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 Fri, Apr 22, 2011 at 1:28 PM, Josh Stone <jistone@redhat.com> wrote:
> That should do it. ?I should have been more clear - the effect I
> intended was just to relieve the need for MAXMAPENTRIES. ?The
> "__devnames" map should now only contain the real device names, and not
> waste any time/space on "N/A" entries. ?Your script's output would not
> be affected though.
>
> Josh
>

Ah. Well, in that case, sorry, still no good. I cut out that bit of
the output, but:

# time stap -DSTP_NO_OVERLOAD devnames.stp -c "sleep 35" | sort | uniq -c
ERROR: Array overflow, check MAXMAPENTRIES near identifier '$file' at
/usr/share/systemtap/tapset/vfs.stp:780:9
WARNING: Number of errors: 1, skipped probes: 2126
Pass 5: run failed.  Try again with another '--vp 00001' option.

Here's an interesting note though: the line numbers are different. The
only difference between the runs is which of the 2 functions is
commented out.

With stock function:
ERROR: Array overflow, check MAXMAPENTRIES near identifier '$file' at
/usr/share/systemtap/tapset/vfs.stp:769:9

With your function:
ERROR: Array overflow, check MAXMAPENTRIES near identifier '$file' at
/usr/share/systemtap/tapset/vfs.stp:780:9


Here's that snippet in vfs.stp (pasting here because the line numbers
are going to be off now, as compared to the stock file):

    761 probe vfs.read = kernel.function("vfs_read")
    762 {
    763         file = $file
    764         pos = $pos
    765         buf = $buf
    766         bytes_to_read = $count
    767         dev = __file_dev($file)
    768         devname = __find_bdevname(dev, __file_bdev($file))
    769         ino = __file_ino($file)
    770
    771         name = "vfs.read"
    772         argstr = sprintf("%d, %d, %p", $count, $pos, $buf)
    773 }
    774
    775 probe vfs.read.return = kernel.function("vfs_read").return
    776 {
    777         name = "vfs.read"
    778         retstr = sprintf("%d", $return)
    779
    780         file = $file
    781         pos = $pos
    782         buf = $buf
    783         bytes_to_read = $count
    784         dev = __file_dev($file)
    785         devname = __find_bdevname(dev, __file_bdev($file))
    786         ino = __file_ino($file)
    787
    788         ret = $return
    789         bytes_read = $return > 0 ? $return : 0
    790         error = $return < 0 ? $return : 0
    791         error_str = error ? errno_str(error) : ""
    792 }

So as you can see, one errors in vfs.read, the other in
vfs.read.return. In both cases something to do with $file gets
assigned to a normal variable.

Side note: I wonder if the "name" variables are messed up... vfs.read*
and vfs.write* don't follow quite the same pattern:

probe vfs.read = kernel.function("vfs_read")
        name = "vfs.read"
probe vfs.read.return = kernel.function("vfs_read").return
        name = "vfs.read"
probe vfs.readv = kernel.function("vfs_readv")
        name = "vfs.read"
probe vfs.readv.return = kernel.function("vfs_readv").return
        name = "vfs.readv"
probe vfs.write = kernel.function("vfs_write")
        name = "vfs.write"
probe vfs.write.return = kernel.function("vfs_write").return
        name = "vfs.write"
probe vfs.writev = kernel.function("vfs_writev")
        name = "vfs.writev"
probe vfs.writev.return = kernel.function("vfs_writev").return
        name = "vfs.writev"


Jake


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