[Bug general/5620] New: RHEL 5.2 occasional backtraces resolve libc as unknown

[Bug general/5620] New: RHEL 5.2 occasional backtraces resolve libc as unknown

[pmuldoon at redhat dot com]

Running testLinuxCoreFileStackTrace(frysk.proc.dead.TestLinuxCore) ...FAIL
  junit.framework.ComparisonFailure: Compare stack traces
expected:<...__libc_start_main ()...> but was:<...[unknown]...>

This can only be replicated via -r 100 (it normally happens deep into the repeat
run). Occasionally, libc_start_main() ends up being resolved as [unknown]. This
does not happen on f8 as far as I can test. 


Time: 0.126
There was 1 failure:
1)
testLinuxCoreFileStackTrace(frysk.proc.dead.TestLinuxCore)junit.framework.ComparisonFailure:
Compare stack traces expected:<...__libc_start_main ()...> but was:<...[unknown]...>
   at frysk.proc.dead.TestLinuxCore.testLinuxCoreFileStackTrace(funit)
   at frysk.junit.Runner.runCases(funit)
   at frysk.junit.Runner.runTestCases(funit)
   at frysk.pkglibdir.funit.main(funit)

-- 
           Summary: RHEL 5.2 occasional backtraces resolve libc as unknown
           Product: frysk
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: general
        AssignedTo: frysk-bugzilla at sourceware dot org
        ReportedBy: pmuldoon at redhat dot com


http://sourceware.org/bugzilla/show_bug.cgi?id=5620

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

[Bug general/5620] RHEL 5.2 occasional backtraces resolve libc as unknown

[pmuldoon at redhat dot com]

------- Additional Comments From pmuldoon at redhat dot com  2008-01-16 08:55 -------
This is the output of the corefile that caused the issue. All other frames
resolve fine.

-bash-3.1$ fstack core.26473 
Task #26473
#0 0x08048383 in fourth ()
#1 0x080483ce in third ()
#2 0x0804840d in second ()
#3 0x0804843e in first ()
#4 0x08048477 in main ()
#5 0x00125dec in [unknown]
#6 0x080482c1 in _start ()


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=5620

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

[Bug general/5620] RHEL 5.2 occasional backtraces resolve libc as unknown

[pmuldoon at redhat dot com]

------- Additional Comments From pmuldoon at redhat dot com  2008-01-16 21:42 -------
After analysis I believe this is the issue:

Occasionally the vdso gets relocated to a location that libc occupies. This
causes the libc libraries to be relocated, and causing the [unknown] library in
the backtrace.  This really designates an unrelocated libc bug. Or in fact, any
solib relocation bug.

In the linkmap table there is a l_addr field which designates absolute location
of a map in memory. In practice this should equal to the l_addr + virt start
address of the LOAD statement of the solib in question. This relocation is not
being taken into account in corefile modeling. It seems a rare occasion. At best
it seems it only occurs when the vdso is located where an solib should live, and
the solib is shifted.

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=5620

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

[Bug general/5620] RHEL 5.2 occasional backtraces resolve libc as unknown

[pmuldoon at redhat dot com]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|frysk-bugzilla at sourceware|pmuldoon at redhat dot com
                   |dot org                     |
             Status|NEW                         |ASSIGNED


http://sourceware.org/bugzilla/show_bug.cgi?id=5620

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