6.8.6 64bit withdrawn

[ozsouth]

I made a 64bit 6.8.6 2018 KIT LOW-LATENCY NON-USRMERGE kernel. Has Overlayfs and Aufs builtin.
This has Full Preemption & Frequency at 1000Hz. It is noticeably more responsive than standard kernels.

Although this would not be a problem for most folk, a missing DOTconfig item should be present, so making a new kernel.

[Marv]

I've run this sucessfully in LxPupSc64 23.01 +6, Voidpup64 04132024 version, and with structure usrmerge adjusted, in Bookworm64 10.0.6. My usual frugal installs to a FAT32 boot partition on SSD, savefiles in the LxPup and VoidPup. ydrv driven -no savefile- in Bookworm64. Fujitsu S761 2nd gen i5 based all intel laptop for hardware. Excellent idle resource use and glxgears FPS in all three. Clean dmesg in Bookworm, one fail in LxPupSc64 and Voidpup64:
[ 11.378202] elogind-uaccess-command[3307]: Failed to apply ACL: Operation not supported
I saw this with the 6.8.4 kernel also but not in any kernels up and through 6.6.25 for those two pups.

Daily use excellent in all three cases, primarily using 'portable' applications.

Thanks,

[ozsouth]

@Marv - I don't have this problem in s15pup64, but it could have serious ramifications - Jasper had issues with re-plugged usbs & ACL failure appears to be related to that. These actions usually originate in /etc/udev/rules.d . I added an extra I2C module to this kernel, hoping to fix it, but seems it didn't. Does unplugging & re-plugging in USB sticks result in re-detection for you?

[Marv]

@ozsouth, Detection of a 32Gb USB3 stick with both FAT32 and EXT2 partitions on it seems normal. Multiple inserts, mounts, unmounts, removals, reinserts, mounts, etc. done. Testbed is a pristine frugal install of LxPupSc64 23.01 +6 with only the kernel swapped to the 6.8.4 one. The ACL error noted above is the only 'fail' in dmesg, both in the pristine install and with my savefile added in. All other operation seems quite normal. both in the pristine install and in my customized install. LxPupSc64 is, of course, LXDE, a possible difference from your S15 install. I'll try a pristine (non-LXDE'd) Voidpup64 just to see if that could be it.

Nope, exactly the same behavior in the pristine install of the current (04/13/2024) VoidPup64. the same dmesg fail, no problems at all with the USB3 stick detection or mounting, repeated several times. Then added in the LXDE ydrv, then my savefile. No change at all wrt the error message or mounting.

Posting from Brave portable in the pristine VoidPup64 install, SNS wifi connection.

[Jasper]

@ozsouth

Apologies, I made a correction to my origonal post here:

viewtopic.php?p=117096&sid=663d4ed62bef ... 91#p117096

The kernel works perfectly.

The low latency kernel will be of benefit to audio producers.

[Marv]

@ozsouth, Setting CONFIG_TMPFS_POSIX_ACL=y in 6.8.4 might be worth a try. See https://blog.tfiu.de/-failed-to-reset-a ... -why-.html for a possible clue. It is not set in 6.8.4 but is set =y in @peebees' 6.6.25 kernel I am using in LxPupSc64 and Voidpup64 at the moment and no ACL errors there. All the other POSIX settings are the same in both.

Just a thought.

[ozsouth]

@Marv - thanks. Link below suggests it is a good idea to enable it.

https://www.kernelconfig.io/CONFIG_TMPF ... sion=6.8.6

Check your pm's in a couple of hours for a test kernel.

[Marv]

Got the PM and kernel. Tested (structure adjusted for slackware) using the same testbed as above and the dmesg fail on boot is gone. USB detection rechecked and seems normal. Posting from that install using Brave portable now. So far good here.

Thanks,

[ozsouth]

As testing has shown CONFIG_TMPFS_POSIX_ACL=y is necessary, I made a 6.8.7 kernel including it. It is not low-latency. See forum Kernels section.
This kernel, although should not trouble most folk, is withdrawn. I will make a 6.6.28 lolat kernel.