[PATCH] Mention some problems with using GDB on FreeBSD

[PATCH] Mention some problems with using GDB on FreeBSD

[Mark Kettenis]

This patch makes PROBLEMS mention two problems with using GDB on
FreeBSD.  I've tried to find a workaround for these problems, but up
until now I've not been successfull.  I submitted patches to solve the
problems to the FreeBSD people but they have not been integrated yet.

Committed.

Mark


Index: ChangeLog
from  Mark Kettenis  <kettenis@gnu.org>

	* PROBLEMS: Refer to GDB 5.3 instead of 5.2.  Mention FreeBSD
	problems.

Index: PROBLEMS
===================================================================
RCS file: /cvs/src/src/gdb/PROBLEMS,v
retrieving revision 1.6
diff -u -p -r1.6 PROBLEMS
--- PROBLEMS 24 May 2002 02:56:56 -0000 1.6
+++ PROBLEMS 23 Aug 2002 22:59:07 -0000
@@ -1,7 +1,24 @@
 
-			Known problems in GDB 5.2
+			Known problems in GDB 5.3
 
 		See also: http://www.gnu.org/software/gdb/bugs/
+
+
+*-*-freebsd*
+---------------
+
+Due to a kernel bug (kern/35175), detaching from an attached process
+will very likely cause the process to be stop or die with a Trace/BPT
+trap.
+
+
+i386-*-freebsd[34]*
+-------------------
+
+There is a bug (bin/41671) in FreeBSD's gcc that causes it to emit bad
+debug information when using the stabs format (which is the default).
+In earlier versions of GDB the effects were rather limited, but
+starting with GDB 5.3 the influence is much more prominent.
 
 
 hppa2.0-hp-hpux10.20

Re: [PATCH] Mention some problems with using GDB on FreeBSD

[Eli Zaretskii]

> Date: Sat, 24 Aug 2002 02:35:30 +0200 (CEST)
> From: Mark Kettenis <kettenis@chello.nl>
> 
> This patch makes PROBLEMS mention two problems with using GDB on
> FreeBSD.

Would it be possible to give users some idea about the phenomena they
will see as the result of these problems?  As the text goes now, I
don't think a user could decide whether some problem they see is due
to this bug or not.

Also, it might be a good idea to tell them how to work around the bug
(I'm guessing "gcc -gdwarf-2" or some such?).

Re: [PATCH] Mention some problems with using GDB on FreeBSD

[Mark Kettenis]

   Date: Sat, 24 Aug 2002 12:36:32 +0300
   From: "Eli Zaretskii" <eliz@is.elta.co.il>

   > Date: Sat, 24 Aug 2002 02:35:30 +0200 (CEST)
   > From: Mark Kettenis <kettenis@chello.nl>
   > 
   > This patch makes PROBLEMS mention two problems with using GDB on
   > FreeBSD.

   Would it be possible to give users some idea about the phenomena they
   will see as the result of these problems?  As the text goes now, I
   don't think a user could decide whether some problem they see is due
   to this bug or not.

   Also, it might be a good idea to tell them how to work around the bug
   (I'm guessing "gcc -gdwarf-2" or some such?).

Thanks for the suggestion.  I checked in the attached patch.

Index: ChangeLog
from  Mark Kettenis  <kettenis@gnu.org>
	* PROBLEMS: Clarify problems with FreeBSD's compiler and suggest
	workaround.

Index: PROBLEMS
===================================================================
RCS file: /cvs/src/src/gdb/PROBLEMS,v
retrieving revision 1.7
diff -u -p -r1.7 PROBLEMS
--- PROBLEMS 23 Aug 2002 23:05:39 -0000 1.7
+++ PROBLEMS 24 Aug 2002 14:27:44 -0000
@@ -17,8 +17,11 @@ i386-*-freebsd[34]*
 
 There is a bug (bin/41671) in FreeBSD's gcc that causes it to emit bad
 debug information when using the stabs format (which is the default).
-In earlier versions of GDB the effects were rather limited, but
-starting with GDB 5.3 the influence is much more prominent.
+As a result GDB tends to place breakpoints on functions before the
+function prologue, and information about function parameters and local
+variables is lost.  In earlier versions of GDB the effects were rather
+limited, but starting with GDB 5.3 the influence is much more
+prominent.  As a workaround, compile your code with -gdwarf-2.
 
 
 hppa2.0-hp-hpux10.20

Re: [PATCH] Mention some problems with using GDB on FreeBSD

[Eli Zaretskii]

> Date: Sat, 24 Aug 2002 16:28:21 +0200 (CEST)
> From: Mark Kettenis <kettenis@chello.nl>
> 
> +As a result GDB tends to place breakpoints on functions before the
> +function prologue, and information about function parameters and local
> +variables is lost.  In earlier versions of GDB the effects were rather
> +limited, but starting with GDB 5.3 the influence is much more
> +prominent.  As a workaround, compile your code with -gdwarf-2.

Perfect; thanks!