[CentOS] Building a kernel without kabichk

[Akemi Yagi]

On Tue, Apr 14, 2009 at 7:33 AM, Terry Hull <tah at nrg-inc.com> wrote:
>
> On 4/14/09 3:12 AM, in article
> bfa89a870904140112s490d223ch440cf009d20e88df at mail.gmail.com, "Akemi Yagi"
> <amyagi at gmail.com> wrote:
>
>> On Mon, Apr 13, 2009 at 8:08 PM, Terry Hull <tah at nrg-inc.com> wrote:
>>>
>>>
>>> On 4/13/09 7:01 PM, in article
>>> bfa89a870904131701i2afd9b41j78b53b0611a9e462 at mail.gmail.com, "Akemi Yagi"
>>> <amyagi at gmail.com> wrote:
>>
>>> The patches I need to make do break kABI.  I just have not yet applied them.
>>> I was trying to get the base kernel to build before I added the complexity
>>> of patching it.
>>
>>>> If the build still fails, please post the last portion of the error
>>>> log (or e-mail the whole log file to me).
>>>
>>> I have attached a compressed log file.  The errors seem to be in the build
>>> of OCFS.
>>
>> Thanks for the err log file. From the looks of it, the error is in the
>> code itself as you noted.  So, that should have nothing to do with
>> kABI.
>>
>> I checked my own log for the kernel 128.1.6 building but did not see
>> anything (warnings and such) relating to ocfs2.  You mentioned " have
>> NOT installed any patches that will break KABI".  Have you installed
>> any patches at all, that is, patches that do not break kABI ?  If so,
>> would any of them touch the ocfs code?
>>
>> Akemi
> No, I have not added any patches at all to the source.  I just modified the
> .spec file and compiled with the --without kabichk option.  I also did not
> have any trouble building the kernel as long as kabichk was enabled.

Terry,

I normally build kABI compatible kernels, so seldom touch the
kabichk-related part of the spec file.  I'm now cc'ing this post to
Alan Bartlett who routinely does it and has amended most part of the
wiki article.

Alan,

This is the beginning of the thread:

http://lists.centos.org/pipermail/centos/2009-April/075066.html

Akemi