On Fri, May 17, 2002 at 11:46:55AM -0600, Michael Lee (TOR) wrote: > > Hi, I'm completely stumped. > > I recently upgraded my Debian system from a 2.2 kernel to 2.4.18, and since > then it has slowed to a near-halt (on a Celeron 733). Where it's most > notable is if I run make. > > For example, "make menuconfig" takes about two full minutes to bring up the > dialog boxes. And then, running "make dep" takes a solid TWO HOURS after > saving my kernel changes. And then (yeah, it gets even better) running > "make-kpkg --revision=custom.1.0 kernel_image" takes a whopping FOUR HOURS > to finish! > > I could be approaching this whole thing all wrong, but I checked the version > of make, and it is: > > shiner:~# make -v > GNU Make version 3.79.1, by Richard Stallman and Roland McGrath. > Built for i386-pc-linux-gnu > > On another Debian machine which runs JUST FINE on a 2.4.18 kernel (a Celeron > 500), the version of make is: > > [EMAIL PROTECTED]:~$ make -v > GNU Make version 3.79.1, by Richard Stallman and Roland McGrath. > Built for i586-pc-linux-gnu > > Is it possible that make is slow on the Celeron 733 because it's the wrong > architecture (i386, instead of i586)? And if so, how on earth do I upgrade > it? I've tried apt-get upgrade and apt-get install (my apt sources are set > to the unstable debian source), but it just keeps telling me I already have > the updated versions. I fear there may be a bigger problem here though. > HELP! (And thanks to anyone who can provide it!).
Try to collect some more information. For example, do hdparm tests on your drive and compare them with what you experienced before. Leave top running in another xterm while you compile to see if resource allocation is what it should be. Try sending your kernel .config file to the list and let us see if there are any things turned off which could affect performance. Most likely, there is something trivial wrong with the new kernel, like not having DMA bugfixes for a particular chipset. It's also possible the hdparm settings for your drive were changed, but that seems less likely. -- NIck -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
