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: <20211013113356.GA1891412@bhelgaas>
Date:   Wed, 13 Oct 2021 06:33:56 -0500
From:   Bjorn Helgaas <helgaas@...nel.org>
To:     Andy Shevchenko <andy.shevchenko@...il.com>
Cc:     Uwe Kleine-König 
        <u.kleine-koenig@...gutronix.de>,
        linux-pci <linux-pci@...r.kernel.org>,
        Sascha Hauer <kernel@...gutronix.de>,
        Alexander Duyck <alexanderduyck@...com>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Andrew Donnellan <ajd@...ux.ibm.com>,
        Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
        Arnaldo Carvalho de Melo <acme@...nel.org>,
        Arnd Bergmann <arnd@...db.de>,
        Benjamin Herrenschmidt <benh@...nel.crashing.org>,
        Bjorn Helgaas <bhelgaas@...gle.com>,
        Borislav Petkov <bp@...en8.de>,
        Boris Ostrovsky <boris.ostrovsky@...cle.com>,
        "David S. Miller" <davem@...emloft.net>,
        Fiona Trahe <fiona.trahe@...el.com>,
        Frederic Barrat <fbarrat@...ux.ibm.com>,
        Giovanni Cabiddu <giovanni.cabiddu@...el.com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Herbert Xu <herbert@...dor.apana.org.au>,
        "H. Peter Anvin" <hpa@...or.com>, Ido Schimmel <idosch@...dia.com>,
        Ingo Molnar <mingo@...hat.com>, Jack Xu <jack.xu@...el.com>,
        Jakub Kicinski <kuba@...nel.org>,
        Jesse Brandeburg <jesse.brandeburg@...el.com>,
        Jiri Olsa <jolsa@...hat.com>, Jiri Pirko <jiri@...dia.com>,
        Juergen Gross <jgross@...e.com>,
        Konrad Rzeszutek Wilk <konrad.wilk@...cle.com>,
        Marco Chiappero <marco.chiappero@...el.com>,
        Mark Rutland <mark.rutland@....com>,
        Mathias Nyman <mathias.nyman@...el.com>,
        Michael Buesch <m@...s.ch>,
        Michael Ellerman <mpe@...erman.id.au>,
        Namhyung Kim <namhyung@...nel.org>,
        Oliver O'Halloran <oohall@...il.com>,
        Paul Mackerras <paulus@...ba.org>,
        Peter Zijlstra <peterz@...radead.org>,
        Rafał Miłecki <zajec5@...il.com>,
        Russell Currey <ruscur@...sell.cc>,
        Salil Mehta <salil.mehta@...wei.com>,
        Sathya Prakash <sathya.prakash@...adcom.com>,
        Simon Horman <simon.horman@...igine.com>,
        Sreekanth Reddy <sreekanth.reddy@...adcom.com>,
        Stefano Stabellini <sstabellini@...nel.org>,
        Suganath Prabu Subramani 
        <suganath-prabu.subramani@...adcom.com>,
        Taras Chornyi <tchornyi@...vell.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Tomaszx Kowalik <tomaszx.kowalik@...el.com>,
        Vadym Kochan <vkochan@...vell.com>,
        Wojciech Ziemba <wojciech.ziemba@...el.com>,
        Yisen Zhuang <yisen.zhuang@...wei.com>,
        Zhou Wang <wangzhou1@...ilicon.com>,
        linux-crypto <linux-crypto@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        linux-perf-users@...r.kernel.org,
        "open list:LINUX FOR POWERPC PA SEMI PWRFICIENT" 
        <linuxppc-dev@...ts.ozlabs.org>,
        linux-scsi <linux-scsi@...r.kernel.org>,
        USB <linux-usb@...r.kernel.org>,
        "open list:TI WILINK WIRELES..." <linux-wireless@...r.kernel.org>,
        MPT-FusionLinux.pdl@...adcom.com, netdev <netdev@...r.kernel.org>,
        oss-drivers@...igine.com, qat-linux@...el.com,
        "maintainer:X86 ARCHITECTURE (32-BIT AND 64-BIT)" <x86@...nel.org>,
        xen-devel@...ts.xenproject.org
Subject: Re: [PATCH v6 00/11] PCI: Drop duplicated tracking of a pci_dev's
 bound driver

On Wed, Oct 13, 2021 at 12:26:42PM +0300, Andy Shevchenko wrote:
> On Wed, Oct 13, 2021 at 2:33 AM Bjorn Helgaas <helgaas@...nel.org> wrote:
> > On Mon, Oct 04, 2021 at 02:59:24PM +0200, Uwe Kleine-König wrote:
> 
> > I split some of the bigger patches apart so they only touched one
> > driver or subsystem at a time.  I also updated to_pci_driver() so it
> > returns NULL when given NULL, which makes some of the validations
> > quite a bit simpler, especially in the PM code in pci-driver.c.
> 
> It's a bit unusual. Other to_*_dev() are not NULL-aware IIRC.

It is a little unusual.  I only found three of 77 that are NULL-aware:

  to_moxtet_driver()
  to_siox_driver()
  to_spi_driver()

It seems worthwhile to me because it makes the patch and the resulting
code significantly cleaner.  Here's one example without the NULL
check:

  @@ -493,12 +493,15 @@ static void pci_device_remove(struct device *dev)
   static void pci_device_shutdown(struct device *dev)
   {
          struct pci_dev *pci_dev = to_pci_dev(dev);
  -       struct pci_driver *drv = pci_dev->driver;

          pm_runtime_resume(dev);

  -       if (drv && drv->shutdown)
  -               drv->shutdown(pci_dev);
  +       if (pci_dev->dev.driver) {
  +               struct pci_driver *drv = to_pci_driver(pci_dev->dev.driver);
  +
  +               if (drv->shutdown)
  +                       drv->shutdown(pci_dev);
  +       }

  static void pci_device_shutdown(struct device *dev)
  {
    struct pci_dev *pci_dev = to_pci_dev(dev);

    pm_runtime_resume(dev);

    if (pci_dev->dev.driver) {
      struct pci_driver *drv = to_pci_driver(pci_dev->dev.driver);

      if (drv->shutdown)
        drv->shutdown(pci_dev);
    }

and here's the same thing with the NULL check:

  @@ -493,7 +493,7 @@ static void pci_device_remove(struct device *dev)
   static void pci_device_shutdown(struct device *dev)
   {
          struct pci_dev *pci_dev = to_pci_dev(dev);
  -       struct pci_driver *drv = pci_dev->driver;
  +       struct pci_driver *drv = to_pci_driver(dev->driver);

  static void pci_device_shutdown(struct device *dev)
  {
    struct pci_dev *pci_dev = to_pci_dev(dev);
    struct pci_driver *drv = to_pci_driver(dev->driver);

    pm_runtime_resume(dev);

    if (drv && drv->shutdown)
      drv->shutdown(pci_dev);

> >  static bool match_id(struct pci_dev *pdev, unsigned short vendor, unsigned short device)
> >  {
> > +       struct pci_driver *drv = to_pci_driver(pdev->dev.driver);
> >         const struct pci_device_id *id;
> >
> >         if (pdev->vendor == vendor && pdev->device == device)
> >                 return true;
> 
> > +       for (id = drv ? drv->id_table : NULL; id && id->vendor; id++)
> > +               if (id->vendor == vendor && id->device == device)
> 
> > +                       break;
> 
> return true;
> 
> >         return id && id->vendor;
> 
> return false;

Good cleanup for a follow-up patch, but doesn't seem directly related
to the objective here.  The current patch is:

  @@ -80,7 +80,7 @@ static struct resource video_rom_resource = {
    */
   static bool match_id(struct pci_dev *pdev, unsigned short vendor, unsigned short device)
   {
  -       struct pci_driver *drv = pdev->driver;
  +       struct pci_driver *drv = to_pci_driver(pdev->dev.driver);
          const struct pci_device_id *id;

          if (pdev->vendor == vendor && pdev->device == device)

> >         device_lock(&vf_dev->dev);
> > -       if (vf_dev->dev.driver) {
> > +       if (to_pci_driver(vf_dev->dev.driver)) {
> 
> Hmm...

Yeah, it could be either of:

  if (to_pci_driver(vf_dev->dev.driver))
  if (vf_dev->dev.driver)

I went back and forth on that and went with to_pci_driver() on the
theory that we were testing the pci_driver * before and the patch is
more of a mechanical change and easier to review if we test the
pci_driver * after.

> > +               if (!pci_dev->state_saved && pci_dev->current_state != PCI_D0
> 
> > +                   && pci_dev->current_state != PCI_UNKNOWN) {
> 
> Can we keep && on the previous line?

I think this is in pci_legacy_suspend(), and I didn't touch that line.
It shows up in the interdiff because without the NULL check in
to_pci_driver(), we had to indent this code another level.  With the
NULL check, we don't need that extra indentation.

> > +                       pci_WARN_ONCE(pci_dev, pci_dev->current_state != prev,
> > +                                     "PCI PM: Device state not saved by %pS\n",
> > +                                     drv->suspend);
> >                 }
> 
> ...
> 
> > +       return drv && drv->resume ?
> > +                       drv->resume(pci_dev) : pci_pm_reenable_device(pci_dev);
> 
> One line?

I don't think I touched that line.

> > +       struct pci_driver *drv = to_pci_driver(dev->dev.driver);
> >         const struct pci_error_handlers *err_handler =
> > -                       dev->dev.driver ? to_pci_driver(dev->dev.driver)->err_handler : NULL;
> > +                       drv ? drv->err_handler : NULL;
> 
> Isn't dev->driver == to_pci_driver(dev->dev.driver)?

Yes, I think so, but not sure what you're getting at here, can you
elaborate?

> >         device_lock(&dev->dev);
> > +       pdrv = to_pci_driver(dev->dev.driver);
> >         if (!pci_dev_set_io_state(dev, state) ||
> > -               !dev->dev.driver ||
> > -               !(pdrv = to_pci_driver(dev->dev.driver))->err_handler ||
> 
> > +               !pdrv ||
> > +               !pdrv->err_handler ||
> 
> One line now?
> 
> >                 !pdrv->err_handler->error_detected) {
> 
> Or this and the previous line?

Could, but the "dev->driver" to "to_pci_driver(dev->dev.driver)"
changes are the heart of this patch, and I don't like to clutter it
with unrelated changes.

> > -       result = PCI_ERS_RESULT_NONE;
> >
> >         pcidev = pci_get_domain_bus_and_slot(domain, bus, devfn);
> >         if (!pcidev || !pcidev->dev.driver) {
> >                 dev_err(&pdev->xdev->dev, "device or AER driver is NULL\n");
> >                 pci_dev_put(pcidev);
> > -               return result;
> > +               return PCI_ERS_RESULT_NONE;
> >         }
> >         pdrv = to_pci_driver(pcidev->dev.driver);
> 
> What about splitting the conditional to two with clear error message
> in each and use pci_err() in the second one?

Could possibly be cleaned up.  Felt like feature creep so I didn't.

> >                 default:
> >                         dev_err(&pdev->xdev->dev,
> > -                               "bad request in aer recovery "
> > -                               "operation!\n");
> > +                               "bad request in AER recovery operation!\n");
> 
> Stray change? Or is it in a separate patch in your tree?

Could be skipped.  The string now fits on one line so I combined it to
make it more greppable.

Bjorn

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ