Date: Sun, 28 Jan 2018 23:54:06 -0700 From: Warner Losh <imp@bsdimp.com> To: Oleksandr Tymoshenko <gonzo@bluezbox.com> Cc: Poul-Henning Kamp <phk@phk.freebsd.dk>, Emmanuel Vadot <manu@bidouilliste.com>, John Baldwin <jhb@freebsd.org>, Ravi Pokala <rpokala@mac.com>, src-committers <src-committers@freebsd.org>, svn-src-all@freebsd.org, svn-src-head@freebsd.org, owner-src-committers@freebsd.org Subject: Re: svn commit: r328257 - in head/sys: arm/broadcom/bcm2835 dts/arm modules Message-ID: <CANCZdfrUT9ZBNqQQo0C=2ydzGHq%2B=c_KS%2BkY5krxaiRGfpJLhw@mail.gmail.com> In-Reply-To: <20180129063950.GA59901@bluezbox.com> References: <201801220710.w0M7AUm9091853@repo.freebsd.org> <CANCZdfpq2QoG4EAj0VW2FF=4VXRv-qQKFfJTJerWH9YOwVoVBA@mail.gmail.com> <90451.1516663240@critter.freebsd.dk> <2987003.eeGRFBb6N8@ralph.baldwin.cx> <CANCZdfrh0NHq7cbkq_genEdzo%2BB3G4TTAcEzpgh11sr%2B82e9aw@mail.gmail.com> <93949.1516733748@critter.freebsd.dk> <20180127210801.37b8001125dd0a2c92372f98@bidouilliste.com> <72042.1517094867@critter.freebsd.dk> <8d8ae9d10058fd72ce3ec467181c9f22@megadrive.org> <13025.1517179897@critter.freebsd.dk> <20180129063950.GA59901@bluezbox.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, Jan 28, 2018 at 11:39 PM, Oleksandr Tymoshenko <gonzo@bluezbox.com> wrote: > > Ignoring value of ofw_bus_status_okay(dev) in probe method is > wrong for number of reasons and pinctl and clocks were brought up > as an example why "status" property is more than just > attach/dont-attach flag. So there is strictly technical reason > why it should be checked. Once this boilerplate code moves to > simplebus bus probe method as I believe Warner proposed drivers > will not be able to control this aspect any more. > I have patches that does exactly this. I have a few more that will allow drivers some control over the situation planned, but they require hammering out some things in newbus first, which I'll write up once I have something workable and it's been blessed by a couple people I've asked to give feedback. The writeup will just be the first round, not the final destination, btw. We need to work out several tricky issues that are interlocking to ensure that we can enable drivers selectively after boot time. That's tricky for a lot of reasons, but the tl;dr version is that embedded is different because you don't just enable a device, but all the dependencies which, unlike x86, are often 'non-linear' and possibly non-orthogonal to other devices (eg, you may either have A B and C enabled, or D E and F, but never any other combination, how do you transition from state to state sanely). It's a tricky ball of wax that we've neglected for too long. Warner
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CANCZdfrUT9ZBNqQQo0C=2ydzGHq%2B=c_KS%2BkY5krxaiRGfpJLhw>