Master Kernel Thread

[masux594]

Well, today is the first of September.. the website is updated .. Still no updates from KC.. Hmmm ..

Sysc, A

Hi, some problems when installing the kernel with guide [step 11, troubles with grub], but 2.6.31 run as a rabbit!

Sysc, A

[oakgrove]

Hi, I'm having some trouble patching the 2.6.30.6 vanilla kernel with the bfs 209 patch. This is a kernel downloaded directly from kernel.org. I haven't modified it in any way other than to unpack it before attempting to apply this patch. The out put I keep getting is:

debian:/usr/src/linux-2.6.30.6# patch -p1 -i ../2.6.30-sched-bfs-209.patch 
patching file Documentation/sysctl/kernel.txt
patching file fs/pipe.c
patching file include/linux/init_task.h
patching file include/linux/sched.h
Hunk #9 succeeded at 1595 (offset 1 line).
patching file kernel/sched.c
Reversed (or previously applied) patch detected!  Assume -R? [n] 
Apply anyway? [n] y
Hunk #1 FAILED at 1.
File kernel/sched.c is not empty after patch, as expected
1 out of 1 hunk FAILED -- saving rejects to file kernel/sched.c.rej
patching file kernel/sysctl.c
patching file kernel/workqueue.c
patching file kernel/sched_fair.c
patching file kernel/sched_idletask.c
patching file kernel/sched_rt.c
Reversed (or previously applied) patch detected!  Assume -R? [n] 
Apply anyway? [n] y
Hunk #1 FAILED at 1.
File kernel/sched_rt.c is not empty after patch, as expected
1 out of 1 hunk FAILED -- saving rejects to file kernel/sched_rt.c.rej
patching file kernel/sched_bfs.c
patching file kernel/Makefile
patching file kernel/kthread.c
patching file kernel/posix-cpu-timers.c
patching file kernel/exit.c
patching file kernel/fork.c
Hunk #1 succeeded at 1191 (offset -13 lines).
patching file mm/oom_kill.c
patching file init/Kconfig
patching file kernel/delayacct.c
patching file kernel/trace/trace.c
patching file fs/proc/base.c
Hunk #1 succeeded at 366 (offset -3 lines).
patching file kernel/sched_debug.c
patching file include/linux/ioprio.h

Any help or reassurances (especially from someone running 30 with this patch) on this would be greatly appreciated as I have patched the 31 kernel with no problems at all but this one just keeps get those Hunk failures.

[stevenpusser]

Try compiling anyway...I got those errors, but the Makefile is changed to not build from those two scheduler source files anyway, so it doesn't matter what's in them.

Hi, I'm having some trouble patching the 2.6.30.6 vanilla kernel with the bfs 209 patch. This is a kernel downloaded directly from kernel.org. I haven't modified it in any way other than to unpack it before attempting to apply this patch. The out put I keep getting is:

debian:/usr/src/linux-2.6.30.6# patch -p1 -i ../2.6.30-sched-bfs-209.patch 
patching file Documentation/sysctl/kernel.txt
patching file fs/pipe.c
patching file include/linux/init_task.h
patching file include/linux/sched.h
Hunk #9 succeeded at 1595 (offset 1 line).
patching file kernel/sched.c
Reversed (or previously applied) patch detected!  Assume -R? [n] 
Apply anyway? [n] y
Hunk #1 FAILED at 1.
File kernel/sched.c is not empty after patch, as expected
1 out of 1 hunk FAILED -- saving rejects to file kernel/sched.c.rej
patching file kernel/sysctl.c
patching file kernel/workqueue.c
patching file kernel/sched_fair.c
patching file kernel/sched_idletask.c
patching file kernel/sched_rt.c
Reversed (or previously applied) patch detected!  Assume -R? [n] 
Apply anyway? [n] y
Hunk #1 FAILED at 1.
File kernel/sched_rt.c is not empty after patch, as expected
1 out of 1 hunk FAILED -- saving rejects to file kernel/sched_rt.c.rej
patching file kernel/sched_bfs.c
patching file kernel/Makefile
patching file kernel/kthread.c
patching file kernel/posix-cpu-timers.c
patching file kernel/exit.c
patching file kernel/fork.c
Hunk #1 succeeded at 1191 (offset -13 lines).
patching file mm/oom_kill.c
patching file init/Kconfig
patching file kernel/delayacct.c
patching file kernel/trace/trace.c
patching file fs/proc/base.c
Hunk #1 succeeded at 366 (offset -3 lines).
patching file kernel/sched_debug.c
patching file include/linux/ioprio.h

Any help or reassurances (especially from someone running 30 with this patch) on this would be greatly appreciated as I have patched the 31 kernel with no problems at all but this one just keeps get those Hunk failures.

[oakgrove]

Hi Stephen, thanks for the heads up. I went on ahead and compiled after reading your post and am using the new scheduler right now. Have to use it some more to see if it really makes much of a difference but, so far, so good.

[fanglinyong]

it's really a nice article!

[Dakiraun]

Excellent advice, though when I tried this out, I got the following at the end of the compile attempt (tried twice to be sure):

dpkg-deb: building package `linux-image-2.6.31-its.m65a.nb.01' in `../linux-image-2.6.31-its.m65a.nb.01_2.6.31-its.m65a.nb.01-10.00.Custom_amd64.deb'.
dpkg-deb: control directory has bad permissions 2755 (must be >=0755 and <=0775)
make[2]: *** [debian/stamp/binary/linux-image-2.6.31-its.m65a.nb.01] Error 2
make[2]: Leaving directory `/usr/src/linux-2.6.31'
make[1]: *** [debian/stamp/binary/pre-linux-image-2.6.31-its.m65a.nb.01] Error 2
make[1]: Leaving directory `/usr/src/linux-2.6.31'
make: *** [kernel_image] Error 2

There are no .deb files left behind either.

EDIT: Solved this myself. I noticed the permissions on the /usr/src directory had the s attribute set, so just ran:

sudo chmod -R a-s /usr/src

Trying a new kernel build again.

[TSJason]

Thanks. Can you tell me how to install it. Am on 2.6.31/JauntyAMD64. Thanks

Colo,

If you look at the first page of this thread, with the instructions, step number 7 shows how to apply patches. You would just use the same method there as with the bfs and autoiso patch (but instead of using 'bzcat', just use 'cat' since the bfs files are not compressed).

During the "make menuconfig" you'll want to use CK's recommendations; specifically:

THESE ARE OPTIONAL FOR LOWEST LATENCY. YOU DO NOT NEED THESE!
Configure your kernel with 1000Hz, preempt ON and disable dynamic ticks.

[go_beep_yourself]

1. Can I still apply a patch to my wireless driver (I compiled as a module) after I have built the kernel?
2. Is it safe to disable most options in menuconfig that says Intel If I have an amd processor?
3. Does anybody really know what all of these options mean? I usually leave them as default if I can't figure it out.
4. How can I do the equivalent of make -j 4 for a multicore processor when using make-kpg?
5. It would be nice to see some explanation of specific options that most of us can disable if we meet the criteria. I want an efficient kernel for performance.

[Dakiraun]

Well, I built the kernel and installed it with no issues. I removed the nvidia-common prior to the install, and put it back afterwards. There is one bad end result though - the nvidia 180.44 driver will not function.

dkms fails to build/find a module for it at start up. So, my question is - how does one fix this? I know I can grab drivers directly from nvidia, but I'd prefer to stick with the packaged ones.

[stevenpusser]

1. Can I still apply a patch to my wireless driver (I compiled as a module) after I have built the kernel?
2. Is it safe to disable most options in menuconfig that says Intel If I have an amd processor?
3. Does anybody really know what all of these options mean? I usually leave them as default if I can't figure it out.
4. How can I do the equivalent of make -j 4 for a multicore processor when using make-kpg?
5. It would be nice to see some explanation of specific options that most of us can disable if we meet the criteria. I want an efficient kernel for performance.

1. If you have the wireless driver source, go ahead and patch, build, and install it over the kernel version. Be aware that the drivers are in a constant state of flux, and new or renamed drivers may complicate your update.

2. You should be able to disable Intel hardware drivers safely.

3. If you aren't sure what the option means after reading the hints, just leave it.

4. For four-core builds, execute

export CONCURRENCY_LEVEL=4

just before the kernel build command, though I have seen some advice to make it (cores + 1)...i.e. 5.

Try 8 if the cores are hyperthreaded.