This is the mail archive of the
gdb@sourceware.org
mailing list for the GDB project.
Cortex-M stack unwinding and stack register naming question
- From: "Fredrik Hederstierna" <fredrik dot hederstierna at verisure dot com>
- To: gdb at sourceware dot org
- Cc: arenquinha at cimeq dot qc dot ca, palves at redhat dot com, chrisfriedt at gmail dot com
- Date: Sat, 16 Jul 2016 15:20:18 +0000
- Subject: Cortex-M stack unwinding and stack register naming question
- Authentication-results: sourceware.org; auth=none
- References:
- Sensitivity:
Hi
I've been trying to follow the great patch from James-Adam Renquinha Henri
https://bugs.launchpad.net/gcc-arm-embedded/+bug/1566054
that makes it possible to unwind stack in exceptions for Cortex-M (today not possible).
It seems to me that it all ended up in a discussion about the MSP and PSP stack pointers,
and how to name and access these registers. Previous discussion here:
http://gdb-patches.sourceware.narkive.com/KPpCBXPB/cortex-m-xml-register-descriptions-for-m-system
Though I'm not sure if all reached any consensus, or how got action point to implement the suggested solution.
Maybe there could be an issue entered in bugzilla that can be assigned/prioritized to/by someone?
So to me it seems that these steps needs to be taken in this order?
1/ Reach consensus on 'system' registers (eg MSP and PSP stack pointers) naming and how to access (XML stuff etc), and fix this?
2/ Add a stack unwinding patch that uses this new API/naming and fix correct use of MSP/PSP for backtracing in exception stacks etc.
3/ Floating point stuff that also the patch address, maybe this can be fixed in a separate third step later (to not block step 2)?
Is this observation/comment correct, or what is the current status, and how are this issues progressing currently?
I'm also willing to help, if I can do any contribution on this, I just need a direction to go to.
Thanks alot for all great work, and Best Regards,
Fredrik Hederstierna