[rfa] qSymbol in remote_wait

[rfa] qSymbol in remote_wait

[Daniel Jacobowitz]

Hi Andrew,

As Amit Kale mentioned in December, to support NPTL gdbserver needs to look
up symbols during remote_wait.  The existing qSymbol model assumes that only
at objfile loads (i.e. during td_ta_new) do we need to look up symbols; NPTL
looks up symbols lazily when it needs them, which includes at the creation
of the first child thread.  This patch (which, I know, needs a matching
change for the manual) allows qSymbol: queries as a response to remote_wait,
in much the same way as the file I/O protocol.

Is this OK?  If so I'll write up the documentation change.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

2004-01-31  Daniel Jacobowitz  <drow@mvista.com>

	* remote.c (handle_remote_qsymbol): New function, broken out
	from remote_check_symbols.
	(remote_check_symbols): Call it.
	(remote_wait): Handle qSymbol.
	(remote_async_wait): Likewise.

Index: remote.c
===================================================================
RCS file: /cvs/src/src/gdb/remote.c,v
retrieving revision 1.128
diff -u -p -r1.128 remote.c
--- remote.c	26 Jan 2004 23:07:00 -0000	1.128
+++ remote.c	1 Feb 2004 01:18:27 -0000
@@ -2075,17 +2075,36 @@ init_all_packet_configs (void)
 /* Symbol look-up. */
 
 static void
-remote_check_symbols (struct objfile *objfile)
+handle_remote_qsymbol (char *reply)
 {
   struct remote_state *rs = get_remote_state ();
-  char *msg, *reply, *tmp;
+  char *msg = alloca (rs->remote_packet_size);
+  char *tmp;
   struct minimal_symbol *sym;
   int end;
 
+  tmp = &reply[8];
+  end = hex2bin (tmp, msg, strlen (tmp) / 2);
+  msg[end] = '\0';
+  sym = lookup_minimal_symbol (msg, NULL, NULL);
+  if (sym == NULL)
+    sprintf (msg, "qSymbol::%s", &reply[8]);
+  else
+    sprintf (msg, "qSymbol:%s:%s", 
+	     paddr_nz (SYMBOL_VALUE_ADDRESS (sym)),
+	     &reply[8]);
+  putpkt (msg);
+}
+
+static void
+remote_check_symbols (struct objfile *objfile)
+{
+  struct remote_state *rs = get_remote_state ();
+  char *reply;
+
   if (remote_protocol_qSymbol.support == PACKET_DISABLE)
     return;
 
-  msg   = alloca (rs->remote_packet_size);
   reply = alloca (rs->remote_packet_size);
 
   /* Invite target to request symbol lookups. */
@@ -2096,18 +2115,8 @@ remote_check_symbols (struct objfile *ob
 
   while (strncmp (reply, "qSymbol:", 8) == 0)
     {
-      tmp = &reply[8];
-      end = hex2bin (tmp, msg, strlen (tmp) / 2);
-      msg[end] = '\0';
-      sym = lookup_minimal_symbol (msg, NULL, NULL);
-      if (sym == NULL)
-	sprintf (msg, "qSymbol::%s", &reply[8]);
-      else
-	sprintf (msg, "qSymbol:%s:%s", 
-		 paddr_nz (SYMBOL_VALUE_ADDRESS (sym)),
-		 &reply[8]);
-      putpkt (msg);
-      getpkt (reply, (rs->remote_packet_size), 0);
+      handle_remote_qsymbol (reply);
+      getpkt (reply, rs->remote_packet_size, 0);
     }
 }
 
@@ -2993,6 +3002,11 @@ Packet: '%s'\n",
 	case 'O':		/* Console output */
 	  remote_console_output (buf + 1);
 	  continue;
+	case 'q':
+	  if (strncmp (buf, "qSymbol:", 8) != 0)
+	    goto invalid;
+	  handle_remote_qsymbol (buf);
+	  continue;
 	case '\0':
 	  if (last_sent_signal != TARGET_SIGNAL_0)
 	    {
@@ -3011,6 +3025,7 @@ Packet: '%s'\n",
 	    }
 	  /* else fallthrough */
 	default:
+invalid:
 	  warning ("Invalid remote reply: %s", buf);
 	  continue;
 	}
@@ -3185,6 +3200,11 @@ remote_async_wait (ptid_t ptid, struct t
              still be waiting on the inferior afterwards. */
           status->kind = TARGET_WAITKIND_IGNORE;
           goto got_status;
+	case 'q':
+	  if (strncmp (buf, "qSymbol:", 8) != 0)
+	    goto invalid;
+	  handle_remote_qsymbol (buf);
+	  continue;
 	case '\0':
 	  if (last_sent_signal != TARGET_SIGNAL_0)
 	    {
@@ -3203,6 +3223,7 @@ remote_async_wait (ptid_t ptid, struct t
 	    }
 	  /* else fallthrough */
 	default:
+invalid:
 	  warning ("Invalid remote reply: %s", buf);
 	  continue;
 	}

Re: [rfa] qSymbol in remote_wait

[Daniel Jacobowitz]

On Sat, Jan 31, 2004 at 08:22:30PM -0500, Daniel Jacobowitz wrote:
> Hi Andrew,
> 
> As Amit Kale mentioned in December, to support NPTL gdbserver needs to look
> up symbols during remote_wait.  The existing qSymbol model assumes that only
> at objfile loads (i.e. during td_ta_new) do we need to look up symbols; NPTL
> looks up symbols lazily when it needs them, which includes at the creation
> of the first child thread.  This patch (which, I know, needs a matching
> change for the manual) allows qSymbol: queries as a response to remote_wait,
> in much the same way as the file I/O protocol.
> 
> Is this OK?  If so I'll write up the documentation change.

Ping.

> 2004-01-31  Daniel Jacobowitz  <drow@mvista.com>
> 
> 	* remote.c (handle_remote_qsymbol): New function, broken out
> 	from remote_check_symbols.
> 	(remote_check_symbols): Call it.
> 	(remote_wait): Handle qSymbol.
> 	(remote_async_wait): Likewise.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: [rfa] qSymbol in remote_wait

[Andrew Cagney]

> On Sat, Jan 31, 2004 at 08:22:30PM -0500, Daniel Jacobowitz wrote:
> 
>> Hi Andrew,
>> 
>> As Amit Kale mentioned in December, to support NPTL gdbserver needs to look
>> up symbols during remote_wait.  The existing qSymbol model assumes that only
>> at objfile loads (i.e. during td_ta_new) do we need to look up symbols; NPTL
>> looks up symbols lazily when it needs them, which includes at the creation
>> of the first child thread.  This patch (which, I know, needs a matching
>> change for the manual) allows qSymbol: queries as a response to remote_wait,
>> in much the same way as the file I/O protocol.
>> 
>> Is this OK?  If so I'll write up the documentation change.
> 
> 
> Ping.

Can you please post the proposed protocol change to gdb@.

Andrew

Re: [rfa] qSymbol in remote_wait

[Daniel Jacobowitz]

On Thu, Oct 14, 2004 at 09:16:25PM -0500, Jim Blandy wrote:
> 
> > As Amit Kale mentioned in December, to support NPTL gdbserver needs to
> > look up symbols during remote_wait.  The existing qSymbol model
> > assumes that only at objfile loads (i.e. during td_ta_new) do we need
> > to look up symbols; NPTL looks up symbols lazily when it needs them,
> > which includes at the creation of the first child thread.  This patch
> > (which, I know, needs a matching change for the manual) allows
> > qSymbol: queries as a response to remote_wait, in much the same way as
> > the file I/O protocol.
> >
> > 2004-01-31  Daniel Jacobowitz  <drow@mvista.com>
> > 
> > 	* remote.c (handle_remote_qsymbol): New function, broken out
> > 	from remote_check_symbols.
> > 	(remote_check_symbols): Call it.
> > 	(remote_wait): Handle qSymbol.
> > 	(remote_async_wait): Likewise.
> 
> I was struggling with this too until I noticed the function
> td_symbol_list, which gives you an à priori list of all the symbols
> libthread_db may ever request.  Unless I've missed something, this is
> exactly what's needed to fit ps_pglobal_lookup into the existing
> remote protocol's qSymbol dance: you can look up all the symbols
> libthread_db needs as soon as the inferior loads libpthread.so, and
> then have ps_pglobal_lookup just provide values from a cache.
> 
> I think this has been present in NPTL's libthread_db from the
> beginning; it's present in some versions of LinuxThreads' as well, but
> I don't think it was there initially.

Woah!  I never noticed that was there.  Thank you very, very much, Jim.

Since I assume that all the symbols in td_symbol_list - or at least,
most of them - will be defined, I can just try the first one every time
that GDB offers to look up symbols for me.

-- 
Daniel Jacobowitz

[rfa] qSymbol in remote_wait

[Jim Blandy]

> As Amit Kale mentioned in December, to support NPTL gdbserver needs to
> look up symbols during remote_wait.  The existing qSymbol model
> assumes that only at objfile loads (i.e. during td_ta_new) do we need
> to look up symbols; NPTL looks up symbols lazily when it needs them,
> which includes at the creation of the first child thread.  This patch
> (which, I know, needs a matching change for the manual) allows
> qSymbol: queries as a response to remote_wait, in much the same way as
> the file I/O protocol.
>
> 2004-01-31  Daniel Jacobowitz  <drow@mvista.com>
> 
> 	* remote.c (handle_remote_qsymbol): New function, broken out
> 	from remote_check_symbols.
> 	(remote_check_symbols): Call it.
> 	(remote_wait): Handle qSymbol.
> 	(remote_async_wait): Likewise.

I was struggling with this too until I noticed the function
td_symbol_list, which gives you an à priori list of all the symbols
libthread_db may ever request.  Unless I've missed something, this is
exactly what's needed to fit ps_pglobal_lookup into the existing
remote protocol's qSymbol dance: you can look up all the symbols
libthread_db needs as soon as the inferior loads libpthread.so, and
then have ps_pglobal_lookup just provide values from a cache.

I think this has been present in NPTL's libthread_db from the
beginning; it's present in some versions of LinuxThreads' as well, but
I don't think it was there initially.