Re: [PATCH] Fix memory-region overlapping checking

[Yao Qi <yao at codesourcery dot com>]

On Thursday, August 02, 2012 02:10:40 PM Wei-cheng Wang wrote:
> Hi,
> 
>   After adding a memory-region with <hi addr> as max CORE_ADDR+1,
>   no more memory-region can be added. It always complains about
> "overlapping"

Thanks for reporting this problem, and drafting a patch to fix it.

> 
>   For example,
> 
> (gdb) mem 0x50 0x80 ro
> (gdb) mem 0xffffff00 0x100000000 ro

However, the real problem is that we don't do range checking for inputs in 
command 'mem'.  The expected behavior is, on a 32-bit target, when user types 
0x100000000, gdb should emit an error saying that it is out of the range of 
CORE_ADDR, or something like that.

> (gdb) info mem
> Using user-defined memory regions.
> Num Enb Low Addr   High Addr  Attrs
> 1   y   0x00000050 0x00000080 ro nocache
> 2   y   0xffffff00 0x100000000 ro nocache
> (gdb) mem 0x100 0x200 ro
> overlapping memory region
> 
>    When checking whether the new memory-region includes a previous one,
>    it should take care special case, where hi == 0 means max CORE_ADDR+1.

The case you gave works well on my x86_64-linux.

-- 
Yao (éå)