F18/rawhide: lockdep consumes a lot of percpu data
by Prarit Bhargava
When booting F18 on a 32 cpu system I repeatedly see errors like:
PERCPU: allocation failed, size=320 align=32, alloc from reserved chunk failed
Pid: 620, comm: udevd Not tainted 3.7.0-0.rc5.git1.2.fc19.x86_64 #1
Call Trace:
[<ffffffff81186c7d>] pcpu_alloc+0x9dd/0xa40
[<ffffffff810e8b5a>] ? mod_verify_sig+0x39a/0x460
[<ffffffff810e1f2e>] ? find_sec+0x5e/0x90
[<ffffffff81186f83>] __alloc_reserved_percpu+0x13/0x20
[<ffffffff810e5b5f>] load_module+0x40f/0x2290
[<ffffffff812e68b2>] ? cred_has_capability+0x62/0x100
[<ffffffff8109a333>] ? up_write+0x23/0x40
[<ffffffff810e7a32>] sys_init_module+0x52/0x230
[<ffffffff816fb799>] system_call_fastpath+0x16/0x1b
kvm: Could not allocate 320 bytes percpu data
This occurs because the lockdep code is allocating a giant chunk of memory in
the percpu data:
On 3.6.6-1:
14940 (D) __per_cpu_end
On 3.7.0-0:
1d5900 (D) __per_cpu_end
and in 3.7.0-0, a dump of the symbol map shows
efc0 (V) tstats_lookup_lock
f020 (V) cpu_lock_stats
1cef40 (V) stop_cpus_work
1cef80 (V) cpu_stopper
ie) the cpu_lock_stats takes up a huge chunk of the per_cpu memory.
I'm not sure what can be done about this; I suppose we could expand the reserved
area to allow for more module percpu allocation ...
P.
11 years, 5 months
Announcing the rawhide kernel nodebug repository
by Justin Forbes
It has been discussed in the past that we should have a repository of
the rawhide kernels with debug turned off to encourage more users to run
the latest upstream snapshots. That repository now exists. You can
enable it by dropping
http://alt.fedoraproject.org/pub/alt/rawhide-kernel-nodebug/fedora-rawhid... into /etc/yum.repos.d and doing a yum update.
This will contain the (almost) daily rawhide updates built with debug
turned off.
Bugs against this kernel should be filed in bugzilla against the rawhide
kernel. Any questions or comments about the repository itself should be
sent to the kernel(a)lists.fedoraproject.org list.
Thanks,
Justin
11 years, 5 months
Microsoft Hyper-V network support in Fedora
by Jóhann B. Guðmundsson
Greetings
I've been digging a bit into running Fedora as a an guest in the most
common virtulzation solution out there most notably running as a guest
in hyperv/kvm/vbox/vmware/xen, which was spurred by a discussion in the
QA community and crawling through the internet I found 1 bug [1] that
affects all of them and one issue limit to hyper-v as in networking not
working out of the box [2].
Digging to the root cause of this leads me to this when users use the
legacy tulip network driver as is mentioned there in that blog post
"The real problem here is that the device does not support carrier
detection via ethtool or mii-tool. If the driver supported carrier
detection, NetworkManager would bring it up automatically."
( With an obvious workaround if that's the case simply disable NM
configure and use initscripts network service instead )
The mentioned issue has been reported upstream here [3]
So my question are these.
1. What's the current status on an fix for that tulip kernel driver bug?
2. What ever happened to the native hyper-v drivers that Microsoft
pushed upstream?
3. If those drivers have not been rejected and simply are stuck in
staging? is there something preventing us pulling into Fedora and
enabling them by default to improve the situation?
JBG
1.https://bugzilla.redhat.com/show_bug.cgi?id=810040
2.http://blog.robertseder.com/2012/08/29/win8hyper-v-what-works-and-what-...
3. https://bugzilla.kernel.org/show_bug.cgi?id=11092
11 years, 5 months
Rough-Patch for F18 kernel.spec: apply modsign patches even for vanilla builds
by Thorsten Leemhuis
Hi!
Find below a rough, illustrative patch that fixes the following error when you try to build a vanilla kernel with kernel.spec from F18 git:
> + echo BUILDING A KERNEL FOR PAE i686...
> BUILDING A KERNEL FOR PAE i686...
> + perl -p -i -e 's/^SUBLEVEL.*/SUBLEVEL = 6/' Makefile
> + perl -p -i -e 's/^EXTRAVERSION.*/EXTRAVERSION = -3.vanilla.stable.knurd.1.fc18.i686.PAE/' Makefile
> + make -s mrproper
> + cp configs/kernel-3.6.6-i686-PAE.config .config
> + cp /builddir/build/SOURCES/x509.genkey .
> + chmod +x scripts/sign-file
> chmod: cannot access 'scripts/sign-file': No such file or directory
> RPM build errors:
> error: Bad exit status from /var/tmp/rpm-tmp.06kdJq (%build)
> Bad exit status from /var/tmp/rpm-tmp.06kdJq (%build)
I didn't check, but I assume the change "Update modsign to what is currently in 3.7-rc2" was the point where the problem was introduced.
HTH & CU
knurd
diff --git a/kernel.spec b/kernel.spec
index d374cf4..e0c3484 100644
--- a/kernel.spec
+++ b/kernel.spec
@@ -648,6 +648,9 @@ Patch05: linux-2.6-makefile-after_link.patch
Patch06: power-x86-destdir.patch
+Patch900: modsign-upstream-3.7.patch
+Patch901: modsign-post-KS-jwb.patch
+
%if !%{nopatches}
@@ -685,8 +688,8 @@ Patch700: linux-2.6-e1000-ich9-montevina.patch
Patch800: linux-2.6-crash-driver.patch
# crypto/
-Patch900: modsign-upstream-3.7.patch
-Patch901: modsign-post-KS-jwb.patch
+#Patch900: modsign-upstream-3.7.patch
+#Patch901: modsign-post-KS-jwb.patch
# secure boot
Patch1000: secure-boot-20121105.patch
@@ -1347,6 +1350,9 @@ ApplyOptionalPatch linux-2.6-compile-fixes.patch
ApplyPatch power-x86-destdir.patch
+ApplyPatch modsign-upstream-3.7.patch
+ApplyPatch modsign-post-KS-jwb.patch
+
%if !%{nopatches}
# revert patches from upstream that conflict or that we get via other means
@@ -1443,8 +1449,8 @@ ApplyPatch linux-2.6-crash-driver.patch
ApplyPatch linux-2.6-e1000-ich9-montevina.patch
# crypto/
-ApplyPatch modsign-upstream-3.7.patch
-ApplyPatch modsign-post-KS-jwb.patch
+#ApplyPatch modsign-upstream-3.7.patch
+#ApplyPatch modsign-post-KS-jwb.patch
# secure boot
ApplyPatch efivarfs-3.6.patch
@@ -2382,6 +2388,9 @@ fi
# ||----w |
# || ||
%changelog
+* Tue Nov 06 2012 Thorsten Leemhuis <fedora(a)leemhuis.info>
+- apply modsign patches even for vanilla builds
+
* Mon Nov 05 2012 Josh Boyer <jwboyer(a)redhat.com> - 3.6.6-3
- Backport efivarfs from efi/next for moktools
- Fix build break without CONFIG_EFI set (reported by Peter W. Bowey)
11 years, 5 months
Fedora Kernel Meeting Minutes for Nov 2, 2012
by Josh Boyer
==============================
#fedora-meeting: Fedora Kernel
==============================
Meeting started by jwb at 18:00:49 UTC. The full logs are available at
http://meetbot.fedoraproject.org/fedora-meeting/2012-11-02/fedora-kernel....
.
Meeting summary
---------------
* init (jwb, 18:01:15)
* F16 (jwb, 18:03:40)
* F16 3.6.5 in updates-testing covers 3.6.2-3.6.5. lots of fixes.
please test and add karma (jwb, 18:05:06)
* F17 (jwb, 18:06:36)
* F17 3.6.5 in updates-testing. please test and give karma (jwb,
18:08:25)
* a number of 'blank screen/hung' bugs reported. no clear cause yet.
git bisect is welcome (jwb, 18:12:00)
* F18 (jwb, 18:15:21)
* 3.6.5-2.fc18 is the current kernel. one more kernel update coming
before beta (jwb, 18:21:50)
* F18 GA kernel level is still expected to be 3.6.x. 3.7 will be a
rebase update (jwb, 18:22:46)
* rawhide (jwb, 18:28:17)
* rawhide is at 3.7-rc3.gitX. will continue through the 3.7 RC
releases (jwb, 18:29:51)
* IDEA: ramp up the 'rawhide-nodebug kernel repo' (jwb, 18:30:48)
* rawhide needs love. much love. (jwb, 18:40:24)
* open floor (jwb, 18:40:42)
Meeting ended at 18:46:31 UTC.
Action Items
------------
Action Items, by person
-----------------------
* **UNASSIGNED**
* (none)
People Present (lines said)
---------------------------
* jwb (83)
* davej (25)
* jforbes (23)
* nirik (19)
* brunowolff (10)
* zodbot (3)
* pknirsch (1)
* jsmith (1)
Generated by `MeetBot`_ 0.1.4
.. _`MeetBot`: http://wiki.debian.org/MeetBot
11 years, 5 months
Second RFC: Repos with vanilla kernel packages
by Thorsten Leemhuis
Hi!
More than nine months ago I did some initial steps to start a repository
with Linux vanilla kernels packaged for Fedora and asked for feedback
here (see
http://thread.gmane.org/gmane.linux.redhat.fedora.kernel/3583/ ). I had
to suspend that effort for a while due to unforeseen real life issues,
but found some time now to do all the work to maintain such a repo it
without to much hassle (IOW: I wrote script to automate much of the work).
Everything I planed for the start is at its place now and I'll likely
announce that repo in some more public channels in a week or two. I
mention this plan here first to give everybody subscribed to this list a
chance to look at the setup first, in case someone here thinks I should
do things differently.
The wiki has usage instructions and a lot of boring details, including
some some warnings why ordinary people should not use those repositories:
https://fedoraproject.org/wiki/Kernel_Vanilla_Repositories
The git repo can be found at:
http://fedorapeople.org/cgit/thl/public_git/kernel.git/
The yum repos for F17 and F18:
http://repos.fedorapeople.org/repos/thl/kernel-vanilla-mainline/
http://repos.fedorapeople.org/repos/thl/kernel-vanilla-stable/
A build log for one of the kernels:
http://kojipkgs.fedoraproject.org//work/tasks/5546/4645546/build.log
A patch that shows the changes to kernel.spec file can be found below.
Cu
knurd
P.S.: Flying to Barcelona for a few vacation days and LinuxCon Europe
early tomorrow, so I might be slower than usual to reply to mails
> diff --git a/kernel.spec b/kernel.spec
> index 92dee5c..cd24dde 100644
> --- a/kernel.spec
> +++ b/kernel.spec
> @@ -32,6 +32,8 @@ Summary: The Linux kernel
> # (Uncomment the '#' and both spaces below to set the buildid.)
> #
> # % define buildid .local
> +%define packager Thorsten Leemhuis https://fedoraproject.org/wiki/Kernel_Vanilla_Repositories
> +%define buildid .vanilla.mainline.knurd.1
> ###################################################################
>
> # The buildid can also be specified on the rpmbuild command line
> @@ -117,7 +119,7 @@ Summary: The Linux kernel
> # kernel-PAE (only valid for i686)
> %define with_pae %{?_without_pae: 0} %{?!_without_pae: 1}
> # kernel-debug
> -%define with_debug %{?_without_debug: 0} %{?!_without_debug: 1}
> +%define with_debug %{?_with_debug: 1} %{?!_with_debug: 0}
> # kernel-doc
> %define with_doc %{?_without_doc: 0} %{?!_without_doc: 1}
> # kernel-headers
> @@ -127,7 +129,7 @@ Summary: The Linux kernel
> # tools
> %define with_tools %{?_without_tools: 0} %{?!_without_tools: 1}
> # kernel-debuginfo
> -%define with_debuginfo %{?_without_debuginfo: 0} %{?!_without_debuginfo: 1}
> +%define with_debuginfo %{?_with_debuginfo: 1} %{?!_with_debuginfo: 0}
> # kernel-bootwrapper (for creating zImages from kernel + initrd)
> %define with_bootwrapper %{?_without_bootwrapper: 0} %{?!_without_bootwrapper: 1}
> # Want to build a the vsdo directories installed
> @@ -158,7 +160,7 @@ Summary: The Linux kernel
> %define with_sparse %{?_with_sparse: 1} %{?!_with_sparse: 0}
> #
> # build a release kernel on rawhide
> -%define with_release %{?_with_release: 1} %{?!_with_release: 0}
> +%define with_release %{?_without_release: 0} %{?!_without_release: 1}
>
> # Set debugbuildsenabled to 1 for production (build separate debug kernels)
> # and 0 for rawhide (all kernels are debug kernels).
> @@ -166,7 +168,7 @@ Summary: The Linux kernel
> %define debugbuildsenabled 0
>
> # Want to build a vanilla kernel build without any non-upstream patches?
> -%define with_vanilla %{?_with_vanilla: 1} %{?!_with_vanilla: 0}
> +%define with_vanilla %{?_without_vanilla: 0} %{?!_without_vanilla: 1}
>
> # Build the kernel-doc package, but don't fail the build if it botches.
> # Here "true" means "continue" and "false" means "fail the build".
> @@ -176,7 +178,7 @@ Summary: The Linux kernel
> %define doc_build_fail true
> %endif
>
> -%define rawhide_skip_docs 1
> +%define rawhide_skip_docs 0
> %if 0%{?rawhide_skip_docs}
> %define with_doc 0
> %define doc_build_fail true
> @@ -228,7 +230,7 @@ Summary: The Linux kernel
>
> %if %{nopatches}
> %define with_bootwrapper 0
> -%define variant -vanilla
> +#define variant -vanilla
> %else
> %define variant_fedora -fedora
> %endif
> @@ -2378,6 +2380,14 @@ fi
> * Sun Oct 21 2012 Justin M. Forbes <jforbes(a)redhat.com> - 3.7.0-0.rc1.git3.2
> - Disable debugging options.
>
> +* Sat Oct 20 2012 Thorsten Leemhuis <fedora(a)leemhuis.info> - 3.7.0-0.rc1.git3.1.vanilla.mainline.knurd.1
> +- add buildid ".vanilla.mainline.knurd.1"
> +- switch defines to build
> + - a vanilla kernel without debuging
> + - no debug kernel and no debuginfo packages, to big to handle for me right now
> +- disable the variant macro as that adds "-vanilla" to %%name, which confuses
> + fedpkg and prevents installonly handling of packages in yum
> +
> * Fri Oct 19 2012 Justin M. Forbes <jforbes(a)redhat.com> - 3.7.0-0.rc1.git3.1
> - Linux v3.7-rc1-154-gc9623de
>
11 years, 5 months