== Comment: #0 - Mahesh M. Hiremath <maheshhi.com> - 2012-10-13 02:13:50 ==
Tool chain GDB tests are failing.

--- Other Details ---

[root@c57f1ju0111 gdb-7.5.0.20120926]# uname -a
Linux c57f1ju0111.ppd.pok.ibm.com 3.6.0-3.fc18.ppc64p7 #1 SMP Wed Oct 10 07:34:22 MST 2012 ppc64 ppc64 ppc64 GNU/Linux



---Steps to Reproduce ---
1.Install gdb-7.5.0.20120926-24.fc18.src.rpm rpm .
2.rpmbuild -bp gdb.spec
3[root@c57f1ju0111 gdb-7.5.0.20120926]# ./configure 
4.[root@c57f1ju0111 gdb-7.5.0.20120926]# make configure-gdb 
5. execute in 32 bit mode:
[root@c57f1ju0111 gdb-7.5.0.20120926]# make check-gdb 'RUNTESTFLAGS=--target_board=unix/-m32 --tool_exec=/usr/bin/gdb' 
6.execute in 64 bit mode
[root@c57f1ju0111 gdb-7.5.0.20120926]#make check-gdb 'RUNTESTFLAGS=--target_board=unix/-m64 --tool_exec=/usr/bin/gdb' 

GDB 32 bit mode test summary : 

                === gdb Summary ===

# of expected passes            21820
# of unexpected failures        269
# of unexpected successes       2
# of expected failures          70
# of known failures             66
# of unresolved testcases       22
# of untested testcases         32
# of unsupported tests          36

GDB 64 bit mode test summary : 

                === gdb Summary ===

# of expected passes            21756
# of unexpected failures        334
# of unexpected successes       2
# of expected failures          80
# of known failures             66
# of unresolved testcases       22
# of untested testcases         24
# of unsupported tests          37

Note: Attached Toolchain_GDB32_log_F18 and Toolchain_GDB64_log_F18 logs for reference.

Thanks
Mahesh

== Comment: #1 - Mahesh M. Hiremath <maheshhi.com> - 2012-10-13 02:15:34 ==


== Comment: #2 - Mahesh M. Hiremath <maheshhi.com> - 2012-10-13 02:16:30 ==


== Comment: #5 - Mahesh M. Hiremath <maheshhi.com> - 2012-10-15 15:34:19 ==


== Comment: #6 - Mahesh M. Hiremath <maheshhi.com> - 2012-10-15 15:35:11 ==


== Comment: #7 - Edjunior Barbosa Machado <emachado.com> - 2012-10-16 13:59:19 ==
Mostly of the failures on both 32 and 64-bit environment are related to gdb.arch/powerpc-power6.exp and powerpc-power7.exp testcases. These were caused by a failure in the 'disassemble' funcionality of gdb, which uses binutils' opcode.
This problem seems to be a regression added by the patch http://sourceware.org/ml/binutils/2012-03/msg00170.html (commit id 5ba09d5182a185d0b7a1cf686d7808131536d281)

== Comment: #11 - Edjunior Barbosa Machado <emachado.com> - 2012-11-28 15:31:39 ==
Alan Modra provided a new patch that fixes this on binutils side. It was already committed upstream:

http://sourceware.org/ml/binutils/2012-11/msg00352.html

Mahesh,
could you please provide access to a machine running Fedora18 so I can create and test a backport patch for gdb?
Thanks.

== Comment: #12 - Mahesh M. Hiremath <maheshhi.com> - 2012-11-28 23:58:32 ==
Hi Edjunior,

Sent you a separate mail with machine access details.

Thanks,
Mahesh

== Comment: #13 - Edjunior Barbosa Machado <emachado.com> - 2012-12-26 17:04:11 ==
Here's the backport patch for latest Fedora 18 gdb (gdb-7.5.1-32.fc18) fixing the disassemble function.

It contains the already accepted fix for binutils:

http://sourceware.org/ml/binutils/2012-11/msg00352.html

and the cleanup on gdb side, also already accepted upstream:

http://sourceware.org/ml/gdb-patches/2012-11/msg00624.html

This also includes a fix in the gdb.arch/powerpc-power6.exp testcase, which is not part of FSF gdb (it appears only on Fedora's gdb source), changing the DFP quad instructions with invalid operands (daddq, dsubq and others).

== Comment: #14 - Edjunior Barbosa Machado <emachado.com> - 2012-12-26 17:05:20 ==
With the attached backport patch, the number of failures was reduced to more reasonable numbers. These are the gdb testsuite results against 64-bit inferior:

# of expected passes            22282
# of unexpected failures        156
# of unexpected successes       3
# of expected failures          82
# of known failures             64
# of untested testcases         18
# of unsupported tests          37

and 32-bit:

# of expected passes            22206
# of unexpected failures        105
# of unexpected successes       2
# of expected failures          70
# of known failures             61
# of untested testcases         38
# of unsupported tests          37