Automating build of stable kernel release candidates
by Chuck Ebbert
There are some manual hacks in the kernel specfile now for building
stable release candidates. However, kernel 2.6.23.9-rc1 ends up being
released as 2.6.23.8-NN because we don't have a good way to change the
version. This update automates that; the only question is whether we
should label the built kernel as an -rc. I didn't add that part, and the
changes are untested:
--- kernel.spec 29 Nov 2007 00:25:06 -0000 1.279
+++ kernel.spec 29 Nov 2007 22:45:41 -0000
@@ -34,6 +34,15 @@
%if 0%{?released_kernel}
# Do we have a 2.6.21.y update to apply?
%define stable_update 9
+# Do we have a stable RC update to apply?
+%define stable_rc 0
+# Stable rc patches are incremental against the previous -stable
+# If this is an rc we need the previous stable patch too
+%if 0%{?stable_rc}
+%define stable_base %(expr %{stable_update} - 1)
+%else
+%define stable_base %{stable_update}
+%endif
# Set rpm version accordingly
%if 0%{?stable_update}
%define stablerev .%{stable_update}
@@ -534,8 +543,12 @@
# Here should be only the patches up to the upstream canonical Linus tree.
# For a stable release kernel
-%if 0%{?stable_update}
-Patch00: patch-2.6.%{base_sublevel}.%{stable_update}.bz2
+%if 0%{?stable_base}
+Patch00: patch-2.6.%{base_sublevel}.%{stable_base}.bz2
+%endif
+%if 0%{?stable_rc}
+Patch01: patch-2.6.%{base_sublevel}.%{stable_update}-rc%{stable_rc}.bz2
+%endif
# non-released_kernel case
# These are automagically defined by the rcrev and gitrev values set up
@@ -1006,8 +1019,12 @@
# Update to latest upstream.
# released_kernel with stable_update available case
-%if 0%{?stable_update}
-ApplyPatch patch-2.6.%{base_sublevel}.%{stable_update}.bz2
+%if 0%{?stable_base}
+ApplyPatch patch-2.6.%{base_sublevel}.%{stable_base}.bz2
+%endif
+%if 0%{?stable_rc}
+ApplyPatch patch-2.6.%{base_sublevel}.%{stable_update}-rc%{stable_rc}.bz2
+%endif
# non-released_kernel case
%else
16 years, 3 months
i686 build on x86_64 fails
by Pete Zaitcev
Just FYI, I found that it's not possible to build an i686 kernel on x86_64
(in 2.6.23.1-42.fc8), because this happens:
+ gcc -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i686 -mtune=generic -fasynchronous-unwind-tables -o scripts/modsign/mod-extract scripts/modsign/mod-extract.c
In file included from /usr/include/features.h:359,
from /usr/include/stdio.h:28,
from scripts/modsign/mod-extract.c:12:
/usr/include/gnu/stubs.h:7:27: error: gnu/stubs-32.h: No such file or directory
It's not important why stubs-32.h does not exist in glibc-headers.
The problem is using -m32 for a host executable, because the kernel.spec
has this:
gcc $RPM_OPT_FLAGS -o scripts/modsign/mod-extract scripts/modsign/mod-extract.c
I suspect that hardcoding gcc -O would work just dandily, but on the other
hand it's unclear if we want to bother fixing it. Thoughts?
-- Pete
16 years, 3 months
Kernel seems to crash in a xen environment and using crossover-office
by Daniel Rindt
Hello Listmembers,
iam running Fedora 8 into a Xen Environment. My Kernel on the domU is
2.6.21-2950.fc8xen. This domU has 1024M of Ram assigned. When iam
starting the Crossover Installer Wizard and choose Install Internet
Explorer 6 SP1 then the Kernel seems to crash. With the virt-manager i
had taken a Screenshot from the console, he is attached. If iam lowering
the Memory to 512M the Crash comes later in the Setup or using.
Additionally i found these messages too:
=== 8< ===
4gb seg fixup, process exe (pid 2457), cs:ip 73:0804d5a1
4gb seg fixup, process exe (pid 2457), cs:ip 73:0804d5a1
4gb seg fixup, process exe (pid 2457), cs:ip 73:00aea736
4gb seg fixup, process exe (pid 2457), cs:ip 73:0804d596
4gb seg fixup, process exe (pid 2457), cs:ip 73:00aebcb0
4gb seg fixup, process exe (pid 2457), cs:ip 73:0804d5a1
4gb seg fixup, process exe (pid 2457), cs:ip 73:00aea736
4gb seg fixup, process exe (pid 2457), cs:ip 73:0804d5a1
4gb seg fixup, process exe (pid 2457), cs:ip 73:00c153e7
4gb seg fixup, process exe (pid 2457), cs:ip 73:0804d596
printk: 171 messages suppressed.
=== 8< ===
If iam using some Windows Software with Crossover then i found after 2-4
minutes the following in my dmesg:
=== 8< ===
4gb seg fixup, process wine-pthread (pid 2857), cs:ip 73:741dfb5a
printk: 164944 messages suppressed.
=== 8< ===
the problem can definitly influenced through the memory dimensioning. I
dont know why the machine knows that he had 4G of memory at all.
In hope for answer and suggestions how to avoid/ solve that - many
thanks in advance. Thanks for reading this mail.
-Daniel
P.S.: i had do a xm dump-core, its bzipped now so i can provide it if
someone request it for analyzing.
16 years, 4 months
ALSA versions in the Fedora kernel
by Chuck Ebbert
Here's a proposal for what version of ALSA we deliver with our kernel:
Fedora 7 will stay with the version it has, 1.0.14, which is being
maintained with stable patches by the ALSA team.
Fedora 8 will stay with 1.0.15 and only merge carefully selected bug fixes
from upstream ALSA.
Fedora 9 / devel will track upstream ALSA development so we can shake out
bugs as soon as possible. This can be done by carrying the git-alsa-mm
patch from the -mm kernel.
16 years, 4 months
Install hang on a Dell PowerEdge SC1430
by Steve Dickson
Are there known issues with F8 (and F7) being
installed on PowerEdges? Booting from DVD (and CDs)
I get the following three lines than nothing...
Loading Vmlinuz........
Loading initrd.img.....
Read.
This happens with both x86 and x86_64 installs
from DVD of either F8 or F7. I also tried burning
the boot.iso from the DVD to a CD and boot from
that with the same results.
Being this is pretty broken, I'm hoping this is a known
problem and there is some type of workaround.
tia,
steved.
16 years, 4 months
vanilla/ changes in devel.
by Dave Jones
I just checked something into devel/ that changes how we 'make prep'.
Before, under kernel-2.6.23/ we had a vanilla dir and a fedora-patched
dir called linux-2.6.23.noarch
The vanilla dir used to be just the unpacked tarball.
With the change I just checked in, that dir is now patched up
to the latest upstream (ie, 2.6.24rc2-git5 right now).
This speeds up subsequent make preps quite a bit as the -rc's increase
in size. The downside (and reason for this heads up) is that anyone
with an existing checkout will find that make prep will now fail
to apply patches as the specfile expects vanilla in the new form.
rm -rf kernel-2.6.23 and make prep again, and it'll all just work out.
Dave
--
http://www.codemonkey.org.uk
16 years, 4 months
How to change the number of processors in a 16-core machine
by Feng Xian
Hi,
I am doing a scalable test on a 16-core machine (8 dual-core amd
processors). So I need to vary the processors from 1 to 16. But I
couldnt remove the cores from motherboard. Is there any way to achieve
this by changing the max number of processors in the kernel? Thanks!
16 years, 4 months
How can I develop loadable module for Fedora 7 and later?
by Cary Fu
When I tried to build a loadable module for Fedora 7, there are lots of error warning came up regarding the line of #include <linux/module.h>. Is there anyone knows where can I find the information about making a loadable module for Fedora 7. Thanks in advance for any input.
Cary Fu
16 years, 4 months