[PATCH v4] makedumpfile: Exclude unnecessary hugepages.
Atsushi Kumagai
kumagai-atsushi at mxc.nes.nec.co.jp
Thu Sep 11 03:04:11 PDT 2014
>On 09/11/14 at 08:52am, Atsushi Kumagai wrote:
>> >Hi Atsushi,
>> >
>> >Since huge pages are included in user pages, I can't think of a way to
>> >make test cases for huge page exclusion. Could you give some suggestions
>> >on this or how did you test it?
>>
>> Before I posted this patch, I tested as below.
>> This idea came from the fact that old makedumpfile can't exclude
>> huge pages except the first page(PG_head).
>>
>> 1. Get the number of hugepages from /proc/meminfo
>> 2. Calculate the number of PG_tail pages
>> 3. Capture the dumpfile without filtering
>> 4. Run makedumpfile and compare the report message between v1.5.6
>> and v1.5.7(rc) to get how many user pages become excludable with
>> this patch.
>> 5. The result of Step2's and Step4's must be same, confirm it.
>>
>> The way above is for THP but you can apply it also for hugetlbfs
>> if you take care of the things that old makedumpfile can't exclude
>> *any* hugetlbfs pages.
>
>But THP is also Anonymous pages, doesn't it do the same for THP between
>1.5.6 and 1.5.7?
Only a PG_head page is marked as an anonymous page, makedumpfile doesn't
distinguish PG_tail pages as anonymous pages. Please see below.
do_huge_pmd_anonymous_page()
+ __do_huge_pmd_anonymous_page()
+ page_add_new_anon_rmap()
+ __page_set_anon_rmap()
Thanks
Atsushi Kumagai
>>
>> I recommend to separate the two cases completely by enabling either
>> THP or hugetlbfs explicitly since it's easier to confirm the results.
>
>For hugeTlbfs, this works, I will try this.
>
>>
>>
>> Thanks
>> Atsushi Kumagai
>>
>> >
>> >Thanks
>> >Baoquan
>> >
>> >
>> >On 08/20/14 at 07:27am, Atsushi Kumagai wrote:
>> >> There are 2 types of hugepages in the kernel, the both should be
>> >> excluded as user pages.
>> >>
>> >> 1. Transparent huge pages (THP)
>> >> All the pages are anonymous pages (at least for now), so we should
>> >> just get how many pages are in the corresponding hugepage.
>> >> It can be gotten from the page->lru.prev of the second page in the
>> >> hugepage.
>> >>
>> >> 2. Hugetlbfs pages
>> >> The pages aren't anonymous pages but kind of user pages, we should
>> >> exclude also these pages in any way.
>> >> Luckily, it's possible to detect these pages by looking the
>> >> page->lru.next of the second page in the hugepage. This idea came
>> >> from the kernel's PageHuge().
>> >> The number of pages can be gotten in the same way as THP.
>> >>
>> >> Changelog:
>> >> v4:
>> >> - Cleaned up according to Petr's and Baoquan's comments.
>> >> v3:
>> >> - Cleaned up according to Petr's comments.
>> >> - Fix misdetection of hugetlb pages.
>> >> v2:
>> >> - Rebased to "Generic multi-page exclusion".
>> >>
>> >> Signed-off-by: Atsushi Kumagai <kumagai-atsushi at mxc.nes.nec.co.jp>
>> >> ---
>> >> makedumpfile.c | 86 ++++++++++++++++++++++++++++++++++++++++++++++++----------
>> >> makedumpfile.h | 7 +++++
>> >> 2 files changed, 78 insertions(+), 15 deletions(-)
>> >>
>> >> diff --git a/makedumpfile.c b/makedumpfile.c
>> >> index 11cd473..b4b6eca 100644
>> >> --- a/makedumpfile.c
>> >> +++ b/makedumpfile.c
>> >> @@ -1180,6 +1180,7 @@ get_symbol_info(void)
>> >> SYMBOL_INIT(vmemmap_list, "vmemmap_list");
>> >> SYMBOL_INIT(mmu_psize_defs, "mmu_psize_defs");
>> >> SYMBOL_INIT(mmu_vmemmap_psize, "mmu_vmemmap_psize");
>> >> + SYMBOL_INIT(free_huge_page, "free_huge_page");
>> >>
>> >> SYMBOL_INIT(cpu_pgd, "cpu_pgd");
>> >> SYMBOL_INIT(demote_segment_4k, "demote_segment_4k");
>> >> @@ -1296,6 +1297,15 @@ get_structure_info(void)
>> >> ENUM_NUMBER_INIT(PG_slab, "PG_slab");
>> >> ENUM_NUMBER_INIT(PG_hwpoison, "PG_hwpoison");
>> >>
>> >> + ENUM_NUMBER_INIT(PG_head_mask, "PG_head_mask");
>> >> + if (NUMBER(PG_head_mask) == NOT_FOUND_NUMBER) {
>> >> + ENUM_NUMBER_INIT(PG_head, "PG_head");
>> >> + if (NUMBER(PG_head) == NOT_FOUND_NUMBER)
>> >> + ENUM_NUMBER_INIT(PG_head, "PG_compound");
>> >> + if (NUMBER(PG_head) != NOT_FOUND_NUMBER)
>> >> + NUMBER(PG_head_mask) = 1UL << NUMBER(PG_head);
>> >> + }
>> >> +
>> >> ENUM_TYPE_SIZE_INIT(pageflags, "pageflags");
>> >>
>> >> TYPEDEF_SIZE_INIT(nodemask_t, "nodemask_t");
>> >> @@ -1530,6 +1540,9 @@ get_value_for_old_linux(void)
>> >> NUMBER(PG_swapcache) = PG_swapcache_ORIGINAL;
>> >> if (NUMBER(PG_slab) == NOT_FOUND_NUMBER)
>> >> NUMBER(PG_slab) = PG_slab_ORIGINAL;
>> >> + if (NUMBER(PG_head_mask) == NOT_FOUND_NUMBER)
>> >> + NUMBER(PG_head_mask) = 1L << PG_compound_ORIGINAL;
>> >> +
>> >> /*
>> >> * The values from here are for free page filtering based on
>> >> * mem_map array. These are minimum effort to cover old
>> >> @@ -1699,6 +1712,7 @@ write_vmcoreinfo_data(void)
>> >> WRITE_SYMBOL("mmu_vmemmap_psize", mmu_vmemmap_psize);
>> >> WRITE_SYMBOL("cpu_pgd", cpu_pgd);
>> >> WRITE_SYMBOL("demote_segment_4k", demote_segment_4k);
>> >> + WRITE_SYMBOL("free_huge_page", free_huge_page);
>> >>
>> >> /*
>> >> * write the structure size of 1st kernel
>> >> @@ -1788,6 +1802,7 @@ write_vmcoreinfo_data(void)
>> >>
>> >> WRITE_NUMBER("PG_lru", PG_lru);
>> >> WRITE_NUMBER("PG_private", PG_private);
>> >> + WRITE_NUMBER("PG_head_mask", PG_head_mask);
>> >> WRITE_NUMBER("PG_swapcache", PG_swapcache);
>> >> WRITE_NUMBER("PG_buddy", PG_buddy);
>> >> WRITE_NUMBER("PG_slab", PG_slab);
>> >> @@ -2040,6 +2055,7 @@ read_vmcoreinfo(void)
>> >> READ_SYMBOL("mmu_vmemmap_psize", mmu_vmemmap_psize);
>> >> READ_SYMBOL("cpu_pgd", cpu_pgd);
>> >> READ_SYMBOL("demote_segment_4k", demote_segment_4k);
>> >> + READ_SYMBOL("free_huge_page", free_huge_page);
>> >>
>> >> READ_STRUCTURE_SIZE("page", page);
>> >> READ_STRUCTURE_SIZE("mem_section", mem_section);
>> >> @@ -2116,6 +2132,7 @@ read_vmcoreinfo(void)
>> >>
>> >> READ_NUMBER("PG_lru", PG_lru);
>> >> READ_NUMBER("PG_private", PG_private);
>> >> + READ_NUMBER("PG_head_mask", PG_head_mask);
>> >> READ_NUMBER("PG_swapcache", PG_swapcache);
>> >> READ_NUMBER("PG_slab", PG_slab);
>> >> READ_NUMBER("PG_buddy", PG_buddy);
>> >> @@ -4643,13 +4660,16 @@ __exclude_unnecessary_pages(unsigned long mem_map,
>> >> mdf_pfn_t pfn_start, mdf_pfn_t pfn_end, struct cycle *cycle)
>> >> {
>> >> mdf_pfn_t pfn;
>> >> + mdf_pfn_t *pfn_counter;
>> >> + mdf_pfn_t nr_pages;
>> >> unsigned long index_pg, pfn_mm;
>> >> unsigned long long maddr;
>> >> mdf_pfn_t pfn_read_start, pfn_read_end;
>> >> unsigned char page_cache[SIZE(page) * PGMM_CACHED];
>> >> unsigned char *pcache;
>> >> - unsigned int _count, _mapcount = 0;
>> >> + unsigned int _count, _mapcount = 0, compound_order = 0;
>> >> unsigned long flags, mapping, private = 0;
>> >> + unsigned long compound_dtor;
>> >>
>> >> /*
>> >> * If a multi-page exclusion is pending, do it first
>> >> @@ -4715,11 +4735,36 @@ __exclude_unnecessary_pages(unsigned long mem_map,
>> >> flags = ULONG(pcache + OFFSET(page.flags));
>> >> _count = UINT(pcache + OFFSET(page._count));
>> >> mapping = ULONG(pcache + OFFSET(page.mapping));
>> >> +
>> >> + if ((index_pg < PGMM_CACHED - 1) &&
>> >> + isCompoundHead(flags)) {
>> >> + compound_order = ULONG(pcache + SIZE(page) + OFFSET(page.lru)
>> >> + + OFFSET(list_head.prev));
>> >> + compound_dtor = ULONG(pcache + SIZE(page) + OFFSET(page.lru)
>> >> + + OFFSET(list_head.next));
>> >> +
>> >> + if ((compound_order >= sizeof(unsigned long) * 8)
>> >> + || ((pfn & ((1UL << compound_order) - 1)) != 0)) {
>> >> + /* Invalid order */
>> >> + compound_order = 0;
>> >> + }
>> >> + } else {
>> >> + /*
>> >> + * The last pfn of the mem_map cache must not be compound page
>> >> + * since all compound pages are aligned to its page order and
>> >> + * PGMM_CACHED is a power of 2.
>> >> + */
>> >> + compound_order = 0;
>> >> + compound_dtor = 0;
>> >> + }
>> >> +
>> >> if (OFFSET(page._mapcount) != NOT_FOUND_STRUCTURE)
>> >> _mapcount = UINT(pcache + OFFSET(page._mapcount));
>> >> if (OFFSET(page.private) != NOT_FOUND_STRUCTURE)
>> >> private = ULONG(pcache + OFFSET(page.private));
>> >>
>> >> + nr_pages = 1 << compound_order;
>> >> + pfn_counter = NULL;
>> >> /*
>> >> * Exclude the free page managed by a buddy
>> >> * Use buddy identification of free pages whether cyclic or not.
>> >> @@ -4727,12 +4772,8 @@ __exclude_unnecessary_pages(unsigned long mem_map,
>> >> if ((info->dump_level & DL_EXCLUDE_FREE)
>> >> && info->page_is_buddy
>> >> && info->page_is_buddy(flags, _mapcount, private, _count)) {
>> >> - int nr_pages = 1 << private;
>> >> -
>> >> - exclude_range(&pfn_free, pfn, pfn + nr_pages, cycle);
>> >> -
>> >> - pfn += nr_pages - 1;
>> >> - mem_map += (nr_pages - 1) * SIZE(page);
>> >> + nr_pages = 1 << private;
>> >> + pfn_counter = &pfn_free;
>> >> }
>> >> /*
>> >> * Exclude the cache page without the private page.
>> >> @@ -4740,8 +4781,7 @@ __exclude_unnecessary_pages(unsigned long mem_map,
>> >> else if ((info->dump_level & DL_EXCLUDE_CACHE)
>> >> && (isLRU(flags) || isSwapCache(flags))
>> >> && !isPrivate(flags) && !isAnon(mapping)) {
>> >> - if (clear_bit_on_2nd_bitmap_for_kernel(pfn, cycle))
>> >> - pfn_cache++;
>> >> + pfn_counter = &pfn_cache;
>> >> }
>> >> /*
>> >> * Exclude the cache page with the private page.
>> >> @@ -4749,23 +4789,39 @@ __exclude_unnecessary_pages(unsigned long mem_map,
>> >> else if ((info->dump_level & DL_EXCLUDE_CACHE_PRI)
>> >> && (isLRU(flags) || isSwapCache(flags))
>> >> && !isAnon(mapping)) {
>> >> - if (clear_bit_on_2nd_bitmap_for_kernel(pfn, cycle))
>> >> - pfn_cache_private++;
>> >> + pfn_counter = &pfn_cache_private;
>> >> }
>> >> /*
>> >> * Exclude the data page of the user process.
>> >> + * - anonymous pages
>> >> + * - hugetlbfs pages
>> >> */
>> >> else if ((info->dump_level & DL_EXCLUDE_USER_DATA)
>> >> - && isAnon(mapping)) {
>> >> - if (clear_bit_on_2nd_bitmap_for_kernel(pfn, cycle))
>> >> - pfn_user++;
>> >> + && (isAnon(mapping) || isHugetlb(compound_dtor))) {
>> >> + pfn_counter = &pfn_user;
>> >> }
>> >> /*
>> >> * Exclude the hwpoison page.
>> >> */
>> >> else if (isHWPOISON(flags)) {
>> >> + pfn_counter = &pfn_hwpoison;
>> >> + }
>> >> + /*
>> >> + * Unexcludable page
>> >> + */
>> >> + else
>> >> + continue;
>> >> +
>> >> + /*
>> >> + * Execute exclusion
>> >> + */
>> >> + if (nr_pages == 1) {
>> >> if (clear_bit_on_2nd_bitmap_for_kernel(pfn, cycle))
>> >> - pfn_hwpoison++;
>> >> + (*pfn_counter)++;
>> >> + } else {
>> >> + exclude_range(pfn_counter, pfn, pfn + nr_pages, cycle);
>> >> + pfn += nr_pages - 1;
>> >> + mem_map += (nr_pages - 1) * SIZE(page);
>> >> }
>> >> }
>> >> return TRUE;
>> >> diff --git a/makedumpfile.h b/makedumpfile.h
>> >> index eba9798..9f90b53 100644
>> >> --- a/makedumpfile.h
>> >> +++ b/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)
>> >> @@ -148,6 +149,9 @@ 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 isCompoundHead(flags) (!!((flags) & NUMBER(PG_head_mask)))
>> >> +#define isHugetlb(dtor) ((SYMBOL(free_huge_page) != NOT_FOUND_SYMBOL) \
>> >> + && (SYMBOL(free_huge_page) == dtor))
>> >> #define isSwapCache(flags) test_bit(NUMBER(PG_swapcache), flags)
>> >> #define isHWPOISON(flags) (test_bit(NUMBER(PG_hwpoison), flags) \
>> >> && (NUMBER(PG_hwpoison) != NOT_FOUND_NUMBER))
>> >> @@ -1218,6 +1222,7 @@ struct symbol_table {
>> >> unsigned long long node_remap_start_vaddr;
>> >> unsigned long long node_remap_end_vaddr;
>> >> unsigned long long node_remap_start_pfn;
>> >> + unsigned long long free_huge_page;
>> >>
>> >> /*
>> >> * for Xen extraction
>> >> @@ -1509,6 +1514,8 @@ struct number_table {
>> >> */
>> >> long PG_lru;
>> >> long PG_private;
>> >> + long PG_head;
>> >> + long PG_head_mask;
>> >> long PG_swapcache;
>> >> long PG_buddy;
>> >> long PG_slab;
>> >> --
>> >> 1.9.0
>> >>
>> >> _______________________________________________
>> >> kexec mailing list
>> >> kexec at lists.infradead.org
>> >> http://lists.infradead.org/mailman/listinfo/kexec
More information about the kexec
mailing list