David Buchanan
David Buchanan
In my implementation, the `low_buffer` keeps track of the current DMA buffer. RE the header, fusee-launcher sends the whole payload at once. I also had a go a while back,...
If I had to guess where the bug is, it's in the REAPURB part, because TBH I don't understand what it does very well. `switch_linux_launcher` does things subtly differently (and...
Yeah, that sounds plausible.
Actually, looking a bit more closely, my usage of REAPURB is completely broken. So yeah, that'd probably be it...
The target process may be owned by a different user altogether, so your proposed fix would not work in that case. I think we should read the uid of the...
Which distro are you running?
If the symbol is no longer exported, then I have a vague idea for a workaround - which is, to locate the function via heuristics. I'll need to see if...
Note to self: investigate `__libc_dlopen_mode` See also: https://github.com/apitrace/apitrace/issues/756 (might need that heuristic approach after all)
Upon further investigation, these changes were made because libdl is now directly part of libc - so I suppose we can just use dlopen directly, on newer libcs? https://lwn.net/Articles/864920/
Until now, I've been leaving the `caller_dlopen` argument undefined. On newer libcs (since this patch: https://patchwork.ozlabs.org/project/glibc/patch/[email protected]/ ) , this doesn't matter. But yours apparently still has the check. To work...