linux-kernel - Re: [PATCH] ALSA: hda/hdmi - Fix duplicate unref of pci

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:   Tue, 10 Dec 2019 17:13:45 +0100
From:   Lukas Wunner <lukas@...ner.de>
To:     "Deucher, Alexander" <Alexander.Deucher@....com>
Cc:     Takashi Iwai <tiwai@...e.de>, Jaroslav Kysela <perex@...ex.cz>,
        Mika Westerberg <mika.westerberg@...ux.intel.com>,
        Bjorn Helgaas <helgaas@...nel.org>,
        Nicholas Johnson <nicholas.johnson-opensource@...look.com.au>,
        "alsa-devel@...a-project.org" <alsa-devel@...a-project.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-pci@...r.kernel.org" <linux-pci@...r.kernel.org>
Subject: Re: [PATCH] ALSA: hda/hdmi - Fix duplicate unref of pci_dev

On Tue, Dec 10, 2019 at 03:53:20PM +0000, Deucher, Alexander wrote:
> > On Tue, Dec 10, 2019 at 03:34:27PM +0000, Deucher, Alexander wrote:
> > > > Nicholas Johnson reports a null pointer deref as well as a refcount
> > > > underflow upon hot-removal of a Thunderbolt-attached AMD eGPU.
> > > > He's bisected the issue down to commit 586bc4aab878 ("ALSA: hda/hdmi
> > > > - fix vgaswitcheroo detection for AMD").
> > > >
> > > > The commit iterates over PCI devices using pci_get_class() and
> > > > unreferences each device found, even though pci_get_class()
> > > > subsequently unreferences the device as well.  Fix it.
> > >
> > > The pci_dev_put() a few lines above should probably be dropped as well.
> > 
> > That one looks fine to me.  The refcount is already increased in the caller
> > get_bound_vga() via pci_get_domain_bus_and_slot() and it's increased
> > again in atpx_present() via pci_get_class().  It needs to be decremented in
> > atpx_present() to avoid leaking a ref.
> 
> I'm not following.  This is part of the same loop as the one you removed.
> All we are doing is checking whether the ATPX method exists or not om the
> platform.  The pdev may not be the same one as the one in
> pci_get_domain_bus_and_slot().  The APTX method in the APU's ACPI namespace,
> not the dGPUs.

Okay.  Still, atpx_present() doesn't pass the found pci_dev back to the
caller, so it would be leaked if the ref isn't returned.

The situation is different for the pci_dev_put() I removed:  The ref is
returned by pci_get_class() on the next loop iteration.

Thanks,

Lukas

> > > > diff --git a/sound/pci/hda/hda_intel.c b/sound/pci/hda/hda_intel.c
> > > > index 35b4526f0d28..b856b89378ac 100644
> > > > --- a/sound/pci/hda/hda_intel.c
> > > > +++ b/sound/pci/hda/hda_intel.c
> > > > @@ -1419,7 +1419,6 @@ static bool atpx_present(void)
> > > >  				return true;
> > > >  			}
> > > >  		}
> > > > -		pci_dev_put(pdev);
> > > >  	}
> > > >  	return false;
> > > >  }
> > > > --
> > > > 2.24.0