICRODUMP" immediately before, include everything from that as well.

Same with my app called AquaMail.

It's a mail app, does use a WebView, and does not try to run in 32-bit mode, so it gets 64-bit.

Apparently no issues on the Nexus 6P, Nexus 5x, or Samsung S6.

On the Galaxy S7, I already got two crashes since buying the phone last night.

Trying to "adb pull" the tombstone I get "access denied", so here is the info from logcat.

All I was able to find. Attached.

Deleted: Galaxy_s7_WebView_crash.txt 54.2 KB

Galaxy_s7_WebView_crash.txt 54.2 KB View Download

Didn't manage to get another crash in Google Search while I had the phone hooked up with ADB. But in stead it crashed in S Calendar (Samsung app) when opening a link. I have attached the entire ADB log of the crash.

I will post a log of Google Search crash if I get one again.

Deleted: s_calendar.txt 41.8 KB

s_calendar.txt 41.8 KB View Download

Crash in #8 looks V8 related:

Operating system: Android
                  samsung/hero2ltexx/hero2lte:6.0.1/MMB29K/G935FXXU1APAW:user/release-keys
CPU: aarch64
     8 CPUs

Crash reason:
Crash address: 0x0
Process uptime: not available

Thread 0 (crashed)
 0  0x439da9017b9d

    Found by: given as instruction pointer in context
 1  0x7eff93f110

    Found by: previous frame's frame pointer
 2  0x7eff939a24

    Found by: previous frame's frame pointer
 3  0x7eff92118c

    Found by: previous frame's frame pointer
 4  libwebviewchromium.so!v8::internal::::Invoke [execution.cc : 98 + 0x14]

    Found by: stack scanning
 5  libwebviewchromium.so!v8::internal::Execution::Call [execution.cc : 164 + 0x20]

    Found by: stack scanning
 6  libwebviewchromium.so!v8::internal::Genesis::CallUtilsFunction [bootstrapper.cc : 1959 + 0x0]

    Found by: stack scanning
 7  libwebviewchromium.so!v8::internal::Genesis::InstallExperimentalNatives [bootstrapper.cc : 3031 + 0xc]

    Found by: stack scanning
 8  libwebviewchromium.so!v8::internal::NativesCollection<(v8::internal::NativeType)1u>::GetBuiltinsCount [natives-external.cc : 198 + 0x0]

    Found by: stack scanning
 9  libwebviewchromium.so!v8::internal::Genesis::Genesis [bootstrapper.cc : 3627 + 0x4]

    Found by: stack scanning
10  libwebviewchromium.so!v8::internal::Bootstrapper::CreateEnvironment [bootstrapper.cc : 328 + 0xc]

    Found by: stack scanning
11  libwebviewchromium.so!v8::Context::New [api.cc : 5494 + 0x0]

    Found by: stack scanning
12  libwebviewchromium.so!blink::WindowProxy::createContext [WindowProxy.cpp : 332 + 0xc]

#9 too. I don't know how much credence we can give to everything before 
v8::internal::::Invoke. There are only 3 SIGBUS crashes in the crash db:

https://crash.corp.google.com/browse?q=product.name%3D%27AndroidWebView%27%20AND%20custom_data.ChromeCrashProto.magic_signature_1.name%3D%27v8%3A%3Ainternal%3A%3A%3A%3AInvoke%27%20AND%20special_protos.user_feedback.mobile_data.crash_data.stack_trace%20like%20%27%25SIGBUS%25%27&ignore_case=false&enable_rewrite=false&omit_field_name=&omit_field_value=&omit_field_opt=#samplereports

Given the PC is 0x439da9017b9d in both cases, and that's nowhere near the address of the other two V8 generated functions, it suggests that the generated code has jumped to a random address. Seems unlikely that this is related to specific JS, though. Are you able to disable JS in your mail app and test to see if that avoids the crash? It would also be useful (but admittedly very difficult) if you were able to run with gdb attached, and when you hit this crash get a disassembly around the point that V8 code jumped from.

Operating system: Android
                  samsung/hero2ltexx/hero2lte:6.0.1/MMB29K/G935FXXU1APB4:user/release-keys
CPU: aarch64
     8 CPUs

Crash reason:
Crash address: 0x0
Process uptime: not available

Thread 0 (crashed)
 0  0x439da9017b9d

    Found by: given as instruction pointer in context
 1  0x7f13139a24

    Found by: previous frame's frame pointer
 2  0x7f1312118c

    Found by: previous frame's frame pointer
 3  libwebviewchromium.so!v8::internal::::Invoke [execution.cc : 98 + 0x14]

    Found by: stack scanning
 4  libwebviewchromium.so!v8::internal::Execution::Call [execution.cc : 164 + 0x20]

    Found by: stack scanning
 5  libwebviewchromium.so!blink::EventPath::calculatePath [EventPath.cpp : 153 + 0x0]

    Found by: stack scanning
 6  libwebviewchromium.so!v8::Function::Call [api.cc : 4391 + 0x18]

    Found by: stack scanning
 7  libwebviewchromium.so!v8::internal::Isolate::FireBeforeCallEnteredCallback [isolate.cc : 2659 + 0x8]

    Found by: stack scanning
 8  libwebviewchromium.so!v8::Function::Call [api.cc : 172 + 0x4]

    Found by: stack scanning
 9  libwebviewchromium.so!v8::Function::Call [log.h : 438 + 0x0]

    Found by: stack scanning
10  libwebviewchromium.so!blink::V8ScriptRunner::callFunction [V8ScriptRunner.cpp : 465 + 0x10]

    Found by: stack scanning
11  libwebviewchromium.so!blink::V8PerIsolateData::runEndOfScopeTasks [ConditionalDestructor.h : 20 + 0x4]

    Found by: stack scanning
12  libwebviewchromium.so!blink::V8ScriptRunner::callFunction [V8ScriptRunner.cpp : 459 + 0x0]

Re: #11

There is no easy way for me to disable JS, the message view depends on it (same as in Gmail, and maybe other mail apps).

I've since gotten a few more crashes, it's *quite* unstable.

Still no crashes on Samsung S6 (also 64-bit Exynos) which same "beta" version, 50.0.2661.35

--

Given that this crash is also reported in other apps, including Google's own, and given that my report seems to be the most detailed --

-- don't you guys want to install my app on an S7 Edge (surely there must be some at Google?) and then you can attach gdb, disable JS or some specific JS engine features, etc?

And if you do, this is the current version:

https://www.aqua-mail.com/download/AquaMail-market-1.6.1.5.apk

Or you can install from Play (it's free):

https://play.google.com/store/apps/details?id=org.kman.AquaMail

( not that the exact version should matter -- what changed was the new phone, the S7 Edge )

Re: #11

I am not able to disable JS in any of these applications since Im not the one who built them. Not sure if I am able to attach gdb if the app is not in debug mode and the device is not rooted?

If you can provide me with a test app, source code for example, I am able to compile and test run it if you want.

There is no new info. Marking dup was correct.

I mananged to extract the microdump and stacktrace from when the Google Search crashed. Not sure if you have any use of it now but I attached it anyways.

Deleted: google_search.txt 60.7 KB

google_search.txt 60.7 KB View Download

Yeah that's the same one still. For some reason I thought google search used 32bit webview so isn't affected by this bug. Clearly I thought wrong