[PATCH v4] kprobes: unpoison stack in jprobe_return() for KASAN

[PATCH v4] kprobes: unpoison stack in jprobe_return() for KASAN

[Dmitry Vyukov]

KASAN stack instrumentation poisons stack redzones on function entry
and unpoisons them on function exit. If a function exits abnormally
(e.g. with a longjmp like jprobe_return()), stack redzones are left
poisoned. Later this leads to random KASAN false reports.

Unpoison stack redzones in the frames we are going to jump over
before doing actual longjmp in jprobe_return().

Signed-off-by: Dmitry Vyukov <dvyukov@google.com>
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Ananth N Mavinakayanahalli <ananth@linux.vnet.ibm.com>
Cc: Anil S Keshavamurthy <anil.s.keshavamurthy@intel.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: x86@kernel.org
Cc: kasan-dev@googlegroups.com

--

Changes since v1:
 - leave kasan_unpoison_remaining_stack() intact
 - instead add kasan_unpoison_stack_above_sp_to()
 - rename kasan_unpoison_remaining_stack() to kasan_unpoison_task_stack_below()

Changes since v2:
 - fix build by adding return type to kasan_unpoison_stack_above_sp_to()
   (tested v2 with it, but forgot to git add)

Changes since v3:
 - fix build warning in kasan_unpoison_stack_above_sp_to() related to
   void*/const void* conversion

I observe false positives due to this in sctp code.
sctp uses jprobe_return() in jsctp_sf_eat_sack().
The stray 0xf4 in shadow memory are stack redzones.

[  376.492209] ==================================================================
[  376.500368] BUG: KASAN: stack-out-of-bounds in memcmp+0xe9/0x150 at addr ffff88005e48f480
[  376.509522] Read of size 1 by task syz-executor/18535
[  376.515249] page:ffffea00017923c0 count:0 mapcount:0 mapping:          (null) index:0x0
[  376.524377] flags: 0x1fffc0000000000()
[  376.528645] page dumped because: kasan: bad access detected
[  376.534939] CPU: 1 PID: 18535 Comm: syz-executor Not tainted 4.8.0+ #28
[  376.542375] Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
[  376.552669]  ffff88005e48f2d0 ffffffff82d2b849 ffffffff0bc91e90 fffffbfff10971e8
[  376.561599]  ffffed000bc91e90 ffffed000bc91e90 0000000000000001 0000000000000000
[  376.570486]  ffff88005e48f480 ffff88005e48f350 ffffffff817d3169 ffff88005e48f370
[  376.579348] Call Trace:
[  376.582196]  [<ffffffff82d2b849>] dump_stack+0x12e/0x185
[  376.588190]  [<ffffffff817d3169>] kasan_report+0x489/0x4b0
[  376.594378]  [<ffffffff82d49529>] ? memcmp+0xe9/0x150
[  376.600099]  [<ffffffff8126377e>] ? update_stack_state.constprop.4+0xde/0x150
[  376.608187]  [<ffffffff817d31a9>] __asan_report_load1_noabort+0x19/0x20
[  376.615620]  [<ffffffff82d49529>] memcmp+0xe9/0x150
[  376.621148]  [<ffffffff82df7486>] depot_save_stack+0x176/0x5c0
[  376.627722]  [<ffffffff85b03f19>] ? skb_free_head+0x79/0xb0
[  376.634006]  [<ffffffff817d2031>] save_stack+0xb1/0xd0
[  376.639805]  [<ffffffff811fc8ab>] ? save_stack_trace+0x1b/0x20
[  376.646369]  [<ffffffff817d1fc6>] ? save_stack+0x46/0xd0
[  376.652366]  [<ffffffff817d27f2>] ? kasan_slab_free+0x72/0xc0
[  376.658837]  [<ffffffff817d05b8>] ? kfree+0xc8/0x2a0
[  376.664454]  [<ffffffff85b03f19>] ? skb_free_head+0x79/0xb0
[  376.670736]  [<ffffffff85b0900a>] ? skb_release_data+0x37a/0x420
[  376.677517]  [<ffffffff85b090ff>] ? skb_release_all+0x4f/0x60
[  376.683992]  [<ffffffff85b11348>] ? consume_skb+0x138/0x370
[  376.690284]  [<ffffffff8676ad7b>] ? sctp_chunk_put+0xcb/0x180
[  376.696761]  [<ffffffff8676ae88>] ? sctp_chunk_free+0x58/0x70
[  376.703234]  [<ffffffff8677fa5f>] ? sctp_inq_pop+0x68f/0xef0
[  376.709616]  [<ffffffff8675ee36>] ? sctp_assoc_bh_rcv+0xd6/0x4b0
[  376.716379]  [<ffffffff8677f2c1>] ? sctp_inq_push+0x131/0x190
[  376.722946]  [<ffffffff867bad69>] ? sctp_backlog_rcv+0xe9/0xa20
[  376.729615]  [<ffffffff85af70dc>] ? __release_sock+0x12c/0x3a0
[  376.736186]  [<ffffffff85af73ae>] ? release_sock+0x5e/0x1c0
[  376.742465]  [<ffffffff8679fc62>] ? sctp_sendmsg+0xd82/0x2e00
[  376.748935]  [<ffffffff85f83c43>] ? inet_sendmsg+0x303/0x4c0
[  376.755318]  [<ffffffff85aeab0f>] ? sock_sendmsg+0xcf/0x110
[  376.761599]  [<ffffffff85aebc1d>] ? SYSC_sendto+0x20d/0x340
[  376.767882]  [<ffffffff85aee175>] ? SyS_sendto+0x45/0x60
[  376.773885]  [<ffffffff81008783>] ? do_syscall_64+0x1d3/0x620
[  376.780372]  [<ffffffff86d8e9cd>] ? entry_SYSCALL64_slow_path+0x25/0x25
[  376.787832]  [<ffffffff86751c79>] ? sctp_do_sm+0x3689/0x4e90
[  376.794215]  [<ffffffff81467040>] ? debug_check_no_locks_freed+0x3c0/0x3c0
[  376.801948]  [<ffffffff8674e5f0>] ? sctp_do_8_2_transport_strike.isra.19+0x900/0x900
[  376.810918]  [<ffffffff81466ed0>] ? debug_check_no_locks_freed+0x250/0x3c0
[  376.818650]  [<ffffffff817d27f2>] kasan_slab_free+0x72/0xc0
[  376.824945]  [<ffffffff817d05b8>] kfree+0xc8/0x2a0
[  376.830374]  [<ffffffff85b03f19>] skb_free_head+0x79/0xb0
[  376.836466]  [<ffffffff85b0900a>] skb_release_data+0x37a/0x420
[  376.843036]  [<ffffffff85b090ff>] skb_release_all+0x4f/0x60
[  376.849319]  [<ffffffff85b11348>] consume_skb+0x138/0x370
[  376.855411]  [<ffffffff8676ad7b>] sctp_chunk_put+0xcb/0x180
[  376.861702]  [<ffffffff8676ae88>] sctp_chunk_free+0x58/0x70
[  376.867990]  [<ffffffff8677fa5f>] sctp_inq_pop+0x68f/0xef0
[  376.874182]  [<ffffffff8675ee36>] sctp_assoc_bh_rcv+0xd6/0x4b0
[  376.880759]  [<ffffffff8677f2c1>] sctp_inq_push+0x131/0x190
[  376.887055]  [<ffffffff867bad69>] sctp_backlog_rcv+0xe9/0xa20
[  376.893532]  [<ffffffff814667dd>] ? trace_hardirqs_on+0xd/0x10
[  376.900107]  [<ffffffff81369bed>] ? __local_bh_enable_ip+0xad/0x190
[  376.907161]  [<ffffffff85af70dc>] __release_sock+0x12c/0x3a0
[  376.913540]  [<ffffffff85af73ae>] release_sock+0x5e/0x1c0
[  376.919922]  [<ffffffff8679fc62>] sctp_sendmsg+0xd82/0x2e00
[  376.926209]  [<ffffffff8679eee0>] ? sctp_id2assoc+0x330/0x330
[  376.932696]  [<ffffffff81467040>] ? debug_check_no_locks_freed+0x3c0/0x3c0
[  376.940417]  [<ffffffff814a6127>] ? debug_lockdep_rcu_enabled+0x77/0x90
[  376.947858]  [<ffffffff814a6127>] ? debug_lockdep_rcu_enabled+0x77/0x90
[  376.955293]  [<ffffffff85f83940>] ? inet_recvmsg+0x4b0/0x4b0
[  376.961675]  [<ffffffff85f83b48>] ? inet_sendmsg+0x208/0x4c0
[  376.968064]  [<ffffffff85f83c43>] inet_sendmsg+0x303/0x4c0
[  376.974261]  [<ffffffff85f839b8>] ? inet_sendmsg+0x78/0x4c0
[  376.980546]  [<ffffffff85f83940>] ? inet_recvmsg+0x4b0/0x4b0
[  376.986929]  [<ffffffff85aeab0f>] sock_sendmsg+0xcf/0x110
[  376.993025]  [<ffffffff85aebc1d>] SYSC_sendto+0x20d/0x340
[  376.999125]  [<ffffffff85aeba10>] ? SYSC_connect+0x2e0/0x2e0
[  377.005510]  [<ffffffff8188cc33>] ? __fd_install+0x233/0x560
[  377.011894]  [<ffffffff81007b93>] ? perf_trace_sys_enter+0x443/0x9d0
[  377.019051]  [<ffffffff81007750>] ? syscall_trace_enter+0xe00/0xe00
[  377.026109]  [<ffffffff8158887f>] ? __sanitizer_cov_trace_pc+0x4f/0x60
[  377.033466]  [<ffffffff85aee175>] SyS_sendto+0x45/0x60
[  377.039278]  [<ffffffff85aee130>] ? SyS_getpeername+0x30/0x30
[  377.045752]  [<ffffffff81008783>] do_syscall_64+0x1d3/0x620
[  377.052035]  [<ffffffff8100501a>] ? trace_hardirqs_on_thunk+0x1a/0x1c
[  377.059281]  [<ffffffff86d8e9cd>] entry_SYSCALL64_slow_path+0x25/0x25
[  377.066509] Memory state around the buggy address:
[  377.071935]  ffff88005e48f380: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
[  377.080024]  ffff88005e48f400: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
[  377.088117] >ffff88005e48f480: f4 f4 00 00 00 00 00 00 00 00 00 00 00 00 00 00
[  377.096205]                    ^
[  377.099900]  ffff88005e48f500: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
[  377.107994]  ffff88005e48f580: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
[  377.116079] ==================================================================
---
 arch/arm64/kernel/sleep.S      |  2 +-
 arch/x86/kernel/kprobes/core.c |  4 ++++
 include/linux/kasan.h          |  2 ++
 mm/kasan/kasan.c               | 21 ++++++++++++++++++---
 4 files changed, 25 insertions(+), 4 deletions(-)

diff --git a/arch/arm64/kernel/sleep.S b/arch/arm64/kernel/sleep.S
index b8799e7..1bec41b 100644
--- a/arch/arm64/kernel/sleep.S
+++ b/arch/arm64/kernel/sleep.S
@@ -135,7 +135,7 @@ ENTRY(_cpu_resume)
 
 #ifdef CONFIG_KASAN
 	mov	x0, sp
-	bl	kasan_unpoison_remaining_stack
+	bl	kasan_unpoison_task_stack_below
 #endif
 
 	ldp	x19, x20, [x29, #16]
diff --git a/arch/x86/kernel/kprobes/core.c b/arch/x86/kernel/kprobes/core.c
index 28cee01..22a462a 100644
--- a/arch/x86/kernel/kprobes/core.c
+++ b/arch/x86/kernel/kprobes/core.c
@@ -50,6 +50,7 @@
 #include <linux/kallsyms.h>
 #include <linux/ftrace.h>
 #include <linux/frame.h>
+#include <linux/kasan.h>
 
 #include <asm/text-patching.h>
 #include <asm/cacheflush.h>
@@ -1080,6 +1081,9 @@ void jprobe_return(void)
 {
 	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
 
+	/* Unpoison stack redzones in the frames we are going to jump over. */
+	kasan_unpoison_stack_above_sp_to(kcb->jprobe_saved_sp);
+
 	asm volatile (
 #ifdef CONFIG_X86_64
 			"       xchg   %%rbx,%%rsp	\n"
diff --git a/include/linux/kasan.h b/include/linux/kasan.h
index d600303..820c0ad 100644
--- a/include/linux/kasan.h
+++ b/include/linux/kasan.h
@@ -44,6 +44,7 @@ static inline void kasan_disable_current(void)
 void kasan_unpoison_shadow(const void *address, size_t size);
 
 void kasan_unpoison_task_stack(struct task_struct *task);
+void kasan_unpoison_stack_above_sp_to(const void *watermark);
 
 void kasan_alloc_pages(struct page *page, unsigned int order);
 void kasan_free_pages(struct page *page, unsigned int order);
@@ -85,6 +86,7 @@ size_t kasan_metadata_size(struct kmem_cache *cache);
 static inline void kasan_unpoison_shadow(const void *address, size_t size) {}
 
 static inline void kasan_unpoison_task_stack(struct task_struct *task) {}
+static inline void kasan_unpoison_stack_above_sp_to(const void *watermark) {}
 
 static inline void kasan_enable_current(void) {}
 static inline void kasan_disable_current(void) {}
diff --git a/mm/kasan/kasan.c b/mm/kasan/kasan.c
index 88af13c..4f0c98c 100644
--- a/mm/kasan/kasan.c
+++ b/mm/kasan/kasan.c
@@ -62,7 +62,7 @@ void kasan_unpoison_shadow(const void *address, size_t size)
 	}
 }
 
-static void __kasan_unpoison_stack(struct task_struct *task, void *sp)
+static void __kasan_unpoison_stack(struct task_struct *task, const void *sp)
 {
 	void *base = task_stack_page(task);
 	size_t size = sp - base;
@@ -77,9 +77,24 @@ void kasan_unpoison_task_stack(struct task_struct *task)
 }
 
 /* Unpoison the stack for the current task beyond a watermark sp value. */
-asmlinkage void kasan_unpoison_remaining_stack(void *sp)
+asmlinkage void kasan_unpoison_task_stack_below(const void *watermark)
 {
-	__kasan_unpoison_stack(current, sp);
+	__kasan_unpoison_stack(current, watermark);
+}
+
+/*
+ * Clear all poison for the region between the current SP and a provided
+ * watermark value, as is sometimes required prior to hand-crafted asm function
+ * returns in the middle of functions.
+ */
+void kasan_unpoison_stack_above_sp_to(const void *watermark)
+{
+	const void *sp = (void *)current_stack_pointer();
+	size_t size = watermark - sp;
+
+	if (WARN_ON(sp > watermark))
+		return;
+	kasan_unpoison_shadow(sp, size);
 }
 
 /*
-- 
2.8.0.rc3.226.g39d4020

Re: [PATCH v4] kprobes: unpoison stack in jprobe_return() for KASAN

[Mark Rutland]

On Fri, Oct 14, 2016 at 01:54:30PM +0200, Dmitry Vyukov wrote:
> KASAN stack instrumentation poisons stack redzones on function entry
> and unpoisons them on function exit. If a function exits abnormally
> (e.g. with a longjmp like jprobe_return()), stack redzones are left
> poisoned. Later this leads to random KASAN false reports.
> 
> Unpoison stack redzones in the frames we are going to jump over
> before doing actual longjmp in jprobe_return().
> 
> Signed-off-by: Dmitry Vyukov <dvyukov@google.com>
> Reviewed-by: Mark Rutland <mark.rutland@arm.com>

... judging by the kbuild test robot I spoke too soon, and should have
been more thorough. :/

> +/*
> + * Clear all poison for the region between the current SP and a provided
> + * watermark value, as is sometimes required prior to hand-crafted asm function
> + * returns in the middle of functions.
> + */
> +void kasan_unpoison_stack_above_sp_to(const void *watermark)
> +{
> +	const void *sp = (void *)current_stack_pointer();

Aargh; it seems current_stack_pointer() is only function-like on some
arches, and not on others (arm64 included). I should have known better;
sorry for the bad suggestion.

I'm not overjoyed about taking the address of a stack variable to
implement this ourselves. Can we use __builtin_frame_address(0) instead?
Or are there cases where that won't work on x86?

> +	size_t size = watermark - sp;
> +
> +	if (WARN_ON(sp > watermark))
> +		return;

... not a new problem, but we should also include <linux/bug.h> for
WARN_ON().

Thanks,
Mark.

Re: [PATCH v4] kprobes: unpoison stack in jprobe_return() for KASAN

[Dmitry Vyukov]

On Fri, Oct 14, 2016 at 3:08 PM, Mark Rutland <mark.rutland@arm.com> wrote:
> On Fri, Oct 14, 2016 at 01:54:30PM +0200, Dmitry Vyukov wrote:
>> KASAN stack instrumentation poisons stack redzones on function entry
>> and unpoisons them on function exit. If a function exits abnormally
>> (e.g. with a longjmp like jprobe_return()), stack redzones are left
>> poisoned. Later this leads to random KASAN false reports.
>>
>> Unpoison stack redzones in the frames we are going to jump over
>> before doing actual longjmp in jprobe_return().
>>
>> Signed-off-by: Dmitry Vyukov <dvyukov@google.com>
>> Reviewed-by: Mark Rutland <mark.rutland@arm.com>
>
> ... judging by the kbuild test robot I spoke too soon, and should have
> been more thorough. :/
>
>> +/*
>> + * Clear all poison for the region between the current SP and a provided
>> + * watermark value, as is sometimes required prior to hand-crafted asm function
>> + * returns in the middle of functions.
>> + */
>> +void kasan_unpoison_stack_above_sp_to(const void *watermark)
>> +{
>> +     const void *sp = (void *)current_stack_pointer();
>
> Aargh; it seems current_stack_pointer() is only function-like on some
> arches, and not on others (arm64 included). I should have known better;
> sorry for the bad suggestion.
>
> I'm not overjoyed about taking the address of a stack variable to
> implement this ourselves. Can we use __builtin_frame_address(0) instead?
> Or are there cases where that won't work on x86?

Mailed v5 with __builtin_frame_address(0).
Built mm/kasan/kasan.o for arm64.

I see that __builtin_frame_address(0) is used on several arches
including x86 and arm64.
So I hope we are good here.


>
>> +     size_t size = watermark - sp;
>> +
>> +     if (WARN_ON(sp > watermark))
>> +             return;
>
> ... not a new problem, but we should also include <linux/bug.h> for
> WARN_ON().

Done