The next field of the DMA descriptor is written after releasing the
descriptor, which may result in a use-after-free issue. Set the value of
the field before it is released to fix the bug.
Signed-off-by: Pan Bian <[email protected]>
---
drivers/usb/gadget/udc/pch_udc.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/usb/gadget/udc/pch_udc.c b/drivers/usb/gadget/udc/pch_udc.c
index 265dab2bbfac..c5b8ec908aab 100644
--- a/drivers/usb/gadget/udc/pch_udc.c
+++ b/drivers/usb/gadget/udc/pch_udc.c
@@ -1518,8 +1518,8 @@ static void pch_udc_free_dma_chain(struct pch_udc_dev *dev,
/* do not free first desc., will be done by free for request */
td = phys_to_virt(addr);
addr2 = (dma_addr_t)td->next;
- dma_pool_free(dev->data_requests, td, addr);
td->next = 0x00;
+ dma_pool_free(dev->data_requests, td, addr);
addr = addr2;
}
req->chain_len = 1;
--
2.7.4
On Wed, Nov 6, 2019 at 10:41 AM Pan Bian <[email protected]> wrote:
>
> The next field of the DMA descriptor is written after releasing the
> descriptor, which may result in a use-after-free issue. Set the value of
> the field before it is released to fix the bug.
>
Had you chance to read the discussion [1]?
I Cc to Gustavo to hear from him about destiny of the change.
[1]: http://lkml.iu.edu/hypermail/linux/kernel/1702.1/00843.html
> Signed-off-by: Pan Bian <[email protected]>
> ---
> drivers/usb/gadget/udc/pch_udc.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/usb/gadget/udc/pch_udc.c b/drivers/usb/gadget/udc/pch_udc.c
> index 265dab2bbfac..c5b8ec908aab 100644
> --- a/drivers/usb/gadget/udc/pch_udc.c
> +++ b/drivers/usb/gadget/udc/pch_udc.c
> @@ -1518,8 +1518,8 @@ static void pch_udc_free_dma_chain(struct pch_udc_dev *dev,
> /* do not free first desc., will be done by free for request */
> td = phys_to_virt(addr);
> addr2 = (dma_addr_t)td->next;
> - dma_pool_free(dev->data_requests, td, addr);
> td->next = 0x00;
> + dma_pool_free(dev->data_requests, td, addr);
> addr = addr2;
> }
> req->chain_len = 1;
> --
> 2.7.4
>
--
With Best Regards,
Andy Shevchenko
On 11/6/19 02:54, Andy Shevchenko wrote:
> On Wed, Nov 6, 2019 at 10:41 AM Pan Bian <[email protected]> wrote:
>>
>> The next field of the DMA descriptor is written after releasing the
>> descriptor, which may result in a use-after-free issue. Set the value of
>> the field before it is released to fix the bug.
>>
>
> Had you chance to read the discussion [1]?
> I Cc to Gustavo to hear from him about destiny of the change.
>
Wow, thanks for reminding me about this thread, Andy.
I had totally forgotten about this for more than two years... :|
Here is the final version:
https://lore.kernel.org/lkml/20191106202821.GA20347@embeddedor/
Thanks
--
Gustavo