After completing the initialization of serdes in uboot, enter the kernel
to perform initialization again
Change-Id: I689c30f55d9e259e5adaa95029f7523026cbed0a
Signed-off-by: Zitong Cai <zitong.cai@rock-chips.com>
After completing the initialization of serdes in uboot, enter the kernel
to perform initialization again
Change-Id: Ia603bd2b8892f1647d0c0fa37f38e1422f78a6b3
Signed-off-by: Zitong Cai <zitong.cai@rock-chips.com>
If the cluster sub wins are enabled, there will be POST_BUF_EMPTY
errors after entering the PSR mode, because cluster main wins are
disabled while the sub wins still are active. The patch can help
avoid these unexpected POST_BUF_EMPTY errors.
Change-Id: I3555e0c08c7c4eacf0ace472e3154693c03c6aa9
Signed-off-by: Damon Ding <damon.ding@rock-chips.com>
Psr mode can help reduce power consumption when using the mcu panel,
which supports to refresh the image on its own while it remains
unchanged.
Change-Id: Ib9bd1bb472c996277eb374cc9ba1433ce3930d55
Signed-off-by: Damon Ding <damon.ding@rock-chips.com>
According to the following commits:
commit 9d0e3cac35 ("drm/atomic: Allow vblank-enabled + self-refresh "disable"")
commit 2bdba9d4a3 ("drm/rockchip: vop: Leave vblank enabled in self-refresh")
It is better to leave vblank enabled after entering the PSR mode, in
order to ensure that the vblank APIs (like DRM_IOCTL_WAIT_VBLANK) can
work in PSR mode. User space is not aware when the display system is
already in PSR mode, so this appears to be an API violation.
Without this patch, there may be a warning according to the following
check in drivers/gpu/drm/drm_atomic_helper.c:
static void
disable_outputs(struct drm_device *dev, struct drm_atomic_state *old_state)
{
......
ret = drm_crtc_vblank_get(crtc);
/*
* Self-refresh is not a true "disable"; ensure vblank remains
* enabled.
*/
if (new_crtc_state->self_refresh_active)
WARN_ONCE(ret != 0,
"driver disabled vblank in self-refresh\n");
else
WARN_ONCE(ret != -EINVAL,
"driver forgot to call drm_crtc_vblank_off()\n");
if (ret == 0)
drm_crtc_vblank_put(crtc);
......
}
Change-Id: I95ee15864c1dfa1045c41c7a9ed2985bb6107f2b
Signed-off-by: Damon Ding <damon.ding@rock-chips.com>
According to the following commits:
commit 9d0e3cac35 ("drm/atomic: Allow vblank-enabled + self-refresh "disable"")
commit 2bdba9d4a3 ("drm/rockchip: vop: Leave vblank enabled in self-refresh")
It is better to leave vblank enabled after entering the PSR mode, in
order to ensure that the vblank APIs (like DRM_IOCTL_WAIT_VBLANK) can
work in PSR mode. User space is not aware when the display system is
already in PSR mode, so this appears to be an API violation.
Without this patch, there may be a warning according to the following
check in drivers/gpu/drm/drm_atomic_helper.c:
static void
disable_outputs(struct drm_device *dev, struct drm_atomic_state *old_state)
{
......
ret = drm_crtc_vblank_get(crtc);
/*
* Self-refresh is not a true "disable"; ensure vblank remains
* enabled.
*/
if (new_crtc_state->self_refresh_active)
WARN_ONCE(ret != 0,
"driver disabled vblank in self-refresh\n");
else
WARN_ONCE(ret != -EINVAL,
"driver forgot to call drm_crtc_vblank_off()\n");
if (ret == 0)
drm_crtc_vblank_put(crtc);
......
}
Change-Id: Ide735636ab53fd6977de5034b8cfc4c8341220af
Signed-off-by: Damon Ding <damon.ding@rock-chips.com>
The module NE160QAM-NX1 support eDP v1.4b with the following
features:
- Support resolutions 3840x2400p60 and 3840x2400p120
- 4 lanes eDP interface with 8.1Gbps link rates
- DPCD v1.4
- PSR(Panel self Refresh)
- DSC(Display Stream Compression)
Change-Id: I41361dc5500cd35234bfdc8436a50e3117c6fba2
Signed-off-by: Damon Ding <damon.ding@rock-chips.com>
The module NE160QAM-NX1 support eDP v1.4b with the following
features:
- Support resolutions 3840x2400p60 and 3840x2400p120
- 4 lanes eDP interface with 8.1Gbps link rates
- DPCD v1.4
- PSR(Panel self Refresh)
- DSC(Display Stream Compression)
Change-Id: Ibd0472e00cef827900ef78320ef08fc372e265d3
Signed-off-by: Damon Ding <damon.ding@rock-chips.com>
The &link_train.link_rate and &link_train.lane_count are used as the
max_link_rate and max_lane_count to check the bandwidth limitation
, select output format and so on. However, they both also are used as
the actual link_rate and lane_count for the eDP controller and phy.
So we add &link_train.max_link_rate and &link_train.max_lane_count
in order to avoid the confusion in the use of above variables.
On the other hand, some unexpected errors will occur without this
patch in some cases.
When I test the PSR function of eDP panel module NE160QAM-NX1, the
link rate will be set to the 8.1G, which is the max supported link
rate of this module, after exiting the PSR state, but it should be
5.4G corresponding to 3840x2400p60 resolution.
analogix_dp_detect(): set the &link_train.link_rate to 8.1G
-> analogix_dp_fast_link_train():
use the &link_train.link_rate as the previous link config
-> analogix_dp_set_link_bandwidth()
-> phy_configure():
error: invalid link_rate config 8.1G
Change-Id: I1a4ae3c1eb308c9e0d25fb7736de52336ea7f880
Signed-off-by: Damon Ding <damon.ding@rock-chips.com>
RK3588/RK3576 hdmi color format setting and dsc mode setting are
the same mask in the same register. If dsc is enabled for hdmi in
uboot, what color is hdmi output in uboot that cannot be obtained
from this register in kernel.
Since hdmi in dsc mode always sends avi infoframe, color format
of hdmi output in uboot can be read from the avi register.
Change-Id: I68157455342302338370feb9176b46e2051b4f01
Signed-off-by: Algea Cao <algea.cao@rock-chips.com>
If edid property is not cleared when hdmi plug out, edid value
maybe keep previous sink's when the hdmi is connected to another
sink.
Change-Id: I8d06de17040573de4ef4706183bd7b46c4002ae4
Signed-off-by: Algea Cao <algea.cao@rock-chips.com>
At atomic check stage, crtc->state is old state and equal to function
drm_atomic_get_old_crtc_state(), After swap state, the crtc->state is
new state and equel to drm_atomic_get_new_crtc_state().
Signed-off-by: Sandy Huang <hjc@rock-chips.com>
Change-Id: I32bf635dbadc7a771cec8d241fe943738f2d2567
vp->output_if will be used at vop2_crtc_atomic_disable(), so we need
init it at vop2_crtc_loader_protect() for logo display.
Signed-off-by: Sandy Huang <hjc@rock-chips.com>
Change-Id: I259a858255abe701f8b8037ac09e84a46f8db489
Otherwise a pulse occur when the pdm resume and suspend.
Change-Id: If5d4d6f3212ce10a647948b72ab171c8740821d9
Signed-off-by: Jason Zhu <jason.zhu@rock-chips.com>
The HCLK_RKRNG_NS clocks must be enabled on the RV1126B
to work properly.
Signed-off-by: Lin Jinhan <troy.lin@rock-chips.com>
Change-Id: I7b86bd02da17372abeb225fc318a1debfa9b60dc
The write back related regs locate at offset 0x280, which is large
than origin reg size 0x200. Access these reg will cause reg bak
out of bounds.
[ 1011.665341] ==================================================================
[ 1011.665361] BUG: KASAN: slab-out-of-bounds in vop_mask_write+0xa1/0xd6
[ 1011.665389] Read of size 4 at addr b22e2ac0 by task vo_render_0/2007
[ 1011.665407]
[ 1011.665421] CPU: 1 PID: 2007 Comm: vo_render_0 Tainted: G B O 6.1.118 #2
[ 1011.665442] Hardware name: Generic DT based system
[ 1011.665462] unwind_backtrace from show_stack+0xb/0xc
[ 1011.665495] show_stack from dump_stack_lvl+0x2b/0x34
[ 1011.665531] dump_stack_lvl from print_report+0x4b/0x304
[ 1011.665562] print_report from kasan_report+0x71/0x80
[ 1011.665589] kasan_report from vop_mask_write+0xa1/0xd6
[ 1011.665615] vop_mask_write from vop_initial+0xbff/0x1bd0
[ 1011.665641] vop_initial from vop_crtc_atomic_enable+0x519/0x6fbc
[ 1011.665669] vop_crtc_atomic_enable from drm_atomic_helper_commit_modeset_enables+0x253/0x634
[ 1011.665705] drm_atomic_helper_commit_modeset_enables from rockchip_drm_atomic_helper_commit_tail_rpm+0xa7/0x7c8
[ 1011.665744] rockchip_drm_atomic_helper_commit_tail_rpm from commit_tail+0xe7/0x200
[ 1011.665780] commit_tail from drm_atomic_helper_commit+0x159/0x164
[ 1011.665811] drm_atomic_helper_commit from drm_atomic_commit+0x181/0x1b0
[ 1011.665851] drm_atomic_commit from drm_mode_atomic_ioctl+0xbb5/0xe28
[ 1011.665888] drm_mode_atomic_ioctl from drm_ioctl+0x4af/0x528
[ 1011.665925] drm_ioctl from vfs_ioctl+0x57/0x64
[ 1011.665956] vfs_ioctl from sys_ioctl+0x29f/0xc1c
[ 1011.665983] sys_ioctl from ret_fast_syscall+0x1/0x54
[ 1011.666009] Exception stack(0xf1233fa8 to 0xf1233ff0)
[ 1011.666033] 3fa0: 8b201608 8bbfe308 00000031 c03864bc 8bbfe308 8bbfe2d8
[ 1011.666056] 3fc0: 8b201608 8bbfe308 c03864bc 00000036 8b202630 8b202640 8b202650 8b202638
[ 1011.666079] 3fe0: 9c302ee4 8bbfe2c8 9c2e6f90 9c78bd18
[ 1011.666094]
[ 1011.666104] Allocated by task 61:
[ 1011.666117] kasan_set_track+0x17/0x1a
[ 1011.666144] ____kasan_kmalloc+0x4b/0x4e
[ 1011.666168] devm_kmalloc+0x1f/0xe8
[ 1011.666191] vop_bind+0xe81/0x2840
[ 1011.666212] component_bind_all+0x211/0x518
[ 1011.666237] rockchip_drm_bind+0x5c5/0x1030
[ 1011.666260] try_to_bring_up_aggregate_device+0x33d/0x3d4
[ 1011.666285] component_master_add_with_match+0x179/0x1b0
[ 1011.666310] rockchip_drm_platform_probe+0x241/0x28c
[ 1011.666334] platform_probe+0xa3/0xf4
[ 1011.666352] really_probe+0x267/0x478
[ 1011.666376] __driver_probe_device+0x225/0x240
[ 1011.666399] driver_probe_device+0x41/0xb4
[ 1011.666423] __device_attach_driver+0x99/0x130
[ 1011.666446] bus_for_each_drv+0xe5/0x100
[ 1011.666467] __device_attach+0x14f/0x1e0
[ 1011.666491] bus_probe_device+0x7b/0x158
[ 1011.666513] deferred_probe_work_func+0x177/0x18c
[ 1011.666537] process_one_work+0x387/0x544
[ 1011.666561] process_scheduled_works+0x37/0x38
[ 1011.666583] worker_thread+0x43d/0x52c
[ 1011.666603] kthread+0x179/0x18c
[ 1011.666623] ret_from_fork+0x11/0x2c
[ 1011.666642]
[ 1011.666653] The buggy address belongs to the object at b22e2800
[ 1011.666653] which belongs to the cache kmalloc-1k of size 1024
[ 1011.666672] The buggy address is located 704 bytes inside of
[ 1011.666672] 1024-byte region [b22e2800, b22e2c00)
[ 1011.666693]
[ 1011.666713] The buggy address belongs to the physical page:
[ 1011.666730] page:8dc7e73e refcount:1 mapcount:0 mapping:00000000 index:0x0 pfn:0x424e0
[ 1011.666756] head:8dc7e73e order:3 compound_mapcount:0 compound_pincount:0
[ 1011.666774] flags: 0x10200(slab|head|zone=0)
[ 1011.666805] raw: 00010200 ef0a2b00 00000002 b1201700 00000000 00100010 ffffffff 00000001
[ 1011.666823] page dumped because: kasan: bad access detected
[ 1011.666835]
[ 1011.666845] Memory state around the buggy address:
[ 1011.666861] b22e2980: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
[ 1011.666877] b22e2a00: 00 00 00 00 00 00 00 00 fc fc fc fc fc fc fc fc
[ 1011.666895] >b22e2a80: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[ 1011.666908] ^
[ 1011.666922] b22e2b00: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[ 1011.666938] b22e2b80: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[ 1011.666953] ==================================================================
Fixes: 83aabef608 ("arm64: dts: rockchip: Add vop support for RV1126B")
Change-Id: I36cf27060350c32d6630be80a5e169cdafdfb26f
Signed-off-by: Chaoyi Chen <chaoyi.chen@rock-chips.com>
