2012-10-13 10:18:06

by Henrik Rydberg

Subject: [PATCH v3] usbdevfs: Fix broken scatter-gather transfer

The handling of large output bulk transfers is broken; the same user
page is read over and over again. Fixed with this patch.

Cc: [email protected]
Acked-by: Peter Stuge <[email protected]>
Acked-by: Hans de Goede <[email protected]>
Acked-by: Alan Stern <[email protected]>
Signed-off-by: Henrik Rydberg <[email protected]>
---
Hi Greg,

Here is the formal and third version of the patch. Version two still
made me nervous, so I moved the increment to where it is clear that
the buffer pointer is never referenced again later in the function. I
still kept the ACKS, hope that is alright with everyone.

Thanks,
Henrik

drivers/usb/core/devio.c | 1 +
1 file changed, 1 insertion(+)

diff --git a/drivers/usb/core/devio.c b/drivers/usb/core/devio.c
index ebb8a9d..7f75343 100644
--- a/drivers/usb/core/devio.c
+++ b/drivers/usb/core/devio.c
@@ -1348,6 +1348,7 @@ static int proc_do_submiturb(struct dev_state *ps, struct usbdevfs_urb *uurb,
ret = -EFAULT;
goto error;
}
+ uurb->buffer += u;
}
totlen -= u;
}
--
1.7.12.2

2012-10-14 15:51:58

by Hans de Goede

Subject: Re: [PATCH v3] usbdevfs: Fix broken scatter-gather transfer

Hi,

On 10/13/2012 12:20 PM, Henrik Rydberg wrote:
> The handling of large output bulk transfers is broken; the same user
> page is read over and over again. Fixed with this patch.
>
> Cc: [email protected]
> Acked-by: Peter Stuge <[email protected]>
> Acked-by: Hans de Goede <[email protected]>
> Acked-by: Alan Stern <[email protected]>
> Signed-off-by: Henrik Rydberg <[email protected]>
> ---
> Hi Greg,
>
> Here is the formal and third version of the patch. Version two still
> made me nervous,

And it should, as as said before I believe it would have broken large
input transfers ...

so I moved the increment to where it is clear that
> the buffer pointer is never referenced again later in the function. I
> still kept the ACKS, hope that is alright with everyone.

Unlike the previous one, this version should work, so yes it may keep
my ACK :)

Regards,

Hans

>
> Thanks,
> Henrik
>
> drivers/usb/core/devio.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/usb/core/devio.c b/drivers/usb/core/devio.c
> index ebb8a9d..7f75343 100644
> --- a/drivers/usb/core/devio.c
> +++ b/drivers/usb/core/devio.c
> @@ -1348,6 +1348,7 @@ static int proc_do_submiturb(struct dev_state *ps, struct usbdevfs_urb *uurb,
> ret = -EFAULT;
> goto error;
> }
> + uurb->buffer += u;
> }
> totlen -= u;
> }
>