linux-kernel - Re: [RFC, PATCH] imx: serial: Take tty->files

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:   Mon, 12 Jun 2017 14:04:54 +0200
From:   Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To:     Fabio Estevam <festevam@...il.com>
Cc:     Peter Senna Tschudin <peter.senna@...labora.com>,
        Alan Cox <gnomes@...rguk.ukuu.org.uk>,
        Rob Herring <robherring2@...il.com>,
        Andrey Smirnov <andrew.smirnov@...il.com>,
        "linux-serial@...r.kernel.org" <linux-serial@...r.kernel.org>,
        Chris Healy <cphealy@...il.com>, Jiri Slaby <jslaby@...e.com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [RFC, PATCH] imx: serial: Take tty->files_lock opportunistically

On Mon, Jun 12, 2017 at 08:55:44AM -0300, Fabio Estevam wrote:
> Hi Greg,
> 
> On Mon, Jun 12, 2017 at 3:46 AM, Greg Kroah-Hartman
> <gregkh@...uxfoundation.org> wrote:
> 
> >> linux-next is good, but mainline is not.
> >>
> >> Actually we need 4dec2f119e86f9c9 ("imx-serial: RX DMA startup
> >> latency") to be applied to 4.12-rc as well.
> >
> > Why?  Can't it wait until 4.13-rc1?
> 
> Because without 4dec2f119e86f9c9  the imx serial driver behaves badly
> in 4.12-rc.
> 
> Just tested 4.12-rc5 and there is scary verbose warning:
> 
> [   11.714904] =====================================================
> [   11.721006] WARNING: HARDIRQ-safe -> HARDIRQ-unsafe lock order detected
> [   11.727628] 4.12.0-rc5 #413 Not tainted
> [   11.731470] -----------------------------------------------------
> [   11.737572] rawtest/219 [HC0[0]:SC0[0]:HE0:SE1] is trying to acquire:
> [   11.744018]  (&(&tty->files_lock)->rlock){+.+...}, at: [<c04d83a0>]
> imx_startup+0x2c4/0x510
> [   11.752403]
> [   11.752403] and this task is already holding:
> [   11.758241]  (&port_lock_key){-.....}, at: [<c04d81e8>]
> imx_startup+0x10c/0x510
> [   11.765571] which would create a new lock dependency:
> [   11.770626]  (&port_lock_key){-.....} -> (&(&tty->files_lock)->rlock){+.+...}
> [   11.777791]
> [   11.777791] but this new dependency connects a HARDIRQ-irq-safe lock:
> [   11.785712]  (&port_lock_key){-.....}
> [   11.785726]
> ....
> 
> Complete stack at: https://pastebin.com/a2gUTwZK
> 
> This problem is caused by commit 18a4208826dd0 ("imx-serial: Reduce RX
> DMA startup latency when opening for reading").
> 
> linux-next has a fix for it: 4dec2f119e86f9c9 ("imx-serial: RX DMA
> startup latency")
> 
> Applying 4dec2f119e86f9c9 ("imx-serial: RX DMA startup latency") into
> 4.12-rc5 makes the issue go away.

Ok, if I end up with some more tty/serial fixes for 4.12-final, I'll
consider queueing this one up as well, thanks.

greg k-h