linux-kernel - Re: [PATCH 0/8] drm/rockchip: Flip wait clean-up

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:   Thu, 15 Sep 2016 10:02:26 -0400
From:   Sean Paul <seanpaul@...omium.org>
To:     Tomasz Figa <tfiga@...omium.org>
Cc:     dri-devel <dri-devel@...ts.freedesktop.org>,
        Linux ARM Kernel <linux-arm-kernel@...ts.infradead.org>,
        linux-rockchip@...ts.infradead.org,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Heiko Stuebner <heiko@...ech.de>,
        David Airlie <airlied@...ux.ie>,
        Mark Yao <mark.yao@...k-chips.com>,
        Daniel Kurtz <djkurtz@...omium.org>
Subject: Re: [PATCH 0/8] drm/rockchip: Flip wait clean-up

On Wed, Sep 14, 2016 at 8:54 AM, Tomasz Figa <tfiga@...omium.org> wrote:
> The display controller found on Rockchip SoCs supported by Rockchip DRM
> driver (VOP) is a bit problematic, because it does not provide hardware
> vblank counter. Because vblank interrupt is used to feed the software
> counter, the driver had custom code to wait for flip completion to avoid
> race between atomic flush and vblank interrupt handler.
>
> This, however, brought a different set of issues. In fact, even with the
> custom wait code, there is stil a race between the handler and driver
> state update (of the values used to compare with registers to determine
> if flip has completed). On top of that, legacy cursor updates are not
> implemented properly and have to wait for vblank to complete, which
> is against the API specification.
>
> This series attempts to clean up the driver from this custom waiting code,
> eliminating related races and bringing correct handling of legacy cursor
> plane. It also gives a nice effect of more than 100 lines of code removed.
>
> This is a forward port of patches from 4.4 kernel used by ChromiumOS and
> tested there. Even though the code base has not changed significantly, it
> would be nice if someone with proper testing environment could give them
> a try.
>
> Based on for-next branch of Sean Paul's dogwood tree:
> https://cgit.freedesktop.org/~seanpaul/dogwood/log/?h=for-next
> git://people.freedesktop.org/~seanpaul/dogwood
>

Thanks for the patches, Tomasz.

I'll queue them up in my rockchip for-next branch.

Sean



> Tomasz Figa (8):
>   drm/rockchip: Clear interrupt status bits before enabling
>   drm/rockchip: Get rid of some unnecessary code
>   drm/rockchip: Avoid race with vblank count increment
>   drm/rockchip: Unreference framebuffers from flip work
>   drm/rockchip: Replace custom wait_for_vblanks with helper
>   drm/rockchip: Do not enable vblank without event
>   drm/rockchip: Always signal event in next vblank after cfg_done
>   drm/rockchip: Kill vop_plane_state
>
>  drivers/gpu/drm/rockchip/rockchip_drm_drv.h |   1 -
>  drivers/gpu/drm/rockchip/rockchip_drm_fb.c  |  64 +------
>  drivers/gpu/drm/rockchip/rockchip_drm_vop.c | 247 +++++++++++-----------------
>  3 files changed, 95 insertions(+), 217 deletions(-)
>
> --
> 2.8.0.rc3.226.g39d4020
>