On Fri, April 15, 2011 3:26 pm, Allen Weiner wrote: > On Fri, 2011-04-15 at 02:33 -0400, Chris Knadle wrote: >> On Thursday, April 14, 2011 20:20:21 Allen Weiner wrote: >> > On Thu, 2011-03-24 at 21:00 -0400, Chris Knadle wrote: >> > > On Thursday, March 24, 2011 13:44:31 Allen Weiner wrote: >> > > > >> > > > >> > > >> > > > >> > F14 yesterday shipped a kernel update. With this thread in mind, I >> > noticed that when I get a kernel update, I also get a kernel-devel. I >> > also recently noticed while browsing through /usr, that although I >> never >> > explicitly downloaded the latest kernel source, my /usr had about 34 >> MB >> > of source for the latest kernel. (I did explicitly download the Fedora >> > SRPM for kernel 2.6.34. When uncompressed it is about 360 MB). So >> maybe >> > the DKMS scheme I'm using downloads just enough of the kernel source >> > needed to compile the nvidia driver. >> >> If you look closer, you're probably going to find that it's the majority >> of >> the code for the Linux kernel, just without the documentation. i.e. at >> least >> code-wise it's probably "the entire kernel"; at least that's what I've >> seen >> ready to get installed when my system tries to install dkms-based >> packages. >> > Oops! At your suggestion, I took a closer look. The directory tree I'm > looking at is /usr/src/kernels/(latest kernel). I was deceived into > thinking it contained source code because the overall directory > structure appears to match that of the kernel SRPM that I downloaded. I > drilled down a random set of half a dozen directories and there was no > source code at all. It is just a collection of KConfig files and > makefiles. So I guess this comes from the kernel-devel package. (Google > search on kernel-devel said it contains makefiles).
Interesting. Dkms packages on Debian depend on a version of the Linux kernel source code being installed; but I guess Fedora doesn't have that same dependency. > Here is a (somewhat related) side note on my other thread "Linux > configuring 133 MB/s HDD for 33 MB/s". As discussed in that thread I > bypassed the problem by adding a kernel parameter libata.force to force > detection of an 80 wire cable. The HDD is now being configured for 133 > MB/s. I did a before (33 MB/s HDD) and after (133 MB/s HDD) compile of > the 2.6.34 kernel gotten from the Fedora kernel SRPM. The elapsed times > were identical (45 minutes)! I was expecting a big improvement. (I used > the config file shipped with F14. I did a "make clean" in between > compiles to ensure an identical test). This doesn't surprise me; the individual reads/writes to the drive in doing a kernel compile are small compared to the CPU load. -- Chris -- Chris Knadle [email protected] _______________________________________________ Mid-Hudson Valley Linux Users Group http://mhvlug.org http://mhvlug.org/cgi-bin/mailman/listinfo/mhvlug Upcoming Meetings (6pm - 8pm) MHVLS Auditorium May 4 - Inkscape Jun 1 - Zimbra Jul 6 - Jul 2011
