Mirror of the lttng-dev mailing list
 help / color / mirror / Atom feed
From: simon.marchi@polymtl.ca (Simon Marchi)
Subject: [lttng-dev] Printing bit fields with LTTng
Date: Wed, 10 Jun 2015 18:20:48 -0400	[thread overview]
Message-ID: <CAFXXi0=kTqTy-V843a3v+6qs0SmOPFsP6-2sVZoE6mi8LB0vhA@mail.gmail.com> (raw)
In-Reply-To: <CAFXXi0k-HfWxPVmmF8=nap20pS-HRQwWwMc0P8jj0-RRhjXkYw@mail.gmail.com>

On 10 June 2015 at 17:28, Simon Marchi <simon.marchi at polymtl.ca> wrote:
> On 10 June 2015 at 17:16, Chidhu R <chid1989 at gmail.com> wrote:
>> Hello,
>>
>> I am getting compilation issues while trying to print bit field values with
>> LTTng.
>>
>> Error:
>>
>> hello.c: In function ?main?:
>> hello.c:36:1: error: ?typeof? applied to a bit-field
>> hello.c:36:1: error: ?typeof? applied to a bit-field
>> hello.c:36:1: error: ?typeof? applied to a bit-field
>> hello.c:36:1: error: ?sizeof? applied to a bit-field
>>
>> Snippet:
>>
>> struct abc{
>>         int a:16;
>>         int b:8;
>> };
>>
>> Line 36: tracepoint(hello_world, my_third_tracepoint, aa.a, "welcome");
>> Line 37: printf("val = %x\n",aa.a);
>>
>> tracepoint line results in compilation error.
>> printf succeeds.
>>
>> Tp definition looks like this.
>>
>> TRACEPOINT_EVENT(
>>     hello_world,
>>     my_third_tracepoint,
>>     TP_ARGS(
>>         uint32_t, arg,
>>         const char * , my_string_arg
>>     ),
>>     TP_FIELDS(
>>         ctf_integer_hex(uint32_t, my_arg, arg)
>>         ctf_string(field, my_string_arg)
>>     )
>> )
>>
>> How to print bit field values?
>>
>> Thanks
>> Chid
>
> I have no idea what the right fix would be, but an easy workaround
> could be to add 0 to your value:
>
>     tracepoint(hello_world, my_third_tracepoint, aa.a + 0, "welcome");
>
> I just tried it and it works.
>
> Simon

Alright, after a bit more investigation (thanks to Philippe Proulx),
we found that this is caused by the SystemTap integration. If the
SystemTap integration is enabled, each LTTng tracepoint also defines a
SystemTap tracepoint.

SystemTap uses sizeof() on the tracepoint's input arguments to find
their sizes and typeof() to find their signedness. It encodes this
info into a special section that then allows tools like gdb to re-use
the tracepoints and get the argument values. For more information,
look at the definitions of _SDT_ARGSIZE and _SDT_ARGSIGNED in
/usr/include/sys/sdt.h.

Pragmatically, I think that the "+ 0" trick is good enough, since it
pleases SystemTap and is otherwise harmless. You could get the same
result by casting your value to an integer type. However, you would
need to carefully choose the size and signedness of the type you cast
to, or it might alter your value. With + 0, it should work with any
type.

Thanks for raising this, I learned something new today.



  reply	other threads:[~2015-06-10 22:20 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-06-10 21:16 Chidhu R
2015-06-10 21:28 ` Simon Marchi
2015-06-10 22:20   ` Simon Marchi [this message]
2015-06-11 18:11     ` Chidhu R

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='CAFXXi0=kTqTy-V843a3v+6qs0SmOPFsP6-2sVZoE6mi8LB0vhA@mail.gmail.com' \
    --to=simon.marchi@polymtl.ca \
    /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