The Latest Kernel Release.

PROBLEMCHYLD · 02-24-2021, 05:57 AM

Does anyone know how they test these kernels? There is no way this bug should have carried over into the new kernel. It has annoyed many of us for a while.

mlangdn · 02-24-2021, 06:10 AM

Quote:

Originally Posted by bassmadrigal

I'm not sure of your hardware, but the 5.11 kernel is supposed to be faster on AMD CPUs.

Code:

uname -a
Linux darkstar.example.net 5.11.1 #1 SMP Tue Feb 23 19:12:11 CST 2021 x86_64 AMD FX(tm)-6300 Six-Core Processor AuthenticAMD GNU/Linux

AMD processors are what I have used forever. Not exactly by design, just what was on sale when I bought or built a new rig.

Chuck56 · 02-24-2021, 08:06 AM

Just installed 5.11.1 on this up to date -current desktop. Running without issue so far.

Code:

bash-5.1$ neofetch
                  :::::::                      my@desktop 
            :::::::::::::::::::                ----------- 
         :::::::::::::::::::::::::             OS: Slackware 14.2 x86_64 (post 14.2 -current) x86_64 
       ::::::::cllcccccllllllll::::::          Kernel: 5.11.1 
    :::::::::lc               dc:::::::        Uptime: 3 mins 
   ::::::::cl   clllccllll    oc:::::::::      Packages: 1611 (pkgtool) 
  :::::::::o   lc::::::::co   oc::::::::::     Shell: bash 5.1.4 
 ::::::::::o    cccclc:::::clcc::::::::::::    Resolution: 2560x1440 
 :::::::::::lc        cclccclc:::::::::::::    DE: Plasma 5.21.0 
::::::::::::::lcclcc          lc::::::::::::   WM: KWin 
::::::::::cclcc:::::lccclc     oc:::::::::::   Theme: Breeze Light [Plasma], GTK+ [GTK2], Breeze [GTK3] 
::::::::::o    l::::::::::l    lc:::::::::::   Icons: breeze [Plasma], breeze [GTK2/3] 
 :::::cll:o     clcllcccll     o:::::::::::    Terminal: konsole 
 :::::occ:o                  clc:::::::::::    CPU: AMD A8-7600 Radeon R7 4C+6G (4) @ 3.100GHz 
  ::::ocl:ccslclccclclccclclc:::::::::::::     GPU: AMD ATI Radeon R7 Graphics 
   :::oclcccccccccccccllllllllllllll:::::      Memory: 1611MiB / 6958MiB 
    ::lcc1lcccccccccccccccccccccccco::::
      ::::::::::::::::::::::::::::::::                                 
        ::::::::::::::::::::::::::::
           ::::::::::::::::::::::
                ::::::::::::

drumz · 02-24-2021, 08:15 AM

Quote:

Originally Posted by PROBLEMCHYLD

Does anyone know how they test these kernels? There is no way this bug should have carried over into the new kernel. It has annoyed many of us for a while.

Maybe try searching just a bit?

Yes, some people test stable-rc kernels and report back. You are free to test stabel-rc kernels as well.

https://lore.kernel.org/stable/c38a6...vidia.com/T/#t

Sometimes people report back that a particular patch shouldn't be backported, and Greg KH drops the patch.

There was a bug in 5.10.y that affected me. By the time I got around to tracking it down it was already queued up for the next stable release:

https://lore.kernel.org/stable/YBgI1...2Zr@kroah.com/

(This was actually the first, and so far only, time I've emailed a kernel mailing list.)

Stable maintainers can't know something is broken if nobody reports it.

If you experience a regression, the best way to help is to do a git bisect to identify the offending commit that broke your setup. This will usually point to exactly what's wrong and the kernel developers will be willing to work with you to fix it.

PROBLEMCHYLD · 02-24-2021, 08:23 AM

Others have reported the bug and even came up with a fix. I did search, is how I found the patch and manually applied it. The bug has been around for months way before 5.11.0 is why it shouldn't have made it in.

https://bugzilla.kernel.org/show_bug.cgi?id=210681#c43

Chuck56 · 02-24-2021, 09:35 AM

Now running 5.11.1 on my laptop. This is a full ROM replacement with rebuilt kernel based on Pat's configs with only PINCTRL_CHERRYVIEW changed to built-in instead of module. Posting from the laptop, running without issue so far.

Code:

bash-5.1$ neofetch
                  :::::::                      my@acer_chromebook_14 / CB3-431 
            :::::::::::::::::::                -------------------- 
         :::::::::::::::::::::::::             OS: Slackware 14.2 x86_64 (post 14.2 -current) x86_64 
       ::::::::cllcccccllllllll::::::          Host: Edgar 1.0 
    :::::::::lc               dc:::::::        Kernel: 5.11.1-acer 
   ::::::::cl   clllccllll    oc:::::::::      Uptime: 6 mins 
  :::::::::o   lc::::::::co   oc::::::::::     Packages: 1557 (pkgtool) 
 ::::::::::o    cccclc:::::clcc::::::::::::    Shell: bash 5.1.4 
 :::::::::::lc        cclccclc:::::::::::::    Resolution: 1920x1080 
::::::::::::::lcclcc          lc::::::::::::   DE: Plasma 5.21.0 
::::::::::cclcc:::::lccclc     oc:::::::::::   WM: KWin 
::::::::::o    l::::::::::l    lc:::::::::::   Theme: Breeze Light [Plasma], Breeze [GTK2/3] 
 :::::cll:o     clcllcccll     o:::::::::::    Icons: breeze [Plasma], breeze [GTK2/3] 
 :::::occ:o                  clc:::::::::::    Terminal: konsole 
  ::::ocl:ccslclccclclccclclc:::::::::::::     CPU: Intel Celeron N3160 (4) @ 2.240GHz 
   :::oclcccccccccccccllllllllllllll:::::      GPU: Intel Atom/Celeron/Pentium Processor x5-E8000/J3xxx/N3xxx 
    ::lcc1lcccccccccccccccccccccccco::::       Memory: 859MiB / 3879MiB 
      ::::::::::::::::::::::::::::::::
        ::::::::::::::::::::::::::::                                   
           ::::::::::::::::::::::                                      
                ::::::::::::

Aeterna · 02-24-2021, 09:58 AM

Quote:

Originally Posted by SCerovec

Then, one of my CPUs is an E3 1225 V2 Xeon with integrated VGA running 3.2 GHz at modest 65W TDP - wonder how much of a difference would that option make on it (non ECC motherboard)?

Well, I don't think that you are losing that much but you can always run default Slackware kernel (5.10.x) that has IDXD enabled.

Why You are not losing much?
1) enabled kernel iDXD option is not enough to take advantage of Xeon DMA. You would have to configure the device:
https://github.com/intel/idxd-config
https://01.org/blogs/jinglin/2020/pe...ol-open-source
2) even if you have xeon and device is configured, it seems that iDXD is very useful in quite specific scenarios when you would have DMA channels saturated (data centers multiple CPU's?)

I am not really interested in iDXD so I just searched a bit for the application so this may not be all that you need to know about iDXD.
Hope that will help a bit.

cwizardone · 02-25-2021, 07:08 AM

Year 2021, Round 15.

Another batch of updates has been scheduled for release on Saturday, 27 February 2021, at approximately 09:00, GMT. If no problems are found while testing the release candidates, they might be available sometime on Friday (depending on your time zone).

The details:

5.11.2-rc1, with 12 patches, http://lkml.iu.edu/hypermail/linux/k...2.3/01386.html

5.10.19-rc1, with 23 patches, http://lkml.iu.edu/hypermail/linux/k...2.3/01399.html

5.4.101-rc1, with 17 patches, http://lkml.iu.edu/hypermail/linux/k...2.3/01443.html

This bluetooth patch appears in both 5.11.2 and 5.10.19, http://lkml.iu.edu/hypermail/linux/k...2.3/01416.html

USB and KVM patches appear in all three updates.

SCerovec · 02-25-2021, 02:52 PM

Quote:

Originally Posted by Aeterna

Well, I don't think that you are losing that much but you can always run default Slackware kernel (5.10.x) that has IDXD enabled.

Why You are not losing much?
1) enabled kernel iDXD option is not enough to take advantage of Xeon DMA. You would have to configure the device:
https://github.com/intel/idxd-config
https://01.org/blogs/jinglin/2020/pe...ol-open-source
2) even if you have xeon and device is configured, it seems that iDXD is very useful in quite specific scenarios when you would have DMA channels saturated (data centers multiple CPU's?)

I am not really interested in iDXD so I just searched a bit for the application so this may not be all that you need to know about iDXD.
Hope that will help a bit.

Thanks a lot!

The benefit indeed is in the minor percentage range - most likely not worth pursuing IMHO.

The most benefit i got from this particular rig was cost savings:

Intel mini ITX H61 2nd hand bough mother board and the 4000P capable CPU - it runs as low as 4W on the wall outlet on standby (Pico PSU and 160W LED SMPSU in cascade = ~0dB only fan is the CPU fan) - it saved it's worth in utility bill over a year of use

PROBLEMCHYLD · 02-25-2021, 04:28 PM

Quote:

Originally Posted by cwizardone

Year 2021, Round 15.

Another batch of updates has been scheduled for release on Saturday, 27 February 2021, at approximately 09:00, GMT. If no problems are found while testing the release candidates, they might be available sometime on Friday (depending on your time zone).

The details:

5.11.2-rc1, with 12 patches, http://lkml.iu.edu/hypermail/linux/k...2.3/01386.html

5.10.19-rc1, with 23 patches, http://lkml.iu.edu/hypermail/linux/k...2.3/01399.html

5.4.101-rc1, with 17 patches, http://lkml.iu.edu/hypermail/linux/k...2.3/01443.html

This bluetooth patch appears in both 5.11.2 and 5.10.19, http://lkml.iu.edu/hypermail/linux/k...2.3/01416.html

USB and KVM patches appear in all three updates.

Thanks. Now the bluetooth should work because that's the patch I use to update the source and it works. Finally.... There must have been other bluetooth issues because the one above is not listed in the past updates.

cwizardone · 02-26-2021, 03:20 AM

Kernel updates 5.11.2, 5.10.19 and 5.4.101 are now available at, https://www.kernel.org

The change logs,

https://cdn.kernel.org/pub/linux/ker...angeLog-5.11.2

https://cdn.kernel.org/pub/linux/ker...ngeLog-5.10.19

https://cdn.kernel.org/pub/linux/ker...ngeLog-5.4.101

AlleyTrotter · 02-26-2021, 09:41 AM

Quote:

Originally Posted by cwizardone

Kernel updates 5.11.2, 5.10.19 and 5.4.101 are now available at,
https://git.kernel.org/pub/scm/linux...able/linux.git

The change logs,

https://cdn.kernel.org/pub/linux/ker...angeLog-5.11.2

5.10.19, not yet posted.*

https://cdn.kernel.org/pub/linux/ker...ngeLog-5.4.101

*It has been an hour and it still hasn't been posted. Looks like someone forgot about it and went out for lunch. When it is posted you will be able to find it here, https://www.kernel.org/
Here is the announcement, http://lkml.iu.edu/hypermail/linux/k...2.3/02693.html

5.11.2 - Built and running fine on Slackware64-14.2
Still had to set "IDXD is not set" to make it build
Seems maybe this is the culprit:

Found this in idxd-config/accfg.spec.in

Code:

Requires:	LNAME%{?_isa} = %{version}-%{release}
BuildRequires:	autoconf
BuildRequires:	asciidoc
BuildRequires:	xmlto
BuildRequires:	automake
BuildRequires:	libtool
BuildRequires:	pkgconfig
BuildRequires:	pkgconfig(libkmod)
BuildRequires:	pkgconfig(uuid)
BuildRequires:	pkgconfig(json-c)
BuildRequires:	systemd

if it needs systemd why does it build on current?
anyway it works fine for me.
HTH
john

Aeterna · 02-26-2021, 10:09 AM

this would be a first systemd requirement in the kernel.

idxd-config/accfg.spec.in is a part of https://github.com/intel/idxd-config
idxd-config is an additional software needed to configure idxd. Without proper configuration, kernel option is useless (it is anyway if you don't have Xeon CPU).

kernel sources from

Quote:

https://www.kernel.org/

do not have

Quote:

idxd-config/accfg.spec.in

Quote:

/usr/src/linux-5.11.2/Documentation/ABI/stable/sysfs-driver-dma-idxd
/usr/src/linux-5.11.2/drivers/dma/idxd
/usr/src/linux-5.11.2/drivers/dma/idxd/Makefile
/usr/src/linux-5.11.2/drivers/dma/idxd/cdev.c
/usr/src/linux-5.11.2/drivers/dma/idxd/device.c
/usr/src/linux-5.11.2/drivers/dma/idxd/dma.c
/usr/src/linux-5.11.2/drivers/dma/idxd/idxd.h
/usr/src/linux-5.11.2/drivers/dma/idxd/init.c
/usr/src/linux-5.11.2/drivers/dma/idxd/irq.c
/usr/src/linux-5.11.2/drivers/dma/idxd/registers.h
/usr/src/linux-5.11.2/drivers/dma/idxd/submit.c
/usr/src/linux-5.11.2/drivers/dma/idxd/sysfs.c
/usr/src/linux-5.11.2/include/uapi/linux/idxd.h

I have no idea what Slackware kernel build script is adding. Never used it

PROBLEMCHYLD · 02-26-2021, 10:45 AM

Confirmation the Bluetooth bug is officially fixed. No more manual patching.

https://bugzilla.kernel.org/show_bug.cgi?id=210681#c50

I'll compile it later today or overnight when I'm sleeping. Thanks c-wiz-1

cwizardone · 02-26-2021, 01:27 PM

A report of possible data loss with early 5.12 builds.

Quote:

Watch Out For Possible Data Loss On Early Linux 5.12 Kernels
Written by Michael Larabel in Linux Kernel on 26 February 2021 at 10:00 AM EST.

As a quick PSA for those that may be eager to test out early Git builds of the Linux 5.12 kernel, I've been hitting a very nasty issue on multiple systems leading to corruption / data loss.......

The full story is here, https://www.phoronix.com/scan.php?pa...ly-Buggy-Issue