drm/rockchip: shutdown drm subsystem on shutdown
Subsystems (5)
PATCH v2 1/1 drm/rockchip: shutdown drm subsystem on shutdown
From: Vicente Bergas <[email protected]>
As explained by Robin Murphy:
Quoted 3 lines
the IOMMU shutdown disables paging, so if the VOP is still scanning out then that will result in whatever IOVAs it was using now going straight out onto the bus as physical addresses.
We had a more radical approach before in commit7f3ef5dedb14 ("drm/rockchip: Allow driver to be shutdown on reboot/kexec") but that resulted in new warnings and oopses on shutdown on rk3399 chromeos devices.
So second try is resurrecting Vicentes shutdown change which should achieve the same result but in a less drastic way.
--- drivers/gpu/drm/rockchip/rockchip_drm_drv.c | 9 +++++++++ 1 file changed, 9 insertions(+)
drivers/gpu/drm/rockchip/rockchip_drm_drv.c +9 -0
@@ -448,6 +448,14 @@ static int rockchip_drm_platform_remove(struct platform_device *pdev)return 0 ; } +static void rockchip_drm_platform_shutdown (struct platform_device *pdev) +{ +struct drm_device *drm =platform_get_drvdata (pdev); + +if (drm) +drm_atomic_helper_shutdown (drm); +} +static const struct of_device_id rockchip_drm_dt_ids[] = { { .compatible ="rockchip,display-subsystem" , }, {/* sentinel */ },
@@ -457,6 +465,7 @@ MODULE_DEVICE_TABLE(of, rockchip_drm_dt_ids);static struct platform_driver rockchip_drm_platform_driver = { .probe = rockchip_drm_platform_probe, .remove = rockchip_drm_platform_remove, + .shutdown = rockchip_drm_platform_shutdown, .driver = { .name ="rockchip-drm" , .of_match_table = rockchip_drm_dt_ids,
Signature · 1 line
2.20.1
Quoted 8 lines · Heiko Stuebner
From: Vicente Bergas <[email protected]>
As explained by Robin Murphy:
Quoted 3 lines
the IOMMU shutdown disables paging, so if the VOP is still scanning out then that will result in whatever IOVAs it was using now going straight out onto the bus as physical addresses.
Quoted 7 lines
We had a more radical approach before in commit
7f3ef5dedb14("drm/rockchip: Allow driver to be shutdown on reboot/kexec") but that resulted in new warnings and oopses on shutdown on rk3399 chromeos devices.So second try is resurrecting Vicentes shutdown change which should achieve the same result but in a less drastic way.
Works without complaints on RK3399 Scarlet devices:
Can't vouch for kexec.
Hi,
Quoted 28 lines · Heiko Stuebner
From: Vicente Bergas <[email protected]>
As explained by Robin Murphy:
Quoted 3 lines
the IOMMU shutdown disables paging, so if the VOP is still scanning out then that will result in whatever IOVAs it was using now going straight out onto the bus as physical addresses.
We had a more radical approach before in commit
7f3ef5dedb14("drm/rockchip: Allow driver to be shutdown on reboot/kexec") but that resulted in new warnings and oopses on shutdown on rk3399 chromeos devices.So second try is resurrecting Vicentes shutdown change which should achieve the same result but in a less drastic way.
Fixes 63238173b2fa ("Revert drm/rockchip: Allow driver to be shutdown on reboot/kexec")Cc Jeffy Chen <[email protected]>Cc Robin Murphy <[email protected]>Cc Marc Zyngier <[email protected]>Cc Brian Norris <[email protected]>Cc Doug Anderson <[email protected]>Suggested-by JeffyChen <[email protected]>Suggested-by Robin Murphy <[email protected]>Signed-off-by Vicente Bergas <[email protected]>[adapted commit message to explain the history]Signed-off-by Heiko Stuebner <[email protected]>
Quoted 3 lines
--- drivers/gpu/drm/rockchip/rockchip_drm_drv.c | 9 +++++++++ 1 file changed, 9 insertions(+)
I picked this back to the Chrome OS 4.19 kernel and tried it on rk3288-veyron-jerry. Rebooting caused no complaints (I confirmed this code was running by adding a printk).
I plugged in to an HDMI monitor. Rebooting caused no complaints.
I then let the display idle off (just in case that mattered). Again no complaints.
Thus, in case it's helpful:
Hi Heiko,
Quoted 15 lines · Heiko Stuebner
From: Vicente Bergas <[email protected]>
As explained by Robin Murphy:
Quoted 3 lines
the IOMMU shutdown disables paging, so if the VOP is still scanning out then that will result in whatever IOVAs it was using now going straight out onto the bus as physical addresses.
We had a more radical approach before in commit
7f3ef5dedb14("drm/rockchip: Allow driver to be shutdown on reboot/kexec") but that resulted in new warnings and oopses on shutdown on rk3399 chromeos devices.So second try is resurrecting Vicentes shutdown change which should achieve the same result but in a less drastic way.
Quoted 12 lines
Fixes 63238173b2fa ("Revert drm/rockchip: Allow driver to be shutdown on reboot/kexec")Cc Jeffy Chen <[email protected]>Cc Robin Murphy <[email protected]>Cc Marc Zyngier <[email protected]>Cc Brian Norris <[email protected]>Cc Doug Anderson <[email protected]>Suggested-by JeffyChen <[email protected]>Suggested-by Robin Murphy <[email protected]>Signed-off-by Vicente Bergas <[email protected]>[adapted commit message to explain the history]Signed-off-by Heiko Stuebner <[email protected]>
Sorry it took so long to test this. I've just given it a go on kevin, and managed to kexec into a secondary kernel. So FWIW:
M.
Signature · 1 line
Jazz is not dead. It just smells funny...
Am Dienstag, 2. April 2019, 13:37:53 CEST schrieb Heiko Stuebner:
Quoted 15 lines
From: Vicente Bergas <[email protected]>
As explained by Robin Murphy:
Quoted 3 lines
the IOMMU shutdown disables paging, so if the VOP is still scanning out then that will result in whatever IOVAs it was using now going straight out onto the bus as physical addresses.
We had a more radical approach before in commit
7f3ef5dedb14("drm/rockchip: Allow driver to be shutdown on reboot/kexec") but that resulted in new warnings and oopses on shutdown on rk3399 chromeos devices.So second try is resurrecting Vicentes shutdown change which should achieve the same result but in a less drastic way.
Quoted 12 lines
Fixes 63238173b2fa ("Revert drm/rockchip: Allow driver to be shutdown on reboot/kexec")Cc Jeffy Chen <[email protected]>Cc Robin Murphy <[email protected]>Cc Marc Zyngier <[email protected]>Cc Brian Norris <[email protected]>Cc Doug Anderson <[email protected]>Suggested-by JeffyChen <[email protected]>Suggested-by Robin Murphy <[email protected]>Signed-off-by Vicente Bergas <[email protected]>[adapted commit message to explain the history]Signed-off-by Heiko Stuebner <[email protected]>
applied to drm-misc-next
Not fixes, to give it more exposure this time. It still has a stable tag so should nevertheless trickle down to stable in time as well.
Thanks for all the testing
Heiko