[Crash-utility] Re: [PATCH v7 1/3] add folio_order function

On Thu, May 14, 2026 at 6:33 PM Huang Shijie <huangsj(a)hygon.cn&gt; wrote: > > The folio_order() has three versions: > 1.) In verion 1, folio_order() uses page[1].compound_order to > get the folio order. We use "folio_batch" as the indicator. > > Please refer to patch set: > https://lore.kernel.org/all/20211208042256.1923824-1-willy@infradead.org/ > > 2.) In version 2, the following patch introduces _folio_order: > c3a15bff46cb5149 "mm: reimplement folio_order() and folio_nr_pages()" > We use "_folio_order" as the indicator. > > Please refer to patch set: > https://lore.kernel.org/all/20220808193430.3378317-1-willy@infradead.org/ > > 3.) In version 3, the following patch replaces the _folio_order with _flags_1: > ebc1baf5c9b46c22 "mm: free up a word in the first tail page" > folio_order() uses _flags_1 to get the folio order. > > We use "free_huge_folio" as the indicator. > Please refer to patch set: > https://lore.kernel.org/linux-mm/20230816151201.3655946-1-willy@infradead... > (Note, we do not use the struct modifications within the patch set as the indicator, > since they have been removed in later kernel.) > > This patch will be used in later patches. > > Signed-off-by: Huang Shijie <huangsj(a)hygon.cn&gt; > --- > defs.h | 8 ++++++ > memory.c | 80 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ > symbols.c | 6 +++++ > 3 files changed, 94 insertions(+) > > diff --git a/defs.h b/defs.h > index 89044b1..bd996ef 100644 > --- a/defs.h > +++ b/defs.h > @@ -2291,6 +2291,9 @@ struct offset_table { /* stash of commonly-used offsets */ > long hrtimer_clock_base_index; > long klp_patch_list; > long tk_data_timekeeper; > + long page_compound_order; > + long folio__folio_order; > + long folio__flags_1; > }; > > struct size_table { /* stash of commonly-used sizes */ > @@ -2470,6 +2473,9 @@ struct size_table { /* stash of commonly-used sizes */ > long cpumask_t; > long task_struct_exit_state; > long bpf_ringbuf_map; > + long page_compound_order; > + long folio__folio_order; > + long folio__flags_1; > }; > > struct array_table { > @@ -5998,6 +6004,8 @@ ulong do_xarray(ulong, int, struct list_pair *); > #define XARRAY_TAG_MASK (3UL) > #define XARRAY_TAG_INTERNAL (2UL) > > +int folio_order(ulong folio); > + > int file_dump(ulong, ulong, ulong, int, int); > #define DUMP_FULL_NAME 0x1 > #define DUMP_INODE_ONLY 0x2 > diff --git a/memory.c b/memory.c > index 17423a5..3deae36 100644 > --- a/memory.c > +++ b/memory.c > @@ -410,6 +410,10 @@ mem_init(void) > DISPLAY_DEFAULT = (sizeof(long) == 8) ? DISPLAY_64 : DISPLAY_32; > } > > +#define FOLIO_ORDER_V1 1 > +#define FOLIO_ORDER_V2 2 > +#define FOLIO_ORDER_V3 3 > +static int folio_order_version; > > /* > * Stash a few popular offsets and some basic kernel virtual memory > @@ -547,6 +551,38 @@ vm_init(void) > MEMBER_OFFSET_INIT(page_freelist, "page", "freelist"); > MEMBER_OFFSET_INIT(page_page_type, "page", "page_type"); > > + /* > + * The "folio_batch" was introduced at patch set: > + * https://lore.kernel.org/all/20211208042256.1923824-1-willy@infradead.org/ > + * Use it as the indicator for folio_order() version 1. > + */ > + if (STRUCT_EXISTS("folio_batch")) > + folio_order_version = FOLIO_ORDER_V1; > + > + MEMBER_OFFSET_INIT(page_compound_order, "page", "compound_order"); > + MEMBER_SIZE_INIT(page_compound_order, "page", "compound_order"); > + > + /* > + * The "_folio_order" was introduced at patch set: > + * https://lore.kernel.org/all/20220808193430.3378317-1-willy@infradead.org/ > + * Use it as the indicator for folio_order() version 2. > + */ > + MEMBER_SIZE_INIT(folio__folio_order, "folio", "_folio_order"); > + MEMBER_OFFSET_INIT(folio__folio_order, "folio", "_folio_order"); > + if (VALID_MEMBER(folio__folio_order)) > + folio_order_version = FOLIO_ORDER_V2; > + > + /* > + * The "free_huge_folio()" was introduced at patch set: > + * https://lore.kernel.org/linux-mm/20230816151201.3655946-1-willy@infradead... > + * Use it as the indicator for folio_order() version 3. > + */ > + MEMBER_OFFSET_INIT(folio__flags_1, "folio", "_flags_1"); > + MEMBER_SIZE_INIT(folio__flags_1, "folio", "_flags_1"); > + if (kernel_symbol_exists("free_huge_folio") || > + THIS_KERNEL_VERSION >= LINUX(6,6,0)) I think the "free_huge_folio" check is enough, THIS_KERNEL_VERSION >= LINUX(6,6,0) is not needed. Again, we should avoid the macro as much as possible.