This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
Re: Systemtap on ARM arch results summary
- From: David Smith <dsmith at redhat dot com>
- To: David Long <dave dot long at linaro dot org>
- Cc: Naresh Kamboju <naresh dot kamboju at linaro dot org>, systemtap at sourceware dot org, William Cohen <wcohen at redhat dot com>
- Date: Thu, 12 Jun 2014 11:11:13 -0500
- Subject: Re: Systemtap on ARM arch results summary
- Authentication-results: sourceware.org; auth=none
- References: <CA+G9fYtV4apghaFPDQcD7-xbvG_Ua1mEUi=jcxVYfOQdwzAPXA at mail dot gmail dot com> <5399C174 dot 2060700 at redhat dot com> <5399C89C dot 6020600 at linaro dot org>
On 06/12/2014 10:34 AM, David Long wrote:
> On 06/12/14 11:04, David Smith wrote:
>> On 06/12/2014 12:37 AM, Naresh Kamboju wrote:
>>> Hi,
>>>
>>> Host: Linux linaro-server 3.15.0-rc5-00020-gf903094-dirty #13 SMP
>>> PREEMPT Wed Jun 4 18:07:32 UTC 2014 armv7l armv7l armv7l GNU/Linux
>>> Snapshot: version 2.6/0.158, commit release-2.5-141-ge56d75e47590 +
>>> changes
>>> GCC: 4.8.2 [gcc (Ubuntu/Linaro 4.8.2-19ubuntu1) 4.8.2]
>>> Distro: Linaro 14.04
>>> target board: arndale
>>>
>>> Tested by: Dave Long <dave.long@linaro.org>
>>>
>>> === systemtap Summary ===
>>>
>>> # of expected passes 3747
>>> # of unexpected failures 335
>>> # of expected failures 287
>>> # of unknown successes 1
>>> # of known failures 51
>>> # of untested testcases 204
>>> # of unsupported tests 7
>>>
>>> Best regards
>>> Naresh Kamboju
>>>
>>
>> That isn't too bad for a newish platform. Could you also send the
>> systemtap.log file (perhaps compressed) so we could look at some of the
>> failures?
>>
>
>
> gzip file attached.
I took a short look at the log file.
75 of those failures are from the nd_syscall.exp testcase. These happen
because the arm version of the _stp_arg() function (in
tapset/register.stp) only provides support for getting the 1st 4
arguments. The 1st 4 arguments are in registers, the rest are on the
stack, and that function doesn't know how to unwind the stack.
53 of those failures are from the syscall.exp testcase. These failures
come from missing syscalls probe hits, like syscall.open. I'd guess that
for some reason the arm open syscall function isn't called 'sys_open'.
--
David Smith
dsmith@redhat.com
Red Hat
http://www.redhat.com
256.217.0141 (direct)
256.837.0057 (fax)