On Tue, Dec 6, 2016 at 7:37 PM, Andrei Gherzan <[email protected]> wrote: > On Tue, Dec 06, 2016 at 06:27:32PM +0000, Andrei Gherzan wrote: >> On Mon, Nov 28, 2016 at 08:37:09PM -0500, Trevor Woerner wrote: >> > On Mon 2016-11-28 @ 03:16:11 PM, Khem Raj wrote: >> > > >> > > > On Nov 28, 2016, at 11:07 AM, Trevor Woerner <[email protected]> >> > > > wrote: >> > > > >> > > > If the PREEMPT_RT patch is applied, the kernel version becomes, say, >> > > > 4.4.32-rt43 (instead of 4.4.32). This confuses the version handling >> > > > code in >> > > > this class. Update how the version string is processed so that >> > > > trailing rt- >> > > > strings are properly handled, in addition to handling the existing >> > > > cases. >> > > > >> > > >> > > This probably will solve the issue I see with 4.9-rcX recipes that are >> > > in my tree on kraj/master >> > >> > I'm not familiar with the issue you're seeing, but the existing and new >> > code >> > are looking for 3 int()s separated by periods. If your recipes have the >> > string >> > "4.9-rcX" then I'm guessing there might still be an issue since the third >> > int() will be "-rcX" in your case. If this is true, you'll need to take a >> > look >> > at where "int(min_ver[2])" is used further down in that bbclass file. >> >> I agreed this is not the best implementation of this. We should only get > > agree* > >> the version using a regex that would get X.Y.Z-R with an optional Z and >> R. >> >> -- >> Andrei Gherzan > -- > Andrei Gherzan > Question is if this version dance in get_dts is still necessary: We don't have kernel < 3.18 or kernel 4.4.<6 around. Why not use what is set in KERNEL_DEVICETREE?
Andreas -- _______________________________________________ yocto mailing list [email protected] https://lists.yoctoproject.org/listinfo/yocto
