Skip to contentSkip to conversation

2.6.Stable and EXTRAVERSION

DISCUSSION Justin M. Forbes 7 participants 9 messages Mar 9, 2005 – Mar 15, 2005 Kbuild

lore
Share & tools
lore
#0 lore raw

With the new stable series kernels, the .x versioning is being added to EXTRAVERSION. This has traditionally been a space for local modification. I know several distributions are using EXTRAVERSION for build numbers, platform and assorted other information to differentiate their kernel releases.
I would propose that the new stable series kernels move the .x version information somewhere more official. I certainly do not mind throwing together a patch to support DOTVERSION or what ever people want to call it. Is anyone opposed to such a change?

Thanks,
Justin M. Forbes

#1 lore raw
Quoted 9 lines · Justin M. Forbes

With the new stable series kernels, the .x versioning is being added to EXTRAVERSION. This has traditionally been a space for local modification. I know several distributions are using EXTRAVERSION for build numbers, platform and assorted other information to differentiate their kernel releases.
I would propose that the new stable series kernels move the .x version information somewhere more official. I certainly do not mind throwing together a patch to support DOTVERSION or what ever people want to call it. Is anyone opposed to such a change?

Distros could conceivably use CONFIG_LOCALVERSION, although it might be cleaner to add another version level.

Chris

#2 lore raw
In reply to Justin M. Forbes
Quoted 5 lines · * Justin M. Forbes ([email protected]) wrote:

With the new stable series kernels, the .x versioning is being added to EXTRAVERSION. This has traditionally been a space for local modification. I know several distributions are using EXTRAVERSION for build numbers, platform and assorted other information to differentiate their kernel releases.

Well, it is the same scheme Linus used with 2.6.8.1...Is it posing a real issue?

thanks,
-chris

Signature · 1 line
Linux Security Modules     http://lsm.immunix.org     http://lsm.bkbits.net
#3 lore raw
In reply to Justin M. Forbes
Quoted 5 lines · Justin M. Forbes

With the new stable series kernels, the .x versioning is being added to EXTRAVERSION. This has traditionally been a space for local modification. I know several distributions are using EXTRAVERSION for build numbers, platform and assorted other information to differentiate their kernel releases.

It's no issue for us. We're using this patch to add in the RPM release number:

Makefile +4 -1
Index: linux-2.6.10/Makefile
===================================================================
@@ -158,8 +158,11 @@ endif
 LOCALVERSION = $(subst $(space),, \
 	       $(shell cat /dev/null $(localversion-files:%~=)) \
 	       $(patsubst "%",%,$(CONFIG_LOCALVERSION)))
+ifneq ($(wildcard $(srctree)/rpm-release),)
+RPM_RELEASE := -$(shell cat $(srctree)/rpm-release)
+endif
 
-KERNELRELEASE=$(VERSION).$(PATCHLEVEL).$(SUBLEVEL)$(EXTRAVERSION)$(LOCALVERSION)
+KERNELRELEASE=$(VERSION).$(PATCHLEVEL).$(SUBLEVEL)$(EXTRAVERSION)$(RPM_RELEASE)$(LOCALVERSION)
 
 # SUBARCH tells the usermode build what the underlying arch is.  That is set
 # first, and if a usermode build is happening, the "ARCH=um" on the command

Cheers,

Signature · 2 lines
Andreas Gruenbacher <[email protected]>
SUSE Labs, SUSE LINUX GMBH
#5 lore raw
In reply to Justin M. Forbes
Quoted 9 lines · Justin M. Forbes

With the new stable series kernels, the .x versioning is being added to EXTRAVERSION. This has traditionally been a space for local modification. I know several distributions are using EXTRAVERSION for build numbers, platform and assorted other information to differentiate their kernel releases.
I would propose that the new stable series kernels move the .x version information somewhere more official. I certainly do not mind throwing together a patch to support DOTVERSION or what ever people want to call it. Is anyone opposed to such a change?

I think it will confuse scripts which expect something local in the 4th field. I confess that I have such, and that field is turned into a directory name during builds, so test results are saved in a proper place. I think vendors and people who care will just keep three digits, and those who want the last can make their EXTRAVERSION
2.Joes_Bar_&_Grill_486
or whatever is needed.

Add to that the confusion of having mainline releases
2.6.x.LOCAL
and stable be
2.6.x.y.LOCAL
and you really make work for people who do things with scripts. This is probably not an issue for humans, and people could program around it, but I think it's a solution to a non-problem. Vendors will have the -stable and their own patches, so they probably will only keep three fields anyway.

And don't even suggest tell Linus to start calling mainline releases 2.6.x.0 because that would probably break even more things.

Leave well enough alone!

Signature · 3 lines
   -bill davidsen ([email protected])
"The secret to procrastination is to put things off until the
 last possible moment - but no longer"  -me
#6 lore raw
Quoted 13 lines · Bill Davidsen
Quoted 9 lines · Justin M. Forbes

With the new stable series kernels, the .x versioning is being added to EXTRAVERSION. This has traditionally been a space for local modification. I know several distributions are using EXTRAVERSION for build numbers, platform and assorted other information to differentiate their kernel releases.
I would propose that the new stable series kernels move the .x version information somewhere more official. I certainly do not mind throwing together a patch to support DOTVERSION or what ever people want to call it. Is anyone opposed to such a change?

Quoted 3 lines

I think it will confuse scripts which expect something local in the 4th field. I confess that I have such, and that field is turned into a directory name during builds, so test results are saved in a proper place. I think vendors and people who care will just keep three digits, and those who want the last can make their EXTRAVERSION
2.Joes_Bar_&_Grill_486
or whatever is needed.

There's also the LOCALVERSION that can be set in the config file. I've switched to using that, since it means one less kernel patch to port.

Chris

#7 lore raw
Quoted 1 line · Chris Friesen

Bill Davidsen wrote:

	[...snip...]
Quoted 10 lines
Quoted 7 lines

I think it will confuse scripts which expect something local in the 4th field. I confess that I have such, and that field is turned into a directory name during builds, so test results are saved in a proper place. I think vendors and people who care will just keep three digits, and those who want the last can make their EXTRAVERSION
2.Joes_Bar_&_Grill_486
or whatever is needed.

There's also the LOCALVERSION that can be set in the config file. I've switched to using that, since it means one less kernel patch to port.

I have to see what that generates. The problem is LOCALVERSION and current use of both 3 and 4 field kernel versions. You need a smarter script to handle that.

Signature · 3 lines
bill davidsen <[email protected]>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.
#4 lore raw
In reply to Andreas Gruenbacher
Quoted 17 lines · Andreas Gruenbacher
Quoted 5 lines · Justin M. Forbes

With the new stable series kernels, the .x versioning is being added to EXTRAVERSION. This has traditionally been a space for local modification. I know several distributions are using EXTRAVERSION for build numbers, platform and assorted other information to differentiate their kernel releases.

It's no issue for us. We're using this patch to add in the RPM release number:

Index: linux-2.6.10/Makefile
===================================================================

Makefile +0 -0
@@ -158,8 +158,11 @@ endif
 LOCALVERSION = $(subst $(space),, \
 	       $(shell cat /dev/null $(localversion-files:%~=)) \

Makefile endif · line 168

Quoted 10 lines
Makefile +4 -1
 	       $(patsubst "%",%,$(CONFIG_LOCALVERSION)))
+ifneq ($(wildcard $(srctree)/rpm-release),)
+RPM_RELEASE := -$(shell cat $(srctree)/rpm-release)
+endif
 
-KERNELRELEASE=$(VERSION).$(PATCHLEVEL).$(SUBLEVEL)$(EXTRAVERSION)$(LOCALVERSION)
+KERNELRELEASE=$(VERSION).$(PATCHLEVEL).$(SUBLEVEL)$(EXTRAVERSION)$(RPM_RELEASE)$(LOCALVERSION)
 
 # SUBARCH tells the usermode build what the underlying arch is.  That is set
 # first, and if a usermode build is happening, the "ARCH=um" on the command

Naming your rpm-release file: localversion00-rpm would do the same without the need to patch the kernel.

Sam

#8 lore raw
In reply to Bill Davidsen
Quoted 4 lines · Bill Davidsen

I have to see what that generates. The problem is LOCALVERSION and current use of both 3 and 4 field kernel versions. You need a smarter script to handle that.

... To which the knee jerk solution would be ...

Name Linus's kernel releases 2.6.z.0 ... ?

Signature · 5 lines
[Name      ]   ::  [Matan I. Peled    ]
[Location  ]   ::  [Israel            ]
[Public Key]   ::  [0xD6F42CA5        ]
[Keyserver ]   ::  [keyserver.kjsl.com]
encrypted/signed  plain text  preferred