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  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]
Date:	Thu, 24 Jul 2014 17:05:07 +0300
From:	Mathias Nyman <mathias.nyman@...ux.intel.com>
To:	Julius Werner <jwerner@...omium.org>,
	Paul Zimmerman <Paul.Zimmerman@...opsys.com>
CC:	Maciej Puzio <mx34567@...il.com>,
	Mathias Nyman <mathias.nyman@...el.com>,
	David Laight <David.Laight@...lab.com>,
	Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
	"linux-usb@...r.kernel.org" <linux-usb@...r.kernel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] usb: xhci: Fix Set TR Dequeue Pointer cycle state for
 quirky xHCs

On 07/17/2014 10:50 PM, Julius Werner wrote:
>> Hmm. Wouldn't it be safer to have a quirk for this, and only enable
>> the workaround if the Asmedia controller is detected? This code is so
>> complicated that it is difficult to see whether this could have a
>> harmful effect on controllers without the bug.
> 
> Sorry for making it complicated, but it kinda has been like that
> before already... I don't think the new patch adds much confusion on
> its own. I would really advise against making this a quirk: checking
> for it in every case essentially doesn't cost us anything (just one
> more register compare that is negligible against all the
> cache-coherent memory accesses of walking the ring), and hardcoding a
> quirk would mean that we have to identify and add every host
> controller that does this individually.
> 
> I still haven't seen anything in the XHCI spec that actually forbids
> this behavior, so it might be a perfectly legal interpretation and who
> knows how many host controllers chose to do it that way. Until we find
> them all, we would have some really bad and hard to track down bugs on
> those controllers (we really just got lucky this time that Maciej was
> able to catch it in a bisect). I think it's better to program the
> driver defensively and able to deal with unexpected situations in
> general.
> --
> To unsubscribe from this list: send the line "unsubscribe linux-usb" in
> the body of a message to majordomo@...r.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 

It's starting to get a bit too complicated.
xhci find_new_dequeue_state() now has four places where it can toggle the cycle bit
in addition to the cycle toggle in find_trb_seg().

In the end we really just want to do it max once.

I'll see if I can simplify the whole cycle bit code somehow.

If not, then we need to take this or revert the original patch

-Mathias

 
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists