Bug 12753 - allow building partial package set from 2.2.16 kernel spec
Summary: allow building partial package set from 2.2.16 kernel spec
Alias: None
Product: Red Hat Raw Hide
Classification: Retired
Component: kernel   
(Show other bugs)
Version: 1.0
Hardware: All
OS: Linux
Target Milestone: ---
Assignee: Michael K. Johnson
QA Contact:
Keywords: FutureFeature
Depends On:
TreeView+ depends on / blocked
Reported: 2000-06-20 08:58 UTC by Michael Tokarev
Modified: 2008-05-01 15:37 UTC (History)
0 users

Fixed In Version:
Doc Type: Enhancement
Doc Text:
Story Points: ---
Clone Of:
Last Closed: 2003-02-19 16:28:51 UTC
Type: ---
Regression: ---
Mount Type: ---
Documentation: ---
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---

Attachments (Terms of Use)
patch to kernel-2.2.16-1.spec file (already described) (4.05 KB, patch)
2000-06-20 09:02 UTC, Michael Tokarev
no flags Details | Diff
diff against 2.2.16-8 (3.18 KB, patch)
2000-07-11 12:39 UTC, Michael Tokarev
no flags Details | Diff
diff against 2.2.16-24 to allow having buildibcs=0 and buildpcmcia=0 (2.17 KB, patch)
2000-10-19 17:41 UTC, Michael Tokarev
no flags Details | Diff

Description Michael Tokarev 2000-06-20 08:58:55 UTC
Kernel source package shipped by RH always produces all the
components (smp kernels, boot kernels, pcmcia package, ibcs
package and so on) on build stage.  Building of one component
is a long operation.  But most users who builds kernel himself
needs only particular configuration (e.g. only up-kernel for
i686 without pcmcia and ibcs -- most common config I guess).
Here is a patch for .spec file (against 2.2.16-1) to allow
such separation.  It introduces some variables (%defined at
top) named as mk_XXX that are default to 1 (%define mk_smp 1).
When changing those variables to 0, corresponding components
are skipped from build process.

One more thing I want here is to allow also easy building
of kernel-headers and kernel-source "for i686 arch", i.e.
so that only one target and one build will require to produce
kernel-xxx-i686.rpm and kernel-source-xxx-i386.rpm.
But this is not so easy with current rpm...

And also, I tried to make easily configurable custom config
and have a patch for this (to .spec also), but this is a bit
ugly for moment.  The idea is to allow definition of some
variable (e.g. --define 'kerncustom someval') and in this case
kernel-xxx-somevel.xxx.rpm will built using kernel-xxx-somevel.config
files, not the original redhat kernel-xxx.config configs.

Please apply (or review and apply) this patch -- it not breaks
anything, it is clean and it is simple.
This all will greatly simplify life for all people who needs
custom kernel builds.

Thank you.

Comment 1 Michael Tokarev 2000-06-20 09:02:05 UTC
Created attachment 620 [details]
patch to kernel-2.2.16-1.spec file (already described)

Comment 2 Michael Tokarev 2000-07-11 12:39:28 UTC
Created attachment 982 [details]
diff against 2.2.16-8

Comment 3 Michael Tokarev 2000-10-19 17:18:26 UTC
I see that at least BOOT/smp/enterprize kernels can be excluded now.
Thank you.  But pcmcia and ibcs still can't.  Is this a principal position?
It may be so, as kernels built without pcmcia/ibcs stuff will not be
"compatible" with original redhat's ones (some modules will be omitted).
If this is a cause, than Ok -- just confirm this and let's close this
whole issue (still new).  If not, here is another patch, against 2.2.16-24.
What I really missed is that kernel is one of almost two packages that
can't be build as non-root, and the only cause is ibcs devices.  So, if
ibcs stuff can be excluded, kernel will be buildable for non-root users.

And also the changes I tried to made here triggered a bug in rpm -- it won't
understand nested `%ifarch' and `%if %{var}' constructs.

Comment 4 Michael Tokarev 2000-10-19 17:41:12 UTC
Created attachment 4390 [details]
diff against 2.2.16-24 to allow having buildibcs=0 and buildpcmcia=0

Comment 5 Doug Ledford 2000-10-26 20:47:15 UTC
Michael can decide if this should be in the src rpm.

Comment 6 Bill Nottingham 2003-02-19 16:28:51 UTC
Closing at this point.

Note You need to log in before you can comment on or make changes to this bug.