Crash on armv7-a using KASAN

[Linus Walleij]

On Tue, Oct 15, 2024 at 7:28 PM Ard Biesheuvel <ardb at kernel.org> wrote:

> > From a superficial look, it sounds like it should be possible to extend
> > that to also handle the KASAN shadow of the vmalloc area (which
> > __check_vmalloc_seq() currently doesn't copy), but I'm not sure of
> > exactly when we initialise the shadow for a vmalloc allocation relative
> > to updating vmalloc_seq.
>
> Indeed. It appears both vmalloc_seq() and arch_sync_kernel_mappings()
> need to take the vmalloc shadow into account specifically.

I'm trying to look into that.

> And we may
> also need the dummy read from the stack's shadow in __switch_to - I am
> pretty sure I added that for a reason.

I added that since it was simple:

>From c3c76df2a9b8132b169809dcdf93488cb43a2688 Mon Sep 17 00:00:00 2001
From: Linus Walleij <linus.walleij at linaro.org>
Date: Tue, 15 Oct 2024 22:50:38 +0200
Subject: [PATCH] ARM: entry: Do a dummy read from VMAP shadow

When switching task, in addition to a dummy read from the new
VMAP stack, also do a dummy read from the VMAP stack's
corresponding KASAN shadow memory to sync things up in
the new MM context.

Signed-off-by: Linus Walleij <linus.walleij at linaro.org>
---
 arch/arm/kernel/entry-armv.S | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/arch/arm/kernel/entry-armv.S b/arch/arm/kernel/entry-armv.S
index 1dfae1af8e31..ed2d0d89e2e9 100644
--- a/arch/arm/kernel/entry-armv.S
+++ b/arch/arm/kernel/entry-armv.S
@@ -25,6 +25,7 @@
 #include <asm/tls.h>
 #include <asm/system_info.h>
 #include <asm/uaccess-asm.h>
+#include <asm/kasan_def.h>

 #include "entry-header.S"
 #include <asm/probes.h>
@@ -561,6 +562,13 @@ ENTRY(__switch_to)
     @ entries covering the vmalloc region.
     @
     ldr    r2, [ip]
+#ifdef CONFIG_KASAN
+    @ Also dummy read from the KASAN shadow memory for the new stack if we
+    @ are using KASAN
+    mov_l    r2, KASAN_SHADOW_OFFSET
+    add    r2, ip, lsr #KASAN_SHADOW_SCALE_SHIFT
+    ldr    r2, [r2]
+#endif
 #endif

     @ When CONFIG_THREAD_INFO_IN_TASK=n, the update of SP itself is what
-- 
2.46.2

We still get crashes in do_translation_fault() so we need the more
thorough solution. Maybe this is needed too.

Yours,
Linus Walleij