net: deal with integer overflows in kmalloc_reserve()
Subsystems (3)
PATCH 1/1 net: deal with integer overflows in kmalloc_reserve()
Blamed commit changed:
ptr = kmalloc(size);
if (ptr)
size = ksize(ptr);to:
size = kmalloc_size_roundup(size);
ptr = kmalloc(size);This allowed various crash as reported by syzbot [1] and Kyle Zeng.
Problem is that if @size is bigger than 0x80000001, kmalloc_size_roundup(size) returns 2^32.
kmalloc_reserve() uses a 32bit variable (obj_size), so 2^32 is truncated to 0.
kmalloc(0) returns ZERO_SIZE_PTR which is not handled by skb allocations.
Following trace can be triggered if a netdev->mtu is set close to 0x7fffffff
We might in the future limit netdev->mtu to more sensible limit (like KMALLOC_MAX_SIZE).
This patch is based on a syzbot report, and also a report and tentative fix from Kyle Zeng.
[1]
BUG: KASAN: user-memory-access in __build_skb_around net/core/skbuff.c:294 [inline] BUG: KASAN: user-memory-access in __alloc_skb+0x3c4/0x6e8 net/core/skbuff.c:527 Write of size 32 at addr 00000000fffffd10 by task syz-executor.4/22554 CPU: 1 PID: 22554 Comm: syz-executor.4 Not tainted 6.1.39-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 07/03/2023 Call trace: dump_backtrace+0x1c8/0x1f4 arch/arm64/kernel/stacktrace.c:279 show_stack+0x2c/0x3c arch/arm64/kernel/stacktrace.c:286 __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x120/0x1a0 lib/dump_stack.c:106 print_report+0xe4/0x4b4 mm/kasan/report.c:398 kasan_report+0x150/0x1ac mm/kasan/report.c:495 kasan_check_range+0x264/0x2a4 mm/kasan/generic.c:189 memset+0x40/0x70 mm/kasan/shadow.c:44 __build_skb_around net/core/skbuff.c:294 [inline] __alloc_skb+0x3c4/0x6e8 net/core/skbuff.c:527 alloc_skb include/linux/skbuff.h:1316 [inline] igmpv3_newpack+0x104/0x1088 net/ipv4/igmp.c:359 add_grec+0x81c/0x1124 net/ipv4/igmp.c:534 igmpv3_send_cr net/ipv4/igmp.c:667 [inline] igmp_ifc_timer_expire+0x1b0/0x1008 net/ipv4/igmp.c:810 call_timer_fn+0x1c0/0x9f0 kernel/time/timer.c:1474 expire_timers kernel/time/timer.c:1519 [inline] __run_timers+0x54c/0x710 kernel/time/timer.c:1790 run_timer_softirq+0x28/0x4c kernel/time/timer.c:1803 _stext+0x380/0xfbc ____do_softirq+0x14/0x20 arch/arm64/kernel/irq.c:79 call_on_irq_stack+0x24/0x4c arch/arm64/kernel/entry.S:891 do_softirq_own_stack+0x20/0x2c arch/arm64/kernel/irq.c:84 invoke_softirq kernel/softirq.c:437 [inline] __irq_exit_rcu+0x1c0/0x4cc kernel/softirq.c:683 irq_exit_rcu+0x14/0x78 kernel/softirq.c:695 el0_interrupt+0x7c/0x2e0 arch/arm64/kernel/entry-common.c:717 __el0_irq_handler_common+0x18/0x24 arch/arm64/kernel/entry-common.c:724 el0t_64_irq_handler+0x10/0x1c arch/arm64/kernel/entry-common.c:729 el0t_64_irq+0x1a0/0x1a4 arch/arm64/kernel/entry.S:584
--- net/core/skbuff.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)
net/core/skbuff.c +8 -2
@@ -550,7 +550,7 @@ static void *kmalloc_reserve(unsigned int *size, gfp_t flags, int node, bool *pfmemalloc) {bool ret_pfmemalloc = false; -unsigned int obj_size; +size_t obj_size;void *obj; obj_size =SKB_HEAD_ALIGN (*size);
@@ -567,7 +567,13 @@ static void *kmalloc_reserve(unsigned int *size, gfp_t flags, int node, obj =kmem_cache_alloc_node (skb_small_head_cache, flags, node); goto out; } - *size = obj_size =kmalloc_size_roundup (obj_size); + + obj_size =kmalloc_size_roundup (obj_size); +/* The following cast might truncate high-order bits of obj_size, this +* is harmless because kmalloc(obj_size >= 2^32) will fail anyway. +*/ + *size = (unsigned int )obj_size; +/* * Try a regular allocation, when that fails and we're not entitled * to the reserves, fail.
Signature · 1 line
2.42.0.283.g2d96d420d3-goog
This patch was applied to netdev/net.git (main) by David S. Miller <[email protected]>:
Hello:
This patch was applied to netdev/net.git (main)
by David S. Miller <[email protected]>:
Quoted 10 lines · On Thu, 31 Aug 2023 18:37:50 +0000 you wrote:
Blamed commit changed:
ptr = kmalloc(size); if (ptr) size = ksize(ptr);to:
size = kmalloc_size_roundup(size); ptr = kmalloc(size);[...]
Here is the summary with links:
- [net] net: deal with integer overflows in kmalloc_reserve()
https://git.kernel.org/netdev/net/c/915d975b2ffa
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html
From: Eric Dumazet
Quoted 11 lines
Sent: 31 August 2023 19:38
Blamed commit changed:
ptr = kmalloc(size); if (ptr) size = ksize(ptr);to:
size = kmalloc_size_roundup(size); ptr = kmalloc(size);
Quoted 8 lines
This allowed various crash as reported by syzbot [1] and Kyle Zeng.
Problem is that if @size is bigger than
0x80000001,kmalloc_size_roundup(size)returns 2^32.
kmalloc_reserve()uses a 32bit variable (obj_size), so 2^32 is truncated to 0.
Can this happen on 32bit arch?
In that case kmalloc_size_roundup() will return 0.
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
Quoted 16 lines · David Laight
From: Eric Dumazet
Quoted 14 lines
Sent: 31 August 2023 19:38
Blamed commit changed:
ptr = kmalloc(size); if (ptr) size = ksize(ptr);to:
size = kmalloc_size_roundup(size); ptr = kmalloc(size);This allowed various crash as reported by syzbot [1] and Kyle Zeng.
Quoted 8 lines
Quoted 5 lines
Problem is that if @size is bigger than
0x80000001,kmalloc_size_roundup(size)returns 2^32.
kmalloc_reserve()uses a 32bit variable (obj_size), so 2^32 is truncated to 0.Can this happen on 32bit arch?
In that casekmalloc_size_roundup()will return 0.
Maybe, but this would be a bug in kmalloc_size_roundup()
I personally can not test 32bit kernels.
From: Eric Dumazet <[email protected]>
Quoted 28 lines
Sent: 04 September 2023 10:06
To: David Laight <[email protected]>
Cc: David S . Miller <[email protected]>; Jakub Kicinski <[email protected]>; Paolo Abeni <[email protected]>; [email protected]; [email protected]; syzbot <[email protected]>; Kyle Zeng <[email protected]>; Kees Cook <[email protected]>; Vlastimil Babka <[email protected]>
Subject: Re: [PATCH net] net: deal with integer overflows inkmalloc_reserve()Quoted 19 lines · David Laight
From: Eric Dumazet
Quoted 17 lines
Sent: 31 August 2023 19:38
Blamed commit changed:
ptr = kmalloc(size); if (ptr) size = ksize(ptr);to:
size = kmalloc_size_roundup(size); ptr = kmalloc(size);This allowed various crash as reported by syzbot [1] and Kyle Zeng.
Problem is that if @size is bigger than
0x80000001,kmalloc_size_roundup(size)returns 2^32.
Quoted 7 lines
Quoted 5 lines
Quoted 2 lines
kmalloc_reserve()uses a 32bit variable (obj_size), so 2^32 is truncated to 0.Can this happen on 32bit arch?
In that casekmalloc_size_roundup()will return 0.Maybe, but this would be a bug in
kmalloc_size_roundup()
That contains:
/* Short-circuit saturated "too-large" case. */ if (unlikely(size == SIZE_MAX)) return SIZE_MAX;
It can also return 0 on failure, I can't remember if kmalloc(0) is guaranteed to be NULL (malloc(0) can do 'other things').
Which is entirely hopeless since MAX_SIZE is (size_t)-1.
IIRC kmalloc() has a size limit (max 'order' of pages) so kmalloc_size_roundup() ought check for that (or its max value).
The final:
/* The flags don't matter since size_index is common to all. */ c = kmalloc_slab(size, GFP_KERNEL); return c ? c->object_size : 0;
probably ought to return size if c is even NULL.
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
Quoted 56 lines · David Laight
From: Eric Dumazet <[email protected]>
Quoted 28 lines
Sent: 04 September 2023 10:06
To: David Laight <[email protected]>
Cc: David S . Miller <[email protected]>; Jakub Kicinski <[email protected]>; Paolo Abeni <[email protected]>; [email protected]; [email protected]; syzbot <[email protected]>; Kyle Zeng <[email protected]>; Kees Cook <[email protected]>; Vlastimil Babka <[email protected]>
Subject: Re: [PATCH net] net: deal with integer overflows inkmalloc_reserve()Quoted 19 lines · David Laight
From: Eric Dumazet
Quoted 17 lines
Sent: 31 August 2023 19:38
Blamed commit changed:
ptr = kmalloc(size); if (ptr) size = ksize(ptr);to:
size = kmalloc_size_roundup(size); ptr = kmalloc(size);This allowed various crash as reported by syzbot [1] and Kyle Zeng.
Problem is that if @size is bigger than
0x80000001,kmalloc_size_roundup(size)returns 2^32.Quoted 7 lines
Quoted 5 lines
Quoted 2 lines
kmalloc_reserve()uses a 32bit variable (obj_size), so 2^32 is truncated to 0.Can this happen on 32bit arch?
In that casekmalloc_size_roundup()will return 0.Maybe, but this would be a bug in
kmalloc_size_roundup()That contains:
/* Short-circuit saturated "too-large" case. */ if (unlikely(size == SIZE_MAX)) return SIZE_MAX;It can also return 0 on failure, I can't remember if kmalloc(0) is guaranteed to be NULL (malloc(0) can do 'other things').
Which is entirely hopeless since
MAX_SIZEis (size_t)-1.IIRC
kmalloc()has a size limit (max 'order' of pages) sokmalloc_size_roundup()ought check for that (or its max value).The final:
/* The flags don't matter since size_index is common to all. */ c = kmalloc_slab(size, GFP_KERNEL); return c ? c->object_size : 0;probably ought to return size if c is even NULL.
Quoted 8 lines
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)It can also return 0 on failure, I can't remember if kmalloc(0) is guaranteed to be NULL (malloc(0) can do 'other things').
kmalloc(0) returns ZERO_SIZE_PTR (16).
My proposed patch is to check the return value of kmalloc, making sure it is neither NULL or ZERO_SIZE_PTR. The patch is attached. It should work for both 32bit and 64bit systems.
From 034ac600c639bfa93c54e317ea3712538c749de6 Mon Sep 17 00:00:00 2001
From: Kyle Zeng <[email protected]>
Date: Mon, 28 Aug 2023 17:53:40 -0700
Subject: [PATCH] properly check for integer overflow in __alloc_skb
kmalloc_reserve may return ZERO_SIZE_PTR(0x10) when integer overflow happens since size is controlled by users.
Make sure we handle this case properly.
--- net/core/skbuff.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
net/core/skbuff.c +1 -1
@@ -642,7 +642,7 @@ struct sk_buff *__alloc_skb(unsigned int size, gfp_t gfp_mask,* Both skb->head and skb_shared_info are cache line aligned. */ data =kmalloc_reserve (&size, gfp_mask, node, &pfmemalloc); -if (unlikely (!data)) +if (unlikely (ZERO_OR_NULL_PTR (data))) goto nodata;/* kmalloc_size_roundup() might give us more room than requested. * Put skb_shared_info exactly at the end of allocated zone,
Signature · 1 line
2.34.1
Quoted 64 lines · Kyle Zeng
Quoted 61 lines · David Laight
From: Eric Dumazet <[email protected]>
Quoted 28 lines
Sent: 04 September 2023 10:06
To: David Laight <[email protected]>
Cc: David S . Miller <[email protected]>; Jakub Kicinski <[email protected]>; Paolo Abeni <[email protected]>; [email protected]; [email protected]; syzbot <[email protected]>; Kyle Zeng <[email protected]>; Kees Cook <[email protected]>; Vlastimil Babka <[email protected]>
Subject: Re: [PATCH net] net: deal with integer overflows inkmalloc_reserve()Quoted 19 lines · David Laight
From: Eric Dumazet
Quoted 17 lines
Sent: 31 August 2023 19:38
Blamed commit changed:
ptr = kmalloc(size); if (ptr) size = ksize(ptr);to:
size = kmalloc_size_roundup(size); ptr = kmalloc(size);This allowed various crash as reported by syzbot [1] and Kyle Zeng.
Problem is that if @size is bigger than
0x80000001,kmalloc_size_roundup(size)returns 2^32.Quoted 7 lines
Quoted 5 lines
Quoted 2 lines
kmalloc_reserve()uses a 32bit variable (obj_size), so 2^32 is truncated to 0.Can this happen on 32bit arch?
In that casekmalloc_size_roundup()will return 0.Maybe, but this would be a bug in
kmalloc_size_roundup()That contains:
/* Short-circuit saturated "too-large" case. */ if (unlikely(size == SIZE_MAX)) return SIZE_MAX;It can also return 0 on failure, I can't remember if kmalloc(0) is guaranteed to be NULL (malloc(0) can do 'other things').
Which is entirely hopeless since
MAX_SIZEis (size_t)-1.IIRC
kmalloc()has a size limit (max 'order' of pages) sokmalloc_size_roundup()ought check for that (or its max value).The final:
/* The flags don't matter since size_index is common to all. */ c = kmalloc_slab(size, GFP_KERNEL); return c ? c->object_size : 0;probably ought to return size if c is even NULL.
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
Quoted 5 lines
Quoted 2 lines
It can also return 0 on failure, I can't remember if kmalloc(0) is guaranteed to be NULL (malloc(0) can do 'other things').
kmalloc(0) returns
ZERO_SIZE_PTR(16).My proposed patch is to check the return value of kmalloc, making sure it is neither NULL or
ZERO_SIZE_PTR. The patch is attached. It should work for both 32bit and 64bit systems.
Again, I do not want this patch, I want to fix the root cause(s).
It makes no sense to allow dev->mtu to be as big as 0x7fffffff and ultimately allow size to be bigger than 0x80000000
I prefer waiting for net-next to open first.
Adding code in the fast path for something that must not ever happen is a no go.
Only CONFIG_DEBUG_NET=y stuff could be accepted.
Quoted 74 lines · Eric Dumazet
Quoted 64 lines · Kyle Zeng
Quoted 61 lines · David Laight
From: Eric Dumazet <[email protected]>
Quoted 28 lines
Sent: 04 September 2023 10:06
To: David Laight <[email protected]>
Cc: David S . Miller <[email protected]>; Jakub Kicinski <[email protected]>; Paolo Abeni <[email protected]>; [email protected]; [email protected]; syzbot <[email protected]>; Kyle Zeng <[email protected]>; Kees Cook <[email protected]>; Vlastimil Babka <[email protected]>
Subject: Re: [PATCH net] net: deal with integer overflows inkmalloc_reserve()Quoted 19 lines · David Laight
From: Eric Dumazet
Quoted 17 lines
Sent: 31 August 2023 19:38
Blamed commit changed:
ptr = kmalloc(size); if (ptr) size = ksize(ptr);to:
size = kmalloc_size_roundup(size); ptr = kmalloc(size);This allowed various crash as reported by syzbot [1] and Kyle Zeng.
Problem is that if @size is bigger than
0x80000001,kmalloc_size_roundup(size)returns 2^32.Quoted 7 lines
Quoted 5 lines
Quoted 2 lines
kmalloc_reserve()uses a 32bit variable (obj_size), so 2^32 is truncated to 0.Can this happen on 32bit arch?
In that casekmalloc_size_roundup()will return 0.Maybe, but this would be a bug in
kmalloc_size_roundup()That contains:
/* Short-circuit saturated "too-large" case. */ if (unlikely(size == SIZE_MAX)) return SIZE_MAX;It can also return 0 on failure, I can't remember if kmalloc(0) is guaranteed to be NULL (malloc(0) can do 'other things').
Which is entirely hopeless since
MAX_SIZEis (size_t)-1.IIRC
kmalloc()has a size limit (max 'order' of pages) sokmalloc_size_roundup()ought check for that (or its max value).The final:
/* The flags don't matter since size_index is common to all. */ c = kmalloc_slab(size, GFP_KERNEL); return c ? c->object_size : 0;probably ought to return size if c is even NULL.
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)Quoted 5 lines
Quoted 2 lines
It can also return 0 on failure, I can't remember if kmalloc(0) is guaranteed to be NULL (malloc(0) can do 'other things').
kmalloc(0) returns
ZERO_SIZE_PTR(16).My proposed patch is to check the return value of kmalloc, making sure it is neither NULL or
ZERO_SIZE_PTR. The patch is attached. It should work for both 32bit and 64bit systems.Again, I do not want this patch, I want to fix the root cause(s).
Quoted 7 lines
It makes no sense to allow
dev->mtuto be as big as0x7fffffffand ultimately allow size to be bigger than0x80000000
I prefer waiting for net-next to open first.Adding code in the fast path for something that must not ever happen is a no go.
Only
CONFIG_DEBUG_NET=ystuff could be accepted.
I will shortly submit this fix for igmpv3_newpack().
It will be useful even if we allow dev->mtu to be in the vicinity of KMALLOC_MAX_SIZE
net/ipv4/igmp.c +0 -0
100644
net/ipv4/igmp.c +0 -0
@@ -353,8 +353,9 @@ static struct sk_buff *igmpv3_newpack(struct
net_device *dev, unsigned int mtu)
struct flowi4 fl4;
int hlen = LL_RESERVED_SPACE(dev);
int tlen = dev->needed_tailroom;- unsigned int size = mtu;
+ unsigned int size;
+ size = min(mtu, IP_MAX_MTU);
while (1) {
skb = alloc_skb(size + hlen + tlen,
GFP_ATOMIC | __GFP_NOWARN);From: Eric Dumazet
Quoted 1 line
Sent: 05 September 2023 04:42
...
Quoted 4 lines
Again, I do not want this patch, I want to fix the root cause(s).
It makes no sense to allow
dev->mtuto be as big as0x7fffffffand ultimately allow size to be bigger than0x80000000
kmem_alloc_reserve() also needs fixing.
It's purpose is to find the size that kmem_alloc() will allocated so that the full size can be allocated rather than later finding out the allocated size.
The latter has issues with the compiler (etc) tracking the sizes of allocates.
So it must never return a smaller size.
Whether the path that returns 0 can happen or not
the correct error return is the original size.
The later kmalloc() will then probably fail and
be checked for.
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
Quoted 10 lines · David Laight
From: Eric Dumazet
Quoted 1 line
Sent: 05 September 2023 04:42
...
Quoted 4 lines
Again, I do not want this patch, I want to fix the root cause(s).
It makes no sense to allow
dev->mtuto be as big as0x7fffffffand ultimately allow size to be bigger than0x80000000
kmem_alloc_reserve()also needs fixing.
Yes, this is what I said. Please provide a patch ?
I stopped caring about 32bit 10 years ago.
Quoted 17 lines
It's purpose is to find the size that
kmem_alloc()will allocated so that the full size can be allocated rather than later finding out the allocated size.
The latter has issues with the compiler (etc) tracking the sizes of allocates.So it must never return a smaller size.
Whether the path that returns 0 can happen or not
the correct error return is the original size.
The laterkmalloc()will then probably fail and
be checked for.David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
Quoted 6 lines · Eric Dumazet
Quoted 4 lines · David Laight
From: Eric Dumazet
Quoted 1 line
Sent: 05 September 2023 04:42
...
Quoted 8 lines
Quoted 6 lines
Quoted 4 lines
Again, I do not want this patch, I want to fix the root cause(s).
It makes no sense to allow
dev->mtuto be as big as0x7fffffffand ultimately allow size to be bigger than0x80000000
kmem_alloc_reserve()also needs fixing.Yes, this is what I said. Please provide a patch ?
Oops, I thought you were speaking about kmalloc_size_roundup()
kmalloc_reserve() is fine, all overflows must be taken care of before reaching it.
From: Eric Dumazet
Quoted 14 lines
Sent: 05 September 2023 13:34
Quoted 11 lines · Eric Dumazet
Quoted 9 lines · David Laight
From: Eric Dumazet
Quoted 1 line
Sent: 05 September 2023 04:42
...
Quoted 4 lines
Again, I do not want this patch, I want to fix the root cause(s).
It makes no sense to allow
dev->mtuto be as big as0x7fffffffand ultimately allow size to be bigger than0x80000000
Quoted 8 lines
Quoted 3 lines
Quoted 1 line
kmem_alloc_reserve()also needs fixing.Yes, this is what I said. Please provide a patch ?
Oops, I thought you were speaking about
kmalloc_size_roundup()
kmalloc_reserve()is fine, all overflows must be taken care of before reaching it.
I was talking about roundup()
I'm in all the wrong place to generate a patch.
And having to send the emails from windows makes the procedure a right PITA.
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)