Regression - Xorg start failed

Alex Riesen raa.lkml at gmail.com
Sun Feb 13 00:15:46 PST 2011


On Sun, Feb 13, 2011 at 07:22, Dave Young <hidave.darkstar at gmail.com> wrote:
> Finally I bisected it, results:
> 47970b1b2aa64464bc0a9543e86361a622ae7c03 is first bad commit
> commit 47970b1b2aa64464bc0a9543e86361a622ae7c03
> Author: Chris Wright <chrisw at sous-sol.org>
> Date:   Thu Feb 10 15:58:56 2011 -0800
>
>    pci: use security_capable() when checking capablities during config space read
>
>    Eric Paris noted that commit de139a3 ("pci: check caps from sysfs file
>    open to read device dependent config space") caused the capability check
>    to bypass security modules and potentially auditing.  Rectify this by
>    calling security_capable() when checking the open file's capabilities
>    for config space reads.
>
>    Reported-by: Eric Paris <eparis at redhat.com>
>    Signed-off-by: Chris Wright <chrisw at sous-sol.org>
>    Signed-off-by: James Morris <jmorris at namei.org>
>

Actually, even reading the PCI capabilities fails with lspci
reporting "Capabilities: <access denied>" if run as root.
"libpciaccess" should have handled this situation, but still
it looks like a regression and it breaks existing systems.


More information about the dri-devel mailing list