On Thu, 2006-06-15 at 11:58 +0900, Isaku Yamahata wrote:
> 1 / 7
> _# HG changeset patch
> # User [EMAIL PROTECTED]
> # Node ID ad418fdb1981be2108d84bafbd294a9db9899396
> # Parent d33add81096b057f98fa740ab88d6c17426f8d68
> mpt_table is accessed concurrently by cpus, so it needs volatile
> qualifier
> PATCHNAME: volatile_mpt_table
> diff -r d33add81096b -r ad418fdb1981 xen/arch/ia64/xen/xenmem.c
> --- a/xen/arch/ia64/xen/xenmem.c Wed Jun 14 16:05:45 2006
-0600
> +++ b/xen/arch/ia64/xen/xenmem.c Thu Jun 15 11:33:14 2006
+0900
> @@ -35,7 +35,7 @@ unsigned long max_page;
> /*
> * Set up the page tables.
> */
> -unsigned long *mpt_table;
> +volatile unsigned long *mpt_table;
>
> void
> paging_init (void)
> @@ -140,7 +140,7 @@ create_mpttable_page_table (u64 start, u
> create_mpttable_page_table (u64 start, u64 end, void *arg)
> {
> unsigned long address, start_page, end_page;
> - unsigned long *map_start, *map_end;
> + volatile unsigned long *map_start, *map_end;
> pgd_t *pgd;
> pud_t *pud;
> pmd_t *pmd;
[snip...]
I don't understand why map_start and map_end need to be
volatile here. They only seem to be copying volatile
values for later use so there is no need for them to be
volatile; did I miss something?
Thanks for the clarification.
--
Ciao,
al
----------------------------------------------------------------------
Al Stone Alter Ego:
Open Source and Linux R&D Debian Developer
Hewlett-Packard Company http://www.debian.org
E-mail: [EMAIL PROTECTED] [EMAIL PROTECTED]
----------------------------------------------------------------------
_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@lists.xensource.com
http://lists.xensource.com/xen-ia64-devel
