Virtio-scsi
by Justin Forbes
It appears that the virtio-scsi changes didn't make the 3.3 merge window,
but have been acked by penberg upstream and should make 3.4. It has been
requested that we carry the patches through the 3.3 cycle for F17, and can
drop them when 3.4 is shipped. Patches are fairly isolated and appear
sane.
Justin
12 years, 2 months
UsrMove and installing kernels on older releases
by Josh Boyer
Hi Kay and Harald,
With UsrMove now landed in rawhide, I'm assuming that you'll be wanting
to commit the patch[1] that makes kernel.spec install modules into
/usr/lib instead of /lib. According to my understanding, this is mostly
a "put things in the new proper location" more than it is a requirement,
because the symlinks should still allow a fully functioning kernel to
work (and they do in my local testing).
One of the things we do quite often during debug is ask users to install
newer kernels on older releases. This lets us easily figure out if an
issue is fixed upstream, etc. However, if the patch to install modules
to /usr/lib goes in, I wonder if that will prevent people from easily
being able to use such kernels on e.g. f15/f16 which obviously do not
have the UsrMove feature. Admittedly I haven't looked into it much
myself yet so perhaps I've missed something obvious that would allow
this to still work.
If there is an issue with that scenario, I'd like to skip the patch
until F17 becomes the oldest supported release. With things obviously
still working as-is today, I don't see much issue in doing so.
josh
[1]
http://harald.fedorapeople.org/downloads/usrmove/kernel-usrmove.patch
12 years, 2 months
Fedora Kernel Meeting 02-03-2012 Minutes
by Josh Boyer
======================================
#fedora-meeting: Fedora Kernel meeting
======================================
Meeting started by jwb at 18:00:34 UTC. The full logs are available at
http://meetbot.fedoraproject.org/fedora-meeting/2012-02-03/fedora-kernel....
.
Meeting summary
---------------
* Release overview (jwb, 18:01:39)
* F15/F16 will be updated to 3.2.3 today (jwb, 18:03:38)
* ACTION: davej to push 3.2.3 updates (jwb, 18:03:50)
* Rawhide/F17 at 3.3-rc2-gitx (jwb, 18:10:01)
* ACTION: jwb to talk to Kay and Harald about usrmove and installing
on older releases (jwb, 18:19:41)
* questions (jwb, 18:21:51)
* ACTION: jwb to pester skvidal about bug 770444 for
kernel-modules-extra (jwb, 18:27:09)
* jforbes will enthrall us with a dictation on kernel-related autoqa
testing next time. (jwb, 18:30:10)
* ACTION: davej will bundle up thoughs on what tests to throw into the
autotest framework (for real this time!) (jwb, 18:30:54)
Meeting ended at 18:31:33 UTC.
Action Items
------------
* davej to push 3.2.3 updates
* jwb to talk to Kay and Harald about usrmove and installing on older
releases
* jwb to pester skvidal about bug 770444 for kernel-modules-extra
* davej will bundle up thoughs on what tests to throw into the autotest
framework (for real this time!)
Action Items, by person
-----------------------
* davej
* davej to push 3.2.3 updates
* davej will bundle up thoughs on what tests to throw into the
autotest framework (for real this time!)
* jwb
* jwb to talk to Kay and Harald about usrmove and installing on older
releases
* jwb to pester skvidal about bug 770444 for kernel-modules-extra
* **UNASSIGNED**
* (none)
People Present (lines said)
---------------------------
* jwb (73)
* davej (39)
* tibbs (10)
* brunowolff (6)
* zodbot (3)
* tomspur (3)
* nirik (1)
* Southern_Gentlem (1)
* drago01 (1)
* jistone (1)
Generated by `MeetBot`_ 0.1.4
.. _`MeetBot`: http://wiki.debian.org/MeetBot
12 years, 2 months
Using merge_config.sh
by Josh Boyer
Hi All,
We've been using merge.pl to merge Fedora's config file fragments for as
long as I can remember. However, in the interest of using what's
upstream, I took a look at the newly added merge_config.sh script and
what it would take to use that instead.
The good:
It actually worked fairly well. The resulting merged config files were
identical to the ones merge.pl creates.
It also provides a lot more output about the options present, what is
overriding what, and final config verification (if we used it to do the
oldconfig step). At some point we could use it to auto-disable new
config options too, but that's a big maybe.
The ok:
The changes to kernel.spec are pretty small. I basically used the same
approach we do today by creating Makefile.merge and calling the script
from within that. Doing similar creation steps in the spec itself was
pretty ugly. The largest change, aside from using a different merge
tool, is that the fragments are merged after all patches are applied. I
don't think there is a reason why they're merged before hand now, but it
is a small difference.
The ugly:
It's slow. All that checking and reporting and verification takes quite
a bit of processing. How much? Well, compare:
Current:
[jwboyer@vader kernel]$ time fedpkg -v prep --arch=noarch >
foo.log
real 0m46.898s
user 0m9.248s
sys 0m35.764s
[jwboyer@vader kernel]$
merge_config.sh (yes, for real):
[jwboyer@vader kernel]$ time fedpkg -v prep --arch=noarch >
foo.log
real 25m16.705s
user 0m58.587s
sys 19m57.125s
[jwboyer@vader kernel]$
merge_config.sh + patch:
[jwboyer@vader kernel]$ time fedpkg -v prep --arch=noarch >
foo.log
real 3m12.711s
user 0m15.655s
sys 2m47.259s
[jwboyer@vader kernel]$
So, ignoring the stock merge_config.sh case which is clearly not usable,
it would make the %prep section roughly 4x longer. The patch I added to
make it at least somewhat reasonable has been sent upstream.
I've attached a log file from a run with merge_config.sh for people to
glance over. Before I do anything else, do people think the output is
useful and this is perhaps beneficial?
josh
Spec diff:
diff --git a/kernel.spec b/kernel.spec
index 858c483..a21eb0a 100644
--- a/kernel.spec
+++ b/kernel.spec
@@ -574,6 +574,7 @@ Source15: merge.pl
Source16: mod-extra.list
Source19: Makefile.release
+Source18: Makefile.merge
Source20: Makefile.config
Source21: config-debug
Source22: config-nodebug
@@ -743,6 +744,8 @@ Patch21091: kmemleak.patch
Patch21092: udlfb-remove-sysfs-framebuffer-device-with-USB-disconnect.patch
+Patch21093: faster-merge.patch
+
# compat-wireless patches
Patch50000: compat-wireless-config-fixups.patch
Patch50001: compat-wireless-pr_fmt-warning-avoidance.patch
@@ -1265,28 +1268,6 @@ make -f %{SOURCE19} config-release
%endif
%endif
-# Dynamically generate kernel .config files from config-* files
-make -f %{SOURCE20} VERSION=%{version} configs
-
-%if %{with_backports}
-# Turn-off bits provided by compat-wireless
-for i in %{all_arch_configs}
-do
- mv $i $i.tmp
- ./merge.pl %{SOURCE200} $i.tmp > $i
- rm $i.tmp
-done
-%endif
-
-# Merge in any user-provided local config option changes
-%if %{?all_arch_configs:1}%{!?all_arch_configs:0}
-for i in %{all_arch_configs}
-do
- mv $i $i.tmp
- ./merge.pl %{SOURCE1000} $i.tmp > $i
- rm $i.tmp
-done
-%endif
ApplyPatch linux-2.6-makefile-after_link.patch
@@ -1439,6 +1420,8 @@ ApplyPatch udlfb-remove-sysfs-framebuffer-device-with-USB-disconnect.patch
#rhbz 783211
ApplyPatch fs-Inval-cache-for-parent-block-device-if-fsync-called-on-part.patch
+ApplyPatch faster-merge.patch
+
# END OF PATCH APPLICATIONS
%endif
@@ -1466,6 +1449,45 @@ done
rm -f kernel-%{version}-*debug.config
%endif
+# Ok, now let's try to do this with merge_config.sh
+chmod +x scripts/kconfig/merge_config.sh
+
+# Dynamically generate kernel .config files from config-* files
+make -f %{SOURCE18} VERSION=%{version} configs
+
+%if %{with_backports}
+# Turn-off bits provided by compat-wireless
+for i in %{all_arch_configs}
+do
+ mv $i $i.tmp
+ scripts/kconfig/merge_config.sh -m -n $i.tmp %{SOURCE200}
+ mv .config $i
+ rm $i.tmp
+done
+%endif
+
+# Merge in any user-provided local config option changes
+%if %{?all_arch_configs:1}%{!?all_arch_configs:0}
+for i in %{all_arch_configs}
+do
+ mv $i $i.tmp
+ scripts/kconfig/merge_config.sh -m -n $i.tmp %{SOURCE1000}
+ mv .config $i
+ rm $i.tmp
+done
+%endif
+
+# Remove configs not for the buildarch
+for cfg in kernel-%{version}-*.config; do
+ if [ `echo %{all_arch_configs} | grep -c $cfg` -eq 0 ]; then
+ rm -f $cfg
+ fi
+done
+
+%if !%{debugbuildsenabled}
+rm -f kernel-%{version}-*debug.config
+%endif
+
# now run oldconfig over all the config files
for i in *.config
do
12 years, 2 months
Fedora Kernel Meeting 02-03-2012
by Josh Boyer
Hi All,
Just a reminder that we'll be holding a Fedora kernel meeting in
#fedora-meeting tomorrow at 18:00 UTC. We'll be chatting about
the current state of the Fedora kernel, where we're heading in F17
and what we're going to be looking at otherwise. Feel free to stop
by and ask questions or make suggestions.
If you want to yell at us, that's fine too but a patch to fix a bug is
the required fee in that case.
josh
12 years, 2 months
