Re: How to recover from a broken kernel upgrade

[M. Cuffaro]

Thanks Tom. Editing wifi_regd.c seems pretty promising. I'll try that.

-Mike


On 2023-09-14 02:21, Tom Dial wrote:



On 9/13/23 17:07, M. Cuffaro wrote:

Hello all,

I have looked around for an answer but was unable to find one, and 
was hoping that someone here could help me. *First, some background 
to my information about my question*. I recently ran


apt full-upgrade

on my Debian bookworm system. Because of what seems like a bug in 
dkms, the upgrade to linux-image-6.1.0-12-amd64 could not complete. 
Running


apt -f install

does not help. The uphshot is that, although I can still boot into my 
laptop using the old version of my kernel (either 
linux-image-6.1.0-11-amd64 or linux-image-6.1.0-10-amd64), I can no 
longer use apt to upgrade other packages on my system. Furthermore 
when I run


needrestart

I am informed that:

The currently running kernel version is 6.1.0-10-amd64 which is not 
the expected kernel version 6.1.0-12-amd64.


*Here is my question: How can I rollback the broken upgrade safely so 
that I can continue to use my system (in particular, apt) normally? 
Any help anyone can provide would be much appreciated.


*Below are some further details. First, here is the result of running 
apt -f install:


$ sudo apt -f install
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
The following packages were automatically installed and are no longer 
required:
   linux-headers-6.1.0-10-amd64 linux-headers-6.1.0-10-common 
linux-image-6.1.0-10-amd64

Use 'sudo apt autoremove' to remove them.
0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
4 not fully installed or removed.
After this operation, 0 B of additional disk space will be used.
Setting up linux-image-6.1.0-12-amd64 (6.1.52-1) ...
/etc/kernel/postinst.d/dkms:
dkms: running auto installation service for kernel 6.1.0-12-amd64.
Sign command: /usr/lib/linux-kbuild-6.1/scripts/sign-file
Signing key: /var/lib/dkms/mok.key
Public certificate (MOK): /var/lib/dkms/mok.pub


(material deleted for brevity)

   CC [M] 
/var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/rtw_android.o

   CC [M] /var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/rtw_proc.o
   CC [M] /var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/nlrtw.o


Your file "wifi_regd.c" appears to be broken. The version at

https://github.com/RinCat/RTL88x2BU-Linux-Driver/tree/master

had the undeclared REGULATORY_IGNORE_STALE_KICKOFF removed a couple of 
months ago.


Perhaps you need to update the driver source dkms is working with.
(I don't know about, or vouch for, that github, or know anything about 
the AC1200 or its Realtek driver.)


Regards

Tom Dial

/var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/wifi_regd.c: In 
function ‘rtw_regd_init’:
/var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/wifi_regd.c:409:36: 
error: ‘REGULATORY_IGNORE_STALE_KICKOFF’ undeclared (first use in 
this function)
   409 | wiphy->regulatory_flags |= 
REGULATORY_IGNORE_STALE_KICKOFF;

   | ^~~
/var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/wifi_regd.c:409:36: 
note: each undeclared identifier is reported only once for each 
function it appears in
   CC [M] 
/var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/rtw_rhashtable.o
make[2]: *** 
[/usr/src/linux-headers-6.1.0-12-common/scripts/Makefile.build:255: 
/var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/wifi_regd.o] Error 1

make[2]: *** Waiting for unfinished jobs
make[1]: *** [/usr/src/linux-headers-6.1.0-12-common/Makefile:2037: 
/var/lib/dkms/rtl88x2bu/5.13.1/build] Error 2

make[1]: Leaving directory '/usr/src/linux-headers-6.1.0-12-amd64'
make: *** [Makefile:2506: modules] Error 2

Here is some information about my system:

** Model information
sys_vendor: Acer
product_name: Swift SF314-56
product_version: V1.08
chassis_vendor: Acer
chassis_version: V1.08
bios_vendor: Insyde Corp.
bios_version: V1.08
board_vendor: WL
board_name: Strongbow_WL
board_version: V1.08

Here is the output of uname -a and lsb_release -a:

[mike@dallben ~]$ uname -a
Linux dallben 6.1.0-10-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.1.38-2 
(2023-07-27) x86_64 GNU/Linux

[mike@dallben ~]$ lsb_release -a
No LSB modules are available.
Distributor ID:    Debian
Description:    Debian GNU/Linux 12 (bookworm)
Release:    12
Codename:    bookworm

Best,
-Mike

Re: How to recover from a broken kernel upgrade

[M. Cuffaro]

Hi Marco,

thanks for the helpful reply. I'll consider this but as a last resort.

Best,
-Mike


On 2023-09-14 00:39, Marco wrote:

Am 13.09.2023 19:07 schrieb M. Cuffaro:


Consult /var/lib/dkms/rtl88x2bu/5.13.1/build/make.log for more
information.

That drivers is incompatible with the current kernel.
How did you install it?
Uninstall it, then run the kernel update again.

Run
dkms status
And try
sudo dkms remove rtl88x2bu/5.13.1 --all

Be aware that this drivers seems to be for wireless networking.
Most likely your WiFi with that Realtek adapter won't work.
We need to find a further solution, but make sure you have an Ethernet
connection to connect o the internet!

Re: How to recover from a broken kernel upgrade

[Tom Dial]

On 9/13/23 17:07, M. Cuffaro wrote:

Hello all,

I have looked around for an answer but was unable to find one, and was hoping 
that someone here could help me. *First, some background to my information 
about my question*. I recently ran

apt full-upgrade

on my Debian bookworm system. Because of what seems like a bug in dkms, the 
upgrade to linux-image-6.1.0-12-amd64 could not complete. Running

apt -f install

does not help. The uphshot is that, although I can still boot into my laptop 
using the old version of my kernel (either linux-image-6.1.0-11-amd64 or 
linux-image-6.1.0-10-amd64), I can no longer use apt to upgrade other packages 
on my system. Furthermore when I run

needrestart

I am informed that:

The currently running kernel version is 6.1.0-10-amd64 which is not the 
expected kernel version 6.1.0-12-amd64.

*Here is my question: How can I rollback the broken upgrade safely so that I 
can continue to use my system (in particular, apt) normally? Any help anyone 
can provide would be much appreciated.

*Below are some further details. First, here is the result of running apt -f 
install:

$ sudo apt -f install
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
The following packages were automatically installed and are no longer required:
   linux-headers-6.1.0-10-amd64 linux-headers-6.1.0-10-common 
linux-image-6.1.0-10-amd64
Use 'sudo apt autoremove' to remove them.
0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
4 not fully installed or removed.
After this operation, 0 B of additional disk space will be used.
Setting up linux-image-6.1.0-12-amd64 (6.1.52-1) ...
/etc/kernel/postinst.d/dkms:
dkms: running auto installation service for kernel 6.1.0-12-amd64.
Sign command: /usr/lib/linux-kbuild-6.1/scripts/sign-file
Signing key: /var/lib/dkms/mok.key
Public certificate (MOK): /var/lib/dkms/mok.pub


(material deleted for brevity)


   CC [M] /var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/rtw_android.o
   CC [M] /var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/rtw_proc.o
   CC [M] /var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/nlrtw.o


Your file "wifi_regd.c" appears to be broken. The version at

https://github.com/RinCat/RTL88x2BU-Linux-Driver/tree/master

had the undeclared REGULATORY_IGNORE_STALE_KICKOFF removed a couple of months 
ago.

Perhaps you need to update the driver source dkms is working with.
(I don't know about, or vouch for, that github, or know anything about the 
AC1200 or its Realtek driver.)

Regards

Tom Dial


/var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/wifi_regd.c: In function 
‘rtw_regd_init’:
/var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/wifi_regd.c:409:36: error: 
‘REGULATORY_IGNORE_STALE_KICKOFF’ undeclared (first use in this function)
   409 | wiphy->regulatory_flags |= REGULATORY_IGNORE_STALE_KICKOFF;
   | ^~~
/var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/wifi_regd.c:409:36: note: 
each undeclared identifier is reported only once for each function it appears in
   CC [M] /var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/rtw_rhashtable.o
make[2]: *** 
[/usr/src/linux-headers-6.1.0-12-common/scripts/Makefile.build:255: 
/var/lib/dkms/rtl88x2bu/5.13.1/build/os_dep/linux/wifi_regd.o] Error 1
make[2]: *** Waiting for unfinished jobs
make[1]: *** [/usr/src/linux-headers-6.1.0-12-common/Makefile:2037: 
/var/lib/dkms/rtl88x2bu/5.13.1/build] Error 2
make[1]: Leaving directory '/usr/src/linux-headers-6.1.0-12-amd64'
make: *** [Makefile:2506: modules] Error 2

Here is some information about my system:

** Model information
sys_vendor: Acer
product_name: Swift SF314-56
product_version: V1.08
chassis_vendor: Acer
chassis_version: V1.08
bios_vendor: Insyde Corp.
bios_version: V1.08
board_vendor: WL
board_name: Strongbow_WL
board_version: V1.08

Here is the output of uname -a and lsb_release -a:

[mike@dallben ~]$ uname -a
Linux dallben 6.1.0-10-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.1.38-2 
(2023-07-27) x86_64 GNU/Linux
[mike@dallben ~]$ lsb_release -a
No LSB modules are available.
Distributor ID:    Debian
Description:    Debian GNU/Linux 12 (bookworm)
Release:    12
Codename:    bookworm

Best,
-Mike

Re: How to recover from a broken kernel upgrade

[Marco]

Am 13.09.2023 19:07 schrieb M. Cuffaro:

> Consult /var/lib/dkms/rtl88x2bu/5.13.1/build/make.log for more
> information.

That drivers is incompatible with the current kernel.
How did you install it?
Uninstall it, then run the kernel update again.

Run
dkms status
And try
sudo dkms remove rtl88x2bu/5.13.1 --all 

Be aware that this drivers seems to be for wireless networking.
Most likely your WiFi with that Realtek adapter won't work.
We need to find a further solution, but make sure you have an Ethernet
connection to connect o the internet!

How to recover from a broken kernel upgrade

[M. Cuffaro]

Hello all,

I have looked around for an answer but was unable to find one, and was 
hoping that someone here could help me. *First, some background to my 
information about my question*. I recently ran


apt full-upgrade

on my Debian bookworm system. Because of what seems like a bug in dkms, 
the upgrade to linux-image-6.1.0-12-amd64 could not complete. Running


apt -f install

does not help. The uphshot is that, although I can still boot into my 
laptop using the old version of my kernel (either 
linux-image-6.1.0-11-amd64 or linux-image-6.1.0-10-amd64), I can no 
longer use apt to upgrade other packages on my system. Furthermore when 
I run


needrestart

I am informed that:

The currently running kernel version is 6.1.0-10-amd64 which is not the 
expected kernel version 6.1.0-12-amd64.


*Here is my question: How can I rollback the broken upgrade safely so 
that I can continue to use my system (in particular, apt) normally? Any 
help anyone can provide would be much appreciated.


*Below are some further details. First, here is the result of running 
apt -f install:


$ sudo apt -f install
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
The following packages were automatically installed and are no longer 
required:
  linux-headers-6.1.0-10-amd64 linux-headers-6.1.0-10-common 
linux-image-6.1.0-10-amd64

Use 'sudo apt autoremove' to remove them.
0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
4 not fully installed or removed.
After this operation, 0 B of additional disk space will be used.
Setting up linux-image-6.1.0-12-amd64 (6.1.52-1) ...
/etc/kernel/postinst.d/dkms:
dkms: running auto installation service for kernel 6.1.0-12-amd64.
Sign command: /usr/lib/linux-kbuild-6.1/scripts/sign-file
Signing key: /var/lib/dkms/mok.key
Public certificate (MOK): /var/lib/dkms/mok.pub

Building module:
Cleaning build area...
'./driverctl' make all(bad exit status: 2)
Error! Bad return status for module build on kernel: 6.1.0-12-amd64 (x86_64)
Consult /var/lib/dkms/rtl88x2bu/5.13.1/build/make.log for more information.
Error! One or more modules failed to install during autoinstall.
Refer to previous errors for more information.
dkms: autoinstall for kernel: 6.1.0-12-amd64 failed!
run-parts: /etc/kernel/postinst.d/dkms exited with return code 11
dpkg: error processing package linux-image-6.1.0-12-amd64 (--configure):
 installed linux-image-6.1.0-12-amd64 package post-installation script 
subprocess returned error exit status 1

Setting up linux-headers-6.1.0-12-amd64 (6.1.52-1) ...
/etc/kernel/header_postinst.d/dkms:
dkms: running auto installation service for kernel 6.1.0-12-amd64.
Sign command: /usr/lib/linux-kbuild-6.1/scripts/sign-file
Signing key: /var/lib/dkms/mok.key
Public certificate (MOK): /var/lib/dkms/mok.pub

Building module:
Cleaning build area...
'./driverctl' make all...(bad exit status: 2)
Error! Bad return status for module build on kernel: 6.1.0-12-amd64 (x86_64)
Consult /var/lib/dkms/rtl88x2bu/5.13.1/build/make.log for more information.
Error! One or more modules failed to install during autoinstall.
Refer to previous errors for more information.
dkms: autoinstall for kernel: 6.1.0-12-amd64 failed!
run-parts: /etc/kernel/header_postinst.d/dkms exited with return code 11
Failed to process /etc/kernel/header_postinst.d at 
/var/lib/dpkg/info/linux-headers-6.1.0-12-amd64.postinst line 11.

dpkg: error processing package linux-headers-6.1.0-12-amd64 (--configure):
 installed linux-headers-6.1.0-12-amd64 package post-installation 
script subprocess returned error exit status 1

dpkg: dependency problems prevent configuration of linux-headers-amd64:
 linux-headers-amd64 depends on linux-headers-6.1.0-12-amd64 (= 
6.1.52-1); however:

  Package linux-headers-6.1.0-12-amd64 is not configured yet.

dpkg: error processing package linux-headers-amd64 (--configure):
 dependency problems - leaving unconfigured
dpkg: dependency problems prevent configuration of linux-image-amd64:
 linux-image-amd64 depends on linux-image-6.1.0-12-amd64 (= 6.1.52-1); 
however:

  Package linux-image-6.1.0-12-amd64 is not configured yet.

dpkg: error processing package linux-image-amd64 (--configure):
 dependency problems - leaving unconfigured
Errors were encountered while processing:
 linux-image-6.1.0-12-amd64
 linux-headers-6.1.0-12-amd64
 linux-headers-amd64
 linux-image-amd64
E: Sub-process /usr/bin/dpkg returned an error code (1)

Here are the contents of /var/lib/dkms/rtl88x2bu/5.13.1/build/make.log:

$ cat /var/lib/dkms/rtl88x2bu/5.13.1/build/make.log
DKMS make.log for rtl88x2bu-5.13.1 for kernel 6.1.0-12-amd64 (x86_64)
Wed 13 Sep 2023 11:09:22 AM EDT
make ARCH=x86_64 CROSS_COMPILE= -C /lib/modules/6.1.0-12-amd64/build 
M=/var/lib/dkms/rtl88x2bu/5.13.1/build  modules

make[1]: Entering directory '/usr/src/linux-headers-6.1.0-12-amd64'
  CC [M] /var/lib/dkms/rtl88x2bu/5.13.1/build/core/rtw_cmd.o
  CC [M] 

Re: Broken kernel?

[Gene Heskett]

On Monday 12 March 2018 06:26:09 Hans wrote:

> Hi Gene,
>
> got into the same problem a long time ago. In my case I got in trouble
> with one of the kernel modules, it was the graphics driver module.
>
Somebody's quoting is making it look as if I'm the OP. I am not, and my 
stuff is running just fine.

But unless I missed the memo, the OP still has a problem.

FWIW, the nvidia driver is a profane word in my machine control world. It 
doesn't like to share its toys, locking out the irq's for hundreds of 
milliseconds at a time, when the cpu is expected to check the machines 
status at millisecond intervals to see if its tripped a limit switch 
etc.

The nouveau driver doesn't do that. And its fast enough to draw a 
backplot of the machines motion in close to realtime.

The ATI drivers are better, but they also do the irq lockouts for too 
long, so the "better" is still relative, and still not good enough for 
good, 100% safe, machine control. ATM, my x86 boxes are all running the 
intel i915 drivers, which seem to be doing things well.

> This one got in trouble with the vesa driver (somehow), and so it
> hanged at boot with a black screen.
>
> The solution was at that time, to dienstall all the Nvidia packages,
> which of course let a new initram-file created, which of course got no
> nvidia kernel module installed.
>
> If there is in your case also a kernel module the problem, you might
> be able, to blacklist the problematic module. If you have access to
> the hardddrive with the kernel modules (i.e. by using a livefile dvd),
> you can either edit the blacklist-modules file, or , if this does not
> work, by removing the modules physically out of the way from the
> modules directory (sorry for my bad English).
>
> The others got right: amd64 means, that the kernel is running on Intel
> compatible 64-bit processors. And you are also right: There was a
> time, when there were also special 64-bit Intel-cpus in the repo (when
> I remember correchtly, these were aimed for Itanium cpu's).

Which somehow seem to have vanished from the used market.
>
> Another hint: If you are able, to start a debian live cd (the actual
> one), you can take a look, which kernel modules are loaded, and also,
> which graphical driver is loaded. However, I believe, you are quite
> happy, when you are able, to ge a shell! So just try a small live
> system.
>
> If the debian live system failes, you may try Knoppix (the version
> with the same kernel version) or maybe another Debian based livesystem
> (I have kali- linux or aptosid in my mind).
>
> Long words, little sense - I hope this helps a little bit.
>
> Good luck and all the best!
>
> Hans

Gonna need it Hans, I made the mistake of looking out the front door at 
my wheelchair ramp construction area this morning, finding 1/4" snow on 
all my nice new lumber. And I still have one, maybe two long days work 
demolishing the deck where the ramp will go. Taking screws out with a 
big prybar. It would be nice if they'd just unscrew, but they've been 
there 15+ years. Takeing 20 calendars off my pile of discards would help 
too as I'm not the man I once was, even once...  Sigh.

-- 
Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

Re: Broken kernel?

[floris]

Gene Heskett schreef op 2018-03-12 10:51:

On Monday 12 March 2018 05:20:07 floris wrote:


Tom Furie schreef op 2018-03-12 09:41:
> On Mon, Mar 12, 2018 at 04:34:56AM -0400, Gene Heskett wrote:
>> I just happen to have 2 machines on my home network that are Intel
>> D525MW
>> motherboards, with dual core intel atom 1.8 GHz cpu's. They run
>> flawlessly from powerbump to powerbump, on an rtai patched but now
>> elderly kernel, a kernel that says its for an i686-pae. Not an
>> AMD64.
>
> Yes, all 64-bit x86 compatible CPUs are also 32-bit compatible.
>
> Cheers,
> Tom

Intel says the Intel Atom CPU N2600 @ 1.60GHz is 64bit[1]. The am64
kernel is the right one.

[1]https://ark.intel.com/products/58916/Intel-Atom-Processor-N2600-1M-
Cache-1_6-GHz ---
Floris
Call off the dogs, you and several other are correct, and I should not 
be

posting bad info at 3 or 4 am when the only reason I'm up at all is to
take the recyclables to the curb. Its just that I've been running the 
32

bit versions so well for more than half a decade I was convinced that
was the way to fly.

But does this not leave the OP scratching his head, wondering why his
machine won't boot? So we are back to the starting square and I'm out 
of

ammo. At this point I'd run memtest86 long enough, at least overnight,
to verify its not bad memory, and if it is, reseat the lappy style 
dimm,

they are getting old enough that if the air is bad, some contact
exercise might be in order. Mine never get turned off except for power
failures, so thermal stress in the atoms mounting has not stuck up a
hand yet. For those that are shut down, thermal cycling might begin to
be a problem.

I'm going back to bed, where anyone my age ought to be this time of the
morning since he's busy building a wheelchair ramp so I can bring my 
xyl
home with a busted osteoporosis riddled leg. If I ever find the guy 
that

called these the golden years, I'd like to see where the gold is in his
house.


I just checked and the dog is still playing in the backyard, so you 
don't have to worry :-)


Back to OP's question. The Intel Atom N2600 is not very Linux/ 
opensource friendly. I have and old laptop with the same processor and 
also have mixed result with different kernels. Fortunately the Debian 
Stable/ Strech amd64 kernel works without big issues. The Testing 
version has problems with the gma500 module. (I know the OP has a 
different issue and doesn't use the build-in GPU.)


@OP
Do you have a custom initramfs? Check 
/etc/initramfs-tools/initramfs.conf and make sure MODULES is set to 
"most".


---
Floris

Re: Broken kernel?

[Hans]

Hi Gene, 

got into the same problem a long time ago. In my case I got in trouble with 
one of the kernel modules, it was the graphics driver module.

This one got in trouble with the vesa driver (somehow), and so it hanged at 
boot with a black screen.

The solution was at that time, to dienstall all the Nvidia packages, which of 
course let a new initram-file created, which of course got no nvidia kernel 
module installed.

If there is in your case also a kernel module the problem, you might be able, 
to blacklist the problematic module. If you have access to the hardddrive with 
the kernel modules (i.e. by using a livefile dvd), you can either edit the 
blacklist-modules file, or , if this does not work, by removing the modules 
physically out of the way from the modules directory (sorry for my bad 
English).

The others got right: amd64 means, that the kernel is running on Intel 
compatible 64-bit processors. And you are also right: There was a time, when 
there were also special 64-bit Intel-cpus in the repo (when I remember 
correchtly, these were aimed for Itanium cpu's).

Another hint: If you are able, to start a debian live cd (the actual one), you 
can take a look, which kernel modules are loaded, and also, which graphical 
driver is loaded. However, I believe, you are quite happy, when you are able, 
to ge a shell! So just try a small live system. 

If the debian live system failes, you may try Knoppix (the version with the 
same kernel version) or maybe another Debian based livesystem (I have kali-
linux or aptosid in my mind).

Long words, little sense - I hope this helps a little bit.

Good luck and all the best!

Hans 

   

Re: Broken kernel?

[Alexander V. Makartsev]

On 12.03.2018 04:36, John ff wrote:
>
> i have an Intel Atom CPU N2600 @ 1.60GHz that I fairly recently
> upgraded from Whezzy to Jessie, but I cannot boot the kernel(s) that
> came with Jessie.
>
> All is OK with vmlinuz-3.2.0-4-amd64, but both 3.16.0-4-amd64 and
> 3.16.0-5-amd64 fail to load. When equipped with a screen I see it
> says Loading .. but it does not complete with no disk activity.
I have an old Atom N450 CPU based NetBook right next to me, running
'jessie' with "3.16.0-5-amd64" kernel.
It was upgraded continuously from "Squeeze" without any issues.
I suggest you to boot with "single" and "debug" kernel parameters, with
"quiet" parameter removed and look into log output.
Hopefully that would help to find out where exactly it stops booting
process.

-- 
With kindest regards, Alexander.

⢀⣴⠾⠻⢶⣦⠀ 
⣾⠁⢠⠒⠀⣿⡁ Debian - The universal operating system
⢿⡄⠘⠷⠚⠋⠀ https://www.debian.org
⠈⠳⣄ 

Re: Broken kernel?

[Tom Furie]

On Mon, Mar 12, 2018 at 05:51:07AM -0400, Gene Heskett wrote:

> Call off the dogs, you and several other are correct, and I should not be 
> posting bad info at 3 or 4 am when the only reason I'm up at all is to 
> take the recyclables to the curb. Its just that I've been running the 32 
> bit versions so well for more than half a decade I was convinced that 
> was the way to fly.

I wasn't aiming no set the dogs on anybody, merely looking to clarify
that amd64 packages aren't only for AMD processors or 32-bit packages
only for 32-bit processors.

Indeed, I still run 32-bit on a couple of 64-bit capable systems due to
the factors that they were originally built on 32-bit hardware and their
configuration and workload really wouldn't benefit in any real terms
from 64-bit, so the work involved in the transfer wouldn't be well
spent.

Cheers,
Tom

-- 
1: No code table for op: ++post


signature.asc
Description: Digital signature

Re: Broken kernel?

[Tom Furie]

On Mon, Mar 12, 2018 at 10:20:07AM +0100, floris wrote:

> Intel says the Intel Atom CPU N2600 @ 1.60GHz is 64bit[1]. The am64 kernel
> is the right one.

Either a 32-bit or 64-bit kernel could be equally valid, depending on
other parameters.

Cheers,
Tom

-- 
All God's children are not beautiful.  Most of God's children are, in fact,
barely presentable.
-- Fran Lebowitz, "Metropolitan Life"


signature.asc
Description: Digital signature

Re: Broken kernel?

[Gene Heskett]

On Monday 12 March 2018 05:20:07 floris wrote:

> Tom Furie schreef op 2018-03-12 09:41:
> > On Mon, Mar 12, 2018 at 04:34:56AM -0400, Gene Heskett wrote:
> >> I just happen to have 2 machines on my home network that are Intel
> >> D525MW
> >> motherboards, with dual core intel atom 1.8 GHz cpu's. They run
> >> flawlessly from powerbump to powerbump, on an rtai patched but now
> >> elderly kernel, a kernel that says its for an i686-pae. Not an
> >> AMD64.
> >
> > Yes, all 64-bit x86 compatible CPUs are also 32-bit compatible.
> >
> > Cheers,
> > Tom
>
> Intel says the Intel Atom CPU N2600 @ 1.60GHz is 64bit[1]. The am64
> kernel is the right one.
>
> [1]https://ark.intel.com/products/58916/Intel-Atom-Processor-N2600-1M-
>Cache-1_6-GHz ---
> Floris
Call off the dogs, you and several other are correct, and I should not be 
posting bad info at 3 or 4 am when the only reason I'm up at all is to 
take the recyclables to the curb. Its just that I've been running the 32 
bit versions so well for more than half a decade I was convinced that 
was the way to fly.

But does this not leave the OP scratching his head, wondering why his 
machine won't boot? So we are back to the starting square and I'm out of 
ammo. At this point I'd run memtest86 long enough, at least overnight, 
to verify its not bad memory, and if it is, reseat the lappy style dimm, 
they are getting old enough that if the air is bad, some contact 
exercise might be in order. Mine never get turned off except for power 
failures, so thermal stress in the atoms mounting has not stuck up a 
hand yet. For those that are shut down, thermal cycling might begin to 
be a problem.

I'm going back to bed, where anyone my age ought to be this time of the 
morning since he's busy building a wheelchair ramp so I can bring my xyl 
home with a busted osteoporosis riddled leg. If I ever find the guy that 
called these the golden years, I'd like to see where the gold is in his 
house.

-- 
Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

Re: Broken kernel?

[floris]

Tom Furie schreef op 2018-03-12 09:41:

On Mon, Mar 12, 2018 at 04:34:56AM -0400, Gene Heskett wrote:

I just happen to have 2 machines on my home network that are Intel 
D525MW

motherboards, with dual core intel atom 1.8 GHz cpu's. They run
flawlessly from powerbump to powerbump, on an rtai patched but now
elderly kernel, a kernel that says its for an i686-pae. Not an AMD64.


Yes, all 64-bit x86 compatible CPUs are also 32-bit compatible.

Cheers,
Tom


Intel says the Intel Atom CPU N2600 @ 1.60GHz is 64bit[1]. The am64 
kernel is the right one.


[1]https://ark.intel.com/products/58916/Intel-Atom-Processor-N2600-1M-Cache-1_6-GHz
---
Floris

Re: Broken kernel?

[Tom Furie]

On Mon, Mar 12, 2018 at 04:34:56AM -0400, Gene Heskett wrote:

> I just happen to have 2 machines on my home network that are Intel D525MW 
> motherboards, with dual core intel atom 1.8 GHz cpu's. They run 
> flawlessly from powerbump to powerbump, on an rtai patched but now 
> elderly kernel, a kernel that says its for an i686-pae. Not an AMD64.

Yes, all 64-bit x86 compatible CPUs are also 32-bit compatible.

Cheers,
Tom

-- 
Support bacteria -- it's the only culture some people have!


signature.asc
Description: Digital signature

Re: Broken kernel?

[Gene Heskett]

On Monday 12 March 2018 02:54:49 Tom Furie wrote:

> On Sun, Mar 11, 2018 at 11:42:44PM -0400, Gene Heskett wrote:
> > AMD64 would be like its built for an AMD phenom or newer. You want a
> > kernel built for an intel cpu.
>
> AMD64, in the context of package naming, is for any 64-bit x86
> compatible CPU.
>
> Cheers,
> Tom

I just happen to have 2 machines on my home network that are Intel D525MW 
motherboards, with dual core intel atom 1.8 GHz cpu's. They run 
flawlessly from powerbump to powerbump, on an rtai patched but now 
elderly kernel, a kernel that says its for an i686-pae. Not an AMD64.

gene@shop:~$ uname -r
3.4-9-rtai-686-pae

gene@lathe:~$ uname -r
3.4-9-rtai-686-pae

This kernel is in fact not a pae kernel, because the rtai patches disable 
that, we found that out after the fact, but those 2 machines only have 
2GB of dram, so the lack of pae is moot. Enabling pae on those wrecks 
the context switching latency anyway. And that is far more important for 
what they are doing (running linuxcnc) than anything else they may be 
asked to do. Yes, they are also 32 bit kernels, again because the 
context switch time for a 32 bit kernel is 1/2 the context switching 
time for a full 64 bit kernel. So its possible we may be compareing 
apples to oranges. What is for sure is that this mobo is the fastest 
mobo ever to come out of silicon valley for running linuxcnc, by a 
factor of at least 2/1 over the next best. I would have several more of 
them if Intel hadn't disco'd it about the time it was discovered by the 
linuxcnc people.

-- 
Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

Re: Broken kernel?

[deloptes]

John  ff wrote:

> All is OK with vmlinuz-3.2.0-4-amd64, but both 3.16.0-4-amd64 and
> 3.16.0-5-amd64 fail to load. When equipped with a screen I see it
> says Loading .. but it does not complete with no disk activity. The
> 3.2.0-4-amd64 kernel gives the same loading message followed shortly by
> saying the /dev/sda2 disk partition is clean and it continues to boot.
> Sorry I cannot remember the exact words after "Loading" but this m/c
> usually runs headless as my firewall and server.

I was forced to run 3.2 for quite a time, because newer did not boot. In my
case AMD cpu in headless server.
I found out that for some reason initramfs-tools did not
package /sbin/switch_root, so I unpacked initrd, copied switch_root
manually and packed again.
You can see more if you remove the "quite" option from the kernel command
line or use the recovery boot option.
If this wouldn't help I would pass init=/bin/sh and work out the boot
process manually to see where it fails.
At one point of time there was something with udev.

regards

Re: Broken kernel?

[deloptes]

Gene Heskett wrote:

> AMD64 would be like its built for an AMD phenom or newer. You want a
> kernel built for an intel cpu.

You are disappointing me, Gene. It is called amd64 because AMD first
released 64 arch cpu to the public. All intel 64 run with this kernel (and
packages).

Re: Broken kernel?

[Tom Furie]

On Sun, Mar 11, 2018 at 11:42:44PM -0400, Gene Heskett wrote:

> AMD64 would be like its built for an AMD phenom or newer. You want a 
> kernel built for an intel cpu.

AMD64, in the context of package naming, is for any 64-bit x86
compatible CPU.

Cheers,
Tom

-- 
A girl's best friend is her mutter.
-- Dorothy Parker


signature.asc
Description: Digital signature

Re: Broken kernel?

[Gene Heskett]

On Sunday 11 March 2018 21:10:16 Eike Lantzsch wrote:

> On Sunday, March 11, 2018 7:47:13 PM -03 Gene Heskett wrote:
> > On Sunday 11 March 2018 19:36:37 John ff wrote:
> > > i have an Intel Atom CPU N2600 @ 1.60GHz that I fairly recently
> > > upgraded from Whezzy to Jessie, but I cannot boot the kernel(s)
> > > that came with Jessie.
> > >
> > > All is OK with vmlinuz-3.2.0-4-amd64, but both 3.16.0-4-amd64 and
> > > 3.16.0-5-amd64 fail to load. When equipped with a screen I see it
> > > says Loading .. but it does not complete with no disk activity.
> > > The 3.2.0-4-amd64 kernel gives the same loading message followed
> > > shortly by saying the /dev/sda2 disk partition is clean and it
> > > continues to boot. Sorry I cannot remember the exact words after
> > > "Loading" but this m/c usually runs headless as my firewall and
> > > server.
> > >
> > > I am looking for any clues as how to continue the upgrade to
> > > Stretch.
> > >
> > > ==John ffitch
> > >
> > >
> > > ⁣Sent from Blue ​
> >
> > First and obvious question: Its an intel atom, so why are you trying
> > to load and exec a kernel that was built to run on an amd64?
>
> because that would be the right kernel to load?
>
AMD64 would be like its built for an AMD phenom or newer. You want a 
kernel built for an intel cpu.

> This might be related:
> Debian Bug report logs - #794410
> debian-installer: Installer hangs during 'select and install software'
> Otherwise I didn't find any similar scenarios with Jessie.
> At this point I'm afraid you maneuvered yourself into a corner and you
> better install Debian Stretch from scratch. This may save time instead
> of trying to figure out what is wrong, fixing and following the
> upgrade-path further. You might want to try a live-DVD (put on a
> USB-stick) first but then don't install from the live-os but use
> netinstall instead.
> Booting the live-os will give you the opportunity to save data and
> config files before starting the install from scratch.
> All the best
> Eike



-- 
Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

Re: Broken kernel?

[Eike Lantzsch]

On Sunday, March 11, 2018 7:47:13 PM -03 Gene Heskett wrote:
> On Sunday 11 March 2018 19:36:37 John ff wrote:
> > i have an Intel Atom CPU N2600 @ 1.60GHz that I fairly recently
> > upgraded from Whezzy to Jessie, but I cannot boot the kernel(s) that
> > came with Jessie.
> > 
> > All is OK with vmlinuz-3.2.0-4-amd64, but both 3.16.0-4-amd64 and
> > 3.16.0-5-amd64 fail to load. When equipped with a screen I see it
> > says Loading .. but it does not complete with no disk activity. The
> > 3.2.0-4-amd64 kernel gives the same loading message followed shortly
> > by saying the /dev/sda2 disk partition is clean and it continues to
> > boot. Sorry I cannot remember the exact words after "Loading" but this
> > m/c usually runs headless as my firewall and server.
> > 
> > I am looking for any clues as how to continue the upgrade to Stretch.
> > 
> > ==John ffitch
> > 
> > 
> > ⁣Sent from Blue ​
> 
> First and obvious question: Its an intel atom, so why are you trying to
> load and exec a kernel that was built to run on an amd64?
because that would be the right kernel to load?

This might be related:
Debian Bug report logs - #794410
debian-installer: Installer hangs during 'select and install software'
Otherwise I didn't find any similar scenarios with Jessie.
At this point I'm afraid you maneuvered yourself into a corner and you better 
install Debian Stretch from scratch. This may save time instead of trying to 
figure out what is wrong, fixing and following the upgrade-path further.
You might want to try a live-DVD (put on a USB-stick) first but then don't 
install from the live-os but use netinstall instead.
Booting the live-os will give you the opportunity to save data and config files 
before starting the install from scratch.
All the best
Eike
-- 
Eike Lantzsch ZP6CGE

Re: Broken kernel?

[Gene Heskett]

On Sunday 11 March 2018 19:36:37 John ff wrote:

> i have an Intel Atom CPU N2600 @ 1.60GHz that I fairly recently
> upgraded from Whezzy to Jessie, but I cannot boot the kernel(s) that
> came with Jessie.
>
> All is OK with vmlinuz-3.2.0-4-amd64, but both 3.16.0-4-amd64 and
> 3.16.0-5-amd64 fail to load. When equipped with a screen I see it
> says Loading .. but it does not complete with no disk activity. The
> 3.2.0-4-amd64 kernel gives the same loading message followed shortly
> by saying the /dev/sda2 disk partition is clean and it continues to
> boot. Sorry I cannot remember the exact words after "Loading" but this
> m/c usually runs headless as my firewall and server.
>
> I am looking for any clues as how to continue the upgrade to Stretch.
>
> ==John ffitch
>
>
> ⁣Sent from Blue ​

First and obvious question: Its an intel atom, so why are you trying to 
load and exec a kernel that was built to run on an amd64?


-- 
Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

Broken kernel?

[John ff]

i have an Intel Atom CPU N2600 @ 1.60GHz that I fairly recently
upgraded from Whezzy to Jessie, but I cannot boot the kernel(s) that
came with Jessie.

All is OK with vmlinuz-3.2.0-4-amd64, but both 3.16.0-4-amd64 and
3.16.0-5-amd64 fail to load. When equipped with a screen I see it
says Loading .. but it does not complete with no disk activity. The
3.2.0-4-amd64 kernel gives the same loading message followed shortly by
saying the /dev/sda2 disk partition is clean and it continues to boot.
Sorry I cannot remember the exact words after "Loading" but this m/c
usually runs headless as my firewall and server.

I am looking for any clues as how to continue the upgrade to Stretch.

==John ffitch


⁣Sent from Blue ​