linux-kernel - Re: [PATCH] drm/i915: fix corruptions on i8xx due to relaxed fencing

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: <2e9ca0040729649ed989a16a07bb2c49.squirrel@webmail.greenhost.nl>
Date:	Thu, 10 Mar 2011 11:36:15 +0100 (CET)
From:	"Indan Zupancic" <indan@....nu>
To:	"Daniel Vetter" <daniel@...ll.ch>
Cc:	"Daniel Vetter" <daniel@...ll.ch>,
	"Daniel Vetter" <daniel.vetter@...ll.ch>,
	"Chris Wilson" <chris@...is-wilson.co.uk>,
	dri-devel@...ts.freedesktop.org,
	"LKML" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] drm/i915: fix corruptions on i8xx due to relaxed 
     fencing

On Thu, March 10, 2011 08:52, Daniel Vetter wrote:
> Am Do, 10.03.2011, 06:06 schrieb Indan Zupancic:
>> This isn't in rc8, can someone make sure it gets into 2.6.38-rc9/2.6.38?
>
> The patch unfortunately broke gen4+ (more precisely: the unwritten abi
> guarantee that userspace can tile buffers that don't have a complete last
> tile row, as long as it promises not to touch it). Hence it got reverted.
> It was just a enforcement check to prevent broken userspace from fooling
> itself. The proper fix is to upgrade your userspace (libdrm +
> xf86-video-intel).
>
> [Aside: This only happens if you have new enough userspace that supports
> relaxed tiling. Old userspace and new kernels are _not_ broken.]

Yes, I noticed it got reverted when digging into git log, but the
commit message only said it broke gen4+, not how gen2 should be
unbroken after the revert.

Which versions fix this, just for reference?

I got libdrm 2.4.23 and xf86-video-intel 2.14.0.

(As a side note, do you have version checking on the kernel side?
If so, you could return 0 for the relaxed fencing feature check
if the driver is the wrong version.)

To keep things manageable I either upgrade the kernel, or userspace,
but never both at once. It seems that that doesn't work with graphics.

Thanks,

Indan


--
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/