[Crash-utility] Re: [PATCH v2] kmem: fix the determination for slab page

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Thank you for the update, Qiwu.

On Wed, Sep 11, 2024 at 10:27 AM <devel-request@xxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
Date: Wed, 11 Sep 2024 02:25:40 -0000
From: qiwu.chen@xxxxxxxxxxxxx
Subject: [Crash-utility] [PATCH v2] kmem: fix the determination for
        slab page
To: devel@xxxxxxxxxxxxxxxxxxxxxxxxxxx
Message-ID: <20240911022540.15869.84683@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
Content-Type: text/plain; charset="utf-8"

The determination for a slab page has changed due to changing
PG_slab from a page flag to a page type since kernel commit
46df8e73a4a3.

Before apply this patch:
crash> kmem -s ffff000002aa4100
kmem: address is not allocated in slab subsystem: ffff000002aa4100

After apply this patch:
crash> kmem -s ffff000002aa4100
CACHE             OBJSIZE  ALLOCATED     TOTAL  SLABS  SSIZE  NAME
ffff00000140f900     4096         94       126     18    32k  task_struct
  SLAB              MEMORY            NODE  TOTAL  ALLOCATED  FREE
  fffffdffc00aa800  ffff000002aa0000     0      7          5     2
  FREE / [ALLOCATED]
  [ffff000002aa4100]

Signed-off-by: qiwu.chen <qiwu.chen@xxxxxxxxxxxxx>
---
 defs.h   |  7 ++++++
 memory.c | 65 +++++++++++++++++++++++++++++++++++++++++++++++---------
 2 files changed, 62 insertions(+), 10 deletions(-)

diff --git a/defs.h b/defs.h
index 2231cb6..e2a9278 100644
--- a/defs.h
+++ b/defs.h
@@ -2243,6 +2243,7 @@ struct offset_table {                    /* stash of commonly-used offsets */
        long vmap_node_busy;
        long rb_list_head;
        long file_f_inode;
+       long page_page_type;
 };

 struct size_table {         /* stash of commonly-used sizes */
@@ -2651,6 +2652,7 @@ struct vm_table {                /* kernel VM-related data */
        ulong max_mem_section_nr;
        ulong zero_paddr;
        ulong huge_zero_paddr;
+       uint page_type_base;
 };


Can you try to dump the values of the above two variables from the dump_vm_table() and dump_offset_table()? We can display their values by help -v and help -o options.
 
 #define NODES                       (0x1)
@@ -2684,6 +2686,11 @@ struct vm_table {                /* kernel VM-related data */
 #define SLAB_CPU_CACHE       (0x10000000)
 #define SLAB_ROOT_CACHES     (0x20000000)
 #define USE_VMAP_NODES       (0x40000000)
+/*
+ * The SLAB_PAGEFLAGS flag is introduced to detect the change of
+ * PG_slab's type from a page flag to a page type.
+ */
+#define SLAB_PAGEFLAGS       (0x80000000)

 #define IS_FLATMEM()           (vt->flags & FLATMEM)
 #define IS_DISCONTIGMEM()      (vt->flags & DISCONTIGMEM)
diff --git a/memory.c b/memory.c
index 967a9cf..48ac627 100644
--- a/memory.c
+++ b/memory.c
@@ -351,6 +351,43 @@ static ulong handle_each_vm_area(struct handle_each_vm_area_args *);

 static ulong DISPLAY_DEFAULT;

+/*
+ * Before kernel commit ff202303c398e, the value is defined as a macro, so copy it here;
+ * After this commit, the value is defined as an enum, which can be evaluated at runtime.
+ */
+#define PAGE_TYPE_BASE 0xf0000000
+#define PageType(page_type, flag)                                              \
+       ((page_type & (vt->page_type_base | flag)) == vt->page_type_base)
+
+static void page_type_init(void)
+{
+       if (!enumerator_value("PAGE_TYPE_BASE", (long *)&vt->page_type_base))
+               vt->page_type_base = PAGE_TYPE_BASE;
+}
+
+/*
+ * The PG_slab's type has changed from a page flag to a page type
+ * since kernel commit 46df8e73a4a3.
+ */
+static bool page_slab(ulong page, ulong flags)
+{
+       if (vt->flags & SLAB_PAGEFLAGS) {
+               if ((flags >> vt->PG_slab) & 1)
+                       return TRUE;
+       }
+
+       if (VALID_MEMBER(page_page_type)) {
+               uint page_type;
+
+               readmem(page+OFFSET(page_page_type), KVADDR, &page_type,
+                       sizeof(page_type), "page_type", FAULT_ON_ERROR);
+               if (PageType(page_type, (uint)vt->PG_slab))
+                       return TRUE;
+       }
+
+       return FALSE;
+}
+
 /*
  *  Verify that the sizeof the primitive types are reasonable.
  */
@@ -504,6 +541,10 @@ vm_init(void)
                ANON_MEMBER_OFFSET_INIT(page_compound_head, "page", "compound_head");
        MEMBER_OFFSET_INIT(page_private, "page", "private");
        MEMBER_OFFSET_INIT(page_freelist, "page", "freelist");
+       if (MEMBER_EXISTS("page", "page_type")) {

When initializing the page_page_type, the above "if" check seems to be redundant.
 
+               MEMBER_OFFSET_INIT(page_page_type, "page", "page_type");
+               page_type_init();

Can this page_type_init() be moved to the page_flags_init()?
 
Other changes are fine to me.

Thanks
Lianbo

+       }
 
        MEMBER_OFFSET_INIT(mm_struct_pgd, "mm_struct", "pgd");

@@ -5931,7 +5972,7 @@ dump_mem_map_SPARSEMEM(struct meminfo *mi)
                                        if ((flags >> v22_PG_Slab) & 1)
                                                slabs++;
                                } else if (vt->PG_slab) {
-                                       if ((flags >> vt->PG_slab) & 1)
+                                       if (page_slab(pp, flags))
                                                slabs++;
                                } else {
                                        if ((flags >> v24_PG_slab) & 1)
@@ -6381,7 +6422,7 @@ dump_mem_map(struct meminfo *mi)
                                        if ((flags >> v22_PG_Slab) & 1)
                                                slabs++;
                                } else if (vt->PG_slab) {
-                                       if ((flags >> vt->PG_slab) & 1)
+                                       if (page_slab(pp, flags))
                                                slabs++;
                                } else {
                                        if ((flags >> v24_PG_slab) & 1)
@@ -6775,6 +6816,9 @@ page_flags_init_from_pageflag_names(void)
                vt->pageflags_data[i].name = nameptr;
                vt->pageflags_data[i].mask = mask;

+               if (!strncmp(nameptr, "slab", 4))
+                       vt->flags |= SLAB_PAGEFLAGS;
+
                if (CRASHDEBUG(1)) {
                        fprintf(fp, "  %08lx %s\n",
                                vt->pageflags_data[i].mask,
@@ -6836,7 +6880,8 @@ page_flags_init_from_pageflags_enum(void)
                        strcpy(nameptr, arglist[0] + strlen("PG_"));
                        vt->pageflags_data[p].name = nameptr;
                        vt->pageflags_data[p].mask = 1 << atoi(arglist[2]);
-
+                       if (!strncmp(nameptr, "slab", 4))
+                               vt->flags |= SLAB_PAGEFLAGS;
                        p++;
                }
        } else
@@ -9736,14 +9781,14 @@ vaddr_to_kmem_cache(ulong vaddr, char *buf, int verbose)
                readmem(page+OFFSET(page_flags), KVADDR,
                        &page_flags, sizeof(ulong), "page.flags",
                        FAULT_ON_ERROR);
-               if (!(page_flags & (1 << vt->PG_slab))) {
+               if (!page_slab(page, page_flags)) {
                        if (((vt->flags & KMALLOC_SLUB) || VALID_MEMBER(page_compound_head)) ||
                            ((vt->flags & KMALLOC_COMMON) &&
                            VALID_MEMBER(page_slab) && VALID_MEMBER(page_first_page))) {
                                readmem(compound_head(page)+OFFSET(page_flags), KVADDR,
                                        &page_flags, sizeof(ulong), "page.flags",
                                        FAULT_ON_ERROR);
-                               if (!(page_flags & (1 << vt->PG_slab)))
+                               if (!page_slab(compound_head(page), page_flags))
                                        return NULL;
                        } else
                                return NULL;
@@ -20195,7 +20240,7 @@ char *
 is_slab_page(struct meminfo *si, char *buf)
 {
        int i, cnt;
-       ulong page_slab, page_flags, name;
+       ulong pg_slab, page_flags, name;
         ulong *cache_list;
         char *retval;

@@ -20210,11 +20255,11 @@ is_slab_page(struct meminfo *si, char *buf)
            RETURN_ON_ERROR|QUIET))
                return NULL;

-       if (!(page_flags & (1 << vt->PG_slab)))
+       if (!page_slab(si->spec_addr, page_flags))
                return NULL;

-       if (!readmem(si->spec_addr + OFFSET(page_slab), KVADDR,
-           &page_slab, sizeof(ulong), "page.slab",
+       if (!readmem(si->spec_addr + OFFSET(page_slab), KVADDR,
+           &pg_slab, sizeof(ulong), "page.slab",
            RETURN_ON_ERROR|QUIET))
                return NULL;

@@ -20222,7 +20267,7 @@ is_slab_page(struct meminfo *si, char *buf)
         cnt = get_kmem_cache_list(&cache_list);

        for (i = 0; i < cnt; i++) {
-               if (page_slab == cache_list[i]) {
+               if (pg_slab == cache_list[i]) {
                        if (!readmem(cache_list[i] + OFFSET(kmem_cache_name),
                            KVADDR, &name, sizeof(char *),
                            "kmem_cache.name", QUIET|RETURN_ON_ERROR))
--
2.25.1
--
Crash-utility mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxxxxxx
To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxxxxxx
https://${domain_name}/admin/lists/devel.lists.crash-utility.osci.io/
Contribution Guidelines: https://github.com/crash-utility/crash/wiki

[Index of Archives]     [Fedora Development]     [Fedora Desktop]     [Fedora SELinux]     [Yosemite News]     [KDE Users]     [Fedora Tools]

 

Powered by Linux