kasan: disable freed user page poisoning with HW tags

Poisoning freed pages protects against kernel use-after-free. The
likelihood of such a bug involving kernel pages is significantly higher
than that for user pages. At the same time, poisoning freed pages can
impose a significant performance cost, which cannot always be justified
for user pages given the lower probability of finding a bug. Therefore,
disable freed user page poisoning when using HW tags. We identify
"user" pages via the flag set GFP_HIGHUSER_MOVABLE, which indicates
a strong likelihood of not being directly accessible to the kernel.

Signed-off-by: Peter Collingbourne <[email protected]>
Reviewed-by: Andrey Konovalov <[email protected]>
Link: https://linux-review.googlesource.com/id/I716846e2de8ef179f44e835770df7e6307be96c9
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Will Deacon <[email protected]>

include/linux/gfp.h

@@ -54,8 +54,9 @@ struct vm_area_struct;
 #define ___GFP_THISNODE		0x200000u
 #define ___GFP_ACCOUNT		0x400000u
 #define ___GFP_ZEROTAGS		0x800000u
+#define ___GFP_SKIP_KASAN_POISON	0x1000000u
 #ifdef CONFIG_LOCKDEP
-#define ___GFP_NOLOCKDEP	0x1000000u
+#define ___GFP_NOLOCKDEP	0x2000000u
 #else
 #define ___GFP_NOLOCKDEP	0
 #endif
@@ -233,17 +234,22 @@ struct vm_area_struct;
  *
  * %__GFP_ZEROTAGS returns a page with zeroed memory tags on success, if
  * __GFP_ZERO is set.
+ *
+ * %__GFP_SKIP_KASAN_POISON returns a page which does not need to be poisoned
+ * on deallocation. Typically used for userspace pages. Currently only has an
+ * effect in HW tags mode.
  */
 #define __GFP_NOWARN	((__force gfp_t)___GFP_NOWARN)
 #define __GFP_COMP	((__force gfp_t)___GFP_COMP)
 #define __GFP_ZERO	((__force gfp_t)___GFP_ZERO)
 #define __GFP_ZEROTAGS	((__force gfp_t)___GFP_ZEROTAGS)
+#define __GFP_SKIP_KASAN_POISON	((__force gfp_t)___GFP_SKIP_KASAN_POISON)
 
 /* Disable lockdep for GFP context tracking */
 #define __GFP_NOLOCKDEP ((__force gfp_t)___GFP_NOLOCKDEP)
 
 /* Room for N __GFP_FOO bits */
-#define __GFP_BITS_SHIFT (24 + IS_ENABLED(CONFIG_LOCKDEP))
+#define __GFP_BITS_SHIFT (25 + IS_ENABLED(CONFIG_LOCKDEP))
 #define __GFP_BITS_MASK ((__force gfp_t)((1 << __GFP_BITS_SHIFT) - 1))
 
 /**
@@ -324,7 +330,8 @@ struct vm_area_struct;
 #define GFP_DMA		__GFP_DMA
 #define GFP_DMA32	__GFP_DMA32
 #define GFP_HIGHUSER	(GFP_USER | __GFP_HIGHMEM)
-#define GFP_HIGHUSER_MOVABLE	(GFP_HIGHUSER | __GFP_MOVABLE)
+#define GFP_HIGHUSER_MOVABLE	(GFP_HIGHUSER | __GFP_MOVABLE | \
+			 __GFP_SKIP_KASAN_POISON)
 #define GFP_TRANSHUGE_LIGHT	((GFP_HIGHUSER_MOVABLE | __GFP_COMP | \
 			 __GFP_NOMEMALLOC | __GFP_NOWARN) & ~__GFP_RECLAIM)
 #define GFP_TRANSHUGE	(GFP_TRANSHUGE_LIGHT | __GFP_DIRECT_RECLAIM)

include/linux/page-flags.h

@@ -137,6 +137,9 @@ enum pageflags {
 #endif
 #ifdef CONFIG_64BIT
 	PG_arch_2,
+#endif
+#ifdef CONFIG_KASAN_HW_TAGS
+	PG_skip_kasan_poison,
 #endif
 	__NR_PAGEFLAGS,
 
@@ -443,6 +446,12 @@ TESTCLEARFLAG(Young, young, PF_ANY)
 PAGEFLAG(Idle, idle, PF_ANY)
 #endif
 
+#ifdef CONFIG_KASAN_HW_TAGS
+PAGEFLAG(SkipKASanPoison, skip_kasan_poison, PF_HEAD)
+#else
+PAGEFLAG_FALSE(SkipKASanPoison)
+#endif
+
 /*
  * PageReported() is used to track reported free pages within the Buddy
  * allocator. We can use the non-atomic version of the test and set

include/trace/events/mmflags.h

@@ -85,6 +85,12 @@
 #define IF_HAVE_PG_ARCH_2(flag,string)
 #endif
 
+#ifdef CONFIG_KASAN_HW_TAGS
+#define IF_HAVE_PG_SKIP_KASAN_POISON(flag,string) ,{1UL << flag, string}
+#else
+#define IF_HAVE_PG_SKIP_KASAN_POISON(flag,string)
+#endif
+
 #define __def_pageflag_names						\
 	{1UL << PG_locked,		"locked"	},		\
 	{1UL << PG_waiters,		"waiters"	},		\
@@ -112,7 +118,8 @@ IF_HAVE_PG_UNCACHED(PG_uncached,	"uncached"	)		\
 IF_HAVE_PG_HWPOISON(PG_hwpoison,	"hwpoison"	)		\
 IF_HAVE_PG_IDLE(PG_young,		"young"		)		\
 IF_HAVE_PG_IDLE(PG_idle,		"idle"		)		\
-IF_HAVE_PG_ARCH_2(PG_arch_2,		"arch_2"	)
+IF_HAVE_PG_ARCH_2(PG_arch_2,		"arch_2"	)		\
+IF_HAVE_PG_SKIP_KASAN_POISON(PG_skip_kasan_poison, "skip_kasan_poison")
 
 #define show_page_flags(flags)						\
 	(flags) ? __print_flags(flags, "|",				\

mm/kasan/hw_tags.c

@@ -246,6 +246,9 @@ void kasan_alloc_pages(struct page *page, unsigned int order, gfp_t flags)
 	 */
 	bool init = !want_init_on_free() && want_init_on_alloc(flags);
 
+	if (flags & __GFP_SKIP_KASAN_POISON)
+		SetPageSkipKASanPoison(page);
+
 	if (flags & __GFP_ZEROTAGS) {
 		int i;
 

mm/page_alloc.c

@@ -394,11 +394,12 @@ static DEFINE_STATIC_KEY_TRUE(deferred_pages);
  * on-demand allocation and then freed again before the deferred pages
  * initialization is done, but this is not likely to happen.
  */
-static inline bool should_skip_kasan_poison(fpi_t fpi_flags)
+static inline bool should_skip_kasan_poison(struct page *page, fpi_t fpi_flags)
 {
 	return static_branch_unlikely(&deferred_pages) ||
 	       (!IS_ENABLED(CONFIG_KASAN_GENERIC) &&
-		(fpi_flags & FPI_SKIP_KASAN_POISON));
+		(fpi_flags & FPI_SKIP_KASAN_POISON)) ||
+	       PageSkipKASanPoison(page);
 }
 
 /* Returns true if the struct page for the pfn is uninitialised */
@@ -449,10 +450,11 @@ defer_init(int nid, unsigned long pfn, unsigned long end_pfn)
 	return false;
 }
 #else
-static inline bool should_skip_kasan_poison(fpi_t fpi_flags)
+static inline bool should_skip_kasan_poison(struct page *page, fpi_t fpi_flags)
 {
 	return (!IS_ENABLED(CONFIG_KASAN_GENERIC) &&
-		(fpi_flags & FPI_SKIP_KASAN_POISON));
+		(fpi_flags & FPI_SKIP_KASAN_POISON)) ||
+	       PageSkipKASanPoison(page);
 }
 
 static inline bool early_page_uninitialised(unsigned long pfn)
@@ -1244,7 +1246,7 @@ static __always_inline bool free_pages_prepare(struct page *page,
 			unsigned int order, bool check_free, fpi_t fpi_flags)
 {
 	int bad = 0;
-	bool skip_kasan_poison = should_skip_kasan_poison(fpi_flags);
+	bool skip_kasan_poison = should_skip_kasan_poison(page, fpi_flags);
 
 	VM_BUG_ON_PAGE(PageTail(page), page);