From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gdb-return-39113-listarch-gdb=sources.redhat.com@sourceware.org>
Received: (qmail 13151 invoked by alias); 26 Apr 2011 10:37:56 -0000
Received: (qmail 12990 invoked by uid 22791); 26 Apr 2011 10:37:55 -0000
X-SWARE-Spam-Status: No, hits=-1.9 required=5.0	tests=AWL,BAYES_00,T_RP_MATCHES_RCVD
X-Spam-Check-By: sourceware.org
Received: from mail.codesourcery.com (HELO mail.codesourcery.com) (38.113.113.100)    by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Tue, 26 Apr 2011 10:37:41 +0000
Received: (qmail 16215 invoked from network); 26 Apr 2011 10:37:40 -0000
Received: from unknown (HELO scottsdale.localnet) (pedro@127.0.0.2)  by mail.codesourcery.com with ESMTPA; 26 Apr 2011 10:37:40 -0000
From: Pedro Alves <pedro@codesourcery.com>
To: gdb@sourceware.org
Subject: Re: bogus extra code in gdbserver detach command support?
Date: Tue, 26 Apr 2011 10:37:00 -0000
User-Agent: KMail/1.13.5 (Linux/2.6.35-28-generic; KDE/4.6.2; x86_64; ; )
Cc: Doug Evans <dje@google.com>
References: <20110425215758.AEED22461A8@ruffy.mtv.corp.google.com>
In-Reply-To: <20110425215758.AEED22461A8@ruffy.mtv.corp.google.com>
MIME-Version: 1.0
Content-Type: Text/Plain;  charset="iso-8859-15"
Content-Transfer-Encoding: 7bit
Message-Id: <201104261137.45851.pedro@codesourcery.com>
X-IsSubscribed: yes
Mailing-List: contact gdb-help@sourceware.org; run by ezmlm
Precedence: bulk
List-Id: <gdb.sourceware.org>
List-Subscribe: <mailto:gdb-subscribe@sourceware.org>
List-Archive: <http://sourceware.org/ml/gdb/>
List-Post: <mailto:gdb@sourceware.org>
List-Help: <mailto:gdb-help@sourceware.org>, <http://sourceware.org/ml/#faqs>
Sender: gdb-owner@sourceware.org
X-SW-Source: 2011-04/txt/msg00147.txt.bz2

On Monday 25 April 2011 22:57:58, Doug Evans wrote:
> Hi.
> 
> Under what circumstances does this code in server.c actually do something?
> 
> 	      /* If we are attached, then we can exit.  Otherwise, we
> 		 need to hang around doing nothing, until the child is
> 		 gone.  */
> 	      for_each_inferior (&all_processes,
> 				 join_inferiors_callback);
> 
> This is for the 'D' packet.
> AIUI, this code will already have emptied all_processes:
> 
>       if (detach_inferior (pid) != 0)
> 	write_enn (own_buf);

Hmm, looks like I broke this with the multi-process changes.  If you
look back at, e.g., gdb 6.8, the join backend implementations would work
because they only relied on the signal_pid global, not any process
structure:

static void
linux_join (void)
{
  extern unsigned long signal_pid;
  int status, ret;

  do {
    ret = waitpid (signal_pid, &status, 0);
    if (WIFEXITED (status) || WIFSIGNALED (status))
      break;
  } while (ret != -1 || errno != ECHILD);
}

/* Wait for inferiors to end.  */
static void
win32_join (void)
{
  extern unsigned long signal_pid;

  HANDLE h = OpenProcess (PROCESS_ALL_ACCESS, FALSE, signal_pid);
  if (h != NULL)
    {
      WaitForSingleObject (h, INFINITE);
      CloseHandle (h);
    }
}


Looks like removing the:

  struct process_info *process;

  process = find_process_pid (pid);
  if (process == NULL)
    return;

bit from linux_join, and call join_inferior
directly on the detachee's pid from server.c
would restore the old behavior.

> 
> Or can multiprocess debugging be used without the extended remote protocol?

Nope.

-- 
Pedro Alves