On 1/13/24 11:22, Vineet Gupta wrote:
Now we just need to debug what in the vdso call frame information is wrong and any pointers to debug that would be appreciated, even if you are able to fix it right away ;-)
The only way is to step through uw_frame_state_for() in libgcc and figure out what's wrong with the unwind information. Probably with a simultaneous debugging of setup_rt_frame() in qemu in order to compare addresses.
I thought I had enough _Static_asserts added so that any mistakes would be caught at compile time, but obviously not. I'll have a look.
r~
