ZeroPain missing a level compared to _kernel_unwind
André Timmermans (100) 655 posts |
My program intercepted an abort and generated a stacktrace which for some reason as one more stack level compared to the ZeroPain log. My backtrace function:
The generated stacktrace, notice the MetaList_Find function:
The zeroPain log:
|
Rick Murray (539) 13840 posts |
Just off the top of my head, going backwards from main are two addresses you can’t resolve to names (but Debugger can – it’s the stuff that gets the runtime and app going); then there’s the quoted address which is in kernel_CallInitProcs with the second address being zero. This indicates the end (head) of the backtrace, doesn’t it? Maybe you shouldn’t be stepping back quite this far. ;-) To be honest, in my backtrace code, I do a match for “main” and abort then. Anything before that is in CLib, out of my control, and not my problem. Going back that far is perhaps useful for an OS level debugger, but for an app backtrace, it’s “clutter”. |
Rick Murray (539) 13840 posts |
BTW, leave a blank space between any text and the <pre>…</pre> blocks. If you do that, it shouldn’t double-space all the quoted text, and maybe if we’re lucky it won’t make quite such a mess of the code as it has. |
Tristan M. (2946) 1039 posts |
Ohhh… So that’s why the double spacing happens sometimes. I learned something today. |
Clive Semmens (2335) 3276 posts |
Don’t forget you can edit – even at this stage. Best not to in terms of content, or people will miss the changes, but for formatting, no problem. |