From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 9337 invoked by alias); 5 May 2003 20:48:12 -0000 Mailing-List: contact gdb-patches-help@sources.redhat.com; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sources.redhat.com Received: (qmail 9281 invoked from network); 5 May 2003 20:48:11 -0000 Received: from unknown (HELO touchme.toronto.redhat.com) (207.219.125.105) by sources.redhat.com with SMTP; 5 May 2003 20:48:11 -0000 Received: from redhat.com (toocool.toronto.redhat.com [172.16.14.72]) by touchme.toronto.redhat.com (Postfix) with ESMTP id CFC2C800030 for ; Mon, 5 May 2003 16:48:10 -0400 (EDT) Message-ID: <3EB6CE0A.4070409@redhat.com> Date: Mon, 05 May 2003 20:48:00 -0000 From: "J. Johnston" Organization: Red Hat Inc. User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.1) Gecko/20020823 Netscape/7.0 X-Accept-Language: en-us, en MIME-Version: 1.0 To: gdb-patches@sources.redhat.com Subject: RFC: patch to refresh prev_pc Content-Type: multipart/mixed; boundary="------------040109060003040503040801" X-SW-Source: 2003-05/txt/msg00073.txt.bz2 This is a multi-part message in MIME format. --------------040109060003040503040801 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Content-length: 1489 The following patch solves a problem on the ia64. The problem exists because of a generic problem to reset the prev_pc value after an inferior function call or after a return command. Because the value is not properly set, the line number used to initialize the ecs is incorrect. On the ia64 this causes a problem because there are extraneous linetable entries generated by the compiler that are within the line (i.e. they don't change the line number). When we apply "next" logic which uses the ecs line number, we end up stopping at the first line table entry past our start position. This often ends up being just a few insns farther in the same line. A specific example of this problem is the next to 1237 test inside call-ar-st.exp. An inferior call is made on line 1236 and upon return we issue a next. I discussed this topic on the gdb forum and a number of attempts were made to ensure the prev_pc value was up to date in init_execution_control_state() in infrun.c. Those attempts failed because the inferior was not guaranteed to be stopped and so we weren't guaranteed that a ptrace to fetch the pc would work. This patch attempts to refresh the prev_pc value just before resuming in proceed(). It works for the ia64 problems cited above and also I have tested it on the x86. Is this patch ok? -- Jeff J. 2003-05-05 Jeff Johnston * infrun.c (prev_pc): Move declaration ahead of proceed(). (proceed): Refresh prev_pc value before resuming. --------------040109060003040503040801 Content-Type: text/plain; name="prev_pc.patch" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="prev_pc.patch" Content-length: 1472 Index: infrun.c =================================================================== RCS file: /cvs/src/src/gdb/infrun.c,v retrieving revision 1.108 diff -u -p -r1.108 infrun.c --- infrun.c 5 May 2003 00:27:07 -0000 1.108 +++ infrun.c 5 May 2003 20:34:16 -0000 @@ -667,6 +667,12 @@ clear_proceed_status (void) bpstat_clear (&stop_bpstat); } + +/* Record the pc of the program the last time it stopped. This is + just used internally by wait_for_inferior, but need to be preserved + over calls to it and cleared when the inferior is started. */ +static CORE_ADDR prev_pc; + /* Basic routine for continuing the program in various fashions. ADDR is the address to resume at, or -1 for resume where stopped. @@ -772,6 +778,10 @@ proceed (CORE_ADDR addr, enum target_sig inferior. */ gdb_flush (gdb_stdout); + /* Refresh prev_pc value which may have been altered by an inferior + function call or a return command. */ + prev_pc = read_pc (); + /* Resume inferior. */ resume (oneproc || step || bpstat_should_step (), stop_signal); @@ -785,11 +795,6 @@ proceed (CORE_ADDR addr, enum target_sig normal_stop (); } } - -/* Record the pc of the program the last time it stopped. This is - just used internally by wait_for_inferior, but need to be preserved - over calls to it and cleared when the inferior is started. */ -static CORE_ADDR prev_pc; /* Start remote-debugging of a machine over a serial link. */ --------------040109060003040503040801--