libera/#sbcl - IRC Chatlog
Search
13:56:17
|3b|
ACTION thought backtrace_from_fp used to work on windows from gdb, not sure if i misremembered (or it was only in specific contexts like after hitting LDB or something) or if it broke
13:57:05
luis
but it's tricky if you're within a C frame since you'll have to figure out the address of the first Lisp frame
14:01:56
luis
teaching Windows how to name Lisp frames is on my... wishlist. I have no idea how that could be done. Perhaps using the elfinator thingie.
14:03:31
luis
Need to tackle CRLF first since that's the main thing blocking us from easily upgrading our version of SBCL. *ugh*
14:05:04
|3b|
ACTION needs to turn my debughlp patches into a contrib one of these days so someone else can maintain it :p
14:16:14
|3b|
ACTION has not, but the infrastructure for that stuff looked nice when i was investigating ways to implement sprof
14:26:37
luis
|3b|: so, teaching windows how to walk lisp frames requires segegrating all code objects onto a code heap, mostly because the metadata can only refer to regions <= 4 GB (I guess it's not a problem if you don't need a heap larger than 4 GB). Alternatively, we could look at the instruction pointer to decide whether we should take a Lisp backtrace or
14:26:37
luis
C backtrace then complete the C backtrace once it hits a Lisp frame and becomes confused.