On Tue, Jul 7, 2009 at 12:20 PM, barry <bha...@gmail.com> wrote: > > There seem to be two problems under discussion in this thread. One is > a "native stack trace" and the other is the keystore problem Fadden > mentions, in a non-public merge. Both go away with the rollback. I > don't think anyone at Google has a good handle on what the "native" > problem is; I don't think it's been seen there. It's pretty suspicious > that they both go away when this one change is rolled back.
I thought the stack trace is printed by debuggerd, right? X86 does not have it. Will this be the reason? > > > Patching the suspect CL into the current Donut candidate seems to > work, or at least not crash quickly. Taking the merged code [with the > keystore issue] and if'fing out the actual use of the bitmap makes the > bug no longer manifest. Even if this code isn't at fault, it's > certainly tickling something else that I don't understand. I'm willing > to trade theories. Unfortunately, I have a few other things to do, and > then vacation coming up, so won't have time to be in on the hunt until > I get back. > > Donut can't have the exposure to risk that this CL presents, so this > CL won't be in donut, unless we figure out the actual cause really > soon, and it's really clear, and even then, probably not. Donut is in > "bugfix only" right now, and shouldn't be cracked for this CL. > > > > --~--~---------~--~----~------------~-------~--~----~ unsubscribe: android-porting+unsubscr...@googlegroups.com website: http://groups.google.com/group/android-porting -~----------~----~----~----~------~----~------~--~---