From mboxrd@z Thu Jan  1 00:00:00 1970
From: Jerry Van Baren <gerald.vanbaren@ge.com>
Date: Fri, 06 Feb 2009 08:59:53 -0500
Subject: [U-Boot] powerpc flush_cache() roll over bug
In-Reply-To: <9EFEF05A-9413-4DF9-9C0F-D6A2D77A401D@kernel.crashing.org>
References: <9EFEF05A-9413-4DF9-9C0F-D6A2D77A401D@kernel.crashing.org>
Message-ID: <498C4259.2080904@ge.com>
List-Id: <u-boot.lists.denx.de>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: u-boot@lists.denx.de

Kumar Gala wrote:
> My brian stopped working but I figured I'd send my issue and hopefully  
> I'll wake up and someone will have a solution for me.

Talk about sweet dreams!  :-D

> The problem is we can call flush_cache(0xfffff000, 0x1000) which will  
> never exit the loop:
> 
> void flush_cache(ulong start_addr, ulong size)
> {
> #ifndef CONFIG_5xx
>          ulong addr, start;
>          u64 end;
> 
>          start = start_addr & ~(CONFIG_SYS_CACHELINE_SIZE - 1);
>          end = start_addr + size - 1;
> 
>          for (addr = start; addr <= end; addr +=  
> CONFIG_SYS_CACHELINE_SIZE) {
>                  asm volatile("dcbst 0,%0" : : "r" (addr) : "memory");
>                  WATCHDOG_RESET();
>          }
> ...
> 
> since end = 0xffffffff.  There are some other situations (like  
> flush_cache(0, 0)) that cause similar issues, but less concerned about  
> that.
> 
> Any suggestions on how to best to re-write the code to correctly  
> handle flush_cache(0xfffff000, 0x1000) are welcome.

You could do a range check between start and end:

for (addr = start;
      (addr <= end) && (addr >= start);
      addr += CONFIG_SYS_CACHELINE_SIZE) {

Best regards,
gvb