[RFA] Avoid internal error in paddress

[RFA] Avoid internal error in paddress

[Pierre Muller]

  I reported an internal error
in gdbtui at startup
concerning the recently modified paddress function:
 
http://sourceware.org/ml/gdb/2009-07/msg00052.html

  I suspect that this internal error could bite
us elsewhere, this is why I propose the following patch:


Pierre Muller
Pascal language support maintainer for GDB

Tested on gcc16 Compiler Farm:
the changes found do not seem to be related to the patch...
< Test Run By muller on Wed Jul  8 09:44:56 2009 (without patch)
---
> Test Run By muller on Wed Jul  8 09:17:05 2009 (with patch)
3512c3512
< PASS: gdb.base/foll-fork.exp: set follow child, hit tbreak
---
> FAIL: gdb.base/foll-fork.exp: (timeout) set follow child, hit tbreak
3524c3524
< PASS: gdb.base/foll-fork.exp: set follow parent, hit tbreak
---
> FAIL: gdb.base/foll-fork.exp: (timeout) set follow parent, hit tbreak

2009-07-08  Pierre Muller  <muller@ics.u-strasbg.fr>

	* utils.c (paddress): Handle case of NULL GDBARCH parameter.

Index: utils.c
===================================================================
RCS file: /cvs/src/src/gdb/utils.c,v
retrieving revision 1.213
diff -u -p -r1.213 utils.c
--- utils.c	2 Jul 2009 17:21:07 -0000	1.213
+++ utils.c	8 Jul 2009 07:04:20 -0000
@@ -2859,7 +2859,12 @@ paddress (struct gdbarch *gdbarch, CORE_
      either zero or sign extended.  Should gdbarch_address_to_pointer or
      some ADDRESS_TO_PRINTABLE() be used to do the conversion?  */
 
-  int addr_bit = gdbarch_addr_bit (gdbarch);
+  int addr_bit;
+  /* Avoid internal error by using CORE_ADDR size if gdbarch is NULL.  */
+  if (gdbarch)
+    addr_bit = gdbarch_addr_bit (gdbarch);
+  else
+    addr_bit = sizeof (CORE_ADDR) * HOST_CHAR_BIT;
 
   if (addr_bit < (sizeof (CORE_ADDR) * HOST_CHAR_BIT))
     addr &= ((CORE_ADDR) 1 << addr_bit) - 1;

Re: [RFA] Avoid internal error in paddress

[Ulrich Weigand]

Hi Pierre,

>   I reported an internal error
> in gdbtui at startup
> concerning the recently modified paddress function:
>  
> http://sourceware.org/ml/gdb/2009-07/msg00052.html
> 
>   I suspect that this internal error could bite
> us elsewhere, this is why I propose the following patch:

I don't really like this, there is no other place in GDB
where a NULL gdbarch is silently accepted.  I think we should
really make sure we know a proper gdbarch ...

I've now checked in this patch:
http://sourceware.org/ml/gdb-patches/2009-07/msg00231.html
to fix the gdbtui crash.

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU Toolchain for Linux on System z and Cell BE
  Ulrich.Weigand@de.ibm.com