linux-kernel - Re: Memory barrier needed with wake_up

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: <Pine.LNX.4.44L0.1701161207000.17898-100000@netrider.rowland.org>
Date:   Mon, 16 Jan 2017 12:09:20 -0500 (EST)
From:   Alan Stern <stern@...land.harvard.edu>
To:     Felipe Balbi <felipe.balbi@...ux.intel.com>
cc:     Peter Zijlstra <peterz@...radead.org>,
        "Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>,
        Ingo Molnar <mingo@...hat.com>,
        USB list <linux-usb@...r.kernel.org>,
        Kernel development list <linux-kernel@...r.kernel.org>,
        Will Deacon <will.deacon@....com>
Subject: Re: Memory barrier needed with wake_up_process()?

On Mon, 16 Jan 2017, Felipe Balbi wrote:

> Sorry for the long delay, I finally have more information on this. All
> this time I was doing something that I never considered to matter: I've
> been running host and peripheral on the same machine. Now that I have
> tracepoints on xHCI as well, I could see that these 30 seconds of
> "nothing" is actuall full of xHCI activity and I can see that for the
> duration of these 30 seconds preempt depth on the CPU that (eventually)
> queues a request on dwc3, is always > 1 (sometimes 2, most of the time
> 1). My conclusion from that is that xHCI (or usbcore ?!?) locks the CPU
> and g_mass_storage is spinning for over 30 seconds at which point
> storage.ko (host side class driver) dequeues the request.
> 
> I'll see if I can capture a fresh trace with both xHCI and dwc3 with
> this happening, but probably not today (testing stuff for -rc).

Does anything change if the host and peripheral are separate machines?

Alan Stern