Date: Thu, 21 Jul 2011 12:21:47 -0400 (EDT) From: Daniel Feenberg <feenberg@nber.org> To: Chad Perrin <perrin@apotheon.com> Cc: freebsd-questions@freebsd.org Subject: Re: 2020: Will BSD and Linux be relevant anymore? Message-ID: <alpine.LFD.2.00.1107211218520.3727@nber7.nber.org> In-Reply-To: <20110721141534.GC59455@guilt.hydra> References: <alpine.BSF.2.00.1107190736560.27391@gwdu60.gwdg.de> <CAGy-%2Bi_phtNbTh7SHhockqTuGrv%2Bd2ZLn0_6A9aKEySYO0MgTw@mail.gmail.com> <24466_1311199850_4E27526A_24466_7987_1_D9B37353831173459FDAA836D3B43499C521866E@WADPMBXV0.waddell.com> <4e28160e.bVryeJCK1esNt615%perryh@pluto.rain.com> <CADGWnjXjtZwXPdFQJ8dFfZ_bG8hSp6LiJt4QvsSbS7umYG=EqQ@mail.gmail.com> <20110721141534.GC59455@guilt.hydra>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 21 Jul 2011, Chad Perrin wrote: > On Thu, Jul 21, 2011 at 10:52:28AM +0200, C. P. Ghost wrote: >> >> I'm not familiar with Windows, but I don't think a typical windows >> driver as written by a hardware vendor would manipulate the windows >> kernel internals (data structures) directly, right? If that's correct, >> we "merely" need to catch the ABI up- and down-calls from and to the >> windows driver, and translate them into regular FreeBSD syscalls (maybe >> augmented by a compat helper library?). >> >> Since this is exactly the approach taken by the Linuxulator, I fail to >> see why a similar method hasn't been tried for those windows kernel >> driver (binary blobs). Maybe some artificial restrictions like, say, >> patents are standing in the way? Or a technical restriction like such >> binary blobs being encrypted with a public key, and only usable from >> Windows kernel with their own secret key? > > It may not be anything so exotic. On a per-release basis, the MS Windows > ABIs and APIs change far more dramatically than the Linux kernel, and are > far less transparent to developers; they must in many cases be discovered > by experimentation, being closed source software. Over a given period of > time, the changes to Linux may be greater in number and magnitude (I'm > not a kernel hacker, so I wouldn't know for sure), but they're spread out > over time rather than bundled in a major collection of changes with a new > marketing campaign. This might make it much more difficult to target the > MS Windows ABIs and APIs. > > I'm just speculating, though. As I said, I'm not a kernel hacker. Doesn't the NDIS specification offer a reasonably stable ABI for wireless drivers? I have often thought that supporting NDIS would offer manufacturers a sort of "halfway house" to ease them into proper support for FreeBSD and Linux. While it is inferior to open source drivers, it would attract users, and with users manufacturers would feel pressure to have better support, which would best be achieved with open-source drivers. Daniel Feenberg > > -- > Chad Perrin [ original content licensed OWL: http://owl.apotheon.org ] >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?alpine.LFD.2.00.1107211218520.3727>