Eric Schrock wrote:
> On Sat, Aug 25, 2007 at 10:45:10AM +0800, zhijun wrote:
>
>> I really appreciate your kindness to explain this for me.
>>
>> ::array ... p | ::eval '<p/Kn | =Kn' gives pure uintptr_t's so it works
>> well with mdb pipeline. What I am suprized is that ::array ... p | /Kn
>> *doesn't* give pure uintptr_t's, as its output has an address in the first
>> line, so I expect it will break the mdb pipeline, but instead it works. Am
>> I missing anything here?
>>
>>
>
> As mentioned in the original mail, dcmds have a way of 'knowing' when
> they are in pipe context (DCMD_PIPE_OUT) and can alter their output
> based on this. For the '/' operator, you can find this check in
> print_arglist():
>
> if (DCMD_HDRSPEC(flags) && (flags & DCMD_PIPE_OUT) == 0) {
>
Hi Eric,
Yes, after trying some dtrace scripts I now understand what you really
mean here.
> Are you just curious, or are you trying to solve a particular problem?
>
Mainly as I am curious, as I didn't even know about the different
behaviour of dcmds with pipeline.
> As Jonathan mentioned, there are more efficient ways to accomplish what
> you have demonstrated thus far.
>
Yes, ::map is a good way to go. Thanks for the help all the way.
BTW: I really wish the document for mdb ( or somewhere else) could
include some examples for dcmds and walkers, it would make our lives
easier than simply doing ::help eval and ::help map and wondering what
are these dcmds for :-)
Best regards,
and have a good day
Zhijun
> - Eric
>
> --
> Eric Schrock, Solaris Kernel Development http://blogs.sun.com/eschrock
> _______________________________________________
> mdb-discuss mailing list
> mdb-discuss at opensolaris.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://mail.opensolaris.org/pipermail/mdb-discuss/attachments/20070826/058ff1e6/attachment.html>
