[Bug uprobes/15785] New: Failure to resolve symbols defined as indirect functions for user-space libraries

[Bug uprobes/15785] New: Failure to resolve symbols defined as indirect functions for user-space libraries

[alexander.hoole at ece dot uvic.ca]

http://sourceware.org/bugzilla/show_bug.cgi?id=15785

            Bug ID: 15785
           Summary: Failure to resolve symbols defined as indirect
                    functions for user-space libraries
           Product: systemtap
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: uprobes
          Assignee: systemtap at sourceware dot org
          Reporter: alexander.hoole at ece dot uvic.ca

Presently, SystemTap is not able to resolve probe points defined on symbols
corresponding to indirect functions defined in user-space libraries.

For, example:
....
# stap -g -v -e 'probe process("/lib64/libc-2.17.so").function("gettimeofday")
{ log("TEST") }'
....

Fails with the following message:
....
semantic error: while resolving probe point: identifier 'process' at
<input>:1:7
        source: probe process("/lib64/libc-2.17.so").function("gettimeofday") {
log("TEST") }
                      ^

semantic error: no match
....

According to the symbol table this is an indirect function:
...
# nm /lib64/libc-2.17.so
...
03bbdcad0f0 i gettimeofday
...

Tests were conducted on Fedora 19 and the debug information for the libraries
etc were installed. This seems to be a problem with defined indirect functions.
 For example, the following works fine:
...
# stap -g -v -e 'probe process("/lib64/libc-2.17.so").function("printf") {
log("TEST") }'
....
# nm /lib64/libc-2.17.so
...
0000003bbdc51dd0 T printf
...

Errata: 
Installed Packages
systemtap.x86_64                        2.2.1-1.fc19            @anaconda       
systemtap-client.x86_64                 2.2.1-1.fc19            @anaconda       
systemtap-devel.x86_64                  2.2.1-1.fc19            @anaconda       
systemtap-initscript.x86_64             2.2.1-1.fc19            @fedora         
systemtap-runtime.x86_64                2.2.1-1.fc19            @anaconda       
systemtap-runtime-java.x86_64           2.2.1-1.fc19            @fedora         
systemtap-sdt-devel.x86_64              2.2.1-1.fc19            @anaconda

Might be related to bug 13222.

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

[Bug uprobes/15785] Failure to resolve symbols defined as indirect functions for user-space libraries

[alexander.hoole at ece dot uvic.ca]

http://sourceware.org/bugzilla/show_bug.cgi?id=15785

alexander.hoole at ece dot uvic.ca changed:

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

--- Comment #1 from alexander.hoole at ece dot uvic.ca ---
Frank, sorry for adding you directly.  I would like to get this bug addressed
and would be happy to help out (but I will need some help myself since I am not
familiar with all the existing code base).

Should this bug be filed here or under the translator?  It has been inactive
for some time (since we were in email communication about it).

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

[Bug translator/15785] Failure to resolve symbols defined as indirect functions for user-space libraries

[fche at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=15785

Frank Ch. Eigler <fche at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
          Component|uprobes                     |translator
         Resolution|---                         |WORKSFORME

--- Comment #2 from Frank Ch. Eigler <fche at redhat dot com> ---
This seems to work with systemtap 2.4 on f19, and yet more should work
with the next version, as we expand support for symbol-table-based
probing.

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

[Bug translator/15785] Failure to resolve symbols defined as IFUNC (indirect functions) for user-space libraries

[fche at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=15785

Frank Ch. Eigler <fche at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|WORKSFORME                  |---
            Summary|Failure to resolve symbols  |Failure to resolve symbols
                   |defined as indirect         |defined as IFUNC (indirect
                   |functions for user-space    |functions) for user-space
                   |libraries                   |libraries

--- Comment #3 from Frank Ch. Eigler <fche at redhat dot com> ---
My mistake.  IFUNC support is still missing.

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

[Bug translator/15785] Failure to resolve symbols defined as IFUNC (indirect functions) for user-space libraries

[mjw at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=15785

Mark Wielaard <mjw at redhat dot com> changed:

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

--- Comment #4 from Mark Wielaard <mjw at redhat dot com> ---
Some GNU_IFUNC background: http://www.airs.com/blog/archives/403

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

[Bug translator/15785] Failure to resolve symbols defined as IFUNC (indirect functions) for user-space libraries

[lberk at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=15785

Lukas Berk <lberk at redhat dot com> changed:

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

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

[Bug translator/15785] Failure to resolve symbols defined as IFUNC (indirect functions) for user-space libraries

[alexander.hoole at ece dot uvic.ca]

http://sourceware.org/bugzilla/show_bug.cgi?id=15785

--- Comment #5 from alexander.hoole at ece dot uvic.ca ---
Thanks Frank. 

During some investigation I also found that GDB, along with the debug package
for glibc, was helpful in tracking down what was actually happening as GDB
properly handles indirect functions and resolves to the optimized function
(e.g. __strcpy_sse2_unaligned).

Here is an old discussion from GDB that may also help:
http://sourceware.org/ml/gdb-patches/2010-02/msg00429.html

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

[Bug translator/15785] Failure to resolve symbols defined as IFUNC (indirect functions) for user-space libraries

[mark at klomp dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=15785

Mark Wielaard <mark at klomp dot org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mark at klomp dot org

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

[Bug translator/15785] Failure to resolve symbols defined as IFUNC (indirect functions) for user-space libraries

[mjw at fedoraproject dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=15785

Mark Wielaard <mjw at fedoraproject dot org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|mjw at fedoraproject dot org       |

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