Weird issues last night

General hardware discussion.
Forum rules
Behave
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Weird issues last night

Post by Grogan »

I came home last night, powered up my computer, started Steam and when the UI appeared, some corrupted lines appeared on the desktop behind it. (My crimson night wallpaper). I figured a redraw of the screen would correct that as usual if something like that ever happens. The reason would be because of the way Steam is an asshole chrome browser, a window with a composited canvas loading over my non-composited X11 window manager. Sync issues etc.

Nope! Not redrawing, not mode switching, not turning the monitor off and on fixed it, it was stuck there. Quit X, drop down to console and the whole screen is dark red with lines, with the console text over top of it. Alright, shut the thing right down, this seems like a hardware failure AGAIN. Even after the PC was shut down and there was no signal, even though the monitor went down and came back to show me the "no signal" overlay, it was the red lines behind it. Monitor failure?

I shut the PC off at the PSU and shut the monitor off at its switch and killed the power. I powered everything back up again and it was all fine :wtf:

I'm not an electronics guy, but from what I understand of it, for the display to have kept that pattern, with the way an LCD works, that could have only been the timing controller left in a bad state. I don't think those things have "memory" or buffers themselves (likely registers though), it's electrical and it kind of works bass ackwards. The liquid crystals block the black backlighting with red, green and blue at various intensities when they get the voltage signals. So it's like "stuck" sending those signals to the lcd diodes because registers are mis-programmed and nothing reset them until power was removed from the unit. Unless somebody smarter than me has a better one, that's my theory.

I started a download (wanted to see the game High on Life again) and played some games with no problems. Went upstairs and did stuff, and by that time my download was complete.

I started up High on Life without incident, but when I got to the settings menu the monitor resolution setting was greyed out at 1280x720. I thought maybe it was because the game starts out as a pixelated minigame (your character is playing a video game) and I'd get access to it later (i.e. game forcing an aspect ratio or something). I didn't remember anything like that, resolution is always the first thing I check when starting a game. So I started the game and about as soon as I got control of my character, I found I couldn't use my arrow keys or click anything I was still panning with the mouse at that point. I tried key combos like alt+tab, ctrl+alt+esc (window list) and ctrl+alt+F2 and by that time things were locked up completely with audio still going.

Press power button, audio stops but there's no shutdown. So hold power button to shut down and power off. Boot to other OS, e2fsck, fix errors (almost always that'll be the systemd "journals"... add one reason why I hate those binary log databases)

When I got back I tried the game again, and this time when I pressed ESC to get to the menu it was the menu proper (not first run), and THAT was in native resolution. So I figure good, my display must be correctly detected this time but no, It's still greyed out at 1280x720. This time the game locks up on me while I'm hovering over menu items. It took me a couple of times pressing ctrl+alt+F2 but I got to another TTY this time and killed the game processes. The audio stopped and the game and all the wine processes died out and it was just Steam processes left in the list. I go back to TTY1 and my X session is gone, it's a console login prompt but defunct. Twas then that I was fucked, and had to hard boot again (couldn't get to another console again and ACPI power button shutdown wasn't responding either. It's actually an "input device" too)

OK, enough of that. Filesystem errors (even understood ones) creep me out. So I get duckduckgo'ing (I refuse to say "googling" anymore!) and from what I've gleaned, that looks CPU core related. I see lots of bullshit solutions but I also see people using affinity type settings. Either taskset or a (patched in... at least valve/proton Wine has this functionality) WINE_CPU_TOPOLOGY variable. It's affinity settings.

I don't think the two issues (monitor corruption and game lockups) were related. You can imagine how that made me feel in the pit of my stomach though, but I don't think it's a graphics card failure, 29 days and 11 hours and 50 minutes after purchase (meaning, too late to send back to Amazon) :lol:

I'm not trying that game again. I removed it and I don't want to play it that badly that I'll risk hard booting. It used to work perfectly on my old system (albeit a bit sluggish in some spots because I refuse to cut graphics settings lol)
User avatar
Zema Bus
Your Co-Host
Posts: 1115
Joined: Sun Feb 04, 2024 1:25 am

Re: Weird issues last night

Post by Zema Bus »

Those were some weird issues, especially with them occurring one right after the other. Hope you don't run into them again.
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

After all that I played Far Cry 6 (my most motherfuckinest game... using DXR and Ultra everything) without any trouble. I then watched youtube videos until I fell asleep and woke up and shut down. I started up without any incident when I got up today.

I also switched back to the beta Steam client. I had gone back to the release because at the time, the beta was acting badly on my new system. It was that issue of the UI not responding until a window change occurred. For example, client defunct until you drag a corner of the window a pixel or two (or some change like that, click on taskbar to iconify and bring it back etc.) So until something happened that pinged the UI, the client UI wasn't showing up. Sometimes like a minute... unnerving. The release client was doing it too to some extent (not showing anything at all until the client window appeared, though not a long delay like the bad beta). So it didn't surprise me that it might cause screen corruption like that. The beta client is fixed now (I knew it because I switched my nephew to it while troubleshooting the slow downloads lol) and it displays its splash screens and zenity progress bars (if applicable) and eventual client UI correctly, and without unusual delays.
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

So I had that display corruption pattern start about 30 seconds after I started up today. I was in XFCE and it was the same reddish, vertical corduroy-like pattern. The red didn't come from the red wallpaper, because I don't use that one in XFCE. The corruption first started in my panel desktop switching pager, then on application windows, then on desktop.

I tried turning off the monitor and yanking the power cord from it. Corruption remained. So logged off, yep, full screen corruption (but usable console over top of it). Reboot and in transition when there's no signal and back, display corruption remained. Grub menu was corrupted like last time too. I booted and the console was corrupted (but usable console over top)

shutdown -h now fixed it, it was corrupted while it was going down, but the display was normal when the graphics card signal activated it again. I didn't need to actually power down the system at the PSU (but I'm using that ErP mode, so there should not be any power to the graphics card)

So it's like, the monitor is fucked until something sends a signal that resets it.

I still think it could be the monitor itself, because if it was the graphics card it would likely show up in a screenshot. This corruption doesn't. (emboldened because I think this is significant). Fuck, I hope this monitor doesn't die, I love this one.

If this happens again I'm going to try yanking the DP cable and see if the behaviour perists in the monitor that has no display connection.
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

Jeeze... I just thought of something. It is probably time to stop using xf86-video-amdgpu (the old X11 driver) and use the generic modesetting driver I'm expected to use these days. I am the one that installs xf86-video-amdgpu but that hasn't had an update for a long time now (it did have some updates but perhaps they were EOL). If I remove it I'll revert to the generic mode setting driver. Why wasn't I using it? Simply because I don't like that word "generic". I thought a more specific driver would be better (not that there was any real reason, I mean, I didn't have a problem). This is newer graphics hardware too, different firmware etc. and that driver may be bollocks.

I found out later, after scouring posts on the internet again, that the reason I could no longer use Zink was because it stopped working with the xf86-video-amdgpu driver. I don't care about that, it was only a curiosity and mystery solved as far as I was concerned, but it serves as an example of it being time to ditch that driver.

That could well be the problem, bad display configuration data (and we're talking about X server, X11 driver, Mesa). Maybe something like OOB signals screwing up the monitor's electronics. I can't see it persisting the way it does it it was the graphics card... unless maybe if the corruption is happening at the interface level (rendering correctly, but not transmitting to the display device correctly) but the way it persists still makes me think it's a monitor screw up.

Both times this happened there was composited canvas involved. The first time in IceWM happened when Steam created its composited window. The second time was using XFCE's compositing. Both were soon after startup. This supports something getting bad display configuration data.
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

Yup... that was at least correct about Zink

Code: Select all

[grogan@nicetry ~]$ MESA_LOADER_DRIVER_OVERRIDE=zink glxgears
Running synchronized to the vertical refresh.  The framerate should be
approximately the same as the monitor refresh rate.
111292 frames in 5.0 seconds = 22258.238 FPS
112513 frames in 5.0 seconds = 22502.510 FPS
111944 frames in 5.0 seconds = 22388.719 FPS
112312 frames in 5.0 seconds = 22462.289 FPS
112043 frames in 5.0 seconds = 22408.523 FPS
^C
(Don't really need Zink, I've got perfectly good OpenGL but I hate mysteries unsolved. I can play with it any time to test games etc. again now too)

P.S. I just noticed there must be some kind of off screen rendering going on in the background (in translation to Vulkan with Zink) as the vertical sync seems to be ignored. I didn't override it, and it's the same if I do:

Code: Select all

[grogan@nicetry ~]$ vblank_mode=0 MESA_LOADER_DRIVER_OVERRIDE=zink glxgears
ATTENTION: default value of option vblank_mode overridden by environment.
108674 frames in 5.0 seconds = 21734.742 FPS
111261 frames in 5.0 seconds = 22252.111 FPS
109257 frames in 5.0 seconds = 21851.393 FPS
109565 frames in 5.0 seconds = 21912.803 FPS
111802 frames in 5.0 seconds = 22360.311 FPS
109064 frames in 5.0 seconds = 21812.771 FPS
^C
If I do it without any tomfoolery, with my actual GPU with native OpenGL, it's double that.

Code: Select all

[grogan@nicetry ~]$ vblank_mode=0 glxgears
ATTENTION: default value of option vblank_mode overridden by environment.
219919 frames in 5.0 seconds = 43983.617 FPS
222503 frames in 5.0 seconds = 44500.551 FPS
222672 frames in 5.0 seconds = 44534.383 FPS
222996 frames in 5.0 seconds = 44599.109 FPS
221619 frames in 5.0 seconds = 44323.750 FPS
219343 frames in 5.0 seconds = 43868.457 FPS
^C
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

This generic modsetting driver DOES do something different, though I can't put my finger on why. When I quit X, it mode switches even though I haven't changed the mode. It's still 1920x1080 @60 Hz in X11 unless I change it with xrandr.

My amdgpu framebuffer console should be (and I think is?) the exact same mode as in X11. When I start X, it does not mode switch, but for some reason when I drop out, it mode switches. That is, turns off the backlight, takes half a second or so to adjust, turns backlight back on and shows the input overlay ("DP" for display port). Typical mode switching behaviour.

So there must be some change in the timings for that to occur.
User avatar
Zema Bus
Your Co-Host
Posts: 1115
Joined: Sun Feb 04, 2024 1:25 am

Re: Weird issues last night

Post by Zema Bus »

Is it looking like xf86-video-amdgpu was the culprit?
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

Oh I don't know, it isn't a frequent enough problem to say. It might be weeks or a month. Hopefully never :-)
User avatar
Zema Bus
Your Co-Host
Posts: 1115
Joined: Sun Feb 04, 2024 1:25 am

Re: Weird issues last night

Post by Zema Bus »

I hate those intermittent issues.
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

I happened again today, but after a while, not that soon after starting up. I was getting ready to compile something, flipping windows, changing window focus with mouse hover in XFCE etc.

Simply unplugging the DP cable from the display and plugging it in again causes it to reset. While it was unplugged and the monitor was off and turned on, the display corruption persisted on the low level full screen "Sceptre" logo and everything until the DP cable was plugged back in.

However, I went for a smoke and came back and the corruption was on my terminal, desktop pager, program window areas etc. just sitting there. Again, unplugging the DP cable and plugging it in got it back. This time I didn't even shut the monitor off. I'm going to shut the whole machine right down now and see if it happens again today. Shit... it just happened again while switching desktops.

I think I can conclude it's definitely the monitor, but I can try a HDMI cable (I'll have to go buy a good one though, nothing but crappy ones, bent at that from moving receivers with them connected etc. Maybe it's just the DisplayPort interface on the monitor that's starting to fail. I could also try a new DP cable, but it seems to be a good quality cable and I'd doubt anything is wrong with it. (Too intermittent to be the cable?)
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

Well... I gotta say. If shutting the computer right down stopped this behaviour from happening every few minutes (soon after unplugging DP cable and re-negotiating), then it's got to be the graphics card sending bad data to the interface in the first place for the display to be acting like this.

That could be drivers/firmware/mesa but is more likely more ominous.

I'm done. If this graphics card fails I'm not buying another one.
User avatar
Zema Bus
Your Co-Host
Posts: 1115
Joined: Sun Feb 04, 2024 1:25 am

Re: Weird issues last night

Post by Zema Bus »

There's still a chance it could be the cable or the port at one end or the other so worth a try. This video suggested some things to try (it's a TV he's talking about in the video):



And I saw a suggestion that it could be caused by an internally loose ribbon cable. It's not exactly the same as what you're experiencing, just a single vertical red line.
User avatar
Zema Bus
Your Co-Host
Posts: 1115
Joined: Sun Feb 04, 2024 1:25 am

Re: Weird issues last night

Post by Zema Bus »

Missed your last post. If it's the card I think you're probably just outside the Amazon return window so you'd have to RMA it to Gigabyte.

Here's an example of Gigabyte's RMA process.
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

Yeah this won't be an internal connection in the monitor with this behaviour (and indeed that's not the same problem), but I still haven't discounted the cable or the port on the graphics card. I haven't got a spare DP cable, or a HDMI cable that I'd trust (I could go and take a bent one from Dad's spare TV lol) so that's going to wait until I go out. It's a busy holiday weekend around here (Victoria Day, a.k.a. "May 24 Weekend", pronounced 2-4 like a "twofour", a 24 pack of beer lol) and the car is staying parked. I'm not going to be driving on the roads with all those people and cops etc.

It boils down to bad data, possibly graphics card getting bad display info at power on. I'm not going to do anything while things are working correctly (hasn't happened again today on this power up and if the behaviour is consistent, won't again until "next time" it gets in this state... I think it won't happen on a reboot even) but next time this happens I'll compare the actual EDID data and modelines in the xorg log. I just copied Xorg.0.log to Xorg.0.log_goodbehaviour to compare to next time this happens. I don't have one to compare right now as it only keeps one renamed copy.

The monitor may be very fussy with bad data/timings. It doesn't really seem like anything is wrong, until the monitor starts to go screwy when it's happening.

P.S. I am acutely aware of the return status of the graphics card lol. The day this first happened (and coincidentally the High On Life game CPU fuckery) it was 10 minutes TO midnight before the deadline this time instead of after :lol:
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

So something to take away from this is, there will not BE an exchange of EDID data until the device that connects to a display powers up. It'll be stored somewhere until then. Powering off/on the display won't do it (which is why you're supposed to have your display turned on before turning on a PC). Yanking a Display Port cable and plugging it in again will force a link retrain, but if the graphics card is sending wrong timings it's not going to fix anything. In this case the signal/timings cause the display to screw up after a few minutes.

That's why the PC needs to be shut down and powered up to correct these symptoms and why I haven't been seeing it again, until next time it gets the wrong data at graphics card powerup time. (I don't mean that's certainly the answer, just that it makes sense)

The next thing I'm going to do is buy both a new DP and HDMI cable and do more testing. First I'll try a new DP cable. If that doesn't permanently solve it, then I'll try the new DP cable in the other DP port on the video card. Then I'll try HDMI etc. Meanwhile I'll be carrying on with kernel/drivers, firmware and mesa updates.
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

I think I'm analyzing this problem about right. I didn't have any more trouble yesterday, and I rebooted a few times in the evening to test my theory that once it's got the correct data for the timings, it's stored. I shut down normally this morning, and when I powered up today everything has still been fine. Until next time. It could be as simple as the pins that are used for the EDID communication getting bad data on the cable, intermittently.

When you think about it, I have never noticed this problem until after I put the system in that ErP power mode. There was always power to the graphics card before that, I could tell by the way the display was behaving at shut down. With USB standby power disabled, there shouldn't be any power to the PCI-E bus at all. So... now there is simply more opportunity for this fault to happen, whatever the cause.

I don't think my graphics card is failing. Possibly the monitor, if it's not the cable or ports, but if so it probably won't be death. Just a pain in the ass, likely.
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

I haven't had it happen since, but when it does I'm ready. I scoped out a utility for decoding the EDID info from a hex dump in a text file (one way to use the utility... it can also read a binary file). It's called "edid-decode" and I had to go get it from git, and installed it to /usr/local because there's no sign of the utility anywhere in Arch's packages.

This is where I got it:

https://git.linuxtv.org/edid-decode.git/

I went to the Xorg.0.log (in ~/.local/share/xorg on a systemd system with rootless X) I renamed and got the hex dump. I cleaned it up so it was only the hex dump, but I don't think you have to according to the man page. If I'm understanding it correctly I could feed it the log file and it would find that hex dump in it, but I can do better than that, and I'll test that some other time.

Edit: Tested and yes indeed, you can just feed it Xorg.0.log files and it will find the dump in the file.

I took that from the xorg log and put it in a text file by itself. This can also be gotten from xrandr with verbose output, but I need to work with historical data, so it's logs.

Code: Select all

00ffffffffffff004e14cd0a01000000
181f0104b53c21783bca75a954459822
1f505421080001010101010101010101
010101010101023a801871382d40582c
3500584a2100001e139580a070382040
30203500584a2100001e000000fd0030
a5bebe3c010a202020202020000000fc
0053636570747265204d32370a2001b2
02031ef24390013f00e200c0e305c000
23097f0783010000e50f00000c00d985
80a070384040302035000a252100001e
8a6f80a070384040302035000a252100
001e0e5c80a070383540302035000a25
2100001e000000000000000000000000
00000000000000000000000000000000
000000000000000000000000000000f1
Then it's simply:

Code: Select all

edid-decode filename.txt
Look at the beautiful information it gives me:

Code: Select all

[grogan@nicetry ~]$ edid-decode edid.txt
edid-decode (hex):

00 ff ff ff ff ff ff 00 4e 14 cd 0a 01 00 00 00
18 1f 01 04 b5 3c 21 78 3b ca 75 a9 54 45 98 22
1f 50 54 21 08 00 01 01 01 01 01 01 01 01 01 01
01 01 01 01 01 01 02 3a 80 18 71 38 2d 40 58 2c
35 00 58 4a 21 00 00 1e 13 95 80 a0 70 38 20 40
30 20 35 00 58 4a 21 00 00 1e 00 00 00 fd 00 30
a5 be be 3c 01 0a 20 20 20 20 20 20 00 00 00 fc
00 53 63 65 70 74 72 65 20 4d 32 37 0a 20 01 b2

02 03 1e f2 43 90 01 3f 00 e2 00 c0 e3 05 c0 00
23 09 7f 07 83 01 00 00 e5 0f 00 00 0c 00 d9 85
80 a0 70 38 40 40 30 20 35 00 0a 25 21 00 00 1e
8a 6f 80 a0 70 38 40 40 30 20 35 00 0a 25 21 00
00 1e 0e 5c 80 a0 70 38 35 40 30 20 35 00 0a 25
21 00 00 1e 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 f1

----------------

Block 0, Base EDID:
  EDID Structure Version & Revision: 1.4
  Vendor & Product Identification:
    Manufacturer: SPT
    Model: 2765
    Serial Number: 1 (0x00000001)
    Made in: week 24 of 2021
  Basic Display Parameters & Features:
    Digital display
    Bits per primary color channel: 10
    DisplayPort interface
    Maximum image size: 60 cm x 33 cm
    Gamma: 2.20
    DPMS levels: Off
    Supported color formats: RGB 4:4:4, YCrCb 4:4:4, YCrCb 4:2:2
    First detailed timing includes the native pixel format and preferred refresh rate
    Display supports continuous frequencies
  Color Characteristics:
    Red  : 0.6630, 0.3281
    Green: 0.2714, 0.5957
    Blue : 0.1337, 0.1240
    White: 0.3134, 0.3291
  Established Timings I & II:
    DMT 0x04:   640x480    59.940476 Hz   4:3     31.469 kHz     25.175000 MHz
    DMT 0x09:   800x600    60.316541 Hz   4:3     37.879 kHz     40.000000 MHz
    DMT 0x10:  1024x768    60.003840 Hz   4:3     48.363 kHz     65.000000 MHz
  Standard Timings: none
  Detailed Timing Descriptors:
    DTD 1:  1920x1080   60.000000 Hz  16:9     67.500 kHz    148.500000 MHz (600 mm x 330 mm)
                 Hfront   88 Hsync  44 Hback  148 Hpol P
                 Vfront    3 Vsync   5 Vback   37 Vpol P
    DTD 2:  1920x1080  164.996368 Hz  16:9    183.476 kHz    381.630000 MHz (600 mm x 330 mm)
                 Hfront   48 Hsync  32 Hback   80 Hpol P
                 Vfront    3 Vsync   5 Vback   24 Vpol P
    Display Range Limits:
      Monitor ranges (Range Limits Only): 48-165 Hz V, 190-190 kHz H, max dotclock 600 MHz
    Display Product Name: 'Sceptre M27'
  Extension blocks: 1
Checksum: 0xb2

----------------

Block 1, CTA-861 Extension Block:
  Revision: 3
  Underscans IT Video Formats by default
  Basic audio support
  Supports YCbCr 4:4:4
  Supports YCbCr 4:2:2
  Native detailed modes: 2
  Video Data Block:
    VIC  16:  1920x1080   60.000000 Hz  16:9     67.500 kHz    148.500000 MHz (native)
    VIC   1:   640x480    59.940476 Hz   4:3     31.469 kHz     25.175000 MHz
    VIC  63:  1920x1080  120.000000 Hz  16:9    135.000 kHz    297.000000 MHz
  Unknown CTA-861 Data Block (tag 0x00, length 0):
  Video Capability Data Block:
    YCbCr quantization: Selectable (via AVI YQ)
    RGB quantization: Selectable (via AVI Q)
    PT scan behavior: No Data
    IT scan behavior: IT video formats not supported
    CE scan behavior: CE video formats not supported
  Colorimetry Data Block:
    BT2020YCC
    BT2020RGB
  Audio Data Block:
    Linear PCM:
      Max channels: 2
      Supported sample rates (kHz): 192 176.4 96 88.2 48 44.1 32
      Supported sample sizes (bits): 24 20 16
  Speaker Allocation Data Block:
    FL/FR - Front Left/Right
  YCbCr 4:2:0 Capability Map Data Block:
    SVD Index 19 is out of range
    SVD Index 20 is out of range
  Detailed Timing Descriptors:
    DTD 3:  1920x1080  143.999630 Hz  16:9    164.736 kHz    342.650000 MHz (522 mm x 293 mm)
                 Hfront   48 Hsync  32 Hback   80 Hpol P
                 Vfront    3 Vsync   5 Vback   56 Vpol P
    DTD 4:  1920x1080  119.998991 Hz  16:9    137.279 kHz    285.540000 MHz (522 mm x 293 mm)
                 Hfront   48 Hsync  32 Hback   80 Hpol P
                 Vfront    3 Vsync   5 Vback   56 Vpol P
    DTD 5:  1920x1080   99.998303 Hz  16:9    113.298 kHz    235.660000 MHz (522 mm x 293 mm)
                 Hfront   48 Hsync  32 Hback   80 Hpol P
                 Vfront    3 Vsync   5 Vback   45 Vpol P
Checksum: 0xf1  Unused space in Extension Block: 43 bytes
So next time this happens (I'd be happy with never), I'll decode the edid block from that log file and compare the differences. The hex dump is the same in every (good behaviour) log I've looked at, so any change in that hex dump would be a sign in itself, but it's better to have human readable information, even if some of it may not be understood correctly by the utility (important stuff will). Also, if this proves to be a problem I can actually feed X a binary file of correct EDID data in an xorg.conf directive (in /etc/xorg.conf.d) obtained using a corresponding read-edid utility (that I don't have yet lol) or I think I can even generate one from that text file with edid-decode if I specify the output file and format.

P.S. I've been saying EDID, but this is actually "E-EDID" (extended edid) nowadays, cough, cough :-)
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

So it happened again today. This time it took two shutdowns for the corruption to go away (didn't power off at PSU in either case)

There goes my theory about EDID data, the hex dump was identical.

I'm off to buy cables today.
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

LOL! I was just coming to this thread to say "It looks like it was the cable, it hasn't happened since".

As soon as I clicked this thread, my browser's program window area went to those reddish verticle lines. I have to go shut down now. Man, if that isn't a coincidence.

Wonder what the fuck it is then... this is the longest it's gone in a while without happening, coincidentally after changing the DP cable. I've been doing lots of shut downs too, trying to trigger it.

It's actually never happened like this before, I've been using the computer for about half an hour. It was always very soon after start up previously, usually before I'd even get a browser open. It's probably only in X11, but I wouldn't be in framebuffer console mode long enough to trigger this. It's not like I use it for hours every day.

P.S. This time it took a complete power off of the PC, I tried shutting down (shutdown -h) normally 3 times to see if it was going to come back, but the display kept that pattern (on black, areas that are supposed to just be backlight... not if there is a black image though or something forcing the colour like on web browser canvas) until PSU switch shut off. That's the first time for that.

I guess the next step is to try the HDMI interface. I also bought that cable. A big, heavy expensive one (it was all they had at the store I went to... $29 for a 6 foot cable lol)

I've also thought about going back to stable mesa, but I want the new features and it might take a long time to know. It really could be, because the X server uses it. I've had bad mesa builds cause TTF font corruption even. It would be a bigger long shot than the cable, because in order for it to persist like that, it would have to be peculiarly corrupting some hardware registers. There's always some standby signal for EDID transfer, even if the monitor is turned off at the switch. The display won't power up if switched off, but there is still communication. They do that because it can otherwise cause problems if the PC is turned on before the display. Yanking the cable would stop it (and re-establish the link) but there's still bad signal from somewhere so it fucks up again soon.
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

I'm starting to wonder if every sharting thing that went wrong with this cursed build boils down to that MSI motherboard. Well, the assrock card failed but I still don't know for 100% sure if that was symptom or cause (of those ESD related faults). Neither of the other graphics cards that have been in here reacted that way to a static zap, but there are underlying problems related to that. For example having to use that ErP power mode because of spurious keyboard signals turning the PC back on. That's motherboard (BIOS), very stupid, and stupid design if there's no distinction between S4 and S5, but there's an ESD problem for that to happen in the first place.

The board is probably the thing I should have replaced right away (and I wasn't completely ignorant of that) when that ESD shit started happening. However I was replacing stuff (case, PSU, cable management) that I didn't like anyway.
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

So I guess this can happen while I'm gaming. I started up when I got home, came to the forum, went right to my game. It was fine for some minutes, but then I started noticing that red vertical corduroy pattern on black areas. I still haven't switched ports, I'm kind of sore and don't want to fuck around moving the computer and twisting etc. Hopefully it'll be good for a few days again.

I could just start powering off at the back and draining capacitance right before starting up. See if it EVER happens if I do that. That would actually be a nice clue if it did. (At this point I'm unconvinced it's the card, the cable, the ports. The monitor's electronics do seem to react badly to whatever bad signal it's getting. It's too intermittent to be something wrong with the monitor, and powering down the PC completely fixes it)
User avatar
Zema Bus
Your Co-Host
Posts: 1115
Joined: Sun Feb 04, 2024 1:25 am

Re: Weird issues last night

Post by Zema Bus »

Have you tried opening the monitor's menu while that's happening to see if the menu is free of those lines, or if it has them as well?
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

No it won't (the overlay wouldn't). It's on backlit black areas. Anything else would cover it, including text on console. It's a brownish red, vertical pattern... like corduroy.
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

Well that was quick... powering right down before starting up evidently doesn't prevent it. I tried that today, power off at PSU, hold power button etc. The corruption happened within 15 minutes.

I guess it's time to try... something.
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

I just switched to the HDMI interface. It's no fucking good, because the display doesn't seem to support 165 Hz, it only advertises up to 144. xrandr doesn't even list it. Not only that, XFCE had a shit fit and lost all my desktop and display settings because of it (it probably broke xml stanzas when the display settings were invalid)

Not only that, I went for a smoke and came back and the lines were there even using the HDMI, just sitting there at the monitor's preferred, default mode of 1920x1080@60Hz. I might as well go back to DisplayPort so I can at least use the capabilities of the monitor. It's not just Linux, the monitor really doesn't support that range on the HDMI interface. From edid-decode:

Code: Select all

Display Range Limits:
      Monitor ranges (GTF): 48-145 Hz V, 30-180 kHz H, max dotclock 600 MHz
So the moral of the story is, we just can't have nice things anymore. Everything is crap these days and you have to jump through hoops for RMA's. I'm done. I haven't got one more dime to spend on this rig. That last 50 bucks I wasted on cables is it, I've had it.
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

Alright, I just made a few observations that point to the monitor as the failing hardware. First of all, the problem is getting worse. I am probably headed for complete failure. It's getting harder to correct and doesn't stay under the usual behaviour of this problem anymore.

After I switched back to the DP interface, the problem occurred very quickly. I shut everything right down, PSU switch and held power button, and turned the display off (so it was blinking red instead of blue, it still communicates). I also yanked the power cable from the monitor and the problem persisted while it was going down. I plugged it back in again, turned it on, and the red pattern was even on the backlit background behind the initial Sceptre logo overlay. I turned the monitor off at the power button once more and on again, and the pattern went away. Note that there was no communication from the graphics card, the PC was off and capacitance drained, this was all the monitor.

The display worked normally for several seconds and then went to the red pattern on the background of my XFCE workspace pager (that's just a pure 000000 black, which means backlight only). So I dropped out of X and when it mode switched back to the framebuffer console, it corrected itself in the link retrain. For the first time ever.

This is erratic monitor behaviour.
User avatar
Zema Bus
Your Co-Host
Posts: 1115
Joined: Sun Feb 04, 2024 1:25 am

Re: Weird issues last night

Post by Zema Bus »

Sorry to hear that. At least you were able to narrow it down to the source. Does your old monitor have HDMI so you could use that while RMAing this one?
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

Strangely enough, it's fine again now for a while. Odd how it could go like 4 days without happening (it seems that's past tense though)

Unfortunately, it is already known to me that the old monitor does have a problem with its HDMI interface, it's not implemented properly. It's an old (12'ish years) Viewsonic "2ms gaming monitor". 19 inch wide, 1920x1080@60Hz.

When I got the R9 380 card, I thought I'd try HDMI since it was a more modern graphics card. I actually had a GREEN backlit screen on console, like an old VAX terminal. OK, back to the DVI interface. When I got the RX 570 card, I tried the HDMI interface again, and while I didn't have artifacts, I had somewhat jerky performance (sync issues) that cleared right up when I went back to DVI. I use the DVI interface for that monitor, but this card only has DP and HDMI.

I guess I'd get a new monitor if I have to. Thing is, I couldn't wait for manufacturer's RMA (which you can't even count on ever going through). I love this one, and it wasn't expensive, I'd consider getting another one if this one is just defective in some way, but that's the thing. A new one just like it could fail in the same way if it came from a production run that uses the same internal third party components.

P.S. I still haven't sent back that Assrock card. It seems their hostile demeanor worked on me, I just know I'm only going to spend more money on shipping and get angry.
User avatar
Grogan
Your Host
Posts: 2048
Joined: Sat Aug 21, 2021 10:04 am
Location: Ontario, Canada

Re: Weird issues last night

Post by Grogan »

Now, before anybody comes along and says "duh, of course it's the monitor", I've always known it was involved in this. It just seemed that the bad state of the monitor's electronics (which at the time was being corrected only by powering off the PC completely or only briefly if yanking the DP cable) was a symptom rather than the cause. I mean, who hasn't had to yank the power from their LCD TV to correct a fault. Sometimes when Dad hollers I go to the breaker panel and flip off all his entertainment stuff (receivers, sound bar, TV, etc.)

When you think about what's happening, and the fact that it's not just the interface or interface cable on either the graphics card or the monitor, the monitor is sending baseline voltage signals constantly to the LCD diodes overriding the negative backlight, if not overridden with other values (like a non 000000 black, or an image overlaying. It will also show up in a semi-transparency, like my shiny black window decoration graphics, faintly in behind. In my game, it showed up in a few bands in the starlit, nebulous gassy sky, and on some pure black surfaces. Subtle, because most of it is not just backlit black)
Post Reply