On 07/01, Oleg Nesterov wrote:
Another attempt. This version tries to decouple utrace and ptrace. This way it is much simpler to follow the upstream changes, afaics.
It turns out, FC15's kernel 2.6.40 has the wrong utrace.patch
See https://bugzilla.redhat.com/show_bug.cgi?id=728379
in short: PTRACE_SYEMU (used by uml) is broken.
TODO:
The single-stepping updates in ptrace_resume() can race with utrace_reset()->user_disable_single_step().
This was fixed by 20/20, but I noticed that this patch is buggy right before sending.
Perhaps PTRACE_SYSEMU/TIF_SYSCALL_EMU logic was broken, I need to recheck.
Yes, and then I sent the fixes for above, but they were not included.
Dave, I am sorry. Looks like I managed to confuse you. I wrongly assumed you will simply pull the "utrace" branch, and thus I did not re-send the full series.
I am (re)sending it now. Please let me now if you need the incremental diff against 2.6.40-*.fc15. Or anything else...
(just in case, I verified that utrace-patch in rawhide is fine, it matches my utrace-3.1 branch).
Oleg.