| 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
| ||
|
Message-ID: <CAOf5uwnf6yF0sksjVRRzxLCD8xaMGhUhkOQ8CjwhYy85zcArLw@mail.gmail.com>
Date: Wed, 12 Feb 2020 12:49:11 +0100
From: Michael Nazzareno Trimarchi <michael@...rulasolutions.com>
To: Kever Yang <kever.yang@...k-chips.com>,
"open list:ARM/Rockchip SoC..." <linux-rockchip@...ts.infradead.org>,
Heiko Stuebner <heiko@...ech.de>
Cc: Philipp Tomsich <philipp.tomsich@...obroma-systems.com>,
LKML <linux-kernel@...r.kernel.org>,
Jagan Teki <jagan@...rulasolutions.com>,
linux-arm-kernel <linux-arm-kernel@...ts.infradead.org>,
Hans de Goede <hdegoede@...hat.com>
Subject: Re: siimple-framebuffer rockchip persistent logo
Hi Heiko and Hans
I manage to have the persistent framebuffer but complete working video subsystem
THe problem is the follow
[ 0.116514] rk_iommu ff930300.iommu: attaching to power domain 'pd_vio'
[ 0.116539] rk_iommu ff930300.iommu: adding clock 'aclk_vop0' to
list of PM clocks
[ 0.116561] rk_iommu ff930300.iommu: adding clock 'hclk_vop0' to
list of PM clocks
[ 0.116989] rk_iommu ff940300.iommu: attaching to power domain 'pd_vio'
[ 0.117013] rk_iommu ff940300.iommu: adding clock 'aclk_vop1' to
list of PM clocks
[ 0.117032] rk_iommu ff940300.iommu: adding clock 'hclk_vop1' to
list of PM clocks
[ 0.117335] rk_iommu ff9a0800.iommu: attaching to power domain 'pd_video'
[ 0.117359] rk_iommu ff9a0800.iommu: adding clock 'aclk_vcodec' to
list of PM clocks
[ 0.117377] rk_iommu ff9a0800.iommu: adding clock 'hclk_vcodec' to
list of PM clocks
[ 0.117773] rockchip-pm-domain
ff730000.power-management:power-controller: Calling pd power off
[ 0.117926] rockchip-pm-domain
ff730000.power-management:power-controller: Calling power domain off
[ 0.117984] rockchip-pm-domain
ff730000.power-management:power-controller: Calling pd power off
[ 0.118016] rockchip-pm-domain
ff730000.power-management:power-controller: Calling power domain off
As you can see the pd_vio domain are off even if simpleframe buffer
own them. The problem is that the
rk_iommu is registered before the simple framebuffer. When pd_io
domain is off everything is done on
hdmi side
+ chosen {
+ #address-cells = <2>;
+ #size-cells = <2>;
+ ranges;
+
+ simplefb_hdmi: framebuffer-lcd0-hdmi {
+ compatible = "rockchip,simple-framebuffer",
+ "simple-framebuffer";
+ rockchip,pipeline = "rockchip,dw_hdmi";
+ clocks = <&cru PCLK_HDMI_CTRL>, <&cru
SCLK_HDMI_HDCP>, <&cru SCLK_HDMI_CEC>,
+ <&cru ACLK_VOP0>, <&cru DCLK_VOP0>,
<&cru HCLK_VOP0>;
+ power-domains = <&power RK3288_PD_VIO>;
+ status = "disabled";
+ };
+ };
Memory area is automatic injected by the bootloader. Now rk iommu is
with runtime enable so basically if I understand
the >sync call will schedule a power off domain if they are not in use.
[ 0.118064] vgaarb: loaded
[ 0.118997] SCSI subsystem initialized
[ 0.119213] libata version 3.00 loaded.
[ 0.119499] usbcore: registered new interface driver usbfs
[ 0.119552] usbcore: registered new interface driver hub
[ 0.119617] usbcore: registered new device driver usb
[ 0.121505] pps_core: LinuxPPS API ver. 1 registered
[ 0.121515] pps_core: Software ver. 5.3.6 - Copyright 2005-2007
Rodolfo Giometti <giometti@...ux.it>
[ 0.121539] PTP clock support registered
[ 0.121777] EDAC MC: Ver: 3.0.0
[ 0.125046] clocksource: Switched to clocksource arch_sys_counter
[ 0.930470] simple-framebuffer 7f800000.framebuffer: attaching to
power domain 'pd_vio'
[ 0.930501] simple-framebuffer 7f800000.framebuffer: adding clock
'pclk_hdmi_ctrl' to list of PM clocks
[ 0.930523] simple-framebuffer 7f800000.framebuffer: adding clock
'sclk_hdmi_hdcp' to list of PM clocks
[ 0.930544] simple-framebuffer 7f800000.framebuffer: adding clock
'sclk_hdmi_cec' to list of PM clocks
[ 0.930564] simple-framebuffer 7f800000.framebuffer: adding clock
'aclk_vop0' to list of PM clocks
[ 0.930585] simple-framebuffer 7f800000.framebuffer: adding clock
'dclk_vop0' to list of PM clocks
[ 0.930606] simple-framebuffer 7f800000.framebuffer: adding clock
'hclk_vop0' to list of PM clocks
[ 0.930642] rockchip-pm-domain
ff730000.power-management:power-controller: Calling pd power on
[ 0.930746] rockchip-pm-domain
ff730000.power-management:power-controller: Calling power domain on
[ 0.931083] simple-framebuffer 7f800000.framebuffer: framebuffer at
0x7f800000, 0x7e9000 bytes, mapped to 0x(ptrval)
[ 0.931101] simple-framebuffer 7f800000.framebuffer:
format=x8r8g8b8, mode=1920x1080x32, linelength=7680
[ 0.931415] simple-framebuffer 7f800000.framebuffer: fb0: simplefb
registered!
Here when everything get registered by offcouse the screen went black already
Michael
On Fri, Feb 7, 2020 at 8:14 PM Michael Nazzareno Trimarchi
<michael@...rulasolutions.com> wrote:
>
> Hi all
>
> I move a bit on this
>
> On Sun, Jan 12, 2020 at 6:16 PM Michael Nazzareno Trimarchi
> <michael@...rulasolutions.com> wrote:
> >
> > Hi Kever
> >
> > Trying to have a persistent banner from uboot to linux-kernel. I'm
> > right now working on linux-rockchip kernel but I think that the
> > problem is even on mainline
> >
> > + framebuffer: framebuffer@...00000 {
> > + compatible = "rockchip,simple-framebuffer",
> > + "simple-framebuffer";
> > + reg = <0x7f800000 (1920 * 1080 * 4)>;
> > + width = <1920>;
> > + height = <1080>;
> > + stride = <(1920 * 4)>;
> > + format = "a8b8g8r8";
> > + clocks = <&cru PCLK_HDMI_CTRL>, <&cru SCLK_HDMI_HDCP>,
> > + <&cru SRST_LCDC0_AXI>, <&cru
> > SRST_LCDC0_AHB>, <&cru SRST_LCDC0_DCLK>,
> > + <&cru ACLK_VOP0>, <&cru HCLK_VOP0>;
> > + status = "okay";
> > + };
> >
>
> Now I can allocate the parameter using the bootloader and create the
> right mapping for the simple framebuffer. I don't even understand
> how sunxi and meson can work if they don't create a reserved memory
> using no-map. This is fixed on my side so the log is totally clean.
> I have added the deregister of simple fb and handover to the drm
>
> Now my boot parameters are:
> Kernel command line: console=ttyS2,115200n8 root=/dev/mmcblk0p1
> rootwait pd_ignore_unused clk_ignore_unused
>
> Still I have display go off on tinker during boot. Any suggestion?
>
> Michael
>
>
> > Seems that it get off before I reach the drm registration
> >
> > [ 2.077495] simple-framebuffer 7f800000.framebuffer: framebuffer at
> > 0x7f800000, 0x7e9000 bytes, mapped to 0xf0900000
> > [ 2.077519] simple-framebuffer 7f800000.framebuffer:
> > format=a8b8g8r8, mode=1920x1080x32, linelength=7680
> > [ 2.161225] simple-framebuffer 7f800000.framebuffer: fb0: simplefb
> > registered!
> > [ 3.433847] fb: switching to rockchip-drm-fb from simple
> >
> > I don't find all the clocks and if those are the only think that I
> > need to stay on. Any suggestion?
--
| Michael Nazzareno Trimarchi Amarula Solutions BV |
| COO - Founder Cruquiuskade 47 |
| +31(0)851119172 Amsterdam 1018 AM NL |
| [`as] http://www.amarulasolutions.com |
Powered by blists - more mailing lists