Re: [PATCH] arch/arm64/kernel/traps: Use find_vma_intersection() in traps for setting si_code

* Catalin Marinas [off-list ref] [210412 13:44]:

On Wed, Apr 07, 2021 at 03:11:06PM +0000, Liam Howlett wrote:

quoted

find_vma() will continue to search upwards until the end of the virtual
memory space.  This means the si_code would almost never be set to
SEGV_MAPERR even when the address falls outside of any VMA.  The result
is that the si_code is not reliable as it may or may not be set to the
correct result, depending on where the address falls in the address
space.

Using find_vma_intersection() allows for what is intended by only
returning a VMA if it falls within the range provided, in this case a
window of 1.

Signed-off-by: Liam R. Howlett <redacted>
---
 arch/arm64/kernel/traps.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/kernel/traps.c b/arch/arm64/kernel/traps.c
index a05d34f0e82a..a44007904a64 100644
--- a/arch/arm64/kernel/traps.c
+++ b/arch/arm64/kernel/traps.c

@@ -383,9 +383,10 @@ void force_signal_inject(int signal, int code, unsigned long address, unsigned i
 void arm64_notify_segfault(unsigned long addr)
 {
 	int code;
+	unsigned long ut_addr = untagged_addr(addr);
 
 	mmap_read_lock(current->mm);
-	if (find_vma(current->mm, untagged_addr(addr)) == NULL)
+	if (find_vma_intersection(current->mm, ut_addr, ut_addr + 1) == NULL)
 		code = SEGV_MAPERR;
 	else
 		code = SEGV_ACCERR;

Thank you for taking the time to thoroughly review this patch.

I don't think your change is entirely correct either. We can have a
fault below the vma of a stack (with VM_GROWSDOWN) and
find_vma_intersection() would return NULL but it should be a SEGV_ACCERR
instead.

I'm pretty sure I am missing something.  From what you said above, I
think this means that there can be a user cache fault below the stack
which should notify the user application that they are not allowed to
expand the stack by sending a SIGV_ACCERR in the si_code?  Is this
expected behaviour or am I missing a code path to this function?

Maybe this should employ similar checks as __do_page_fault() (with
expand_stack() and VM_GROWSDOWN).

You mean the code needs to detect endianness and to check if this is an
attempt to expand the stack for both cases?

Thanks,
Liam