This is the mail archive of the gdb-prs@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug gdb/16511] New: Display of function names fails on AIX when analysing 64bit core dump


https://sourceware.org/bugzilla/show_bug.cgi?id=16511

            Bug ID: 16511
           Summary: Display of function names fails on AIX when analysing
                    64bit core dump
           Product: gdb
           Version: 7.6
            Status: NEW
          Severity: normal
          Priority: P2
         Component: gdb
          Assignee: unassigned at sourceware dot org
          Reporter: m.koeppendoerfer at phoenixgroup dot eu
              Host: powerpc-ibm-aix7.1.0.0
            Target: powerpc-ibm-aix7.1.0.0
             Build: powerpc-ibm-aix7.1.0.0

Created attachment 7376
  --> https://sourceware.org/bugzilla/attachment.cgi?id=7376&action=edit
Patch for displaying function names when analysing 64bit core dumps on AIX

When analysing a core dump of a 64bit process on AIX, gdb displays ?? instead
of function name on AIX 7.1 (7100-02-02-1316). Core dump was generated from a
binary compiled with gcc 4.6.3 with g++ -maix64. gdb was also built with gcc
4.6.3 on the same host.

To reproduce:
1) create core dump on AIX - in my case, I used a mini application causing a
segmentation fault
2) "gdb executable-file core-file" - since gdb prints the failing function,
only the address and ?? will be printed, e.g.
#0  0x000000010000074c in ?? ()
3) issue "where" or "bt" command in gdb. Will also print ?? instead of function
names, e.g.
#0  0x000000010000074c in ?? ()
#1  0x000000010000072c in ?? ()
#2  0x00000001000007b4 in ?? ()
#3  0x00000001000007f8 in ?? ()
#4  0x000000010000084c in ?? ()
#5  0x00000001000002b4 in ?? ()

Actual: see above

Expected: Function names and line numbers

Build / Hardware: gdb 7.6.2 as released on AIX 7.1 (7100-02-02-1316) with gcc
4.6.3

Additional Builds: gdb 7.5 shows same behaviour

Additional Information: I could track the issue to the SOLIB_ADD-macro in
gdb/config/rs6000/nm-rs6000.h. On line 32, the if-clause is always true, as a
dummy pid (1) gets passed in case of a core file. Testing this to be not equal
to one solves the problem with missing function names.
This fix causes the notification of observers in gdb/rs6000-nat.c which leads
to some error messages. In my test build, I've simply removed (commented) the
call to observer_notify_new_object_file. I've attached my patch for my test
build.

Regards,

Marco

-- 
You are receiving this mail because:
You are on the CC list for the bug.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]