[Starlingx-discuss] CentOS 7.5 upgrade status update

Saul Wold sgw at linux.intel.com
Tue Sep 18 16:07:41 UTC 2018


Jason,

So if I am understand this correct the test_kernel_module_signatures 
failure listed below is a not an issue due to the way WR built the 
kernel and modules?

Why was this build differently by WR than you would normally build an 
image for testing?  Wouldn't you always have this signature problem then?

I am just trying to understand what differences in the way WR builds and 
the way that is documented, just as we had the issue with the installer 
setup.

Nimalini: Given this information, do we have any blockers for the 7.5 
upgrade?

Thanks
Sau!


On 09/18/2018 07:27 AM, McKenna, Jason wrote:
> Hi folks,
> 
> I do not think the kernel signature item is a real issue, but an artifact of how we built the test image.
> 
> The warning reported is basically saying that the driver was compiled against a different build of the kernel than is currently running.  This is typical for intermediate builds where a kernel was compiled, the driver was compiled, and the kernel compiled again (which was the case when I was attempting to get the CentOS 7.5 test load built as the contents in the repos were changing).  I will clean the kernel driver packages the next time I perform a build and we can verify that the error disappears.
> 
> -J
> 
> For anyone interested in technical details, the kernel regenerates a key pair every time it is built, and compiles one half into the kernel and puts the other half in the kernel-devel package.  Packages (like kernel drivers) building against the kernel-devel package will sign themselves with the key from the package.  Upon loading a driver, the kernel will validate that the signature on the driver matches the half of the key which the kernel has access to.  A mismatched key indicates that the kernel running and kernel-devel package which the driver was built against came from 2 different builds of the kernel, and marks the kernel tainted as a warning.
> 
>> -----Original Message-----
>> From: Rasa, Nimalini
>> Sent: September 18, 2018 9:50 AM
>> To: Lin, Shuicheng <shuicheng.lin at intel.com>; Xie, Cindy
>> <cindy.xie at intel.com>; McKenna, Jason <Jason.McKenna at windriver.com>;
>> Martinez Monroy, Elio <elio.martinez.monroy at intel.com>; Waheed, Numan
>> <Numan.Waheed at windriver.com>; starlingx-discuss at lists.starlingx.io
>> Cc: Khalil, Ghada <Ghada.Khalil at windriver.com>
>> Subject: RE: [Starlingx-discuss] CentOS 7.5 upgrade status update
>>
>> Hi Shuicheng,
>> It does have the correct version:
>> controller-0:~$ uname -a
>> Linux controller-0 3.10.0-862.11.6.el7.36.tis.x86_64 #1 SMP PREEMPT Thu Sep
>> 13 10:48:38 EDT 2018 x86_64 x86_64 x86_64 GNU/Linux
>>
>> Thanks,
>> Nimalini
>>
>> -----Original Message-----
>> From: Lin, Shuicheng [mailto:shuicheng.lin at intel.com]
>> Sent: Monday, September 17, 2018 9:16 PM
>> To: Rasa, Nimalini; Xie, Cindy; McKenna, Jason; Martinez Monroy, Elio;
>> Waheed, Numan; starlingx-discuss at lists.starlingx.io
>> Cc: Khalil, Ghada
>> Subject: RE: [Starlingx-discuss] CentOS 7.5 upgrade status update
>>
>> Hi Nimalini,
>> For the kernel signature issue, could you help confirm whether the kernel
>> version is 3.10.0-862.11.6 or not?
>> You could check it with "uname -a" or search it in the kern.log file.
>> Thanks.
>>
>> Best Regards
>> Shuicheng
>>
>>
>> -----Original Message-----
>> From: Rasa, Nimalini [mailto:Nimalini.Rasa at windriver.com]
>> Sent: Tuesday, September 18, 2018 12:21 AM
>> To: Lin, Shuicheng <shuicheng.lin at intel.com>; Xie, Cindy
>> <cindy.xie at intel.com>; McKenna, Jason <Jason.McKenna at windriver.com>;
>> Martinez Monroy, Elio <elio.martinez.monroy at intel.com>; Waheed, Numan
>> <Numan.Waheed at windriver.com>; starlingx-discuss at lists.starlingx.io
>> Cc: Khalil, Ghada <Ghada.Khalil at windriver.com>
>> Subject: RE: [Starlingx-discuss] CentOS 7.5 upgrade status update
>>
>> Hi Shuicheng,
>> In this test case, basically trying to allocate double the amount of 2M pages.
>>
>> For the kernel signature, I tried it on 3 different hardware labs and got the
>> following:
>> controller-0:~$ cat /proc/sys/kernel/tainted
>> 12288
>> controller-0:~$ grep --color=never -i "module verification failed"
>> /var/log/kern.log
>> 2018-09-14T19:55:12.495 localhost kernel: notice [    8.016748] e1000e:
>> module verification failed: signature and/or required key missing - tainting
>> kernel
>> 2018-09-14T20:49:52.937 controller-0 kernel: notice [    9.517476] e1000e:
>> module verification failed: signature and/or required key missing - tainting
>> kernel
>>
>>
>> Thanks,
>> Nimalini
>>
>> -----Original Message-----
>> From: Lin, Shuicheng [mailto:shuicheng.lin at intel.com]
>> Sent: Friday, September 14, 2018 9:13 PM
>> To: Rasa, Nimalini; Xie, Cindy; McKenna, Jason; Martinez Monroy, Elio;
>> Waheed, Numan; starlingx-discuss at lists.starlingx.io
>> Cc: Khalil, Ghada
>> Subject: RE: [Starlingx-discuss] CentOS 7.5 upgrade status update
>>
>> Hi Nimalini,
>> Could you share me the detail reproduce step for "
>> test_schedule_vm_mempage_config"?
>>
>> For the 2nd failure, I tried "'cat /proc/sys/kernel/tainted'" both in my virtual
>> and bare metal environment, and both of them show 4096. Then what
>> should I do next?
>> "
>> controller-0:~$ cat /proc/sys/kernel/tainted
>> 4096
>> controller-0:~$ grep --color=never -i "module verification failed"
>> /var/log/kern.log controller-0:~$ "
>>
>> Best Regards
>> Shuicheng
>>
>> -----Original Message-----
>> From: Rasa, Nimalini [mailto:Nimalini.Rasa at windriver.com]
>> Sent: Saturday, September 15, 2018 3:02 AM
>> To: Xie, Cindy <cindy.xie at intel.com>; McKenna, Jason
>> <Jason.McKenna at windriver.com>; Martinez Monroy, Elio
>> <elio.martinez.monroy at intel.com>; Waheed, Numan
>> <Numan.Waheed at windriver.com>; starlingx-discuss at lists.starlingx.io
>> Cc: Khalil, Ghada <Ghada.Khalil at windriver.com>; Lin, Shuicheng
>> <shuicheng.lin at intel.com>
>> Subject: RE: [Starlingx-discuss] CentOS 7.5 upgrade status update
>>
>> Hi Cindy,
>> Adding more clarification of the test cases that are failed:
>>
>> FAIL       20180914 06:54:06
>> test_schedule_vm_mempage_config[1048576]
>>    Test case is trying to over allocate 2M pages, expected to reject the cli, in
>> this case it's accepted.
>>
>> FAIL       20180914 07:57:34            test_kernel_module_signatures
>> Test case is doing the following:
>>          - 'cat /proc/sys/kernel/tainted', ensure value is 4096. If not, do following
>> steps:
>>          - 'grep --color=never -i "module verification failed" /var/log/kern.log' to
>> find out failed modules
>>          - 'modinfo <failed_module> | grep --color=never -E "sig|filename" to
>> display signing info for each module
>>
>>
>> Hope it clarifies bit more as to what the test cases it trying to do. If you have
>> any questions please let me know.
>>
>> Thanks,
>> Nimalini
> _______________________________________________
> Starlingx-discuss mailing list
> Starlingx-discuss at lists.starlingx.io
> http://lists.starlingx.io/cgi-bin/mailman/listinfo/starlingx-discuss
> 



More information about the Starlingx-discuss mailing list