lkml.org 
[lkml]   [2017]   [Sep]   [21]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
SubjectRe: usb/gadget: copy_to_user called with spinlock held
On Thu, 21 Sep 2017, Andrey Konovalov wrote:

> Hi!
>
> I've got the following report while fuzzing the kernel with syzkaller.
>
> On commit ebb2c2437d8008d46796902ff390653822af6cc4 (Sep 18).
>
> Line numbers might be a little off, due to some local changes to
> gadgetfs code but the issue is AFAIU with calling copy_to_user() with
> spinlock held in ep0_read(). I see that there's a FIXME exactly about
> that and I was wondering if there's any chance of getting this fixed?
>
> Thanks!

The patch below, which goes on top of the gadgetfs patch from a few
days ago, should eliminate this problem.

Alan Stern



Index: usb-4.x/drivers/usb/gadget/legacy/inode.c
===================================================================
--- usb-4.x.orig/drivers/usb/gadget/legacy/inode.c
+++ usb-4.x/drivers/usb/gadget/legacy/inode.c
@@ -986,11 +986,14 @@ ep0_read (struct file *fd, char __user *
retval = -EIO;
else {
len = min (len, (size_t)dev->req->actual);
-// FIXME don't call this with the spinlock held ...
+ ++dev->udc_usage;
+ spin_unlock_irq(&dev->lock);
if (copy_to_user (buf, dev->req->buf, len))
retval = -EFAULT;
else
retval = len;
+ spin_lock_irq(&dev->lock);
+ --dev->udc_usage;
clean_req (dev->gadget->ep0, dev->req);
/* NOTE userspace can't yet choose to stall */
}
\
 
 \ /
  Last update: 2017-09-21 19:39    [W:0.035 / U:1.748 seconds]
©2003-2017 Jasper Spaans. hosted at Digital OceanAdvertise on this site