On Tue, 2008-08-26 at 13:10 -0700, MHR wrote: > On Tue, Aug 26, 2008 at 1:03 PM, Akemi Yagi <amyagi at gmail.com> wrote: > > > > Well... if you go for the fuse/dkms route, you will lose the chance to > > build the module upon kernel update. :-) > > > > True - decisions, decisions,.... Well you can always do what I did a few years back. "In days of yore, long ago" I had a LAN using some coax and ethernet cards leftover from my early Lantastic installation. Cards were a mixed bag Including some IBM ones that had both RJ-45 and BNC ports. At the time the drivers (for Crystal chips) had no support for the BNC ports and didn't allow use of all the chipset features. I jumped in to the source and I WAS AGHAST! It was the crappiest, least structured, most obfuscated mess I had ever seen. Falling back to my former life, I hollered "REWRITE, YES" (accompanied by rapid pumping of right fist upon slightly bent knee! :-)) So I went to the web site, downloaded the docs, learned how it was supposed to work and started re-work. I didn't do the whole job, but annotated and restructured the parts I had to touch. I then maintained it of many version of the kernel and driver. Since I was doing LFS at the time, it sometimes got interesting. But it did keep my in practice building kernels, modules and making patches. Maybe that's a possibility for you? A nice side-effect is being able to help occasionally when some poses a question about what you been working on. For me, a lot of satisfaction came when I was able to help the LFS list users when they had networking problems - "peripheral learning effect". > > mhr > <snip> -- Bill