[PATCH 2/2] riscv: Fix CONFIG_KASAN_STACK build

Alexandre Ghiti alexandre.ghiti at canonical.com
Tue Oct 26 21:58:43 PDT 2021


Now that CONFIG_KASAN_SHADOW_OFFSET is correctly defined, the boot
hung while populating the shadow memory right after the call to
kasan_populate_early_shadow: when calling this function, all the shadow
memory is already populated with kasan_early_shadow_pte which has
PAGE_KERNEL protection.

kasan_populate_early_shadow write-protects the mapping of the range
of addresses passed in argument in zero_pte_populate, which actually
write-protects all the shadow memory mapping since
kasan_early_shadow_pte is used for all the shadow memory at this point.
And then when using memblock API to populate the shadow memory, the first
write access to the kernel stack triggers a trap.

We already manually populate all the shadow memory in kasan_early_init
and we write-protect kasan_early_shadow_pte at the end of kasan_init
which makes the call to kasan_populate_early_shadow superfluous so
we can remove it.

Signed-off-by: Alexandre Ghiti <alexandre.ghiti at canonical.com>
---
 arch/riscv/mm/kasan_init.c | 7 -------
 1 file changed, 7 deletions(-)

diff --git a/arch/riscv/mm/kasan_init.c b/arch/riscv/mm/kasan_init.c
index 8175e98b9073..8df937902630 100644
--- a/arch/riscv/mm/kasan_init.c
+++ b/arch/riscv/mm/kasan_init.c
@@ -175,13 +175,6 @@ void __init kasan_init(void)
 	phys_addr_t p_start, p_end;
 	u64 i;
 
-	/*
-	 * Populate all kernel virtual address space with kasan_early_shadow_page
-	 * except for the linear mapping and the modules/kernel/BPF mapping.
-	 */
-	kasan_populate_early_shadow((void *)KASAN_SHADOW_START,
-				    (void *)kasan_mem_to_shadow((void *)
-								VMEMMAP_END));
 	if (IS_ENABLED(CONFIG_KASAN_VMALLOC))
 		kasan_shallow_populate(
 			(void *)kasan_mem_to_shadow((void *)VMALLOC_START),
-- 
2.30.2




More information about the linux-riscv mailing list