Hello Michael,
On Fri, 19 Oct 2012 18:29:12 +0200
Michael Holzheu <holzheu at linux.vnet.ibm.com> wrote:
> On s390 the zEC12 machines support 2GB frames. In order to walk page
> tables correctly add support to the page table walker function so it
> detects 2GB frames.
>
> Signed-off-by: Heiko Carstens <heiko.carstens at de.ibm.com>
> Signed-off-by: Michael Holzheu <holzheu at linux.vnet.ibm.com>
Thank you for your work. I have a question.
> ---
> arch/s390x.c | 6 ++++++
> makedumpfile.h | 1 +
> 2 files changed, 7 insertions(+)
>
> --- a/arch/s390x.c
> +++ b/arch/s390x.c
> @@ -247,6 +247,11 @@ vtop_s390x(unsigned long vaddr)
> return NOT_PADDR;
> }
> table = entry & _REGION_ENTRY_ORIGIN;
> + if ((entry & _REGION_ENTRY_LARGE) && (level == 1)) {
> + table &= ~0x7fffffffUL;
> + paddr = table + (vaddr & 0x7ffffffUL);
^^^^^^^^^^^
There is 0x7ffffff(27bit) but the patch for crash uses 0x7fffffff(31bit).
Which is correct ?
Thanks
Atsushi Kumagai
> + return paddr;
> + }
> len = RSG_TABLE_LENGTH(entry);
> level--;
> }
> @@ -257,6 +262,7 @@ vtop_s390x(unsigned long vaddr)
> * if no, then get the page table entry using PX index.
> */
> if (entry & _SEGMENT_ENTRY_LARGE) {
> + table &= ~_PAGE_BYTE_INDEX_MASK;
> paddr = table + (vaddr & _PAGE_BYTE_INDEX_MASK);
> } else {
> entry = _kl_pg_table_deref_s390x(vaddr,
> --- a/makedumpfile.h
> +++ b/makedumpfile.h
> @@ -563,6 +563,7 @@ do { \
> #define _REGION_ENTRY_TYPE_MASK 0x0c /* region table type mask */
> #define _REGION_ENTRY_INVALID 0x20 /* invalid region table entry */
> #define _REGION_ENTRY_LENGTH 0x03 /* region table length */
> +#define _REGION_ENTRY_LARGE 0x400
> #define _REGION_OFFSET_MASK 0x7ffUL /* region/segment table offset mask */
>
> #define RSG_TABLE_LEVEL(x) (((x) & _REGION_ENTRY_TYPE_MASK) >> 2)
>
>
> _______________________________________________
> kexec mailing list
> kexec at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/kexec
