That certainly looks like a good result! I don't see anything clearly
wrong in the patch. All the logic in syscall_trace() is squirrelly enough
that it's hard to be real confident. But frankly that's mostly because it
always looked so wonky to begin with that we've been scared to touch it
much.
Hi Roland,
After testing on ia64 with real example of SIGKILL
I made a few fixes and now it works like this:
Unpatched:
[r...@hp-rx6600-03 strace.2]# strace ./k
execve("./k", ["./k"], [/* 22 vars */]) = 0
uname({sys="Linux", node="hp-rx6600-03.rhts.bos.redhat.com", ...}) = 0
brk(0)
