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: [PATCH] Fix memory-region overlapping checking


Given the example you posted,

> Let me correct my previous example. It should be
> (gdb) mem 0x50 0x80 ro
> (gdb) mem 0xffffff00 0 ro
> (gdb) mem 0x100 0x200 ro
> overlapping memory region
> 

It is a bug.


On Thursday, August 02, 2012 02:10:40 PM Wei-cheng Wang wrote:
> Wei-cheng
> 
> 2012-08-02  Wei-cheng Wang  <cole945@gmail.com>
> 
>         * memattr.c (create_mem_region): Fix memory-region overlapping
> checking in special case.
> 
> diff --git a/gdb/memattr.c b/gdb/memattr.c
> --- a/gdb/memattr.c
> +++ b/gdb/memattr.c
> @@ -207,7 +207,7 @@ create_mem_region (CORE_ADDR lo, CORE_ADDR hi,
> 
>        if ((lo >= n->lo && (lo < n->hi || n->hi == 0))
>           || (hi > n->lo && (hi <= n->hi || n->hi == 0))
> -         || (lo <= n->lo && (hi >= n->hi || hi == 0)))
> +         || (lo <= n->lo && ((hi >= n->hi && n->hi != 0) || hi == 0)))
>         {
>           printf_unfiltered (_("overlapping memory region\n"));
>           return;

I read your patch, and draw some charts on paper to show the relationship of 
these four variables here for overlapping.

This line is to check the overlapping like,

  lo  n->lo  n->hi  hi

However, without your fix, the following case is treated as overlapping by 
mistake,

  lo  hi  n->lo  n->hi (0)

Your patch looks correct to me, however I am not the people to approve it.

This line of code was written in 2002, so a 10-year-old bug is fixed! :)

-- 
Yao (éå)


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