linux-kernel - Re: [PATCH 01/40] PCI: fix default vga ref

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 21 Sep 2012 14:52:13 -0600
From:	Bjorn Helgaas <bhelgaas@...gle.com>
To:	Yinghai Lu <yinghai@...nel.org>
Cc:	Len Brown <lenb@...nel.org>,
	Taku Izumi <izumi.taku@...fujitsu.com>,
	Jiang Liu <jiang.liu@...wei.com>, x86 <x86@...nel.org>,
	Andrew Morton <akpm@...ux-foundation.org>,
	Linus Torvalds <torvalds@...ux-foundation.org>,
	Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
	linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org,
	linux-acpi@...r.kernel.org, Dave Airlie <airlied@...hat.com>,
	Julia Lawall <julia@...u.dk>, Matthew Garrett <mjg@...hat.com>
Subject: Re: [PATCH 01/40] PCI: fix default vga ref_count

On Wed, Sep 19, 2012 at 12:54 PM, Yinghai Lu <yinghai@...nel.org> wrote:
> when __ARCH_HAS_VGA_DEFAULT_DEVICE is not defined, aka EFIFB is not used,
> for static path, vga_default setting is through vga_arbiter_add_pci_device.
> and later x86 pci_fixup_video, will skip setting again.
> - subsys_initcall(vga_arb_device_init) come first to call vga_arbiter_add_pci_device. It will call pci_get_dev to hold one reference.
>
> for hotplug add path, even vga_arbiter_add_pci_device is called via notifier,
> but it will check VGA_RSRC_LEGACY_MASK that is not set for hotplug path.
> So x86 pci_fixup_video will take over to call vga_set_default_device().
> It will not hold one refrence.
>
> Later for hotplug remove path, vga_arbiter_del_pci_device that does not check
> VGA_RSRC_LEGACY_MASK will call put_device and it will cause ref_count to
> decrease extra. that will have that pci device get deleted early wrongly.
>
> Need to make get/put balance for both cases.
>
> -v2: According to Bjorn, update vga_set_default_device instead...
>
> Signed-off-by: Yinghai Lu <yinghai@...nel.org>
> Cc: x86@...nel.org
> Cc: Dave Airlie <airlied@...hat.com>
> Cc: Andrew Morton <akpm@...ux-foundation.org>
> Cc: Julia Lawall <julia@...u.dk>
> Cc: Matthew Garrett <mjg@...hat.com>
> ---
>  drivers/gpu/vga/vgaarb.c |   14 +++++++++-----
>  1 files changed, 9 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/gpu/vga/vgaarb.c b/drivers/gpu/vga/vgaarb.c
> index b6852b7..0888951e 100644
> --- a/drivers/gpu/vga/vgaarb.c
> +++ b/drivers/gpu/vga/vgaarb.c
> @@ -141,6 +141,12 @@ EXPORT_SYMBOL_GPL(vga_default_device);
>
>  void vga_set_default_device(struct pci_dev *pdev)
>  {
> +       if (vga_default)
> +               pci_dev_put(vga_default);
> +
> +       if (pdev)
> +               pdev = pci_dev_get(pdev);
> +
>         vga_default = pdev;

I think this is equivalent to:

    pci_dev_put(vga_default);
    vga_default = pci_dev_get(pdev);

I haven't seen an answer to Matthew's question about whether we want
to check for "vga_default == pdev".

>  }
>  #endif
> @@ -577,7 +583,7 @@ static bool vga_arbiter_add_pci_device(struct pci_dev *pdev)
>  #ifndef __ARCH_HAS_VGA_DEFAULT_DEVICE
>         if (vga_default == NULL &&
>             ((vgadev->owns & VGA_RSRC_LEGACY_MASK) == VGA_RSRC_LEGACY_MASK))
> -               vga_default = pci_dev_get(pdev);
> +               vga_set_default_device(pdev);
>  #endif
>
>         vga_arbiter_check_bridge_sharing(vgadev);
> @@ -613,10 +619,8 @@ static bool vga_arbiter_del_pci_device(struct pci_dev *pdev)
>         }
>
>  #ifndef __ARCH_HAS_VGA_DEFAULT_DEVICE
> -       if (vga_default == pdev) {
> -               pci_dev_put(vga_default);
> -               vga_default = NULL;
> -       }
> +       if (vga_default == pdev)
> +               vga_set_default_device(NULL);
>  #endif
>
>         if (vgadev->decodes & (VGA_RSRC_LEGACY_IO | VGA_RSRC_LEGACY_MEM))
> --
> 1.7.7
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/