public inbox for systemtap@sourceware.org
 help / color / mirror / Atom feed
From: Stan Cox <scox@redhat.com>
To: Siva N <harnan@outlook.com>
Cc: systemtap@sourceware.org
Subject: Re: stap --dyninst with C++ applications
Date: Tue, 19 Jan 2021 14:38:03 -0500	[thread overview]
Message-ID: <9e49ec35-f18b-2a52-39c9-4be844905ed8@redhat.com> (raw)
In-Reply-To: <CS1PR8401MB0903E324E417D28ED3108F22C6E90@CS1PR8401MB0903.NAMPRD84.PROD.OUTLOOK.COM>

> I am curious to understand how C++ object arguments are handled with USDT.
> 
> I have a C++ application. I have added probe points using the DTRACE_PROBE[n] macro. If I pass a reference/pointer to a C++ object as one of the arguments, I wonder how one accesses this inside the probe handler. I could treat the argument(s) as a void pointer, and if I managed create the actual C++ object layout

--dyninst should handle probe("NAME").mark("LABEL") probes fine.  If I 
have the small c++ sample:

> #include <string>
> #include <iostream>
> #include <tstclass.h>
> using namespace std;
> 
> struct A
> {
>   A();
>   long aa;
>   long bb;
>   string string1;
>   string string2;
> };
> 
> A::A ()
> {
>   aa = 10;
>   bb = 20;
>   string1 = "abc";
>   string2 = "xyz";
> }
> 
> int main()
> {
>   struct A anA;
>   SDT_MISC_TEST_PROBE_4(&anA);
>   std::cout << anA.aa << ' ' << anA.bb << ' ' << anA.string1 << ' ' << anA.string2 << ' ' << '\n';
> }

and the .d file:
> provider sdt_misc {
> 	probe test_probe_4 (struct A arg);
> };

Then the arg struct corresponding to SDT_MISC_TEST_PROBE_4 can be 
accessed via:
> function get_string:string (str)
> %{
>   char astr[32];
>   strcpy (astr,(char*)STAP_ARG_str);
>   STAP_RETURN (astr);
> %}
> 
> probe process("./tstclass.x").mark("test_probe_4") 
> { 
>   printf("%s aa=%#lx string1=%s\n",pp(),@cast($arg1,"struct A")->aa, 
> 	 get_string(@cast($arg1,"struct A")->string1->_M_dataplus->_M_p))
> }

The @cast interprets arg1 as a struct A so the members can be accessed. 
  The probes are translated to C so it uses internal knowledge to get at 
string1.  (Is there an easier way to do this?)  So that yields:
% stap --dyninst -g --disable-cache tstclass.stp -c ./tstclass.x
process("/this/dir/tstclass.x").statement(0x401283) aa=0xa string1=abc
10 20 abc xyz


      reply	other threads:[~2021-01-19 19:38 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-11-10 20:19 Siva N
2021-01-19 19:38 ` Stan Cox [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=9e49ec35-f18b-2a52-39c9-4be844905ed8@redhat.com \
    --to=scox@redhat.com \
    --cc=harnan@outlook.com \
    --cc=systemtap@sourceware.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).