[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <ddcb3b03-90c6-876e-31ee-005be6eebd1a@linux.intel.com>
Date: Fri, 30 Aug 2019 16:40:25 +0300
From: Mathias Nyman <mathias.nyman@...ux.intel.com>
To: Ikjoon Jang <ikjn@...omium.org>
Cc: Mathias Nyman <mathias.nyman@...el.com>,
Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
linux-usb@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] xhci: fix memleak on setup address fails.
On 26.8.2019 9.41, Ikjoon Jang wrote:
> On Wed, Aug 14, 2019 at 9:57 PM Mathias Nyman
> <mathias.nyman@...ux.intel.com> wrote:
>>
>> On 11.8.2019 11.22, Ikjoon Jang wrote:
>>> Xhci re-enables a slot on transaction error in set_address using
>>> xhci_disable_slot() + xhci_alloc_dev().
>>>
>>> But in this case, xhci_alloc_dev() creates debugfs entries upon an
>>> existing device without cleaning up old entries, thus memory leaks.
>>>
>>> So this patch simply moves calling xhci_debugfs_free_dev() from
>>> xhci_free_dev() to xhci_disable_slot().
>>>
>>
>> Othwerwise this looks good, but xhci_alloc_dev() will call xhci_disable_slot()
>> in some failure cases before the slot debugfs entry is created.
>>
>> In these cases xhci_debugfs_remove_slot() will be called without
>> xhci_debugfs_create_slot() ever being called.
>>
>> This might not be an issue as xhci_debugfs_remove_slot() checks
>> if (!dev || !dev->debugfs_private) before doing anything, but should
>> be checked out.
>>
>
> I checked out the case by adding simple fault injection on xhci_alloc_dev(),
> to simulate xhci_debugfs_remove_slot() can be called without
> xhci_debugfs_create_slot() being called.
>
Thanks, patch sent forward
-Mathias
Powered by blists - more mailing lists