Hi,

I sent this patch with the autoload-breakpoints patches because I found 
this issue when I test the autoload-breakpoints.  But I think sent it 
together with a feature patch list is not a good idea.  So I move it out 
for review.

This is bug was found when I when I test autoload-breakpint code.  And I 
found that it affect target-async too.
It can be reproduce:
(gdb) set target-async on
(gdb) start
Temporary breakpoint 1 at 0x4004c8: file 1.c, line 4.
Starting program: /home/teawater/tmp/a.out

Temporary breakpoint 1, main () at 1.c:4
4		sleep (20);
(gdb) disassemble
Dump of assembler code for function main:
    0x00000000004004c4 <+0>:	push   %rbp
    0x00000000004004c5 <+1>:	mov    %rsp,%rbp
=> 0x00000000004004c8 <+4>:	mov    $0x14,%edi
    0x00000000004004cd <+9>:	mov    $0x0,%eax
    0x00000000004004d2 <+14>:	callq  0x4003d0 <sleep@plt>
    0x00000000004004d7 <+19>:	mov    $0x0,%eax
    0x00000000004004dc <+24>:	pop    %rbp
    0x00000000004004dd <+25>:	retq
End of assembler dump.
(gdb) list
1	int
2	main()
3	{
4		sleep (20);
5	
6		return 0;
7	}
8	
(gdb) b 6
Breakpoint 2 at 0x4004d7: file 1.c, line 6.
(gdb) c&
Continuing.
(gdb) d
Delete all breakpoints? (y or n) y
warning: Error removing breakpoint 2
(gdb)
Program received signal SIGTRAP, Trace/breakpoint trap.
0x00000000004004d8 in main () at 1.c:6
6		return 0;
c
Continuing.

Program received signal SIGSEGV, Segmentation fault.
0x00000000004004d8 in main () at 1.c:6
6		return 0;
(gdb) info reg pc
pc: 0x4004d8
(gdb) disassemble main
Dump of assembler code for function main:
    0x00000000004004c4 <+0>:	push   %rbp
    0x00000000004004c5 <+1>:	mov    %rsp,%rbp
    0x00000000004004c8 <+4>:	mov    $0x14,%edi
    0x00000000004004cd <+9>:	mov    $0x0,%eax
    0x00000000004004d2 <+14>:	callq  0x4003d0 <sleep@plt>
    0x00000000004004d7 <+19>:	int3
=> 0x00000000004004d8 <+20>:	add    %al,(%rax)
    0x00000000004004da <+22>:	add    %al,(%rax)
    0x00000000004004dc <+24>:	pop    %rbp
    0x00000000004004dd <+25>:	retq
End of assembler dump.

This because is when GDB got fail when it remove the breakpoint, it give 
up the control of this breakpoint.
There are 2 issues about it:
1. When the GDB stop, this breakpoint is not be removed.
2. If inferior is stoped by this breakpoint, adjust_pc_after_break 
didn't know this stop is beauce the breakpoint, so the pc address will 
not be adjust to the right value.

I add a list called bp_location_remove_fail_chain, when GDB got fail 
with remove a breakpoint, add it to this list.  When 
adjust_pc_after_break check if this address is the breakpint, check this 
list too.  And when gdb remve all breakpoints, try remove breakpint in 
this list.

Thanks,
Hui

2012-04-11  Hui Zhu  <hui_zhu@mentor.com>

	* breakpoint.c (ALL_BP_LOCATION_REMOV_FAIL): New macro.
	(ALL_BP_LOCATION_REMOV_FAIL_SAFE): New macro.
	(bp_location_remove_fail_chain_inserted_here_p): New function.
	(bp_location_remove_fail_chain_insert): New function.
	(bp_location_remove_fail_chain_remove): New function.
	(remove_breakpoints): Call remove_breakpoint with the bp_locations
	inside the ALL_BP_LOCATION_REMOV_FAIL_SAFE.
	(software_breakpoint_inserted_here_p): Call
	bp_location_remove_fail_chain_inserted_here_p.
	(update_global_location_list): Call
	bp_location_remove_fail_chain_insert.
	* breakpoint.h (bp_location_remove_fail_chain_remove): New extern.
	* target.c (target_kill): Call bp_location_remove_fail_chain_remove.
	(target_detach): Ditto.
	(target_disconnect): Ditto.