[Bug translator/5890] sys.stp andtestsuite/systemtap.syscall/sys.stp fail on 2.6.25 x86_64

["jistone at redhat dot com" <sourceware-bugzilla at sourceware dot org>]

------- Additional Comments From jistone at redhat dot com  2009-10-09 02:56 -------
(In reply to comment #6)
> Unfortunately, the one who fixed it is also the one who broke it.
> [...]
> Josh, introducing status quo here is a really bad idea. ;)

I'll take full responsibility, of course. :)  What kernel is this?  On my
2.6.30.8-64.fc11.x86_64 it works fine.

It's possible that this is (unfortunately) correct.  Commit b7478964 was made to
discover cases where a function exists multiple times in a CU.  For example, an
extern function might be generated normally, but then also inlined somewhere
within the same file.  Or, a single function might get multiple generated
outputs with different optimizations.

My guess is that on your kernel has multiple copies of compat_sys_recvmsg.  It's
just a shallow wrapper for sys_recvmsg, so I wouldn't be surprised if the call
in compat_sys_socketcall is inlined.  It's not static though, so the compiler
would still need to generate a full function output too.

That it worked before b7478964 might be just because the full copy with an
accessible $fd was shadowing the inline where $fd is less accessible.  If this
is the case, then I think our failure here is actually insufficient debuginfo
(bug #1155), and the previous success was wrong.

Anyway, I'm just guessing, so let's check what is actually the case.  Please try
with more -v and see where the probe is actually being resolved.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |WAITING


http://sourceware.org/bugzilla/show_bug.cgi?id=5890

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.