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: Linux VFS cache hit rate script


On Thu, Apr 21, 2011 at 4:45 PM, Josh Stone <jistone@redhat.com> wrote:
> Oy - it's possible that the "dev" is meaningless (or means something
> different) when the name is returned "N/A" (bdev==0), so we may be
> caching noise as the index. ?Or in any case, caching "N/A" in so many
> different forms is not helpful. ?Would you be willing to try this
> modified function?
>
> function __find_bdevname:string(dev:long, bdev:long)
> {
> ? ? ? ?if (bdev == 0)
> ? ? ? ? ? ? ? ?return "N/A"
> ? ? ? ?if (dev in __devnames)
> ? ? ? ? ? ? ? ?return __devnames[dev]
> ? ? ? ?return __devnames[dev] = bdevname(bdev)
> }

No problem... seems to have no effect though:

 339869 	dev: 16	devname: N/A
    345 	dev: 18	devname: N/A
   4569 	dev: 22	devname: N/A
      5 	dev: 23	devname: N/A
    798 	dev: 265289728	devname: dm-0
    640 	dev: 3	devname: N/A
   2594 	dev: 5	devname: N/A
 250861 	dev: 6	devname: N/A

I just commented out the existing version in
/usr/share/systemtap/tapsets/vfs.stp, and put your function in
instead. Do I need to do anything more than that, like rebuild
something? (not a systemtap dev)

Jake


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