Date: Wed, 21 Oct 2009 20:40:16 +0100 From: Rui Paulo <rpaulo@freebsd.org> To: Bernhard Schmidt <bschmidt@techwires.net> Cc: freebsd-net@freebsd.org, David Horn <dhorn2000@gmail.com> Subject: Re: Intel WiFi 5100/5300 Message-ID: <A3F61DF2-4587-4950-BA87-1A7156796DCC@freebsd.org> In-Reply-To: <200910211012.15474.bschmidt@techwires.net> References: <20091009170839.142800@gmx.net> <200910210833.44121.bschmidt@techwires.net> <25ff90d60910210029t5f8f67d0nd17b537ecaacdee9@mail.gmail.com> <200910211012.15474.bschmidt@techwires.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On 21 Oct 2009, at 09:12, Bernhard Schmidt wrote: > On Wednesday 21 October 2009 09:29:13 David Horn wrote: >>>> The only new issue I have found so far is that I must manually load >>>> iwnfw.ko before loading if_iwn.ko (the module depend used to work >>>> on >>>> the in-tree driver) >>> >>> Hmm.. that is probably related to the rename of the firmware image, >>> iwnfw-5000 instead of iwnfw. Is MODULE_DEPEND(iwn, iwnfw, 1, 1, >>> 1); an >>> option there? >> >> MODULE_DEPEND(iwn, iwnfw_fw, 1, 1, 1) >> >> added to if_iwn.c fixes it nicely (note: iwnfw_fw not just iwnfw). >> It >> turns out the original driver loaded the iwnfw.ko module as part of >> firmware_get() since the firmware module name matched the first >> firmware image name (see firmware.h comments). Looking at the other >> drivers, the other option is to break up the firmware images into >> unique kernel modules (e.g. ral or iwi), and allow firmware_get() to >> do the load. I would think that this would reduce kernel memory >> usage >> as well (several individual firmware modules vs all firmware images >> in >> one module). Just a thought. > > Any "offical" opinions on that one? Should we break iwnfw up into > individual > modules? I believe so. Thanks for your work. I hope this can be in HEAD soon. Regards, -- Rui Paulo
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?A3F61DF2-4587-4950-BA87-1A7156796DCC>