This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [RFA] windows: do not crash if inferior


On Wed, Jan 20, 2010 at 11:15:49AM -0500, Christopher Faylor wrote:
>On Tue, Jan 19, 2010 at 01:22:27PM +0400, Joel Brobecker wrote:
>>> 2010-01-13  gingold  <gingold@adacore.com>
>>> 
>>>         * windows-nat.c (do_initial_windows_stuff): Call error() if the
>>>         inferior doesn't exist anymore.
>>
>>Just some thoughts, as Chris usually reviews Windows patches...
>>
>>I don't see a "generic" way of dealing with this situation.  So the type
>>of approach you took (returning early from do_initial_windows_stuff)
>>seems to be the only approach I can see. I was initially a little
>>reluctant about throwing an error: As far as I can tell from the code,
>>the debugger should have already printed an error message such as
>>"Inferior exited with code ..." (is that correct?) - and so an extra
>>"inferior exited early" message could be considered superfluous. However,
>>if you do not error-out now, core GDB will assume that target_create_inferior
>>succeeded and thus possibly do something unexpected as well.
>>
>>Bottom line - I cannot propose a better approach short of revamping
>>a bit the target_create_inferior routine to add error-handling,
>>I think the patch is fine.
>>
>>I would suggest a different wording, to explain that we were trying
>>to run the program when the error occured.
>>
>>    error (_("cannot run program, inferior exited prematurely during startup"));
>
>Does inferior_thread really need the assert()?  If not, we could jsut test
>tp for NULL.
>
>Otherwise, I agree with Joel's assessment.

Actually, how about something like this instead?  I used the same wording as fork-child.c
after seeing Pedro's note.

cgf

Index: windows-nat.c
===================================================================
RCS file: /cvs/src/src/gdb/windows-nat.c,v
retrieving revision 1.196.4.1
diff -d -u -r1.196.4.1 windows-nat.c
--- windows-nat.c	30 Sep 2009 07:40:10 -0000	1.196.4.1
+++ windows-nat.c	20 Jan 2010 17:35:47 -0000
@@ -41,6 +41,7 @@
 #include <psapi.h>
 #ifdef __CYGWIN__
 #include <sys/cygwin.h>
+#include <cygwin/version.h>
 #endif
 #include <signal.h>
 
@@ -123,6 +128,8 @@
 static uintptr_t dr[8];
 static int debug_registers_changed;
 static int debug_registers_used;
+
+static int windows_initialization_done;
 #define DR6_CLEAR_VALUE 0xffff0ff0
 
 /* The string sent by cygwin when it processes a signal.
@@ -1399,6 +1407,8 @@
 		     (unsigned) current_event.dwProcessId,
 		     (unsigned) current_event.dwThreadId,
 		     "EXIT_PROCESS_DEBUG_EVENT"));
+      if (!windows_initialization_done)
+	error (_("During startup program exited with code 0x%x."), (unsigned int) current_event.u.ExitProcess.dwExitCode);
       if (saw_create != 1)
 	break;
       ourstatus->kind = TARGET_WAITKIND_EXITED;
@@ -1597,6 +1607,7 @@
   terminal_init_inferior_with_pgrp (pid);
   target_terminal_inferior ();
 
+  windows_initialization_done = 0;
   inf->stop_soon = STOP_QUIETLY;
   while (1)
     {
@@ -1609,6 +1620,7 @@
 	break;
     }
 
+  windows_initialization_done = 1;
   inf->stop_soon = NO_STOP_QUIETLY;
   stop_after_trap = 0;
   return;


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]