mm,kfence: decouple kfence from page granularity mapping judgement

[ Upstream commit bfa7965b33 ]

Kfence only needs its pool to be mapped as page granularity, if it is
inited early. Previous judgement was a bit over protected. From [1], Mark
suggested to "just map the KFENCE region a page granularity". So I
decouple it from judgement and do page granularity mapping for kfence
pool only. Need to be noticed that late init of kfence pool still requires
page granularity mapping.

Page granularity mapping in theory cost more(2M per 1GB) memory on arm64
platform. Like what I've tested on QEMU(emulated 1GB RAM) with
gki_defconfig, also turning off rodata protection:
Before:
[root@liebao ]# cat /proc/meminfo
MemTotal:         999484 kB
After:
[root@liebao ]# cat /proc/meminfo
MemTotal:        1001480 kB

To implement this, also relocate the kfence pool allocation before the
linear mapping setting up, arm64_kfence_alloc_pool is to allocate phys
addr, __kfence_pool is to be set after linear mapping set up.

LINK: [1] https://lore.kernel.org/linux-arm-kernel/Y+IsdrvDNILA59UN@FVFF77S0Q05N/
Suggested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Zhenhua Huang <quic_zhenhuah@quicinc.com>
Reviewed-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Reviewed-by: Marco Elver <elver@google.com>
Link: https://lore.kernel.org/r/1679066974-690-1-git-send-email-quic_zhenhuah@quicinc.com
Signed-off-by: Will Deacon <will@kernel.org>
Stable-dep-of: acfa60dbe0 ("arm64: mm: Fix "rodata=on" when CONFIG_RODATA_FULL_DEFAULT_ENABLED=y")
Signed-off-by: Sasha Levin <sashal@kernel.org>

arch/arm64/include/asm/kfence.h

@@ -19,4 +19,14 @@ static inline bool kfence_protect_page(unsigned long addr, bool protect)
 	return true;
 }
 
+#ifdef CONFIG_KFENCE
+extern bool kfence_early_init;
+static inline bool arm64_kfence_can_set_direct_map(void)
+{
+	return !kfence_early_init;
+}
+#else /* CONFIG_KFENCE */
+static inline bool arm64_kfence_can_set_direct_map(void) { return false; }
+#endif /* CONFIG_KFENCE */
+
 #endif /* __ASM_KFENCE_H */

arch/arm64/mm/mmu.c

@@ -24,6 +24,7 @@
 #include <linux/mm.h>
 #include <linux/vmalloc.h>
 #include <linux/set_memory.h>
+#include <linux/kfence.h>
 
 #include <asm/barrier.h>
 #include <asm/cputype.h>
@@ -38,6 +39,7 @@
 #include <asm/ptdump.h>
 #include <asm/tlbflush.h>
 #include <asm/pgalloc.h>
+#include <asm/kfence.h>
 
 #define NO_BLOCK_MAPPINGS	BIT(0)
 #define NO_CONT_MAPPINGS	BIT(1)
@@ -521,12 +523,67 @@ static int __init enable_crash_mem_map(char *arg)
 }
 early_param("crashkernel", enable_crash_mem_map);
 
+#ifdef CONFIG_KFENCE
+
+bool __ro_after_init kfence_early_init = !!CONFIG_KFENCE_SAMPLE_INTERVAL;
+
+/* early_param() will be parsed before map_mem() below. */
+static int __init parse_kfence_early_init(char *arg)
+{
+	int val;
+
+	if (get_option(&arg, &val))
+		kfence_early_init = !!val;
+	return 0;
+}
+early_param("kfence.sample_interval", parse_kfence_early_init);
+
+static phys_addr_t __init arm64_kfence_alloc_pool(void)
+{
+	phys_addr_t kfence_pool;
+
+	if (!kfence_early_init)
+		return 0;
+
+	kfence_pool = memblock_phys_alloc(KFENCE_POOL_SIZE, PAGE_SIZE);
+	if (!kfence_pool) {
+		pr_err("failed to allocate kfence pool\n");
+		kfence_early_init = false;
+		return 0;
+	}
+
+	/* Temporarily mark as NOMAP. */
+	memblock_mark_nomap(kfence_pool, KFENCE_POOL_SIZE);
+
+	return kfence_pool;
+}
+
+static void __init arm64_kfence_map_pool(phys_addr_t kfence_pool, pgd_t *pgdp)
+{
+	if (!kfence_pool)
+		return;
+
+	/* KFENCE pool needs page-level mapping. */
+	__map_memblock(pgdp, kfence_pool, kfence_pool + KFENCE_POOL_SIZE,
+			pgprot_tagged(PAGE_KERNEL),
+			NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS);
+	memblock_clear_nomap(kfence_pool, KFENCE_POOL_SIZE);
+	__kfence_pool = phys_to_virt(kfence_pool);
+}
+#else /* CONFIG_KFENCE */
+
+static inline phys_addr_t arm64_kfence_alloc_pool(void) { return 0; }
+static inline void arm64_kfence_map_pool(phys_addr_t kfence_pool, pgd_t *pgdp) { }
+
+#endif /* CONFIG_KFENCE */
+
 static void __init map_mem(pgd_t *pgdp)
 {
 	static const u64 direct_map_end = _PAGE_END(VA_BITS_MIN);
 	phys_addr_t kernel_start = __pa_symbol(_stext);
 	phys_addr_t kernel_end = __pa_symbol(__init_begin);
 	phys_addr_t start, end;
+	phys_addr_t early_kfence_pool;
 	int flags = NO_EXEC_MAPPINGS;
 	u64 i;
 
@@ -539,6 +596,8 @@ static void __init map_mem(pgd_t *pgdp)
 	 */
 	BUILD_BUG_ON(pgd_index(direct_map_end - 1) == pgd_index(direct_map_end));
 
+	early_kfence_pool = arm64_kfence_alloc_pool();
+
 	if (can_set_direct_map())
 		flags |= NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS;
 
@@ -604,6 +663,8 @@ static void __init map_mem(pgd_t *pgdp)
 		}
 	}
 #endif
+
+	arm64_kfence_map_pool(early_kfence_pool, pgdp);
 }
 
 void mark_rodata_ro(void)

arch/arm64/mm/pageattr.c

@@ -11,6 +11,7 @@
 #include <asm/cacheflush.h>
 #include <asm/set_memory.h>
 #include <asm/tlbflush.h>
+#include <asm/kfence.h>
 
 struct page_change_data {
 	pgprot_t set_mask;
@@ -22,12 +23,14 @@ bool rodata_full __ro_after_init = IS_ENABLED(CONFIG_RODATA_FULL_DEFAULT_ENABLED
 bool can_set_direct_map(void)
 {
 	/*
-	 * rodata_full, DEBUG_PAGEALLOC and KFENCE require linear map to be
+	 * rodata_full and DEBUG_PAGEALLOC require linear map to be
 	 * mapped at page granularity, so that it is possible to
 	 * protect/unprotect single pages.
+	 *
+	 * KFENCE pool requires page-granular mapping if initialized late.
 	 */
 	return (rodata_enabled && rodata_full) || debug_pagealloc_enabled() ||
-		IS_ENABLED(CONFIG_KFENCE);
+		arm64_kfence_can_set_direct_map();
 }
 
 static int change_page_range(pte_t *ptep, unsigned long addr, void *data)

mm/kfence/core.c

@@ -828,6 +828,10 @@ void __init kfence_alloc_pool(void)
 	if (!kfence_sample_interval)
 		return;
 
+	/* if the pool has already been initialized by arch, skip the below. */
+	if (__kfence_pool)
+		return;
+
 	__kfence_pool = memblock_alloc(KFENCE_POOL_SIZE, PAGE_SIZE);
 
 	if (!__kfence_pool)