R1/beta4 booting hangs with no icons lighting up

[Carl_Miller]

When trying to boot the R1/beta4 live environment on this computer, Haiku hangs with zero boot screen icons lit.

[humdinger]

Plesse attach yor Info instead of linking some website. If this isn't just spam...

[Carl_Miller]

• CPU: AMD Ryzen 5 5600X
• Motherboard: MSI B550-A PRO
• Memory: 16 GB (8 GB×2) G.Skill Ripjaws V DDR4-3200 RAM
• Storage:
  • Western Digital WDS100T2B0C 1TB M.2 NVMe SSD (Linux);
  • Kingston SA400S37480G 480 GB SATA SSD (Windows);
  • Western Digital WD80EDAZ-11TA3A08 8 TB SATA HDD (data)
• Video Card: MSI GAMING X Radeon RX 570 4 GB GPU

[humdinger]

You may want to check ​https://www.haiku-os.org/docs/welcome/en/bugreports.html#hardware to see if you can capture where it hangs (on-screen debug).

So, from the description it has nothing to do with the installer. Ticket title adjusted, correct me if I guessed wrong.

[waddlesplash]

Please test with a nightly build.

[Carl_Miller]

@humdinger With on screen debug output enabled in beta4, I see no debug output whatsoever, just the regular boot screen hanging with zero lights.

@waddlesplash My next post will contain the results of my tests with and without on screen debug output using the latest nightly (hrev57104).

[Carl_Miller]

@waddlesplash I get the same result in hrev57104, no matter whether or not I have on screen debug output enabled; it persists even when I ask it to boot in safe mode.

Moreover, I can't close my microphone (which defaults to being open) in this state, which leads me to believe that it's *freezing*, not hanging.

[Carl_Miller]

After talking to jessicah in IRC this morning, I still wasn't able to find a combination of safe mode flags and BIOS settings that would advance the boot process any further. :(

Please advise as to advanced debug or other options.

[Carl_Miller]

First screenful of on-screen debug output with hrev57156

[Carl_Miller]

When booting hrev57156 with safe mode and fail-safe graphics drivers enabled, I now get one screenful of on-screen debug output before Haiku freezes (picture attached).

I'm waiting for my next disability check to come in before I get the equipment necessary to do serial debugging.

(last minute addition: oops, forgot to disable on-screen debug paging. Will report back with that disabled.)

[Carl_Miller]

This is weird… when I turn off on-screen debug paging, I get no on-screen debug output whatsoever before it freezes.

[waddlesplash]

That's extremely strange. This area of the bootloader is not my expertise. Perhaps jessicah could shed some light here?

[tqh]

We had similar issues when refactoring serial output. Not sure if on screen paging only writes to serial when a page is done, but might be an issue here.

Can you see what UEFI version you are using, should be reported in firmware (bios) settings. Also try enabling CSM mode (compability mode), as that might init (memory map) any serial ports or other hardware that we need early on. I can't think of anything else than CPU and interrupts this early though.

[Carl_Miller]

I was on version AC0, but on a whim I checked for updates, and found version AE0, to which I updated.

Now, the system tries to boot, but crashes unless I enable safe mode graphics (which also means my second monitor isn't detected). I'll post a picture of the crash message presently.

[Carl_Miller]

Crash message when trying to boot with normal options under UEFI AC0

[Carl_Miller]

Typo; the description of that attachment should say AE0.

[waddlesplash]

That's a problem with the Radeon HD graphics driver which is already tracked in other tickets.

Does it boot all the way to desktop with failsafe graphics?

[Carl_Miller]

It does.

[waddlesplash]

Well, there's nothing more that needs to be done here, then :)

[pulkomandy]

"in other tickets" is not sufficient, please provide links to them.

Also, the Radeon driver should not try to handle hardware it doesn't know what to do with. If it leads to a black screen, we should remove the affected PCI IDs from it.

Leaving the ticket open until one of these things is done.

[Carl_Miller]

It leads not to a black screen, but to KDL (see most recent attachment), even in the installed system and with the updated UEFI, unless I blacklist radeon and radeon_hd or use fail-safe graphics drivers.

[waddlesplash]

It's a variation specifically of #17664.

[pulkomandy]

I am confused because the issue with a 0 size framebuffer in #17664 was already solved, and the patches merged. How is it possible that this is the same problem? It surely is a different one with similar consequences. I think it would be easier to handle it as a separate ticket instead of appending it to a loosely related one.

Carl_Miller, I suggest you open a new ticket for this new problem (since it is not directly related to what was first solved here). Please attach a syslog if you can, as well as the panic screenshot already attached here, that will be helpful for understanding the problem.

[Carl_Miller]

@pulkomandy Done. #18530