mm, page_owner: decouple freeing stack trace from debug_pagealloc
authorVlastimil Babka <vbabka@suse.cz>
Mon, 14 Oct 2019 21:11:44 +0000 (14:11 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Mon, 14 Oct 2019 22:04:00 +0000 (15:04 -0700)
Commit 8974558f49a6 ("mm, page_owner, debug_pagealloc: save and dump
freeing stack trace") enhanced page_owner to also store freeing stack
trace, when debug_pagealloc is also enabled.  KASAN would also like to
do this [1] to improve error reports to debug e.g. UAF issues.

Kirill has suggested that the freeing stack trace saving should be also
possible to be enabled separately from KASAN or debug_pagealloc, i.e.
with an extra boot option.  Qian argued that we have enough options
already, and avoiding the extra overhead is not worth the complications
in the case of a debugging option.  Kirill noted that the extra stack
handle in struct page_owner requires 0.1% of memory.

This patch therefore enables free stack saving whenever page_owner is
enabled, regardless of whether debug_pagealloc or KASAN is also enabled.
KASAN kernels booted with page_owner=on will thus benefit from the
improved error reports.

[1] https://bugzilla.kernel.org/show_bug.cgi?id=203967

[vbabka@suse.cz: v3]
Link: http://lkml.kernel.org/r/20191007091808.7096-3-vbabka@suse.cz
Link: http://lkml.kernel.org/r/20190930122916.14969-3-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Qian Cai <cai@lca.pw>
Suggested-by: Dmitry Vyukov <dvyukov@google.com>
Suggested-by: Walter Wu <walter-zh.wu@mediatek.com>
Suggested-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Suggested-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Suggested-by: Qian Cai <cai@lca.pw>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Documentation/dev-tools/kasan.rst
mm/page_owner.c

index b72d07d70239c421e683c9df06411f728e114a9e..525296121d8953add4af9e7144a2eb2d6ba27b1e 100644 (file)
@@ -41,6 +41,9 @@ smaller binary while the latter is 1.1 - 2 times faster.
 Both KASAN modes work with both SLUB and SLAB memory allocators.
 For better bug detection and nicer reporting, enable CONFIG_STACKTRACE.
 
+To augment reports with last allocation and freeing stack of the physical page,
+it is recommended to enable also CONFIG_PAGE_OWNER and boot with page_owner=on.
+
 To disable instrumentation for specific files or directories, add a line
 similar to the following to the respective kernel Makefile:
 
index d3cf5d336ccf4b3dcb5f16181338d1cd03f3953a..de1916ac3e242c0f10ab5706fcbcceaadef5ba7a 100644 (file)
@@ -24,12 +24,10 @@ struct page_owner {
        short last_migrate_reason;
        gfp_t gfp_mask;
        depot_stack_handle_t handle;
-#ifdef CONFIG_DEBUG_PAGEALLOC
        depot_stack_handle_t free_handle;
-#endif
 };
 
-static bool page_owner_disabled = true;
+static bool page_owner_enabled = false;
 DEFINE_STATIC_KEY_FALSE(page_owner_inited);
 
 static depot_stack_handle_t dummy_handle;
@@ -44,7 +42,7 @@ static int __init early_page_owner_param(char *buf)
                return -EINVAL;
 
        if (strcmp(buf, "on") == 0)
-               page_owner_disabled = false;
+               page_owner_enabled = true;
 
        return 0;
 }
@@ -52,10 +50,7 @@ early_param("page_owner", early_page_owner_param);
 
 static bool need_page_owner(void)
 {
-       if (page_owner_disabled)
-               return false;
-
-       return true;
+       return page_owner_enabled;
 }
 
 static __always_inline depot_stack_handle_t create_dummy_stack(void)
@@ -84,7 +79,7 @@ static noinline void register_early_stack(void)
 
 static void init_page_owner(void)
 {
-       if (page_owner_disabled)
+       if (!page_owner_enabled)
                return;
 
        register_dummy_stack();
@@ -148,25 +143,18 @@ void __reset_page_owner(struct page *page, unsigned int order)
 {
        int i;
        struct page_ext *page_ext;
-#ifdef CONFIG_DEBUG_PAGEALLOC
        depot_stack_handle_t handle = 0;
        struct page_owner *page_owner;
 
-       if (debug_pagealloc_enabled())
-               handle = save_stack(GFP_NOWAIT | __GFP_NOWARN);
-#endif
+       handle = save_stack(GFP_NOWAIT | __GFP_NOWARN);
 
        page_ext = lookup_page_ext(page);
        if (unlikely(!page_ext))
                return;
        for (i = 0; i < (1 << order); i++) {
                __clear_bit(PAGE_EXT_OWNER_ACTIVE, &page_ext->flags);
-#ifdef CONFIG_DEBUG_PAGEALLOC
-               if (debug_pagealloc_enabled()) {
-                       page_owner = get_page_owner(page_ext);
-                       page_owner->free_handle = handle;
-               }
-#endif
+               page_owner = get_page_owner(page_ext);
+               page_owner->free_handle = handle;
                page_ext = page_ext_next(page_ext);
        }
 }
@@ -450,7 +438,6 @@ void __dump_page_owner(struct page *page)
                stack_trace_print(entries, nr_entries, 0);
        }
 
-#ifdef CONFIG_DEBUG_PAGEALLOC
        handle = READ_ONCE(page_owner->free_handle);
        if (!handle) {
                pr_alert("page_owner free stack trace missing\n");
@@ -459,7 +446,6 @@ void __dump_page_owner(struct page *page)
                pr_alert("page last free stack trace:\n");
                stack_trace_print(entries, nr_entries, 0);
        }
-#endif
 
        if (page_owner->last_migrate_reason != -1)
                pr_alert("page has been migrated, last migrate reason: %s\n",