Menu
▾
▴
Re: [Sbcl-devel] [Sbcl-commits] CVS: sbcl/src/code debug-int.lisp, 1.108, 1.109 interr.lisp, 1.41, 1.42
|
From: Cyrus H. <ch...@bo...> - 2007-04-05 23:09:51
|
While what Alastair describes is true, we (or at least I) don't understand why calling compute-calling-frame with the fp and pc from the context works, while calling it with the fp and pc frame %caller- frame-and-pc doesn't work. So the question I have is what's the best way to get the top frame? Amazingly, when I described the problem to Juho, he pretty much knew right where to look and I think he had some ideas on why %caller-frame-and-pc is tripping us up here. So there are really two separate issues, one is doing the back trace and the other is getting the proper top-frame. Even if we fix the calling conventions such that the backtrace mechanism is happy, we might still be running into problems with getting the proper top-frame. Cyrus On Apr 5, 2007, at 12:19 PM, Alastair Bridgewater wrote: > Nikodemus Siivola writes: > >> ...and when we do add it, let's describe the problem it addresses >> properly: what does "trouble tracing through signal handler frames on >> some platforms" mean? Does it mean the backtrace is truncated, or >> that >> we have the signal handler frames visible, or that we get mysterious >> memory faults, or what? > > As I understand it, "trouble tracing through signal handler frames > on some > platforms" means that, on x86-oid systems, the backtrace mechanism > occasionally misinterprets a C stack frame as a Lisp one, and > within two > frames after that is generating no-longer-mysterious memory faults > (because > we now know what causes them). > > The agreed-upon "proper" solution for this is to change the Lisp > stack frame > layout to match the C one in terms of placement of return address > and old > frame pointer relative to the current frame pointer. > >> Cheers, >> >> -- Nikodemus > > --Alastair Bridgewater > > > ---------------------------------------------------------------------- > --- > Take Surveys. Earn Cash. Influence the Future of IT > Join SourceForge.net's Techsay panel and you'll get the chance to > share your > opinions on IT & business topics through brief surveys-and earn cash > http://www.techsay.com/default.php? > page=join.php&p=sourceforge&CID=DEVDEV > _______________________________________________ > Sbcl-devel mailing list > Sbc...@li... > https://lists.sourceforge.net/lists/listinfo/sbcl-devel |