Brian Utterback
2005-Jul-21 16:43 UTC
[dtrace-discuss] How to limits stack frames in pid provider with ustack?
I am trying to implement a call analyzing graph tool, and I am
trying to get dtrace to capture counts of caller-callee pairs.
My thought on this was to use an aggregate for this, indexing by
the output of ustack. I though to limit the stack frames to only
2, but for some reason, dtrace doesn''t like it when I add the
frame limit:
# dtrace -n pid6886::gettimeofday:entry''{ustack();}''
dtrace: description ''pid6886::gettimeofday:entry'' matched 3
probes
CPU ID FUNCTION:NAME
1 44286 gettimeofday:entry
libc.so.1`gettimeofday
libumem.so.1`umem_update_thread+0x1e4
libc.so.1`_lwp_start
^C
#dtrace -n pid6886::gettimeofday:entry''{ustack(2);}''
dtrace: invalid probe specifier pid6886::gettimeofday:entry: extraneous
argument ''{ustack(2);}'' ($1 is not referenced)
Any ideas?
--
blu
Remember when SOX compliant meant they were both the same color?
----------------------------------------------------------------------
Brian Utterback - OP/N1 RPE, Sun Microsystems, Inc.
Ph:877-259-7345, Em:brian.utterback-at-ess-you-enn-dot-kom
Jonathan Adams
2005-Jul-21 16:50 UTC
[dtrace-discuss] How to limits stack frames in pid provider with ustack?
On Thu, Jul 21, 2005 at 12:43:53PM -0400, Brian Utterback wrote:> #dtrace -n pid6886::gettimeofday:entry''{ustack(2);}'' > dtrace: invalid probe specifier pid6886::gettimeofday:entry: extraneous > argument ''{ustack(2);}'' ($1 is not referenced)Somehow, dtrace is getting three arguments: arg[0] dtrace arg[1] -n arg[2] pid6886::gettimeofday:entry arg[3] {ustack(2);} I have no idea how or why your shell would be producing this kind of madness. Generally, I always quote the entire -n argument: # dtrace -n ''pid6886::gettimeofday:entry{ustack(2);}'' In any case, I can''t reproduce your error message. Cheers, - jonathan -- Jonathan Adams, Solaris Kernel Development
Bryan Cantrill
2005-Jul-21 17:24 UTC
[dtrace-discuss] How to limits stack frames in pid provider with ustack?
On Thu, Jul 21, 2005 at 12:43:53PM -0400, Brian Utterback wrote:> I am trying to implement a call analyzing graph tool, and I am > trying to get dtrace to capture counts of caller-callee pairs. > My thought on this was to use an aggregate for this, indexing by > the output of ustack. I though to limit the stack frames to only > 2, but for some reason, dtrace doesn''t like it when I add the > frame limit: > > # dtrace -n pid6886::gettimeofday:entry''{ustack();}'' > dtrace: description ''pid6886::gettimeofday:entry'' matched 3 probes > CPU ID FUNCTION:NAME > 1 44286 gettimeofday:entry > libc.so.1`gettimeofday > libumem.so.1`umem_update_thread+0x1e4 > libc.so.1`_lwp_start > ^C > > #dtrace -n pid6886::gettimeofday:entry''{ustack(2);}'' > dtrace: invalid probe specifier pid6886::gettimeofday:entry: extraneous > argument ''{ustack(2);}'' ($1 is not referenced)Are you absolutely sure there wasn''t a space between the entry and the tick in the above? This works for me on both S10/GA and Nevada/latest: # dtrace -n BEGIN''{ustack(2)}'' dtrace: description ''BEGIN'' matched 1 probe CPU ID FUNCTION:NAME 1 1 :BEGIN libc.so.1`ioctl+0xa libdtrace.so.1`dtrace_go+0xef ^C But the following will _always_ generate the message that you saw: # dtrace -n BEGIN ''{ustack(2)}'' dtrace: invalid probe specifier BEGIN: extraneous argument ''{ustack(2)}'' ($1 is not referenced) You can naturally see why; dtrace(1M) can''t differentiate that clause from an argument (one could argue that we could have a slightly better message when the extraneous argument looks suspiciously like a valid D clause, but it''s an error at any rate). Anyway, verify that there wasn''t a space -- and if it''s still busted, please let us know what bits you''re running on... - Bryan -------------------------------------------------------------------------- Bryan Cantrill, Solaris Kernel Development. http://blogs.sun.com/bmc
Brian Utterback
2005-Jul-21 18:39 UTC
[dtrace-discuss] How to limits stack frames in pid provider with ustack?
Bryan Cantrill wrote:> On Thu, Jul 21, 2005 at 12:43:53PM -0400, Brian Utterback wrote: > >>I am trying to implement a call analyzing graph tool, and I am >>trying to get dtrace to capture counts of caller-callee pairs. >>My thought on this was to use an aggregate for this, indexing by >>the output of ustack. I though to limit the stack frames to only >>2, but for some reason, dtrace doesn''t like it when I add the >>frame limit: >> >># dtrace -n pid6886::gettimeofday:entry''{ustack();}'' >>dtrace: description ''pid6886::gettimeofday:entry'' matched 3 probes >>CPU ID FUNCTION:NAME >> 1 44286 gettimeofday:entry >> libc.so.1`gettimeofday >> libumem.so.1`umem_update_thread+0x1e4 >> libc.so.1`_lwp_start >>^C >> >>#dtrace -n pid6886::gettimeofday:entry''{ustack(2);}'' >>dtrace: invalid probe specifier pid6886::gettimeofday:entry: extraneous >>argument ''{ustack(2);}'' ($1 is not referenced) > > > Are you absolutely sure there wasn''t a space between the entry and the > tick in the above? This works for me on both S10/GA and Nevada/latest: > > # dtrace -n BEGIN''{ustack(2)}'' > dtrace: description ''BEGIN'' matched 1 probe > CPU ID FUNCTION:NAME > 1 1 :BEGIN > libc.so.1`ioctl+0xa > libdtrace.so.1`dtrace_go+0xef > ^C > > But the following will _always_ generate the message that you saw: > > # dtrace -n BEGIN ''{ustack(2)}'' > dtrace: invalid probe specifier BEGIN: extraneous argument ''{ustack(2)}'' > ($1 is not referenced) > > You can naturally see why; dtrace(1M) can''t differentiate that clause from > an argument (one could argue that we could have a slightly better message > when the extraneous argument looks suspiciously like a valid D clause, > but it''s an error at any rate). > > Anyway, verify that there wasn''t a space -- and if it''s still busted, > please let us know what bits you''re running on...Okay, quoting the whole thing made it work. I cannot verify that that I didn''t have a space, so I will assume that I did and that was the problem. -- blu Remember when SOX compliant meant they were both the same color? ---------------------------------------------------------------------- Brian Utterback - OP/N1 RPE, Sun Microsystems, Inc. Ph:877-259-7345, Em:brian.utterback-at-ess-you-enn-dot-kom