[Bug translator/15937] New: On arm fc19 unable to find $whence argment for kernel.function("sys_llseek").call

[Bug translator/15937] New: On arm fc19 unable to find $whence argment for kernel.function("sys_llseek").call

[wcohen at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=15937

            Bug ID: 15937
           Summary: On arm fc19 unable to find $whence argment for
                    kernel.function("sys_llseek").call
           Product: systemtap
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: translator
          Assignee: systemtap at sourceware dot org
          Reporter: wcohen at redhat dot com

When reviewing the test results on arm machines running fc19 I found that
systemtap wasn't able to find the $whence argument for
kernel.function(sys_llseek").call and this was causing a number of tests to
fail.

On a trimslice arm machine running:

$ uname -a
Linux dhcp129-3.rdu.redhat.com 3.10.10-200.fc19.armv7hl #1 SMP Fri Aug 30
00:21:28 UTC 2013 armv7l armv7l armv7l GNU/Linux


I can replicate the problem with:

$ stap -e 'probe kernel.function("sys_llseek").call{ printf("whence =%d\n",
$whence)}'
semantic error: No cfa_ops supplied, but needed by DW_OP_call_frame_cfa:
identifier '$whence' at <input>:1:66
        source: probe kernel.function("sys_llseek").call{ printf("whence
=%d\n", $whence)}
                                                                               
 ^

Pass 2: analysis failed.  [man error::pass2]


However, the -L command does list it as available:

$ stap -L 'kernel.function("sys_llseek").call'
kernel.function("SyS_llseek@fs/read_write.c:264").call $fd:long int
$offset_high:long int $offset_low:long int $result:long int $whence:long int

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

[Bug translator/15937] On arm fc19 unable to find $whence argment for kernel.function("sys_llseek").call

[mjw at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=15937

Mark Wielaard <mjw at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mjw at redhat dot com

--- Comment #1 from Mark Wielaard <mjw at redhat dot com> ---
In elfutils-0.156 we decided not to lie about whether or not we knew the
default cfa (cfi state) for a particular abi. See this thread:
http://thread.gmane.org/gmane.comp.sysutils.elfutils.devel/2554
Previously we would just guess and hope for the best (which actually was
normally OKish, so one could say the above was a little paranoid). But we only
had that knowledge for i386 and x86_64.

The backend knowledge has been added upstream for arm, ppc and s390 upstream,
which has been backported for fedora (f20 and f21):

* Fri Sep 06 2013 Mark Wielaard <mjw@redhat.com> 0.156-5
- Add elfutils-0.156-abi_cfi-ppc-s390-arm.patch.
 Sets up initial CFI return register, CFA location expression and
 register rules for PPC, S390 and ARM (dwarf_cfi_addrframe support).

I'll also create an update for f19.

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

[Bug translator/15937] On arm fc19 unable to find $whence argment for kernel.function("sys_llseek").call

[mjw at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=15937

--- Comment #2 from Mark Wielaard <mjw at redhat dot com> ---
https://admin.fedoraproject.org/updates/elfutils-0.156-5.fc19

If you could try that package out and provide some karma that would be
appreciated.

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

[Bug translator/15937] On arm fc19 unable to find $whence argment for kernel.function("sys_llseek").call

[wcohen at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=15937

William Cohen <wcohen at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #3 from William Cohen <wcohen at redhat dot com> ---
Using this new version of elfutils-0.156-5 eliminated about 40 of the test
failures due to the cfa_ops problem. Below is URL comparing old run with older
version of elfutils vs a run with elfutils-0.156-5:

https://web.elastic.org/~dejazilla/viewrgdiff.php?rg1=94511&rg2=650838&_sort=0A&_limit=1000

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