Fedora Linux Thread

[ClassicHasClass]

What does ls -l /dev/ipmi0 look like? On this system,

Code Select Expand


% ls -l /dev/ipmi0
crw-rw----. 1 root ipmi 239, 0 Feb  5 09:05 /dev/ipmi0


I think you have the groups right, but either your udev rule is wrong or it hasn't taken effect yet.

[MPC7500]

Code Select Expand


$ ls -l /dev/ipmi0
crw-r-----. 1 root ipmi 241, 0  5. Feb 18:05 /dev/ipmi0


Code Select Expand


$ cat /etc/udev/rules.d/99_my.rules
# allow ipmi to be seen by ipmi group
KERNEL=="ipmi*", GROUP="ipmi", MODE="0640"


Code Select Expand


chgrp ipmi /dev/ipmi0


Code Select Expand


chmod 0640 /dev/ipmi0


What have I done wrong and how do I correct the permissions?

[ClassicHasClass]

Oh, my bad. Yes, it does need to be writeable to send the commands to the BMC. Try 0660 instead of 0640. I wasn't thinking there.

[MPC7500]

Thanks, that's it :-)

For future revisions:
Link between IPMI fan 0/1/2 and mainboard fan connectors

[ClassicHasClass]

Yes, but I don't know if that mapping applies for the T2. IPMI reports six fans on this machine. I haven't sat down to figure out which are which.

Incidentally, what is the CPU monitor tool you're using there? I like the clockspeed display.

[MPC7500]

Yes, the mapping is for the Blackbird only.
I'm using cpufreq.

[MPC7500]

It is a pity that cpufreq does not distinguish between physical cores and threads. Then it would be possible to switch off single cores and you could also switch between SMT1, 2 and 4.

[MauryG5]

Guys finally fixed the Fedora Software application problem. In recent updates have sent the fix to the problem.

[MPC7500]

Which repository have I to enable to get Kernel 5.6?

I currently have a problem with the 8x PCIe slot. There is a NVMe adopter (with Corsair MP600) in this slot and I get the error:

Code Select Expand


nvme nvme0: frozen state error detected, reset. 
Removing after probe failure status: -19.
can't change power state from D3hot to D0 (config space inaccessible)

I guess this could be a kernel problem?
This error only occurs if the 16x slot is used (RX5500XT).

Has anyone an idea how to locate the error?

Edit: Here I found someone with a similar issue.
His conclusion in the last post: Mainboard change to a different model.

[MPC7500]

One further quwestion, I'm following this guide: Installing Kernel

When I perform: sudo make INSTALL_MOD_STRIP=1 -j`nproc` install, I get this:

Code Select Expand


  LDS     arch/powerpc/boot/zImage.lds
  WRAP    arch/powerpc/boot/zImage.pseries
  WRAP    arch/powerpc/boot/zImage.epapr
sh -x ./arch/powerpc/boot/install.sh "5.6.0-rc3+" vmlinux System.map "/boot"
+ set -e
+ '[' -x /root/bin/installkernel ']'
+ '[' -x /sbin/installkernel ']'
++ basename vmlinux
+ image_name=vmlinux
+ '[' -f /boot/vmlinux ']'
+ mv /boot/vmlinux /boot/vmlinux.old
+ '[' -f /boot/System.map ']'
+ mv /boot/System.map /boot/System.old
+ cat vmlinux
+ cp System.map /boot/System.map
+ path=/boot
+ shift 4
+ '[' 0 -ne 0 ']'


Instead of this:

Code Select Expand


  LDS     arch/powerpc/boot/zImage.lds
  WRAP    arch/powerpc/boot/zImage.pseries
  WRAP    arch/powerpc/boot/zImage.epapr
sh -x ./arch/powerpc/boot/install.sh "5.6.0-rc3+" vmlinux System.map "/boot"
+ set -e
+ '[' -x /root/bin/installkernel ']'
+ '[' -x /sbin/installkernel ']'
+ exec /sbin/installkernel 5.6.0-rc3+ vmlinux System.map /boot


What's my error?

[MauryG5]

but sorry, this 5500 XT graphics card, can you make it work with the same patch as the 5.5.0.rc1 kernel that we installed some time ago? I know that is for ships 10 and this is ships 14 ... How do you make it work let me understand ...

[MPC7500]

It's the same patch. Because Navi 14 and Navi 10 are almost the same GPUs.

[MauryG5]

Got it all clear. And why did you decide to install Kernel 5.6? The patch is then provided by default in these new Kernels or should we always install it ourselves?

[MPC7500]

Yes, I thought the patch is applied to Kernel 5.6. But with Kernel 5.6 it doesn't work.

[MauryG5]

So I think you should wait now. In fact, I have not yet understood what will happen with the new Kernels regarding the operation of Navi10 / 14 because obviously we must always be able to use the graphics cards otherwise it doesn't make sense ... I am currently doing very well and I hold on to 5.5.0.rc1 with patches, then when they give me guarantees of being able to use new Kernels that support Navi10 / 14 then I will be happy to install the new Kernels.