linux-kernel - Re: [PATCH 5/5] USB: serial: f81232: fix bulk

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20180204015026.GC28684@localhost>
Date:   Sun, 4 Feb 2018 12:50:26 +1100
From:   Johan Hovold <johan@...nel.org>
To:     "Ji-Ze Hong (Peter Hong)" <hpeter@...il.com>
Cc:     Johan Hovold <johan@...nel.org>, gregkh@...uxfoundation.org,
        linux-usb@...r.kernel.org, linux-kernel@...r.kernel.org,
        peter_hong@...tek.com.tw,
        "Ji-Ze Hong (Peter Hong)" <hpeter+linux_kernel@...il.com>
Subject: Re: [PATCH 5/5] USB: serial: f81232: fix bulk_in/out size

On Thu, Feb 01, 2018 at 01:50:55PM +0800, Ji-Ze Hong (Peter Hong) wrote:
> Hi Johan,
> 
> Johan Hovold 於 2018/1/30 下午 12:11 寫道:
> > On Mon, Jan 22, 2018 at 03:58:47PM +0800, Ji-Ze Hong (Peter Hong) wrote:
> >> diff --git a/drivers/usb/serial/f81232.c b/drivers/usb/serial/f81232.c
> >> index a054f69446fd..f3ee537d643c 100644
> >> --- a/drivers/usb/serial/f81232.c
> >> +++ b/drivers/usb/serial/f81232.c
> >> @@ -769,8 +769,7 @@ static struct usb_serial_driver f81232_device = {
> >>   	},
> >>   	.id_table =		id_table,
> >>   	.num_ports =		1,
> >> -	.bulk_in_size =		256,
> >> -	.bulk_out_size =	256,
> >> +	.bulk_out_size =	16,
> > 
> > So it seems you should really be setting bulk_in_size to 64 here (and
> > possibly leave bulk_out_size unset) as that would appear to match your
> > device buffer sizes.
> 
> Yes, we want to set the bulk_in_size as 64. The public datasheet has
> some error with bulk in/out, the correct size is 64.
> 
> We had test the bulk_out_size set the same with internal TX FIFO will
> make the best performance in tests, but it's ok to set 64. In my opinion
> , I'll prefer to set 16.

Having larger URB buffers than the endpoint size is typically more
efficient, but sometimes there are hardware issues that needs to be
worked around.

Johan