Hi,
good idea. Thorsten, can you implement this feature?
Thanks
Waldemar
On Thu, 14 Jun 2007 at 18:17 +0200, Lothar Gesslein wrote:
> On Thu, 2007-06-14 at 10:38 +0200, [EMAIL PROTECTED] wrote:
> > Author: tg
> > Date: 2007-06-14 10:38:37 +0200 (Thu, 14 Jun 2007)
> > New Revision: 2789
> >
> > Modified:
> > trunk/freewrt/rules.mk
> > Log:
> > simplify kernel version selection
> > this code depends on Config.in always defining either
> > FWRT_LINUX_2_4 xor FWRT_LINUX_2_4 to 'y'
> >
> >
> > Modified: trunk/freewrt/rules.mk
> > ===================================================================
> > --- trunk/freewrt/rules.mk 2007-06-14 08:31:45 UTC (rev 2788)
> > +++ trunk/freewrt/rules.mk 2007-06-14 08:38:37 UTC (rev 2789)
> > @@ -82,18 +82,10 @@
> > # XXX derive these from FWRT_* vars
> > UNAME_S:= Linux
> >
> > -ifeq ($(strip ${FWRT_LINUX_2_4_BRCM}),y)
> > +ifeq ($(strip ${FWRT_LINUX_2_4}),y)
> > UNAME_R:= 2.4.34
> > -else ifeq ($(strip ${FWRT_LINUX_2_6_BRCM}),y)
> > +else ifeq ($(strip ${FWRT_LINUX_2_6}),y)
> > UNAME_R:= 2.6.19.1
> > -else ifeq ($(strip ${FWRT_LINUX_2_4_RB}),y)
> > -UNAME_R:= 2.4.34
> > -else ifeq ($(strip ${FWRT_LINUX_2_4_X86}),y)
> > -UNAME_R:= 2.4.34
> > -else ifeq ($(strip ${FWRT_LINUX_2_6_RB}),y)
> > -UNAME_R:= 2.6.19.1
> > -else ifeq ($(strip ${FWRT_LINUX_2_6_X86}),y)
> > -UNAME_R:= 2.6.19.1
> > else
> > $(error Cannot determine target "uname -r" output)
> > endif
>
> I think it would be nice to be able to have different kernel versions
> for different targets. I'm building 2.6.21 for ar7-2.6 (WIP), because i
> want to be in sync with OpenWrt.
> Each platform-maintainer should be able to use the version he/she wants,
> without having to wait for the other platforms to update their patches.
>
> Unfortunatelly, this currently works (worked, you just kicked it
> completely ;)) only very limited, because the patches in
> target/linux/generic-2.x are always shared and might have to be upgraded
> for a newer kernel. My proposed solution for this is to check for
> existance of generic-${UNAME_R}, if a patch with the same name, e.g.
> generic-${UNAME_R}/100-foobarpatch exists, it is used in favor of
> generic-2.x/100-foobarpatch, but all other patches from generic-2.x/ are
> used nevertheless. This way patches that still apply can left there, and
> not too much code is redundant.
>
> If we want to be consequent, this also has to be done for
> {brcm,x86,rb,foobar}-2.x, though i rather don't think this is necessary.
> But if the work to also implement it here is negligible it should be
> done.
>
> Comments, please. And i'd be glad if someone skilled could take the work
> of implementing it nicely =)
>
> Bye,
> Lothar
>
> PS: Chaos Communication Camp 2007 anyone? I'll be there quite sure.
> PPS: Trac is dead, the sqlite db in the background seems to be borked.
>
> _______________________________________________
> freewrt-developers mailing list
> [email protected]
> https://www.freewrt.org/lists/listinfo/freewrt-developers
--
All embedded development kits suck. This one just sucks less.
http://www.freewrt.org
_______________________________________________
freewrt-developers mailing list
[email protected]
https://www.freewrt.org/lists/listinfo/freewrt-developers
