On Sat, 27 Apr 2019 14:56:24 +0000 (UTC) Emmanuel Vadot <[email protected]> wrote:
> Author: manu > Date: Sat Apr 27 14:56:24 2019 > New Revision: 346788 > URL: https://svnweb.freebsd.org/changeset/base/346788 > > Log: > arm64: allwinner: ccu_de2: Remove H5 compatible > > We don't have the display engine driver commited in FreeBSD yet so it is > useless to expose the clocks yet (and also it have not been tested on H5). My initial though were wrong. I though I've never tested the clocks on H5 but I did (but the full display engine wasn't tested). What happened is that since r346271 aw_rtc now attaches at BUS_PASS_BUS + MIDDLE as it exposes clocks used by the ccu (clock and control unit) and the CCU attaches at BUS_PASS_BUS + LAST which is the same bus pass as the de2_ccu. Since the de2ccu node is before the ccu node in the DTB it is processed first. A proper fix would be to check if the needed clocks are already registed in the probe function. And this I guess a lots of driver will need this I'll see how to make a generic function that drivers could use. > Reported by: Manuel Stühn ([email protected]) > PR: 237571 > MFC after: 1 week > > Modified: > head/sys/arm/allwinner/clkng/ccu_de2.c > > Modified: head/sys/arm/allwinner/clkng/ccu_de2.c > ============================================================================== > --- head/sys/arm/allwinner/clkng/ccu_de2.c Sat Apr 27 14:48:27 2019 > (r346787) > +++ head/sys/arm/allwinner/clkng/ccu_de2.c Sat Apr 27 14:56:24 2019 > (r346788) > @@ -115,7 +115,6 @@ static struct aw_ccung_clk de2_ccu_clks[] = { > > static struct ofw_compat_data compat_data[] = { > {"allwinner,sun50i-a64-de2-clk", 1}, > - {"allwinner,sun50i-h5-de2-clk", 1}, > {NULL, 0} > }; > -- Emmanuel Vadot <[email protected]> <[email protected]> _______________________________________________ [email protected] mailing list https://lists.freebsd.org/mailman/listinfo/svn-src-all To unsubscribe, send any mail to "[email protected]"
