ARM: 9350/1: fault: Implement copy_from_kernel_nofault_allowed()

Under PAN emulation when dumping backtraces from things like the
LKDTM EXEC_USERSPACE test[1], a double fault (which would hang a CPU)
would happen because of dump_instr() attempting to read a userspace
address. Make sure copy_from_kernel_nofault() does not attempt this
any more.

Closes: https://lava.sirena.org.uk/scheduler/job/497571
Link: https://lore.kernel.org/all/202401181125.D48DCB4C@keescook/
Reported-by: Mark Brown <broonie@kernel.org>
Suggested-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Tested-by: Mark Brown <broonie@kernel.org>
Cc: Wang Kefeng <wangkefeng.wang@huawei.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Ben Hutchings <ben@decadent.org.uk>
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>

diff --git a/arch/arm/mm/fault.c b/arch/arm/mm/fault.c
index e96fb40b9cc32a64b5a0379e5ee42bd9e38f1aa5..ec16907a551c411627b47b8491b1cc7ae362cc18 100644 (file)
--- a/arch/arm/mm/fault.c
+++ b/arch/arm/mm/fault.c
@@ -25,6 +25,13 @@
 
 #include "fault.h"
 
+bool copy_from_kernel_nofault_allowed(const void *unsafe_src, size_t size)
+{
+       unsigned long addr = (unsigned long)unsafe_src;
+
+       return addr >= TASK_SIZE && ULONG_MAX - addr >= size;
+}
+
 #ifdef CONFIG_MMU
 
 /*