New kernels

[xeno74]

Tested final 6.18 and installed to Xenial for testing. Found it works fine with Xenial. No problems encountered.

Great!

Tested latest 6.19 alphas 1 and 2 on my MATE 2017 and Mint 2025 installs. While keeping in mind latest changes of move to AMD driver for Radeon chipsets. So far, on my R7 250, both initial alphas boot fine, run well, appear stable, are fluent on desktop and present no immediate issues.

I think you use the fbdev driver.

Does it mean that amdgpu kernel driver really work on Big Endian ? If yes, it is very good news!

Our kernel packages don’t have the AMDGPU driver because it doesn’t have big endian support. (byte swapping of firmwares)

[xeno74]

The latest Git kernel doesn't boot anymore on our e5500 machines (QEMU, X5000, Mirari etc) after the PowerPC updates 6.19-1.

Error messages (QEMU e5500 VM):

Code: Select all

Bad kernel stack pointer 4d694c01a88270 at c0000000000003bc
Oops: Bad kernel stack pointer, sig: 6 [#1]
BE PAGE_SIZE=4K  SMP NR_CPUS=4 QEMU e500
Modules linked in:
CPU: 2 UID: 0 PID: 2788 Comm: udevd Not tainted 6.19.0-a4-powerpc64-smp-g416f99c3b16f-dirty #1 VOLUNTARY
Hardware name: QEMU ppce500 e5500 0x80240012 QEMU e500
NIP:  c0000000000003bc LR: 0000000000230760 CTR: 0000000000230754
REGS: c00000003ff9bd20 TRAP: 0300   Not tainted (6.19.0-a4-powerpc64-smp-g416f99c3b16f-dirty)
MSR:  0000000090029000 <CE,EE,ME>  CR: 48842888  XER: 20000000
DEAR: 004d694c01a882f0 ESR: 0000000000000000 IRQMASK: 1
GPR00: 01a56ad000000000 004d694c01a88270 ffd86ca00046c958 ffd86c89ffffffff
GPR04: 0000000000000000 00000000ffffffff 0000000000000000 0000000000000007
GPR08: 000000000037a60c 0000000000000000 0000000000000000 0000000000230754
GPR12: 0000000004000000 c00000003fff7900 0000000000000004 0000000001a88270
GPR16: 00000000020397f0 0000000000bc287c 00000000004d009c 0000000000000000
GPR20: 0000000000000000 0000000000000000 00000000ffd86d5c 0000000001a4a130
GPR24: 00000000004d0040 00000000ffd86bdc 0000000000000000 00000000ffd86ab4
GPR28: 00000000004d009c 0000000001a807e0 0000000000379fc0 0000000001a84b50
NIP [c0000000000003bc] interrupt_return_srr_user+0x80/0xf4
LR [0000000000230760] 0x230760
Call Trace:
Code: 38000000 e8e100a8 e90100b0 e92100b8 e94100c0 e96100c8 e98100d0 e9a100d8 7c6ff120 7c8803a6 7ca903a6 7cc103a6 <e8410080> e8610088 e8810090 e8a10098
---[ end trace 0000000000000000 ]---

note: udevd[2788] exited with irqs disabled

It boots on the X1000.

I reported this issue to the kernel developers today.

Link: [PPC] [e5500] Boot issues after the PowerPC updates 6.19-1

[xeno74]

Hi All,

Here is the alpha4 of kernel 6.19 with the first PowerPC updates for X1000 and e5500 machines (X5000/20, X5000/40, Mirari, and QEMU VMs).

Download and further information: github.com



Unfortunately our e5500 machines don't boot with the alpha4 with the first PowerPC updates. I created a second package for the e5500 machines without the first PowerPC updates.

Please test the kernels.

Thanks,
Christian

[Hypex]

Does it mean that amdgpu kernel driver really work on Big Endian ? If yes, it is very good news!

No, unfortunately, not yet. Not in my case. My R7 250 lacks 3d support even with the Radeon driver, since it's too new. And lacks 2d acceleration from what I know. The speed is fine on the desktop, including tasks like browsing. But xeno74 would be right, it would be using the fbdev driver.

[Hypex]

It boots on the X1000.

I reported this issue to the kernel developers today.

Link: [PPC] [e5500] Boot issues after the PowerPC updates 6.19-1

So just tested alphas 3 and 4 on my Mate and Mint installs. Can confirm it boots and works fine here on my X1000 as well. For the X1000 this series is looking very good.

I accidentally tried to install from the e5500 specific archive. Then realised my mistake. Which explained why it downloaded so fast.

Last few nights I found Hyperion forums off line and back on tonight. This has been happening more frequently lately. But I tried to look at the PowerPC kernel updates from your GH page with Articfox on Mint, and it won't load any more, it just got stuck on the Anubis page.

[xeno74]

So just tested alphas 3 and 4 on my Mate and Mint installs. Can confirm it boots and works fine here on my X1000 as well. For the X1000 this series is looking very good.

Great! That sounds good!

Last few nights I found Hyperion forums off line and back on tonight. This has been happening more frequently lately. But I tried to look at the PowerPC kernel updates from your GH page with Articfox on Mint, and it won't load any more, it just got stuck on the Anubis page.

Please try SeaLion on MintPPC.

[xeno74]

Hi All,

Here is the alpha5 of kernel 6.19 for X1000 and e5500 machines (X5000/20, X5000/40, Mirari, and QEMU VMs).

I have found the first bad commit because of the boot issue on our e5500 machines and created a patch for reverting this commit.

Download and further information: github.com



Please test the kernels.

Thanks,
Christian

[xeno74]

Hi All,

I have released the alpha6 of kernel 6.19 with a smaller patch because of the e5500 boot issue.

Download and further information: github.com



Cheers,
Christian

[xeno74]

New for the RC1:

• Matrox G200: Fix bigendian handling in mgag200 (CONFIG_DRM_MGAG200)
• DRM fixes
• PCI fix
• MM updates
• Linux Git log
• Phoronix articles, reviews and news stories covering Linux 6.19

We still need the interrupt.patch because of the boot issue #18 on our e5500 machines.

[Hypex]

Please try SeaLion on MintPPC.

Okay, so SeaLion can load the Anubis page, but it takes a few minutes to process. Whatever it is doing the progress bar got there in the end and the final page loaded. But the delay in determining my browser wasn't some AI bot makes it practically unusable on PPC.

In other news, I gave the alpha 5 and 6 a test, with all looking good. It looked like some issue with shutdown, seeing some delay, but last night the alpha 6 manged it fine. Rebooting worked fine in my tests.

The RC1 will soon be upon us.