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: little confuse about kernel trace point


On 03/26/2012 11:06 PM, ch huang wrote:
> i see the following code in example script
> 
> 
> probe kernel.trace("mm_pdflush_kupdate") {
>   if (!log_event()) next
>   kupdate[pid()] <<< $count
>   command[pid()] = execname()
> }
> 
> so when i search "mm_pdflush_kupdate" i find something about trace
> point define named mm_pdflush_kupdate
> 
> in file include/trace/mm.h
> 
>  62 DEFINE_TRACE(mm_pdflush_kupdate,
>  63     TPPROTO(unsigned long count),
>  64     TPARGS(count));
> 
> it seems mm_pdflush_kupdate trace a variable in kernel code,but how
> can i find exactly name ? and in which function?

To see how any tracepoint is used, look in the kernel source for a
trace_NAME() call -- in this case you'll find in mm/page-writeback.c:

422 static void wb_kupdate(unsigned long arg)
423 {
...
445 	nr_to_write = global_page_state(NR_FILE_DIRTY) +
446 			global_page_state(NR_UNSTABLE_NFS) +
447 			(inodes_stat.nr_inodes - inodes_stat.nr_unused);
448 	trace_mm_pdflush_kupdate(nr_to_write);


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