Re: stap results on powerpc on latest systemtap snapshot.

[Mark Wielaard <mjw at redhat dot com>]

Hi Srikar,

On Thu, 2011-10-27 at 13:39 +0530, Srikar Dronamraju wrote:
> As promised, here are results from couple of runs of SystemTap on
> powerpc on the latest systemtap snapshot. 

Thanks. Analysis of the failures below.

> This one has utrace/uprobes support.
> 
> Host: Linux trillian.ltc.austin.ibm.com 2.6.32-131.0.15.el6.ppc64 #1 SMP Tue May 10 15:44:09 EDT 2011 ppc64 ppc64 ppc64 GNU/Linux
> Snapshot: 430f532
> GCC: 4.4.5 [gcc (GCC) 4.4.5 20110214 (Red Hat 4.4.5-6)]
> 
> FAIL: backtrace (1 0)
> FAIL: backtrace-unwindsyms (1 0)

This is because the powerpc kernel doesn't have a dump_trace()
implementation, which we use as fallback if no other backtrace option is
available. See runtime/autoconf-save-stack-trace.c.

> FAIL: cast-scope

This one fails on other arches too.

> FAIL: compiling sdt.c c89 -pedantic uprobe

That is surprising. Could you send the systemtap.log output for this
one?

> FAIL: ptr_char_var (0x0 != 0x7e)
> FAIL: ptr_short_int_var (0xfff != 0x7fff)
> FAIL: unsigned_short_int_var (0x8001 != 0x4860)
> FAIL: ptr_int_var (0xffffffff916cff40 != 0x10000)
> FAIL: unsigned_int_var (0xffff8001 != 0x916d2a40)
> FAIL: ptr_long_long_int_var (0x1 != 0x10000)
> FAIL: arr_char_var (0x0 != "!~")
> FAIL: arr_struct_var (4095 != 1)
> FAIL: constants ( !+ "constants")
> FAIL: ptr_char_var (0x0 != 0x7e)
> FAIL: ptr_short_int_var (0xfff != 0x7fff)
> FAIL: unsigned_short_int_var (0x8001 != 0x7f50)
> FAIL: ptr_int_var (0xffffffffa296ff40 != 0x10000)
> FAIL: unsigned_int_var (0xffff8001 != 0xa2972a40)
> FAIL: ptr_long_long_int_var (0x1 != 0x10000)
> FAIL: arr_char_var (0x0 != "!~")
> FAIL: arr_struct_var (4095 != 1)
> FAIL: constants ( !+ "constants")

That is strange, this seems to relate to the following XFAILed tests on
my setup:
XFAIL: sdt_misc types types  ptr_char_var  ptr_short_int_var
unsigned_short_int_var  ptr_int_var  unsigned_int_var
ptr_long_long_int_var  arr_char_var  arr_struct_var V1_uprobe -O2
XFAIL: sdt_misc types types  ptr_char_var  ptr_short_int_var
unsigned_short_int_var  ptr_int_var  unsigned_int_var
ptr_long_long_int_var  arr_char_var  arr_struct_var V1_uprobe -O3

Don't know why they fail on your setup.

> FAIL: sdt_misc compiling types V3_uprobe c89 -pedantic

This one also seems fine on my setup. Could you lookup the systemtap.log
output for it?

> FAIL: stmt_rel stmt_rel (0)

This seems to be bad debuginfo related.

> FAIL: vma_vdso.c compile -m32
> FAIL: vta-test.c compile -m32
> FAIL: unable to compile usymbols_lib.c -m32

I assume these fail because glibc-devel.ppc and libgcc-devel.ppc (the
32bit variants) aren't installed on your system.

> FAIL: systemtap.examples/network/netdev build
> FAIL: systemtap.examples/network/netdev run
> FAIL: semok/thirtynine.stp

These too seem related to bad debuginfo/

> FAIL: shared buffer guest
> FAIL: shared buffer guest2
> FAIL: buffer sharing (0, 0)

These are caused by:
WARNING:
".stp_print_flush_test1" [/tmp/stap5nJjAB/stap_510307a32109df80c2768652ebe2cc8e_796.ko] undefined!
Error inserting module
'/tmp/stap5nJjAB/stap_510307a32109df80c2768652ebe2cc8e_796.ko': Unknown
symbol in module

I don't know why that is.

> FAIL: Valid Server Client Arguments: --client-options
> FAIL: Valid Server Client Arguments: --client-options -a i386
> FAIL: Valid Server Client Arguments: --client-options -D X=Y
> FAIL: Valid Server Client Arguments: --client-options -I foo
> FAIL: Valid Server Client Arguments: --client-options -m test
> FAIL: Valid Server Client Arguments: --client-options -r 2.6.32-131.0.15.el6.ppc64
> FAIL: Valid Server Client Arguments: --client-options -a i386 -D X=Y -I foo -m test -r 2.6.32-131.0.15.el6.ppc64
> FAIL: Valid Server Client Arguments: --unprivileged --client-options
> FAIL: Valid Server Client Arguments: --client-options --unprivileged
> FAIL: Valid Server Client Arguments: --unprivileged -a i386 --client-options
> FAIL: Valid Server Client Arguments: --unprivileged -B X=Y --client-options
> FAIL: Valid Server Client Arguments: --unprivileged -D X=Y --client-options
> FAIL: Valid Server Client Arguments: --unprivileged -I foo --client-options
> FAIL: Valid Server Client Arguments: --unprivileged -m test --client-options
> FAIL: Valid Server Client Arguments: --unprivileged -R foo --client-options
> FAIL: Valid Server Client Arguments: --unprivileged -r 2.6.32-131.0.15.el6.ppc64 --client-options
> FAIL: Valid Server Client Arguments: --unprivileged -a i386 -B X=Y -D X=Y -I foo -m test -R foo -r 2.6.32-131.0.15.el6.ppc64 --client-options
> FAIL: Valid Server Client Arguments: -a i386 --unprivileged --client-options
> FAIL: Valid Server Client Arguments: -B X=Y --unprivileged --client-options
> FAIL: Valid Server Client Arguments: -D X=Y --unprivileged --client-options
> FAIL: Valid Server Client Arguments: -I foo --unprivileged --client-options
> FAIL: Valid Server Client Arguments: -m test --unprivileged --client-options
> FAIL: Valid Server Client Arguments: -R foo --unprivileged --client-options
> FAIL: Valid Server Client Arguments: -r 2.6.32-131.0.15.el6.ppc64 --unprivileged --client-options
> FAIL: Valid Server Client Arguments: -a i386 -B X=Y -D X=Y -I foo -m test -R foo -r 2.6.32-131.0.15.el6.ppc64 --unprivileged --client-options
> FAIL: Valid Server Client Arguments: -a i386 --client-options --unprivileged
> FAIL: Valid Server Client Arguments: -B X=Y --client-options --unprivileged
> FAIL: Valid Server Client Arguments: -D X=Y --client-options --unprivileged
> FAIL: Valid Server Client Arguments: -I foo --client-options --unprivileged
> FAIL: Valid Server Client Arguments: -m test --client-options --unprivileged
> FAIL: Valid Server Client Arguments: -R foo --client-options --unprivileged
> FAIL: Valid Server Client Arguments: -r 2.6.32-131.0.15.el6.ppc64 --client-options --unprivileged
> FAIL: Valid Server Client Arguments: -a i386 -B X=Y -D X=Y -I foo -m test -R foo -r 2.6.32-131.0.15.el6.ppc64 --client-options --unprivileged

I am not seeing these, it could be related to firewalls and/or avahi
setup.

> FAIL: systemtap.stress/current.stp startup (timeout)

I am not really sure what is going on here. This test generates a lot of
WARNINGs and has some specific ia64, x86_64 and i386 code in it. Might
need porting to powerpc.

> FAIL: 64-bit dup nd_syscall
> FAIL: 64-bit eventfd nd_syscall
> FAIL: 64-bit inotify nd_syscall
> FAIL: 64-bit net1 nd_syscall
> FAIL: 64-bit pipe nd_syscall
> FAIL: 64-bit poll nd_syscall
> FAIL: 64-bit signal nd_syscall
> FAIL: 64-bit signalfd nd_syscall

I am seeing these too. Probably need tweaks to the argstr parsing for
powerpc. Most of these also fail on my x86_64 setup though.

> FAIL: unprivileged myproc: not tested: process(string).library(string).plt
> FAIL: unprivileged myproc: not tested: process(string).library(string).plt.statement(number)
> FAIL: unprivileged myproc: not tested: process(string).library(string).plt(string)
> FAIL: unprivileged myproc: not tested: process(string).library(string).plt(string).statement(number)
> FAIL: unprivileged myproc: not tested: process(string).plt
> FAIL: unprivileged myproc: not tested: process(string).plt.statement(number)
> FAIL: unprivileged myproc: not tested: process(string).plt(string)
> FAIL: unprivileged myproc: not tested: process(string).plt(string).statement(number)
> FAIL: unprivileged probes: not tested: process(string).library(string).plt
> FAIL: unprivileged probes: not tested: process(string).library(string).plt.statement(number)
> FAIL: unprivileged probes: not tested: process(string).library(string).plt(string)
> FAIL: unprivileged probes: not tested: process(string).library(string).plt(string).statement(number)
> FAIL: unprivileged probes: not tested: process(string).plt
> FAIL: unprivileged probes: not tested: process(string).plt.statement(number)
> FAIL: unprivileged probes: not tested: process(string).plt(string)
> FAIL: unprivileged probes: not tested: process(string).plt(string).statement(number)

These are "correct". New tests need to be written for these new
constructs. http://sourceware.org/bugzilla/show_bug.cgi?id=13373

Thanks,

Mark