I'm attempting to upgrade to kernel-smp-2.4.21-27.0.4.EL (x86_64) on a Dell 1850. The new kernel is panicing on reboot. When comparing against the current working kernel, I just noticed that we're currently running the "UP" kernel:
[root@polaris root]# uname -srvp Linux 2.4.21-27.0.2.EL #1 SMP Wed Jan 19 11:49:43 CST 2005 x86_64
First question... why is the kernel version reporting as SMP when this is the UP kernel package? I thought it might have something to do with HT, but no, it's seeing both processors:
[root@polaris root]# cat /proc/cpuinfo | grep proc processor : 0 processor : 1 processor : 2 processor : 3
Second question... has anyone else experienced problems with kernel-smp-2.4.21-27.0.4.EL on x86_64? I installed the newer kernel and it paniced on reboot. I'm offsite without console, so I'll be going in tomorrow to take a closer look. Just FYI, here is the information on each kernel we're working with:
[root@polaris root]# rpm -qi kernel Name : kernel Relocations: (not relocatable) Version : 2.4.21 Vendor: CentOS Release : 27.0.2.EL Build Date: Wed 19 Jan 2005 12:56:58 PM EST Install Date: Fri 08 Apr 2005 05:44:16 AM EDT Build Host: head Group : System Environment/Kernel Source RPM: kernel-2.4.21-27.0.2.EL.src.rpm Size : 33381631 License: GPLv2 Signature : DSA/SHA1, Thu 20 Jan 2005 10:52:57 AM EST, Key ID 7049e44d025e513b Summary : The Linux kernel (the core of the Linux operating system) Description : The kernel package contains the Linux kernel (vmlinuz), the core of your CentOS operating system. The kernel handles the basic functions of the operating system: memory allocation, process allocation, device input and output, etc.
[root@polaris root]# rpm -qi kernel-smp Name : kernel-smp Relocations: (not relocatable) Version : 2.4.21 Vendor: CentOS-3 Release : 27.0.4.EL Build Date: Fri 22 Apr 2005 06:47:22 PM EDT Install Date: Wed 11 May 2005 10:16:23 AM EDT Build Host: tru64 Group : System Environment/Kernel Source RPM: kernel-2.4.21-27.0.4.EL.src.rpm Size : 26669758 License: GPLv2 Signature : DSA/SHA1, Fri 22 Apr 2005 07:12:58 PM EDT, Key ID 7049e44d025e513b Packager : Pasi Pirhonen upi@iki.fi - http://iki.fi/upi/ Summary : The Linux kernel compiled for SMP machines. Description : This package includes a SMP version of the Linux kernel. It is required only on machines with two or more CPUs.
Install the kernel-smp package if your machine uses two or more CPUs.
Thanks in advance,
-- Jason Dixon DixonGroup Consulting http://www.dixongroup.net
On Wed, May 11, 2005 at 10:59:51AM -0400, Jason Dixon wrote:
I'm attempting to upgrade to kernel-smp-2.4.21-27.0.4.EL (x86_64) on a Dell 1850.
On CentOS-3, you should be using the ia32e (EM64T Intel)kernels not the x86_64 (AMD opteron). Can you retry with the ia32e kernels (UP/SMP) ?
Cheers,
Tru
On May 11, 2005, at 11:14 AM, Tru Huynh wrote:
On Wed, May 11, 2005 at 10:59:51AM -0400, Jason Dixon wrote:
I'm attempting to upgrade to kernel-smp-2.4.21-27.0.4.EL (x86_64) on a Dell 1850.
On CentOS-3, you should be using the ia32e (EM64T Intel)kernels not the x86_64 (AMD opteron). Can you retry with the ia32e kernels (UP/SMP) ?
Yes, I see that now:
[root@polaris root]# rpm -q --qf "%{ARCH} %{VERSION}-%{RELEASE}\n" kernel ia32e 2.4.21-27.0.2.EL
[root@polaris root]# rpm -q --qf "%{ARCH} %{VERSION}-%{RELEASE}\n" kernel-smp x86_64 2.4.21-27.0.4.EL
However, I wonder why "yum install kernel-smp" chose the x86_64 kernel? Looking in the CentOS update repository, there is no ia32e directory. Inside the x86_64 directory, I see the following kernels:
kernel-2.4.21-27.0.4.EL.ia32e.rpm kernel-2.4.21-27.0.4.EL.x86_64.rpm kernel-smp-2.4.21-27.0.4.EL.x86_64.rpm kernel-smp-unsupported-2.4.21-27.0.4.EL.x86_64.rpm kernel-unsupported-2.4.21-27.0.4.EL.ia32e.rpm kernel-unsupported-2.4.21-27.0.4.EL.x86_64.rpm
Why is there no kernel-smp for ia32e? Based on the current kernel's behavior, I guess it's safe to assume the UP kernel will work for SMP, but why isn't it labeled accordingly? Is this just an oversight by the package maintainers? It would be great if someone could post a FAQ regarding these kernels for CentOS 3.4.
Thanks,
-- Jason Dixon DixonGroup Consulting http://www.dixongroup.net
On Wed, May 11, 2005 at 11:22:17AM -0400, Jason Dixon wrote:
However, I wonder why "yum install kernel-smp" chose the x86_64 kernel?
because the kernel-smp.ia32e is missing :( (our fault) I am rebuilding it now, sorry.
Looking in the CentOS update repository, there is no ia32e directory.
That is expected. The ia32e kernel is the only addition to the x86_64 which was there first. On CentOS-4, PNAELV just merged the ia32e and x86_64 kernels.
Tru
On May 11, 2005, at 11:49 AM, Tru Huynh wrote:
On Wed, May 11, 2005 at 11:22:17AM -0400, Jason Dixon wrote:
However, I wonder why "yum install kernel-smp" chose the x86_64 kernel?
because the kernel-smp.ia32e is missing :( (our fault) I am rebuilding it now, sorry.
No problem, I'm just glad it's not my fault. :)
Would you mind posting the location when it's uploaded? I would like to try it out ASAP.
Looking in the CentOS update repository, there is no ia32e directory.
That is expected. The ia32e kernel is the only addition to the x86_64 which was there first. On CentOS-4, PNAELV just merged the ia32e and x86_64 kernels.
Just curious, where does the ia32e naming schema originate? It's not an Itanium, and it's a 64bit architecture (let's not start that argument again, please).
Thanks,
-- Jason Dixon DixonGroup Consulting http://www.dixongroup.net
On Wed, May 11, 2005 at 05:49:28PM +0200, Tru Huynh wrote:
On Wed, May 11, 2005 at 11:22:17AM -0400, Jason Dixon wrote:
However, I wonder why "yum install kernel-smp" chose the x86_64 kernel?
because the kernel-smp.ia32e is missing :( (our fault)
replying to myself :( I checked the spec file, it's upstream :(
%ifarch ia32e %define buildsmp 0 %endif
So there is no kernel-smp.ia32e (none also on PNAELV)
Tru
On May 11, 2005, at 11:58 AM, Tru Huynh wrote:
On Wed, May 11, 2005 at 05:49:28PM +0200, Tru Huynh wrote:
On Wed, May 11, 2005 at 11:22:17AM -0400, Jason Dixon wrote:
However, I wonder why "yum install kernel-smp" chose the x86_64 kernel?
because the kernel-smp.ia32e is missing :( (our fault)
replying to myself :( I checked the spec file, it's upstream :(
%ifarch ia32e %define buildsmp 0 %endif
So there is no kernel-smp.ia32e (none also on PNAELV)
Weird. Any idea why the UP processor reports SMP then?
Thanks,
-- Jason Dixon DixonGroup Consulting http://www.dixongroup.net
On Wed, May 11, 2005 at 12:00:51PM -0400, Jason Dixon wrote:
So there is no kernel-smp.ia32e (none also on PNAELV)
Weird. Any idea why the UP processor reports SMP then?
because the ia32e kernel is already SMP :P
CONFIG_SMP=y in the kernel-2.4.21-ia32e.config (cf /usr/src/linux-2.4.21-27.0.4.EL/configs from kernel-source-2.4.21-27.0.4.EL)
You shouldn't need the kernel-smp on EM64T.
Cheers,
Tru
On May 11, 2005, at 12:39 PM, Tru Huynh wrote:
On Wed, May 11, 2005 at 12:00:51PM -0400, Jason Dixon wrote:
So there is no kernel-smp.ia32e (none also on PNAELV)
Weird. Any idea why the UP processor reports SMP then?
because the ia32e kernel is already SMP :P
CONFIG_SMP=y in the kernel-2.4.21-ia32e.config (cf /usr/src/linux-2.4.21-27.0.4.EL/configs from kernel-source-2.4.21-27.0.4.EL)
You shouldn't need the kernel-smp on EM64T.
Someone needs to beat the Red Hat kernel packager with a clue stick.
Thanks for your help, the new kernel works fine. :)
-- Jason Dixon DixonGroup Consulting http://www.dixongroup.net
On Wed, 2005-05-11 at 12:42 -0400, Jason Dixon wrote:
On May 11, 2005, at 12:39 PM, Tru Huynh wrote:
On Wed, May 11, 2005 at 12:00:51PM -0400, Jason Dixon wrote:
So there is no kernel-smp.ia32e (none also on PNAELV)
Weird. Any idea why the UP processor reports SMP then?
because the ia32e kernel is already SMP :P
CONFIG_SMP=y in the kernel-2.4.21-ia32e.config (cf /usr/src/linux-2.4.21-27.0.4.EL/configs from kernel-source-2.4.21-27.0.4.EL)
You shouldn't need the kernel-smp on EM64T.
Someone needs to beat the Red Hat kernel packager with a clue stick.
Thanks for your help, the new kernel works fine. :)
--
You are not alone ... I screwed up my kernel too. {And I build this stuff :)}
Their logic (silly as it is) is that there is no such thing as a non-SMP kernel for ia32e (since it has hyperthreading) ... SO ... no need for an ia32e SMP and non-SMP kernel. So, for a reason that I am not possibly able to understand, they decided that the ia32e kernel would be compiled with SMP and not be labeled as SMP.
go figure :)
On May 11, 2005, at 12:59 PM, Johnny Hughes wrote:
You are not alone ... I screwed up my kernel too. {And I build this stuff :)}
And a damn fine job you guys do. </brown nose>
Their logic (silly as it is) is that there is no such thing as a non-SMP kernel for ia32e (since it has hyperthreading) ... SO ... no need for an ia32e SMP and non-SMP kernel. So, for a reason that I am not possibly able to understand, they decided that the ia32e kernel would be compiled with SMP and not be labeled as SMP.
Wow, makes perfect sense. Ouch! My brain just popped a synapse.
Thanks,
-- Jason Dixon DixonGroup Consulting http://www.dixongroup.net