* Re: [PATCH] makedumpfile: Petr Tesarik's hugepage filtering
[not found] <E1UcggB-0005gy-4i@eag09.americas.sgi.com>
@ 2013-05-24 1:11 ` Atsushi Kumagai
2013-05-24 2:24 ` HATAYAMA Daisuke
0 siblings, 1 reply; 4+ messages in thread
From: Atsushi Kumagai @ 2013-05-24 1:11 UTC (permalink / raw)
To: cpw; +Cc: kexec
On Wed, 15 May 2013 13:43:59 -0500
Cliff Wickman <cpw@sgi.com> wrote:
> From: Cliff Wickman <cpw@sgi.com>
>
> This fix comes from Petr Tesarik <ptesarik@suse.cz>, and was applied
> to the SuSE version of makedumpfile.
> I don't see it in the mmap version.
> It is important to filtering out a great deal of user memory.
>
> But this from Petr on 5/15:
> > This patch depends on exporting the relevant PG_* flags from the
> > kernel (in VMCOREINFO), and that's where I got stuck, because depending
> > on the number of available bits for the page flags, the kernel either
> > has PG_head and PG_tail, or only PG_compound, so I needed a #ifdef, and
> > the kernel maintainers didn't like the conditional.
> > I can restart the discussion with kernel maintainers and see what I can do
> So I'm not sending to the kexec list until he's ready.
I send this mail to kexec list to open the discussion below.
> His description:
>
> > This filters out pages that have the compound page flag set.
> > The precise position of the flag(s) depends on the kernel version
> > and configuration, so an additional patch is needed for the kernel.
> >
> > It is still possible to filter out huge pages on unpatched kernels
> > if the debuginfo is available. The page flag can also be saved
> > into a generated vmcoreinfo file.
> >
> > To sum it up:
> >
> > a. any kernel with the appropriate patch includes the PG_xxx values in
> > VMCOERINFO, and these are automatically picked up by makedumpfile
> > (planned for a regular SLES11 kernel maint update)
> > b. for unpatched 2.6.26+ kernels, the info can be pre-generated with
> > "makedumpfile -g", or you can pass the "-x" option with the kernel
> > debuginfo file when filtering
> > NOTE: running "makedumpfile" without either "-g" or "-x" will NOT
> > filter out huge pages; a warning message should be issued
> > c. for pre-2.6.26 kernels, makedumpfile will assume the legacy value of
> > 14 for PG_compound (this also covers SLES10 kernels)
> > d. ancient pre-2.6.13 kernels will not be supported; no change here
> >
> > Signed-off-by: Petr Tesarik <ptesarik@suse.cz>
> ---
> makedumpfile.c | 43 ++++++++++++++++++++++++++++++++++++++++++-
> makedumpfile.h | 8 ++++++++
> 2 files changed, 50 insertions(+), 1 deletion(-)
>
> Index: makedumpfile.mmap/makedumpfile.c
> ===================================================================
> --- makedumpfile.mmap.orig/makedumpfile.c
> +++ makedumpfile.mmap/makedumpfile.c
> @@ -173,6 +173,13 @@ is_in_same_page(unsigned long vaddr1, un
> return FALSE;
> }
>
> +/* Test whether flags belong to a huge page */
> +static inline int
> +isHugePage(unsigned long flags)
> +{
> + return flags & info->hugepage_flags;
> +}
> +
> #define BITMAP_SECT_LEN 4096
> static inline int is_dumpable(struct dump_bitmap *, unsigned long long);
> static inline int is_dumpable_cyclic(char *bitmap, unsigned long long);
> @@ -1092,11 +1099,19 @@ get_structure_info(void)
>
> ENUM_NUMBER_INIT(PG_lru, "PG_lru");
> ENUM_NUMBER_INIT(PG_private, "PG_private");
> + ENUM_NUMBER_INIT(PG_head, "PG_head");
> + ENUM_NUMBER_INIT(PG_tail, "PG_tail");
> + ENUM_NUMBER_INIT(PG_compound, "PG_compound");
> ENUM_NUMBER_INIT(PG_swapcache, "PG_swapcache");
> ENUM_NUMBER_INIT(PG_buddy, "PG_buddy");
> ENUM_NUMBER_INIT(PG_slab, "PG_slab");
> ENUM_NUMBER_INIT(PG_hwpoison, "PG_hwpoison");
>
> + if (NUMBER(PG_head) == NOT_FOUND_NUMBER &&
> + NUMBER(PG_compound) == NOT_FOUND_NUMBER)
> + /* Pre-2.6.26 kernels did not have pageflags */
> + NUMBER(PG_compound) = PG_compound_ORIGINAL;
> +
> ENUM_TYPE_SIZE_INIT(pageflags, "pageflags");
>
> TYPEDEF_SIZE_INIT(nodemask_t, "nodemask_t");
> @@ -1537,6 +1552,12 @@ write_vmcoreinfo_data(void)
>
> WRITE_NUMBER("PG_lru", PG_lru);
> WRITE_NUMBER("PG_private", PG_private);
> + if (NUMBER(PG_head) != NOT_FOUND_NUMBER)
> + WRITE_NUMBER("PG_head", PG_head);
> + if (NUMBER(PG_tail) != NOT_FOUND_NUMBER)
> + WRITE_NUMBER("PG_tail", PG_tail);
> + if (NUMBER(PG_compound) != NOT_FOUND_NUMBER)
> + WRITE_NUMBER("PG_compound", PG_compound);
> WRITE_NUMBER("PG_swapcache", PG_swapcache);
> WRITE_NUMBER("PG_buddy", PG_buddy);
> WRITE_NUMBER("PG_slab", PG_slab);
> @@ -1839,6 +1860,9 @@ read_vmcoreinfo(void)
>
> READ_NUMBER("PG_lru", PG_lru);
> READ_NUMBER("PG_private", PG_private);
> + READ_NUMBER("PG_head", PG_head);
> + READ_NUMBER("PG_tail", PG_tail);
> + READ_NUMBER("PG_compound", PG_compound);
> READ_NUMBER("PG_swapcache", PG_swapcache);
> READ_NUMBER("PG_slab", PG_slab);
> READ_NUMBER("PG_buddy", PG_buddy);
> @@ -3079,6 +3103,23 @@ sync_bitmap(struct dump_bitmap *bitmap)
> off_t offset;
> offset = bitmap->offset + BUFSIZE_BITMAP * bitmap->no_block;
>
> + /* Get page flags for compound pages */
> + if (NUMBER(PG_head) != NOT_FOUND_NUMBER) {
> + info->hugepage_flags = 1UL << NUMBER(PG_head);
> + if (NUMBER(PG_tail) != NOT_FOUND_NUMBER)
> + info->hugepage_flags |= 1UL << NUMBER(PG_tail);
> + else
> + /* If PG_tail is not explicitly saved, then assume
> + * that it immediately follows PG_head.
> + */
> + info->hugepage_flags |= info->hugepage_flags << 1;
> + } else if (NUMBER(PG_compound) != NOT_FOUND_NUMBER) {
> + info->hugepage_flags = 1UL << NUMBER(PG_compound);
> + } else if (info->dump_level & DL_EXCLUDE_USER_DATA) {
> + MSG("Compound page bit could not be determined: ");
> + MSG("huge pages will NOT be filtered.\n");
> + }
> +
> /*
> * The bitmap buffer is not dirty, and it is not necessary
> * to write out it.
> @@ -4345,7 +4386,7 @@ __exclude_unnecessary_pages(unsigned lon
> * Exclude the data page of the user process.
> */
> else if ((info->dump_level & DL_EXCLUDE_USER_DATA)
> - && isAnon(mapping)) {
> + && (isAnon(mapping) || isHugePage(flags))) {
> if (clear_bit_on_2nd_bitmap_for_kernel(pfn))
> pfn_user++;
This code will filter out both transparent huge pages and hugetlbfs pages.
But I think hugetlbfs pages shouldn't be filtered out as anonymous pages
because anonymous pages and hugetlbfs pages are certainly different kind
of pages.
Therefore, I plan to add a new dump level to filter out hugetlbfs pages.
My basic idea is like below:
* Exclude the data page of the user process.
*/
else if ((info->dump_level & DL_EXCLUDE_USER_DATA)
&& isAnon(mapping)) {
if (clear_bit_on_2nd_bitmap_for_kernel(pfn))
pfn_user++;
+
+ if (isPageHead(flags)) {
+ int i, nr_pages = 1 << compound_order;
+
+ for (i = 1; i < nr_pages; ++i) {
+ if (clear_bit_on_2nd_bitmap_for_kernel(pfn + i))
+ pfn_user++;
+ }
+ pfn += nr_pages - 2;
+ mem_map += (nr_pages - 1) * SIZE(page);
+ }
+ }
+ /*
+ * Exclude the hugetlbfs pages.
+ */
+ else if ((info->dump_level & DL_EXCLUDE_HUGETLBFS_DATA)
+ && (!isAnon(mapping) && isPageHead(flags))) {
+ int i, nr_pages = 1 << compound_order;
+
+ for (i = 0; i < nr_pages; ++i) {
+ if (clear_bit_on_2nd_bitmap_for_kernel(pfn + i))
+ pfn_hugetlb++;
+ }
+ pfn += nr_pages - 1;
+ mem_map += (nr_pages - 1) * SIZE(page);
}
/*
* Exclude the hwpoison page.
But before that, I would like to hear the opinions about the new dump level
to make sure that the level is worthy or not.
Thanks
Atsushi Kumagai
> }
> Index: makedumpfile.mmap/makedumpfile.h
> ===================================================================
> --- makedumpfile.mmap.orig/makedumpfile.h
> +++ makedumpfile.mmap/makedumpfile.h
> @@ -74,6 +74,7 @@ int get_mem_type(void);
> #define PG_lru_ORIGINAL (5)
> #define PG_slab_ORIGINAL (7)
> #define PG_private_ORIGINAL (11) /* Has something at ->private */
> +#define PG_compound_ORIGINAL (14) /* Is part of a compound page */
> #define PG_swapcache_ORIGINAL (15) /* Swap page: swp_entry_t in private */
>
> #define PAGE_BUDDY_MAPCOUNT_VALUE_v2_6_38 (-2)
> @@ -140,6 +141,9 @@ test_bit(int nr, unsigned long addr)
>
> #define isLRU(flags) test_bit(NUMBER(PG_lru), flags)
> #define isPrivate(flags) test_bit(NUMBER(PG_private), flags)
> +#define isHead(flags) test_bit(NUMBER(PG_head), flags)
> +#define isTail(flags) test_bit(NUMBER(PG_tail), flags)
> +#define isCompound(flags) test_bit(NUMBER(PG_compound), flags)
> #define isSwapCache(flags) test_bit(NUMBER(PG_swapcache), flags)
> #define isHWPOISON(flags) (test_bit(NUMBER(PG_hwpoison), flags) \
> && (NUMBER(PG_hwpoison) != NOT_FOUND_NUMBER))
> @@ -908,6 +912,7 @@ struct DumpInfo {
> unsigned long vmalloc_end;
> unsigned long vmemmap_start;
> unsigned long vmemmap_end;
> + unsigned long hugepage_flags; /* page flags for a compound page */
>
> /*
> * Filter config file containing filter commands to filter out kernel
> @@ -1377,6 +1382,9 @@ struct number_table {
> */
> long PG_lru;
> long PG_private;
> + long PG_head;
> + long PG_tail;
> + long PG_compound;
> long PG_swapcache;
> long PG_buddy;
> long PG_slab;
_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec
^ permalink raw reply [flat|nested] 4+ messages in thread