Mirror of the gdb-patches mailing list
 help / color / mirror / Atom feed
From: Sergio Durigan Junior <sergiodj@redhat.com>
To: Simon Marchi <simon.marchi@ericsson.com>
Cc: Tom Tromey <tom@tromey.com>,  <gdb-patches@sourceware.org>
Subject: Re: [RFA 3/3] Remove cleanups from break-catch-syscall.c
Date: Wed, 06 Dec 2017 18:58:00 -0000	[thread overview]
Message-ID: <87efo7elaa.fsf@redhat.com> (raw)
In-Reply-To: <280a7994-38a6-2138-b199-8902ee2651b0@ericsson.com> (Simon	Marchi's message of "Fri, 20 Oct 2017 16:26:15 -0400")

On Friday, October 20 2017, Simon Marchi wrote:

> On 2017-10-18 12:06 AM, Tom Tromey wrote:
>> This removes the remaining cleanups from break-catch-syscall.c by
>> storing temporary strings in a vector.
>> 
>> ChangeLog
>> 2017-10-17  Tom Tromey  <tom@tromey.com>
>> 
>> 	* break-catch-syscall.c (catch_syscall_completer): Use
>> 	std::string, gdb::unique_xmalloc_ptr.
>> ---
>>  gdb/ChangeLog             |  5 +++++
>>  gdb/break-catch-syscall.c | 35 ++++++++++++++++++-----------------
>>  2 files changed, 23 insertions(+), 17 deletions(-)
>> 
>> diff --git a/gdb/break-catch-syscall.c b/gdb/break-catch-syscall.c
>> index 01e761ce37..2bbfee0ac4 100644
>> --- a/gdb/break-catch-syscall.c
>> +++ b/gdb/break-catch-syscall.c
>> @@ -560,9 +560,7 @@ catch_syscall_completer (struct cmd_list_element *cmd,
>>                           const char *text, const char *word)
>>  {
>>    struct gdbarch *gdbarch = get_current_arch ();
>> -  struct cleanup *cleanups = make_cleanup (null_cleanup, NULL);
>> -  const char **group_list = NULL;
>> -  const char **syscall_list = NULL;
>> +  gdb::unique_xmalloc_ptr<const char *> group_list;
>>    const char *prefix;
>>    int i;
>>  
>> @@ -575,34 +573,37 @@ catch_syscall_completer (struct cmd_list_element *cmd,
>>    if (startswith (prefix, "g:") || startswith (prefix, "group:"))
>>      {
>>        /* Perform completion inside 'group:' namespace only.  */
>> -      group_list = get_syscall_group_names (gdbarch);
>> +      group_list.reset (get_syscall_group_names (gdbarch));
>>        if (group_list != NULL)
>> -	complete_on_enum (tracker, group_list, word, word);
>> +	complete_on_enum (tracker, group_list.get (), word, word);
>>      }
>>    else
>>      {
>>        /* Complete with both, syscall names and groups.  */
>> -      syscall_list = get_syscall_names (gdbarch);
>> -      group_list = get_syscall_group_names (gdbarch);
>> +      gdb::unique_xmalloc_ptr<const char *> syscall_list
>> +	(get_syscall_names (gdbarch));
>> +      group_list.reset (get_syscall_group_names (gdbarch));
>> +
>> +      const char **group_ptr = group_list.get ();
>> +
>> +      /* Hold on to strings while we're using them.  */
>> +      std::vector<std::string> holders;
>>  
>>        /* Append "group:" prefix to syscall groups.  */
>> -      for (i = 0; group_list[i] != NULL; i++)
>> +      for (i = 0; group_ptr[i] != NULL; i++)
>>  	{
>> -	  char *prefixed_group = xstrprintf ("group:%s", group_list[i]);
>> +	  std::string prefixed_group = string_printf ("group:%s",
>> +						      group_ptr[i]);
>>  
>> -	  group_list[i] = prefixed_group;
>> -	  make_cleanup (xfree, prefixed_group);
>> +	  group_ptr[i] = prefixed_group.c_str ();
>> +	  holders.push_back (prefixed_group);
>>  	}
>
> Err, I think there's something that doesn't make sense here actually.  We
> record in group_ptr[i] a pointer to the buffer of the temporary std::string,
> that gets deleted when we go out of scope (end of the iteration).  That
> causes this fail:
>
> Running /home/emaisin/src/binutils-gdb/gdb/testsuite/gdb.base/catch-syscall.exp ...
> FAIL: gdb.base/catch-syscall.exp: complete catch syscall group suggests 'group:' prefix (pattern 2)

Hey guys,

Any news on this?  I was about to report this regression, but it seems
you were already dealing with it here.

Thanks,

> By hand, you can do
>
> (gdb) catch syscall g<TAB>
>
> There should be many entries starting with group:, in the failing case there's only
> one.  Presumably because in group_ptr all the pointers point to the same location,
> that contains the last group added.  The completion mechanism then removes duplicates.
>
> It is not enough to assign holders.back ().c_str () (after having pushed the string in
> the vector), because when the vector gets reallocated it can now point to stale memory.
> I think we have to do it in two pass, prepare the vector of std::string, and then get
> pointers to the strings.  Something like this:
>
>
> commit f9cab673480425a130b73394c3a63d256eadf314
> Author: Simon Marchi <simon.marchi@ericsson.com>
> Date:   Fri Oct 20 16:22:40 2017 -0400
>
>     Fix syscall group completion
>
> diff --git a/gdb/break-catch-syscall.c b/gdb/break-catch-syscall.c
> index 82d3e36..095284c 100644
> --- a/gdb/break-catch-syscall.c
> +++ b/gdb/break-catch-syscall.c
> @@ -562,7 +562,6 @@ catch_syscall_completer (struct cmd_list_element *cmd,
>    struct gdbarch *gdbarch = get_current_arch ();
>    gdb::unique_xmalloc_ptr<const char *> group_list;
>    const char *prefix;
> -  int i;
>
>    /* Completion considers ':' to be a word separator, so we use this to
>       verify whether the previous word was a group prefix.  If so, we
> @@ -590,14 +589,11 @@ catch_syscall_completer (struct cmd_list_element *cmd,
>        std::vector<std::string> holders;
>
>        /* Append "group:" prefix to syscall groups.  */
> -      for (i = 0; group_ptr[i] != NULL; i++)
> -	{
> -	  std::string prefixed_group = string_printf ("group:%s",
> -						      group_ptr[i]);
> +      for (int i = 0; group_ptr[i] != NULL; i++)
> +	holders.push_back (string_printf ("group:%s", group_ptr[i]));
>
> -	  group_ptr[i] = prefixed_group.c_str ();
> -	  holders.push_back (std::move (prefixed_group));
> -	}
> +      for (int i = 0; group_ptr[i] != NULL; i++)
> +	group_ptr[i] = holders[i].c_str ();
>
>        if (syscall_list != NULL)
>  	complete_on_enum (tracker, syscall_list.get (), word, word);

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/


  parent reply	other threads:[~2017-12-06 18:58 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-10-18  4:06 [RFA 1/3] Remove cleanups from prepare_execute_command Tom Tromey
2017-10-18  4:06 ` [RFA 2/3] Remove cleanup from call_function_by_hand_dummy Tom Tromey
2017-10-18  4:06 ` [RFA 3/3] Remove cleanups from break-catch-syscall.c Tom Tromey
2017-10-19 21:00   ` Simon Marchi
2017-10-19 21:57     ` Tom Tromey
2017-10-19 21:57       ` Tom Tromey
2017-10-20 20:26   ` Simon Marchi
2017-10-25  4:42     ` Tom Tromey
2017-10-25 10:38       ` Pedro Alves
2017-10-25 14:50         ` Tom Tromey
2017-12-06 18:58     ` Sergio Durigan Junior [this message]
2017-12-06 21:41       ` Simon Marchi

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=87efo7elaa.fsf@redhat.com \
    --to=sergiodj@redhat.com \
    --cc=gdb-patches@sourceware.org \
    --cc=simon.marchi@ericsson.com \
    --cc=tom@tromey.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