System hang with display power mgmt set

[gpsamu]

Hello,

Running Buster with kernel 4.4. I've enabled power management for the display via the "Power management" xfce widget. When the power management on the display is activated after the specified period of inactivity the built in screen goes black and displays the messages:

[365.911055] it6251 2-005c: error -11 writing to edp addr 0x5 
[365.911063] dummy 2-005e: it6251.c:1t6251_init:285 error -1 writing 255 to 5
[365.919987] dummy 2-005e: error -11 writing to lvds addr 0x5 
[365.919993] dummy 2-005e: it6251.c:it6251 init:295 error -1 writing 255 to 5
[366.022805] it6251 2-005c: System status: 0x3e [366.024196] it6251 2-005C: RPCLKCnt: 1520
[ 366.027010] it6251 2-005c: RPC Req: Ox24
[366.025597] it6251 2-005c: Clock: Ox198 
[366.026306] 1t6251 2-005C: Ref Link State: Ox00
[366.028416] 1t6251 2-005c: hactive: 1920 
[366.029818] it6251 2-005c: vactive: 1080

At this state the system appears to be hung. It does not respond to mouse or keyboard input and the display doesn't wake up. For now, I'm using the screensaver instead and blanking the screen after the specified interval. Wondering if anybody else has seen this or if that power management functionality simply isn't supported on Novena?

[Jookia]

Hmm. This shouldn't hang. Does it reply to pings or serial out?

[gpsamu]

It does not reply to pings if I recall correctly. I usually tried to ssh in when this happens, with no luck. Unfortunately I don't have the necessary serial cable. I've ordered one but it hasn't arrived yet.

[Jookia]

Hmm. I think I may hit an issue similiar to this, but I haven't been able to reproduce it as I don't Interesting! I've never had a Novena with a battery, but I have hit an issue like this repeatedly but rarely where the display would go blank and won't respond to serial, pings whatever. I'm honestly not sure what's happening here, but logically it would make sense if this were some kind of power issue that's knocking out the CPU somehow. I haven't scoped this yet since it's a very but finding a more reproducible way to hit this when connected to AC would be very, VERY helpful.

[gpsamu]

It seems easy to reproduce with the power management enabled for the display. I will disable the screensaver and re-enable the display power mgmt later this evening to see if I can get it to hang again.

[Jookia]

Are you on a laptop, desktop or single board?

…

On Fri, Jan 15, 2021 at 04:18:21PM -0800, Gábor Samu wrote: It *seems* easy to reproduce with the power management enabled for the display. I will disable the screensaver and re-enable the display power mgmt later this evening to see if I can get it to hang again. -- You are receiving this because you commented. Reply to this email directly or view it on GitHub: #26 (comment)

[gpsamu]

It's a desktop model. So it has the attached display but no battery or battery board.

[Jookia]

Oh great, I have that. If you can reliably reproduce it could you write some instructions and I'll try to replicate it?

…

On Fri, Jan 15, 2021 at 04:22:45PM -0800, Gábor Samu wrote: It's a desktop model. So it has the attached display but no battery or battery board. -- You are receiving this because you commented. Reply to this email directly or view it on GitHub: #26 (comment)

[gpsamu]

Sure, not a problem. I will try to get to it this evening. Otherwise tomorrow AM.

[Jookia]

No hurry, I'm not exactly actively working on the Novena at the moment. Partly because of this exact issue among others.

…

On Fri, Jan 15, 2021 at 04:34:07PM -0800, Gábor Samu wrote: Sure, not a problem. I will try to get to it this evening. Otherwise tomorrow AM. -- You are receiving this because you commented. Reply to this email directly or view it on GitHub: #26 (comment)

[gpsamu]

Scanning the Kosagi forum I came across this: https://www.kosagi.com/forums/viewtopic.php?id=464

My display works fine though.

[Jookia]

I've had this happen on the bare board without the eDP adapter board. Assuming it's the same bug.

…

On Fri, Jan 15, 2021 at 05:40:31PM -0800, Gábor Samu wrote: Scanning the Kosagi forum I came across this: https://www.kosagi.com/forums/viewtopic.php?id=464 My display works fine though. -- You are receiving this because you commented. Reply to this email directly or view it on GitHub: #26 (comment)

[gpsamu]

Ok, I was wrong about not being able to ping the system. I had the display power settings configured as followed (see screenshot).

When the display went blank, it displayed the following:
...
[ 503.446974] it6251 2-005c: error -11 writing to edp addr 0x5
[ 503.446985] dummy 2-005e: it6251.c:it6251_init:285 error -1 writing 255 to 5
[ 503.456541] dummy 2-005e: error -11 writing to lvds addr 0x5
[ 503.456553] dummy 2-005e: it6251.c:it6251_init:295 error -1 writing 255 to 5
[ 503.550632] it6251 2-005c: System status: 0x3e
[ 503.552120] it6251 2-005c: RPCLKCnt: 1520
[ 503.553609] it6251 2-005c: Clock: 0x197
[ 503.554356] it6251 2-005c: Ref Link State: 0x00
[ 503.555111] it6251 2-005c: RPC Req: 0x24
[ 503.556599] it6251 2-005c: hactive: 1920
[ 503.558083] it6251 2-005c: vactive: 1080

However mouse and keyboard input did not wake up the display.  The only thing I could do was ssh to the system and issue a reboot.  

Note that dmesg showed me the following: 

...
[  503.429746] imx-ldb 2000000.aips-bus:ldb@020e0008: 0:imx_ldb_set_clock: LDB_CTRL: 0x000001fd
[  503.429760] imx-ldb 2000000.aips-bus:ldb@020e0008: 0:imx_ldb_set_clock: now: 475200000 want: 484750000
[  503.429785] imx-ldb 2000000.aips-bus:ldb@020e0008: 0:imx_ldb_set_clock after: 475200000
[  503.429794] imx-ldb 2000000.aips-bus:ldb@020e0008: 0:imx_ldb_set_clock: now: 135771428 want: 138500000
[  503.429808] imx-ldb 2000000.aips-bus:ldb@020e0008: 0:imx_ldb_set_clock after: 135771428
[  503.429825] imx-ldb 2000000.aips-bus:ldb@020e0008: 1:imx_ldb_set_clock: LDB_CTRL: 0x000001fd
[  503.429834] imx-ldb 2000000.aips-bus:ldb@020e0008: 1:imx_ldb_set_clock: now: 475200000 want: 484750000
[  503.429848] imx-ldb 2000000.aips-bus:ldb@020e0008: 1:imx_ldb_set_clock after: 475200000
[  503.429856] imx-ldb 2000000.aips-bus:ldb@020e0008: 1:imx_ldb_set_clock: now: 135771428 want: 138500000
[  503.429867] imx-ldb 2000000.aips-bus:ldb@020e0008: 1:imx_ldb_set_clock after: 135771428
[  503.436828] imx-ldb 2000000.aips-bus:ldb@020e0008: imx_ldb_encoder_commit: LDB_CTRL: 0x000001fd
[  503.446974] it6251 2-005c: error -11 writing to edp addr 0x5
[  503.446985] dummy 2-005e: it6251.c:it6251_init:285 error -1 writing 255 to 5
[  503.456541] dummy 2-005e: error -11 writing to lvds addr 0x5
[  503.456553] dummy 2-005e: it6251.c:it6251_init:295 error -1 writing 255 to 5
[  503.550632] it6251 2-005c: System status: 0x3e
[  503.552120] it6251 2-005c: RPCLKCnt: 1520
[  503.553609] it6251 2-005c: Clock: 0x197
[  503.554356] it6251 2-005c: Ref Link State: 0x00
[  503.555111] it6251 2-005c: RPC Req: 0x24
[  503.556599] it6251 2-005c: hactive: 1920
[  503.558083] it6251 2-005c: vactive: 1080
...

[Jookia]

Ah, it might be a different bug then. Could you try this with my 4.19 buster stuff?

[gpsamu]

I can give it a whirl tomorrow. Assuming I can easily revert back to 4.4 correct? I read of the audio and power button light issues in 4.19.

[Jookia]

No, you can't revert back to 4.4 easily. You would need to make an SD card backup unfortunately. Yeah there are still issues.

…

On Fri, Jan 15, 2021 at 08:23:47PM -0800, Gábor Samu wrote: I can give it a whirl tomorrow. Assuming I can easily revert back to 4.4 correct? I read of the audio and power button light issues in 4.19. -- You are receiving this because you commented. Reply to this email directly or view it on GitHub: #26 (comment)

[gpsamu]

It's taken me a while, but I just confirmed the same issue with 4.19. I prepared a new SD card and ran through all of the upgrade steps including 5, which installs the 4.19 kernel. I set the display power management as per the screenshot above and hit the same issue. The display would not wake up with any keyboard/mouse activity, but I could still ping it.

[gpsamu]

Was just thinking. My goal here was to have the display shut off entirely (backlight off) due to inactivity. Is using the power management widget the correct way to achieve this. Logically yes, but not sure if there is another way? Not really concerned with longevity as I think it's using LED backlights...

[Jookia]

I'm not too sure what the stack here is. I believe the correct way to do this that the power management widget should use is DPMS.

…

On Thu, Jan 28, 2021 at 05:27:18PM -0800, Gábor Samu wrote: Was just thinking. My goal here was to have the display shut off entirely (backlight off) due to inactivity. Is using the power management widget the correct way to achieve this. Logically yes, but not sure if there is another way? Not really concerned with longevity as it's using LED backlights... -- You are receiving this because you commented. Reply to this email directly or view it on GitHub: #26 (comment)