Almost perfect. It worked correctly for the first test cases I tried, and
there's a nice touch on the print that separates groups (I like it!). However,
when I tried adding the -f switch, just to see what would happen, I get this:
% ./task -f -e PM_RUN_CYC,PM_CYC -e PM_RUN_CYC,PM_FXU_IDLE,PM_CYC,PM_CYC,PM_CYC
md5sum `which gdb`
825b15d7279ef21d6c9d018d775758ae /usr/bin/gdb
task: could not read event 0, tried to read 40 bytes, but got 32
task: cannot read values event PM_RUN_CYC: No space left on device
>From the source code, it looks like maybe the .name string is not getting
>terminated properly, and that ": No space left on device" is left over from
>some other call.
This wouldn't explain the error, but the fds index being used on
perf_utilc.s:105 isn't right:
ret = read(fds[evt].fd, values, sz);
if (ret != sz) { /* unsigned */
if (ret == -1)
err(1, "cannot read values event %s",
fds[0].name);
That should be fds[evt].name
Any idea how the string terminator may be getting stomped on?
- Corey
On 5/4/2010 2:21 PM, stephane eranian wrote:
> On Tue, May 4, 2010 at 11:17 PM, Corey Ashford
> <cjash...@linux.vnet.ibm.com> wrote:
>>
>>
>> On 5/4/2010 2:13 PM, stephane eranian wrote:
>>> Corey,
>>>
>>> I have not yet pushed the patch, I wanted to get your opinion first.
>>
>> Push away! it looks good to me.
>>
> Done, try now.
> Thanks for your contribution.
>
>> Thanks,
>>
>> - Corey
>>
>>
--
Regards,
- Corey
Corey Ashford
Software Engineer
IBM Linux Technology Center, Linux Toolchain
Beaverton, OR
503-578-3507
cjash...@us.ibm.com
------------------------------------------------------------------------------
_______________________________________________
perfmon2-devel mailing list
perfmon2-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/perfmon2-devel
