Dan Albert
Dan Albert
I'll defer to the LLVM folks on whether or not that's innocuous enough to cherry-pick to r27. It's already in QA, so the bar for cherry-picks of non-regressions has gone...
(and there almost always is another bug, so it's likely)
Duplicate of #929
Sorry for the delay, there wasn't anyone to look after the samples for a while. What does this actually fix? What's a "large screen device"? I don't know how to...
I can see that one thing it does fix is that the timer doesn't restart when the screen rotates, but I'm still not sure what this has to do with...
Working for me. Presumably was an issue with that version of Android Studio, and is fixed on newer versions?
The first release that it could be shipped in (r23) is also the release where we plan to stop using that unwinder, so getting that change submitted won't buy us...
Spoke with @rprichard to clarify and the LLVM unwinder will need a similar improvement. LLVM's unwinder should be quite fast on the newer Android releases that include the libdl improvements...
My inclination is to do this in r27 with an announcement in the changelog so it doesn't surprise anyone. iirc false positives are possible, but only for hand written assembler...
Ah, that's right, I'd remembered wrong. So the only place this is likely to cause trouble is for anyone that has inactive PAC/BTI code that's wrong, and everywhere else it...