Date: Thu, 24 Dec 1998 12:44:26 +0900 From: Kazutaka YOKOTA <yokota@zodiac.mech.utsunomiya-u.ac.jp> To: mike@smith.net.au, des@flood.ping.uio.no Cc: current@FreeBSD.ORG, yokota@zodiac.mech.utsunomiya-u.ac.jp Subject: Re: future of syscons Message-ID: <199812240344.MAA22380@zodiac.mech.utsunomiya-u.ac.jp> In-Reply-To: Your message of "Tue, 22 Dec 1998 15:56:18 PST." <199812222356.PAA01687@dingo.cdrom.com> References: <199812222356.PAA01687@dingo.cdrom.com>
next in thread | previous in thread | raw e-mail | index | archive | help
des>> The kernel configuration file should look: des>> des>> controller kbdc0 at isa? port IO_KBD tty des>> device atkbd0 at isa? tty flags xxxx des>> device psm0 at isa? tty flags yyy des> des>Shouldn't it be: des> des>device atkbd0 at kbdc0 flags xxxx des>device psm0 at kbdc0 flags yyyy msmith>> controller kbdc0 at isa? port IO_KBD tty msmith>> device atkbd0 at isa? tty flags xxxx msmith>> device psm0 at isa? tty flags yyy msmith> msmith>Just to note that ultimately you won't specify the 'tty' mask here, but msmith>rather in the interrupt connection function, so: msmith> msmith>controller atkbdc0 at isa? port IO_KBD msmith>device atkbd0 at kbdc? msmith>device psm0 at kbdc? That's exactly what I want to do! What I proposed was merely an interim, compromise measure. Because of the current config(8) and the ISA bus code in i386, we cannot write flags like the following. device psm0 at kbdc? flags yyy (In alpha, this shouldn't be a problem, should it?) msmith>Also, since the atkbd and psm device interrupt handlers can only be msmith>called from the atkbdc driver, they don't register interrupts and thus msmith>don't have/need masks. There is a compromise here too. In i386, if atkbdc wants to have TWO interrupt handlers, it has to call register_intr() directly (and make sure that there won't be conflicts), rather than letting isa_configure() to do the deed, because isa_configure() can assign only one interrupt to a device. Am I wrong? I expect this can be solved once resource management code is fully utilized in i386. But, this is not yet happening. For the time being, psm and atkbd have to register their own interrupt handlers independently and share keyboard controller I/O routines in atkbdc, (just as syscons and psm have been doing until now via routines in kbdio). Again, alpha should be able to do it better, right? msmith>> device sc0 at isa? tty flags zzz msmith> msmith>What's this? Don't you mean: msmith> msmith>pseudo-device syscons 1 I thought I will do it like this first. But, again, I realized that we currently don't have a means to pass flags to the pseudo-device. The current syscons code still uses some flags bits. Once we eliminate these, we can declare it as a pseudo device. Kazu To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199812240344.MAA22380>