Messages in this thread | | | Date | Thu, 16 Apr 1998 13:06:17 -0400 (EDT) | From | Greg Zornetzer <> | Subject | Re: another (!) new kmod.c |
| |
On Thu, 16 Apr 1998, Mikael Pettersson wrote:
> Another new kmod? Do we really need that.. > Well, this one (sales pitch mode on): > * Is reentrant. > * Is implemented using a very small extension of the stock 2.1.9x > kmod: instead of having the kmod thread itself spawn modprobe > and wait for it to finish (which is why it isn't reentrant), > kmod spawns a "loader" thread, which in turns carries out all > the necessary steps. In the mean time, kmod may pick up new work. > * Doesn't bring any new problems, like having to fiddle with > work-queues or having to close() file descriptors. > * As of yesterday, it uses the CLONE_FS trick to allow it to work > properly even if you boot via initrd. (As does the stock 2.1.96) > * Has been running without a single hitch on my UP PC since Sunday. > (Testers on SMP machines are most welcome.) > * Is quite elegant, IMHO of course :-) > > Patch against vanilla 2.1.96 follows below. > (If you've patched that bogus character on line 116, > the diff will probably fail.) > > Cheers, > > /Mikael Hmm, this bears a lot of resemblance to the patch that I was working on, however, Adam Richter redid the kmod code in IMO a better way that does not require the persistent kernel thread - and is quite small. He sent it to Linus a day ago. I'm surprised that you didn't see it.
-Greg Zornetzer gaz@andrew.cmu.edu
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |