Give good ol' method some skas lovin in gentoo-sources or mjc-sources or something http://prdownloads.sourceforge.net/user-mode-linux/host-skas3.patch skas is shared kernel address space, roughly doubles the performance of UML's running under it
It's a real shame that gentoo-sources doesn't include this patch (it doesn't patch cleanly today, I'm not sure why).
Patching gentoo-sources 2.4.19-gentoo-r10 results in: patch -p1 < ~/download/uml_patches/host-skas3.patch patching file arch/i386/config.in Hunk #1 succeeded at 380 (offset 89 lines). patching file arch/i386/kernel/ldt.c Hunk #3 succeeded at 166 (offset 16 lines). patching file arch/i386/kernel/process.c Hunk #1 succeeded at 548 (offset -3 lines). patching file arch/i386/kernel/ptrace.c Hunk #1 succeeded at 154 (offset 7 lines). Hunk #2 succeeded at 423 (offset 3 lines). patching file arch/i386/kernel/sys_i386.c Hunk #1 succeeded at 41 (offset 1 line). Hunk #2 FAILED at 56. Hunk #3 succeeded at 85 (offset 16 lines). Hunk #4 succeeded at 101 (offset 1 line). 1 out of 4 hunks FAILED -- saving rejects to file arch/i386/kernel/sys_i386.c.rej patching file include/asm-i386/processor.h Hunk #1 succeeded at 446 (offset 10 lines). patching file include/asm-i386/ptrace.h patching file include/linux/mm.h Hunk #1 succeeded at 562 (offset 70 lines). Hunk #2 succeeded at 542 with fuzz 2. Hunk #3 FAILED at 555. 1 out of 3 hunks FAILED -- saving rejects to file include/linux/mm.h.rej patching file include/linux/proc_mm.h patching file mm/Makefile Hunk #1 succeeded at 17 with fuzz 1. patching file mm/mmap.c Hunk #1 FAILED at 390. 1 out of 1 hunk FAILED -- saving rejects to file mm/mmap.c.rej patching file mm/mprotect.c Hunk #1 succeeded at 344 with fuzz 2 (offset 80 lines). Hunk #2 FAILED at 362. Hunk #3 FAILED at 413. 2 out of 3 hunks FAILED -- saving rejects to file mm/mprotect.c.rej patching file mm/proc_mm.c
Created attachment 9278 [details] Failed patches to skas3 kernel patch See last comment..
I now have a functional skas3 kernel; these failed hunks have been merged by hand into 2.4.20-gentoo-r2. Now, will somebody tell me the correct method for creating a patch & ebuild file that will get accepted by the kernel team? :) Thanks, Eric
attach the patch to this bug and i'll look @ it for inclusion into the next pfeifer-sources. if all looks good, i may move it into the gentoo-sources for the next release. Jay
ok, i haven't lost sight of this, just had other things that were higher priority. Jay
Created attachment 18028 [details] 2.4.20-gentoo-r5/999_skas Woohoo! Functional skas patch for gentoo-sources-2.4.20-r7. name this file '999_skas', and stick it in patches-2.4.20-gentoo-r5.tar.bz2. the standard host-skas3.patch file doesn't work with the gentoo kernel; applied a one line patch from <http://www.mathematik.uni-marburg.de/~odb/usermodelinux-howto.html#k1>. SKAS increases UML speed QUITE a bit. It's a very good think to have, and I hope this gets merged into the next 2.4.2x release. This patch should have no effect if /proc/mm is not built into the kernel.
Patch is in gentoo-sources-2.4.20-r11. Thanks!
