Mirror of the gdb-patches mailing list
 help / color / mirror / Atom feed
From: Simon Marchi <simon.marchi@polymtl.ca>
To: Tim Wiederhake <tim.wiederhake@intel.com>
Cc: gdb-patches@sourceware.org, markus.t.metzger@intel.com
Subject: Re: [PATCH v3 05/12] btrace: Use function segment index in insn  iterator.
Date: Wed, 10 May 2017 02:20:00 -0000	[thread overview]
Message-ID: <509c21522fecdf146903a231e97d531c@polymtl.ca> (raw)
In-Reply-To: <1494312929-22749-6-git-send-email-tim.wiederhake@intel.com>

On 2017-05-09 02:55, Tim Wiederhake wrote:
> Remove FUNCTION pointer in struct btrace_insn_iterator and use an index 
> into
> the list of function segments instead.
> 
> 2017-05-09  Tim Wiederhake  <tim.wiederhake@intel.com>
> 
> gdb/ChangeLog:
> 
> 	* btrace.c: (btrace_insn_get, btrace_insn_get_error, 
> btrace_insn_number,
> 	btrace_insn_begin, btrace_insn_end, btrace_insn_next, 
> btrace_insn_prev,
> 	btrace_find_insn_by_number): Replace function segment pointer with
> 	index.
> 	(btrace_insn_cmp): Simplify.
> 	* btrace.h: (struct btrace_insn_iterator) Rename index to
> 	insn_index.  Replace function segment pointer with index into function
> 	segment vector.
> 	* record-btrace.c (record_btrace_call_history): Replace function
> 	segment pointer use with index.
> 	(record_btrace_frame_sniffer): Retrieve function call segment through
> 	vector.
> 	(record_btrace_set_replay): Remove defunc't safety check.

Looks good, just a few comments below.

> @@ -2468,12 +2474,21 @@ int
>  btrace_insn_cmp (const struct btrace_insn_iterator *lhs,
>  		 const struct btrace_insn_iterator *rhs)
>  {
> -  unsigned int lnum, rnum;
> +  gdb_assert (lhs->btinfo == rhs->btinfo);
> 
> -  lnum = btrace_insn_number (lhs);
> -  rnum = btrace_insn_number (rhs);
> +  if (lhs->call_index > rhs->call_index)
> +    return 1;
> +
> +  if (lhs->call_index < rhs->call_index)
> +    return -1;
> +
> +  if (lhs->insn_index > rhs->insn_index)
> +    return 1;
> +
> +  if (lhs->insn_index < rhs->insn_index)
> +    return -1;
> 
> -  return (int) (lnum - rnum);
> +  return 0;
>  }

I the number of comparisons could be reduced by doing:

   int
   btrace_insn_cmp (const struct btrace_insn_iterator *lhs,
                    const struct btrace_insn_iterator *rhs)
   {
     gdb_assert (lhs->btinfo == rhs->btinfo);

     if (lhs->call_index != rhs->call_index)
       return lhs->call_index - rhs->call_index;

     return lhs->insn_index - rhs->insn_index;
   }

> 
>  /* See btrace.h.  */
> @@ -2522,8 +2537,8 @@ btrace_find_insn_by_number (struct
> btrace_insn_iterator *it,
>      }
> 
>    it->btinfo = btinfo;
> -  it->function = bfun;
> -  it->index = number - bfun->insn_offset;
> +  it->call_index = bfun->number - 1;
> +  it->insn_index = number - bfun->insn_offset;
>    return 1;
>  }
> 
> diff --git a/gdb/btrace.h b/gdb/btrace.h
> index ef2c781..ca79667 100644
> --- a/gdb/btrace.h
> +++ b/gdb/btrace.h
> @@ -195,12 +195,11 @@ struct btrace_insn_iterator
>    /* The branch trace information for this thread.  Will never be 
> NULL.  */
>    const struct btrace_thread_info *btinfo;
> 
> -  /* The branch trace function segment containing the instruction.
> -     Will never be NULL.  */
> -  const struct btrace_function *function;

Just an idea, you could factor out those

   it->btinfo->functions[it->call_index]

in a small helper method in btrace_insn_iterator:

btrace_function *function ()
{
   return this->btinfo->functions[this->call_index];
}

> @@ -2691,7 +2691,7 @@ record_btrace_set_replay (struct thread_info *tp,
> 
>    btinfo = &tp->btrace;
> 
> -  if (it == NULL || it->function == NULL)
> +  if (it == NULL)

Not sure, but wouldn't the equivalent check be that call_index < 
btinfo->functions.size () ?

Thanks,

Simon


  reply	other threads:[~2017-05-10  2:20 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-05-09  7:01 [PATCH v3 00/12] btrace: Turn linked list of function call segments into vector Tim Wiederhake
2017-05-09  7:01 ` [PATCH v3 02/12] btrace: Transfer ownership of pointers Tim Wiederhake
2017-05-09 12:21   ` Simon Marchi
2017-05-09  7:01 ` [PATCH v3 09/12] btrace: Remove struct btrace_thread_info::flow Tim Wiederhake
2017-05-10  3:46   ` Simon Marchi
2017-05-10 11:46     ` Wiederhake, Tim
2017-05-10 13:59       ` Simon Marchi
2017-05-09  7:01 ` [PATCH v3 03/12] btrace: Add btinfo to instruction interator Tim Wiederhake
2017-05-09  7:01 ` [PATCH v3 11/12] btrace: Remove bfun_s vector Tim Wiederhake
2017-05-10  4:27   ` Simon Marchi
2017-05-10 11:46     ` Wiederhake, Tim
2017-05-09  7:01 ` [PATCH v3 08/12] btrace: Replace struct btrace_thread_info::up Tim Wiederhake
2017-05-10  3:26   ` Simon Marchi
2017-05-10 11:46     ` Wiederhake, Tim
2017-05-09  7:01 ` [PATCH v3 06/12] btrace: Remove constant arguments Tim Wiederhake
2017-05-10  2:45   ` Simon Marchi
2017-05-10 11:46     ` Wiederhake, Tim
2017-05-09  7:01 ` [PATCH v3 04/12] btrace: Use function segment index in call iterator Tim Wiederhake
2017-05-09 12:50   ` Simon Marchi
2017-05-09 13:14     ` Wiederhake, Tim
2017-05-09 14:29       ` Simon Marchi
2017-05-09  7:01 ` [PATCH v3 12/12] btrace: Store function segments as objects Tim Wiederhake
2017-05-10  5:10   ` Simon Marchi
2017-05-10 11:46     ` Wiederhake, Tim
2017-05-10 14:16       ` Simon Marchi
2017-05-09  7:01 ` [PATCH v3 07/12] btrace: Remove struct btrace_thread_info::{begin,end} Tim Wiederhake
2017-05-10  3:06   ` Simon Marchi
2017-05-09  7:01 ` [PATCH v3 10/12] btrace: Replace struct btrace_thread_info::segment Tim Wiederhake
2017-05-10  4:14   ` Simon Marchi
2017-05-10 11:46     ` Wiederhake, Tim
2017-05-10 14:13       ` Simon Marchi
2017-05-09  7:01 ` [PATCH v3 01/12] btrace: Use std::vector in struct btrace_thread_information Tim Wiederhake
2017-05-09 12:10   ` Simon Marchi
2017-05-09  7:01 ` [PATCH v3 05/12] btrace: Use function segment index in insn iterator Tim Wiederhake
2017-05-10  2:20   ` Simon Marchi [this message]
2017-05-10 11:46     ` Wiederhake, Tim

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=509c21522fecdf146903a231e97d531c@polymtl.ca \
    --to=simon.marchi@polymtl.ca \
    --cc=gdb-patches@sourceware.org \
    --cc=markus.t.metzger@intel.com \
    --cc=tim.wiederhake@intel.com \
    /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