linux-kernel - Re: XHCI vs PCM2903B/PCM2904 part 2

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] [day] [month] [year] [list]

Date:   Tue, 30 Jun 2020 14:52:01 -0400
From:   Rik van Riel <riel@...riel.com>
To:     Mathias Nyman <mathias.nyman@...ux.intel.com>,
        Alan Stern <stern@...land.harvard.edu>
Cc:     linux-usb <linux-usb@...r.kernel.org>, alsa-devel@...a-project.org,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        Mathias Nyman <mathias.nyman@...el.com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Jaroslav Kysela <perex@...ex.cz>, Takashi Iwai <tiwai@...e.com>
Subject: Re: XHCI vs PCM2903B/PCM2904 part 2

On Tue, 2020-06-30 at 17:27 +0300, Mathias Nyman wrote:
> On 30.6.2020 16.08, Rik van Riel wrote:
> > I misread the code, it's not a bitfield, so state 1 means an
> > endpoint marked with running state. The next urb is never getting a
> > response, though.
> > 
> > However, the xhci spec says an endpoint is halted upon a babble
> > error.
> 
> I was looking at the same, so according to specs this state shouldn't
> be possible. 

The PCM2903B chip, and potentially the hub it is behind,
are USB2 devices, though. Does USB2 know about halted
endpoints?

> > The code right above the babble handling case adds halted into the
> > endpoint state itself. Does the code handling the babble error need
> > to do something similar to trigger cleanup elsewhere? 
> 
> It's a flag to prevent ringing the doorbell for a halted endpoint.
> Anyway, reset endpoint is meant to recover an endpoint in a halted
> state.
> Resetting non-halted endpoints will just lead to a context state
> error, and
> besides, isoc endpoints shouldn't halt.
> 
> Anyways, I haven't got any better idea at the moment.
> You can try and see what a forced reset does with:

So close. Looks like something in the XHCI/USB stack isn't
doing the reset because something is in an unexpected state?

[   51.706798] xhci_hcd 0000:00:14.0: WARN Set TR Deq Ptr cmd failed
due to incorrect slot or ep state.
[   51.706802] got overflow, ep->flags = 2
[   51.932550] usb 3-9.7.5: reset high-speed USB device number 18 using
xhci_hcd
[   68.830396] xhci_hcd 0000:00:14.0: WARN Set TR Deq Ptr cmd failed
due to incorrect slot or ep state.
[   68.830409] got overflow, ep->flags = 2
[   70.077981] rfkill: input handler disabled
[  157.992374] got overflow, ep->flags = 2
[  157.992406] xhci_hcd 0000:00:14.0: WARN Set TR Deq Ptr cmd failed
due to incorrect slot or ep state.

> diff --git a/drivers/usb/host/xhci-ring.c b/drivers/usb/host/xhci-
> ring.c
> index 2c255d0620b0..d79aca0df6d4 100644
> --- a/drivers/usb/host/xhci-ring.c
> +++ b/drivers/usb/host/xhci-ring.c
> @@ -1900,8 +1900,7 @@ static int
> xhci_requires_manual_halt_cleanup(struct xhci_hcd *xhci,
>                  * endpoint anyway.  Check if a babble halted the
>                  * endpoint.
>                  */
> -               if (GET_EP_CTX_STATE(ep_ctx) == EP_STATE_HALTED)
> -                       return 1;
> +               return 1;
>  
>         return 0;
>  }
> 
> Traces also showed thet endpoint doorbell was rang after th babble
> error, so
> we know that didn't help restarting the endpoint.
> 
> -Mathias
> 
-- 
All Rights Reversed.

Download attachment "signature.asc" of type "application/pgp-signature" (489 bytes)