Re: KASAN: slab-out-of-bounds Read in ld_usb_read (3)

[Johan Hovold]

On Thu, Oct 17, 2019 at 12:25:13PM +0200, Johan Hovold wrote:
> On Wed, Oct 16, 2019 at 06:42:10PM -0700, syzbot wrote:
> > Hello,
> > 
> > syzbot found the following crash on:
> > 
> > HEAD commit:    22be26f7 usb-fuzzer: main usb gadget fuzzer driver
> > git tree:       https://github.com/google/kasan.git usb-fuzzer
> > console output: https://syzkaller.appspot.com/x/log.txt?x=14f6dc5f600000
> > kernel config:  https://syzkaller.appspot.com/x/.config?x=387eccb7ac68ec5
> > dashboard link: https://syzkaller.appspot.com/bug?extid=6fe95b826644f7f12b0b
> > compiler:       gcc (GCC) 9.0.0 20181231 (experimental)
> > syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=102c3227600000
> > C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=12a503a0e00000
> > 
> > IMPORTANT: if you fix the bug, please add the following tag to the commit:
> > Reported-by: syzbot+6fe95b826644f7f12b0b@xxxxxxxxxxxxxxxxxxxxxxxxx
> > 
> > ldusb 1-1:0.28: Read buffer overflow, -131383859965943 bytes dropped
> 
> This warning should be using %zu and thus read 18446612689849585673.
> 
> Apparently, *actual_buffer holds a kernel address ffff8881cfb42000
> instead of a transfer length. Possibly a stale value since the buffer is
> not cleared on allocation.
> 
> > ==================================================================
> > BUG: KASAN: slab-out-of-bounds in _copy_to_user+0x124/0x150  
> > lib/usercopy.c:28
> > Read of size 102391 at addr ffff8881cfb40008 by task syz-executor372/1737
> 
> And due to missing sanity checks the driver proceeds to access data
> beyond the ring-buffer entry.
> 
> > CPU: 0 PID: 1737 Comm: syz-executor372 Not tainted 5.4.0-rc3+ #0
> > Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS  
> > Google 01/01/2011
> > Call Trace:
> >   __dump_stack lib/dump_stack.c:77 [inline]
> >   dump_stack+0xca/0x13e lib/dump_stack.c:113
> >   print_address_description.constprop.0+0x36/0x50 mm/kasan/report.c:374
> >   __kasan_report.cold+0x1a/0x33 mm/kasan/report.c:506
> >   kasan_report+0xe/0x20 mm/kasan/common.c:634
> >   check_memory_region_inline mm/kasan/generic.c:185 [inline]
> >   check_memory_region+0x128/0x190 mm/kasan/generic.c:192
> >   _copy_to_user+0x124/0x150 lib/usercopy.c:28
> >   copy_to_user include/linux/uaccess.h:152 [inline]
> >   ld_usb_read+0x329/0x760 drivers/usb/misc/ldusb.c:492
> >   __vfs_read+0x76/0x100 fs/read_write.c:425
> >   vfs_read+0x1ea/0x430 fs/read_write.c:461
> >   ksys_read+0x1e8/0x250 fs/read_write.c:587
> >   do_syscall_64+0xb7/0x580 arch/x86/entry/common.c:290
> >   entry_SYSCALL_64_after_hwframe+0x49/0xbe
> > RIP: 0033:0x4421d9
> > Code: e8 7c e7 ff ff 48 83 c4 18 c3 0f 1f 80 00 00 00 00 48 89 f8 48 89 f7  
> > 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff  
> > ff 0f 83 bb 07 fc ff c3 66 2e 0f 1f 84 00 00 00 00
> > RSP: 002b:00007ffc08eb0888 EFLAGS: 00000246 ORIG_RAX: 0000000000000000
> > RAX: ffffffffffffffda RBX: 00007ffc08eb0ae0 RCX: 00000000004421d9
> > RDX: 0000000000018ff7 RSI: 0000000020000a80 RDI: 0000000000000004
> > RBP: 0000000000000000 R08: 000000000000000f R09: 0000000000402eb0
> > R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
> > R13: 0000000000402eb0 R14: 0000000000000000 R15: 0000000000000000
> > 
> > The buggy address belongs to the page:
> > page:ffffea00073ed000 refcount:1 mapcount:0 mapping:0000000000000000  
> > index:0x0 compound_mapcount: 0
> > flags: 0x200000000010000(head)
> > raw: 0200000000010000 dead000000000100 dead000000000122 0000000000000000
> > raw: 0000000000000000 0000000000000000 00000001ffffffff 0000000000000000
> > page dumped because: kasan: bad access detected
> > 
> > Memory state around the buggy address:
> >   ffff8881cfb55500: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> >   ffff8881cfb55580: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> > > ffff8881cfb55600: fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe
> >                     ^
> >   ffff8881cfb55680: fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe
> >   ffff8881cfb55700: fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe
> > ==================================================================
> 
> The drivers ring-buffer implementation is broken so this could
> potentially be a memory-ordering issue (e.g. the entry is read before it
> has been updated).

Heh, it doesn't even need to race to trigger this; the driver sets
dev->interrupt_in_done when the URB is killed on disconnect() and read
happily continues with reading the next entry which was never updated.

> Let's try the below.

The below should work, but should be amended so that a blocking read()
returns after disconnect().

Johan

> #syz test: https://github.com/google/kasan.git 22be26f7
> 
> From 61c87ec9b1311949e313f633e84aab5c73975078 Mon Sep 17 00:00:00 2001
> From: Johan Hovold <johan@xxxxxxxxxx>
> Date: Thu, 17 Oct 2019 11:53:04 +0200
> Subject: [PATCH] wip: USB: lsusb: fix ring-buffer locking
> 
> Signed-off-by: Johan Hovold <johan@xxxxxxxxxx>
> ---
>  drivers/usb/misc/ldusb.c | 17 ++++++++++++-----
>  1 file changed, 12 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/usb/misc/ldusb.c b/drivers/usb/misc/ldusb.c
> index f3108d85e768..c8d09ac7ade9 100644
> --- a/drivers/usb/misc/ldusb.c
> +++ b/drivers/usb/misc/ldusb.c
> @@ -467,7 +467,7 @@ static ssize_t ld_usb_read(struct file *file, char __user *buffer, size_t count,
>  
>  	/* wait for data */
>  	spin_lock_irq(&dev->rbsl);
> -	if (dev->ring_head == dev->ring_tail) {
> +	while (dev->ring_head == dev->ring_tail) {
>  		dev->interrupt_in_done = 0;
>  		spin_unlock_irq(&dev->rbsl);
>  		if (file->f_flags & O_NONBLOCK) {
> @@ -477,8 +477,8 @@ static ssize_t ld_usb_read(struct file *file, char __user *buffer, size_t count,
>  		retval = wait_event_interruptible(dev->read_wait, dev->interrupt_in_done);
>  		if (retval < 0)
>  			goto unlock_exit;
> -	} else {
> -		spin_unlock_irq(&dev->rbsl);
> +
> +		spin_lock_irq(&dev->rbsl);
>  	}
>  
>  	/* actual_buffer contains actual_length + interrupt_in_buffer */
> @@ -487,17 +487,19 @@ static ssize_t ld_usb_read(struct file *file, char __user *buffer, size_t count,
>  	if (bytes_to_read < *actual_buffer)
>  		dev_warn(&dev->intf->dev, "Read buffer overflow, %zd bytes dropped\n",
>  			 *actual_buffer-bytes_to_read);
> +	spin_unlock_irq(&dev->rbsl);
>  
>  	/* copy one interrupt_in_buffer from ring_buffer into userspace */
>  	if (copy_to_user(buffer, actual_buffer+1, bytes_to_read)) {
>  		retval = -EFAULT;
>  		goto unlock_exit;
>  	}
> -	dev->ring_tail = (dev->ring_tail+1) % ring_buffer_size;
> -
>  	retval = bytes_to_read;
>  
>  	spin_lock_irq(&dev->rbsl);
> +
> +	dev->ring_tail = (dev->ring_tail+1) % ring_buffer_size;
> +
>  	if (dev->buffer_overflow) {
>  		dev->buffer_overflow = 0;
>  		spin_unlock_irq(&dev->rbsl);
> @@ -693,12 +695,17 @@ static int ld_usb_probe(struct usb_interface *intf, const struct usb_device_id *
>  		dev_warn(&intf->dev, "Interrupt out endpoint not found (using control endpoint instead)\n");
>  
>  	dev->interrupt_in_endpoint_size = usb_endpoint_maxp(dev->interrupt_in_endpoint);
> +	dev_info(&intf->dev, "%s - interrupt_in_endpoint_size = %zu\n",
> +			__func__, dev->interrupt_in_endpoint_size);
>  	dev->ring_buffer =
>  		kmalloc_array(ring_buffer_size,
>  			      sizeof(size_t) + dev->interrupt_in_endpoint_size,
>  			      GFP_KERNEL);
>  	if (!dev->ring_buffer)
>  		goto error;
> +	dev_info(&intf->dev, "%s - ring_buffer = %px\n", __func__,
> +			dev->ring_buffer);
> +
>  	dev->interrupt_in_buffer = kmalloc(dev->interrupt_in_endpoint_size, GFP_KERNEL);
>  	if (!dev->interrupt_in_buffer)
>  		goto error;