Re: testsuite summary report - s390x

[David Smith <dsmith at redhat dot com>]

Here's an update...

On 06/28/2010 12:58 PM, David Smith wrote:
> I just finished analyzing the full testsuite results on s390x.  I've
> attached the
> 'sum' file (and sent it to dejazilla).  Here are the highlights.

...

> FAIL: flightrec1 (failed to switch output file)
> FAIL: flightrec4 (failed to switch output file)
> 
> Both of the above failures don't happen when the tests
> (systemtap.base/flightrec[14].exp) are run by themselves.  I believe the
> testcases have a timing problem.

I filed and fixed PR11762 on this issue.

...

> FAIL: warnings (0)
> 
> This is a testcase problem.  The warnings.exp testcase fails with:
> 
> semantic error: no match while resolving probe point
> kernel.statement("bio_init@fs/bio.c+5")
> 
> On s390x, the optimizer has really compressed that function.  If I
> change it to 'kernel.statement("bio_init@fs/bio.c+1")' the test passes fine.

I've fixed this one.

> FAIL: backtrace of yyy_func2 (2)
> FAIL: print_stack of yyy_func2 (2)
> FAIL: backtrace of yyy_func3 (2)
> FAIL: print_stack of yyy_func3 (2)
> FAIL: backtrace of yyy_func4 (2)
> FAIL: print_stack of yyy_func4 (2)
> FAIL: fib (2 0)
> FAIL: fib (10 0)
> 
> All of the above are backtrace problems.  Basically the backtraces
> aren't correct:
> 
> --- yyy_func4 ---
> the stack is 0x0000000020b1403c 0x0000000000000000 0x0000000000000000
> 0x00000000001a286e 0x000000000010e878 0x000000000010e878
>  0x0000000020b1403c : yyy_func4+0x0/0xc [systemtap_test_module2]
>  0x0000000000000000 : __mod_license91+0x0/0xc [systemtap_test_module2]
>  0x0000000000000000 : __mod_license91+0x0/0xc [systemtap_test_module2]
>  0x00000000001a286e
>  0x000000000010e878
>  0x000000000010e878
> 
> yyy_func4 should show yyy_func[321] above it on the stack.

This is (old) PR5450.

> FAIL: systemtap.stress/current.stp compilation
> 
> current.exp failed with:
> 
> semantic error: no match while resolving probe point
> kernel.function("__switch_to").call
> semantic error: no match while resolving probe point
> kernel.function("__switch_to").return
> 
> I'll look into this one more.

This one is (old) PR4331.

> FAIL: 64-bit alarm nd_syscall
> FAIL: 64-bit clock nd_syscall
> FAIL: 64-bit forkwait nd_syscall
> FAIL: 64-bit inotify nd_syscall
> FAIL: 64-bit mmap nd_syscall
> FAIL: 64-bit select nd_syscall
> FAIL: 32-bit access nd_syscall
> FAIL: 32-bit alarm nd_syscall
> FAIL: 32-bit clock nd_syscall
> FAIL: 32-bit forkwait nd_syscall
> FAIL: 32-bit inotify nd_syscall
> FAIL: 32-bit link nd_syscall
> FAIL: 32-bit mmap nd_syscall
> FAIL: 32-bit select nd_syscall
> FAIL: 32-bit signal nd_syscall
> 
> These are failing because of 2 main reasons.
> 
> 1) The testcase keeps getting errors of "ERROR: Cannot access arg(6)" on
> syscalls that don't have 6 args.

I just filed PR11763 on this one.

> 2) Compile errors, like:
> 
> /root/src/testsuite/systemtap.syscall/eventfd.c:2:25: error:
> sys/eventfd.h: No such file or directory

This one is actually OK, since the eventfd() syscall isn't supported on
s390x.

-- 
David Smith
dsmith@redhat.com
Red Hat
http://www.redhat.com
256.217.0141 (direct)
256.837.0057 (fax)