linux-kernel - Re: [PATCH] usb: dwc3: Clear DWC3_EVENT

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]

Date:   Thu, 2 Feb 2023 20:06:56 +0000
From:   Thinh Nguyen <Thinh.Nguyen@...opsys.com>
To:     Linyu Yuan <quic_linyyuan@...cinc.com>
CC:     Thinh Nguyen <Thinh.Nguyen@...opsys.com>,
        정재훈 <jh0801.jung@...sung.com>,
        'Felipe Balbi' <balbi@...nel.org>,
        'Greg Kroah-Hartman' <gregkh@...uxfoundation.org>,
        "'open list:USB XHCI DRIVER'" <linux-usb@...r.kernel.org>,
        'open list' <linux-kernel@...r.kernel.org>,
        'Seungchull Suh' <sc.suh@...sung.com>,
        'Daehwan Jung' <dh10.jung@...sung.com>
Subject: Re: [PATCH] usb: dwc3: Clear DWC3_EVENT_PENDING when count is 0

On Thu, Feb 02, 2023, Linyu Yuan wrote:
> 
> On 2/2/2023 2:57 AM, Thinh Nguyen wrote:
> > On Tue, Jan 31, 2023, Linyu Yuan wrote:
> > > hi Thinh,
> > > 
> > > 
> > > regarding your suggestion, assume it is not PCIe type,  still have one
> > > question,
> > > 
> > > 
> > > -       if (evt->flags & DWC3_EVENT_PENDING)
> > > +       if (evt->flags & DWC3_EVENT_PENDING) {
> > > +               if (!evt->count) {
> > > +                       u32 reg = dwc3_readl(dwc->regs, DWC3_GEVNTSIZ(0));
> > > +
> > > +                       if (!(reg & DWC3_GEVNTSIZ_INTMASK))
> > > +                               evt->flags &= ~DWC3_EVENT_PENDING;
> > > 
> > > do we need to return IRQ_WAKE_THREAD  ?
> > No, if evt->count is 0, but GEVNTCOUNT is > 0, the controller will
> > generate interrupt. The evt->count will be updated and the events will
> > be handled on the next interrupt.
> 
> 
> when will next interrupt happen ?

Immediately after. You can test this by just return IRQ_HANDLED and not
clear the GEVNTCOUNT to see its behavior.

> 
> as when enter here, i guess GEVENTCOUNT is already > 0, but we didn't read
> it.

GEVNTCOUNT is always updating as new events are generated. We only clear
however many events we process, but that doesn't stop it from
incrementing.

BR,
Thinh

> 
> 
> > 
> > > +               }
> > >                  return IRQ_HANDLED;
> > > 
> > > as here return IRQ HANDLED, how can we make sure a new IRQ will be handled
> > > after previous IRQ thread clean PENDING flag ?
> > If evt->count > 0, that means the bottom half is still running. So,
> > leave it be. If evt->count == 0, then the cached events are processed,
> > we're safe to clear the PENDING flag. New interrupt will be generated if
> > GEVNTCOUNT is > 0.
> > 
> > > +       }
> > > 
> > > 
> > > also for non-PCIe controller, consider IRQ mask register working correctly,
> > > 
> > > consider a case IRQ happen before IRQ thread exit,  here just return
> > > IRQ_HANDLED.
> > > 
> > > once IRQ thread exit, it will clean PENDING flag, so next IRQ event will run
> > > normally.
> > > 
> > > if 정재훈 saw PENDING flag is not cleared, does it mean IRQ thread have no
> > > chance to exit ?
> > The PENDING flag should be cleared eventually when the bottom half
> > completes. I don't expect the interrupt storm to block the IRQ thread
> > forever, but I can't guarantee the device behavor. 정재훈 can confirm.
> > This change should resolve the interrupt storm.
> > 
> > BR,
> > Thinh