This is the mail archive of the systemtap@sourceware.org mailing list for the systemtap project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: problems displaying offsets...


On 05/20/2010 03:12 PM, David Smith wrote:
> On 05/18/2010 08:20 AM, Steve Dickson wrote:
>> In need of a second pair of eyes... 
>>
>> probe kernel.function("kernel_sendpage")
>> {
>>     size = ulong_arg(4)
>>     printf("kernel_sendpage: offset %d size %d\n", $offset, size);
>> }
>>
>> From some reason the offset displayed is always zero in the
>> above probe and I know for fact its not always zero... 
>>
>> I'm have the same problem in I/O probes in the nfsd.proc.commit 
>> probe... Does anybody see what I'm doing wrong?
> 
> Hmm.  I don't really see anything too odd here.  Why are you using 'size
> = ulong_arg(4)' instead of just $size?
> 

Yes, it seems like $size would be the easier way to go. Looking
through the code size_t could be "unsigned long" (x86_64) or "unsigned
int" (i386):

http://rhkernel.org/RHEL6+2.6.32-19.el6/arch/x86/include/asm/posix_types_32.h#L18
http://rhkernel.org/RHEL6+2.6.32-19.el6/arch/x86/include/asm/posix_types_64.h#L18

Portability might be a problem using ulong_arg() in this case.


-Will


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]