* Michael Halcrow [Thu, 23 Sep 2004 at 18:12 -0500] > On Thu, Sep 23, 2004 at 05:39:51PM -0600, Von Fugal wrote: > > * Michael Halcrow [Thu, 23 Sep 2004 at 07:37 -0500] > > > On Wed, Sep 22, 2004 at 10:10:27PM -0700, stuporglue wrote: > > > > > Unloading modules is, at best, a dubious practice, and you should > > > > > avoid doing so in general. The kernel developers have even > > > > > entertained the idea of completely removing the ability to rmmod, > > > > > because it unnecessarily complicates the kernel. > > > > > > > > That may be the case, but if I couldn't I'd have to reboot after each > > > > suspend. On my iBook, if I want scan/monitor mode, I have to compile > > > > the airport modules separately, and have them inserted at boot (or > > > > when I need them). When I put it to sleep, and then use it again, I > > > > won't be able to use wireless till I remove the module, insert it > > > > again, and restart the service. Just restarting the service won't do > > > > it either. > > > > > > That sounds like a bug in the driver, and that should be reported to > > > LKML so that it is fixed. You should never be constrained to unload a > > > module. > > > > Nevertheless, sometimes it's what you have to do. I have a similar > > problem using ndiswrapper. Sometimes my driver just wigs out, and I > > have to rmmod and re- modprobe the driver to get it working again. > > I'd hate to see that ability dissapear. > > The developers could really care less about proprietary, binary-only > drivers, and for good reason. They are frequently the source of > instability, they break whenever the kernel internal ABI changes > (which is all the time), and the developers can't do a thing about it > without the source code. > > For a poignant illustration of this fact, refer to the recent pwc > issue. I support Greg K-H's actions attitude wholeheartedly, and I > have *no* sympathy for people who buy hardware that requires > proprietary, closed-source drivers. Tough luck; you should have > patronized a company that ``gets it.'' Do not expect the kernel > developers to bend over backwards to help companies who keep their > specs secret at the detriment of their customers.
I agree with your arguments there. I'm all against closed specs and propriety drivers that break. My point was merely that S**t Happens(TM) and I wan't the ability to remove broken stuff from my kernel without a reboot. Now I don't think I would ask for it if it wasn't already there. But since it is, I would submit the plea to keep it there, if reasonable. (If it isn't broke, don't fix it.) If it's truly A Real Pain to keep it in there and the benifits of removing it make it worth the effort, then I'm not one to argue against that. OTOH, I'd hate to see it removed and then later on down the road decide that, you know what, that was pretty useful, but it's gonna be a lot of work to put back in something that we put a lot of work into taking out... Food for Thought. Von Fugal
signature.asc
Description: Digital signature
____________________ BYU Unix Users Group http://uug.byu.edu/ ___________________________________________________________________ List Info: http://uug.byu.edu/cgi-bin/mailman/listinfo/uug-list
