Package Details: linux-pf 4.6.1-1

Git Clone URL: https://aur.archlinux.org/linux-pf.git (read-only)
Package Base: linux-pf
Description: Linux kernel and modules with the pf-kernel patch [-ck patchset (BFS included), TuxOnIce, BFQ] and aufs3
Upstream URL: http://pf.natalenko.name/
Licenses: GPL2
Groups: base
Conflicts: kernel26-pf, linux-pf
Provides: aufs3, linux-pf=4.6
Replaces: aufs3, kernel26-pf
Submitter: nous
Maintainer: Thaodan
Last Packager: Thaodan
Votes: 172
Popularity: 3.416264
First Submitted: 2011-07-24 12:01
Last Updated: 2016-06-10 21:08

Dependencies (18)

Sources (9)

Latest Comments

kogone commented on 2016-06-10 23:42

is tuxonice not included anymore?

tuankiet65 commented on 2016-06-05 15:55

Hashes for config and config.x86_64 haven't been updated I think.

anaveragehuman commented on 2016-04-24 18:48

Hi, I'm noticing a few problems:

1. modprobed_db is now modprobed-db, so that needs to be changed in the PKGBUILD.
2. When I do "L" to `make localmodconfig`, it seems to pause, but there's a question there that I don't see until I hit enter. But by then it assumes I wanted the default, which is 'n'.
3. Both menuconfig and nconfig seem to be broken because entering the keystrokes seem to enter the keystroke into the terminal rather than it being intercepted by menuconfig or ncurses. I am using xterm, but this also occurs on TTYs.

ruggfrancesco commented on 2016-04-06 10:30

as @aucubin notes, in order to work, change the sha256sums to:

sha256sums=('401d7c8fef594999a460d10c72c5a94e9c2e1022f16795ec51746b0d165418b2'
'20a7dcee51530bdcda4c20a5a99de1527ed639c276315cf02d18cb372ff41f37'
'17a63957522907f2b268c3d49798f2dc5c25822e8cc7d1116d8bad4357f88029'
'82d660caa11db0cd34fd550a049d7296b4a9dcd28f2a50c81418066d6e598864'
'1256b241cd477b265a3c2d64bdc19ffe3c9bbcee82ea3994c590c2c76e767d99'
'b09352ee835724951aaf6a83b74ddb2964261e82ce6c800edf4fa9e8bdb11bc9'
'SKIP'
'5313df7cb5b4d005422bd4cd0dae956b2dadba8f3db904275aaf99ac53894375')

aucubin commented on 2016-03-22 13:52

seems like you forgot to change the sha256sums of config and config.x86_64 because I needed to change them manually in the PKGBUILD to get past the checksums

lectus commented on 2015-10-21 08:18

@dlh: Touch the void, Luke! Just add fourth parameter "void" in AuStub definition and all will be fine.
File "fs/aufs/loop.h", row 44: AuStub(struct file *, loop_backing_file, return NULL, void).
ps: "Touching the Void"- book/movie about one "nearly fatal climb" of the mountain Siula Grande in the Peruvian Andes in 1985.

dlh commented on 2015-10-14 09:29

Does not compile
fs/aufs/loop.h:44:23: error: function declaration isn’t a prototype [-Werror=strict-prototypes]
AuStub(struct file *, loop_backing_file, return NULL)

post-factum commented on 2015-09-11 08:34

BTW, BFS issue has been fixed, and you may try latest -pf for 4.1 branch.

Thaodan commented on 2015-08-06 02:43

sorry for the delay about linux-pf-4.1 see here what the issue:
https://bbs.archlinux.org/viewtopic.php?pid=1551004#p1551004

Det commented on 2015-06-08 05:36

Wouldn't it make more sense to version it as something like '4.0_pf5-1' or '4.0.pf5-1' for clarity with aur4.archlinux.org?

Thaodan commented on 2015-05-21 12:07

linux-pf: upkg, synched with ARCH configs

plasma commented on 2015-05-18 18:35

i stumble upon this when trying to compile:

CC kernel/sched/bfs.o
kernel/sched/bfs.c:528:20: error: function declaration isn’t a prototype [-Werror=strict-prototypes]
static inline void grq_priodl_lock()
^
kernel/sched/bfs.c:533:20: error: function declaration isn’t a prototype [-Werror=strict-prototypes]
static inline void grq_priodl_unlock()
^
cc1: some warnings being treated as errors
scripts/Makefile.build:258: recipe for target 'kernel/sched/bfs.o' failed

Thaodan commented on 2015-05-14 15:54

upkg, synched configs with arch-linux-pf except for tomoyo which is enabled but makes no performance downsides

quequotion commented on 2015-05-14 03:08

build keeps failing in package_linux-pf() where it makes a symlink to extramodules..... why?

jaro3 commented on 2015-05-03 05:01

These are the performance suspects taken from the diff between 3.19.3-3-ARCH and 3.19.4-pf:

*********** PF specific ************
-CONFIG_CGROUP_SCHED=y
-CONFIG_FAIR_GROUP_SCHED=y
-CONFIG_CFS_BANDWIDTH=y
-CONFIG_RT_GROUP_SCHED=y
+CONFIG_SCHED_BFS=y
-CONFIG_SCHED_AUTOGROUP=y
-CONFIG_IOSCHED_DEADLINE=y
-CONFIG_IOSCHED_CFQ=y
+CONFIG_IOSCHED_DEADLINE=m
+CONFIG_IOSCHED_CFQ=m
-CONFIG_DEFAULT_CFQ=y
+CONFIG_IOSCHED_BFQ=y
+CONFIG_CGROUP_BFQIO=y
+CONFIG_DEFAULT_BFQ=y
-CONFIG_DEFAULT_IOSCHED="cfq"
+CONFIG_DEFAULT_IOSCHED="bfq"

-CONFIG_KERNEL_GZIP=y
+CONFIG_KERNEL_LZ4=y

+CONFIG_UKSM=y

-CONFIG_HZ_300=y
-CONFIG_HZ=300
+CONFIG_HZ_1000=y
+CONFIG_HZ=1000

+CONFIG_TOI_CORE=y
+CONFIG_TOI_FILE=m
+CONFIG_TOI_SWAP=m
+CONFIG_TOI_CRYPTO=m
+CONFIG_TOI_USERUI=m
+CONFIG_TOI_USERUI_DEFAULT_PATH="/usr/sbin/tuxoniceui"
+CONFIG_TOI_DEFAULT_IMAGE_SIZE_LIMIT=-2
+CONFIG_TOI_REPLACE_SWSUSP=y
+CONFIG_TOI_IGNORE_LATE_INITCALL=y
+CONFIG_TOI_DEFAULT_WAIT=25
+CONFIG_TOI_DEFAULT_EXTRA_PAGES_ALLOWANCE=7000
+CONFIG_TOI=y
+CONFIG_TOI_EXPORTS=y
+CONFIG_TOI_ZRAM_SUPPORT=y

+CONFIG_AUFS_FS=m
+CONFIG_AUFS_BRANCH_MAX_127=y
+CONFIG_AUFS_SBILIST=y
+CONFIG_AUFS_HNOTIFY=y
+CONFIG_AUFS_HFSNOTIFY=y
+CONFIG_AUFS_EXPORT=y
+CONFIG_AUFS_INO_T_64=y
+CONFIG_AUFS_RDU=y
+CONFIG_AUFS_SHWH=y
+CONFIG_AUFS_BR_RAMFS=y
+CONFIG_AUFS_BR_FUSE=y
+CONFIG_AUFS_POLL=y
+CONFIG_AUFS_BR_HFSPLUS=y
+CONFIG_AUFS_BDEV_LOOP=y
*************************************************

****** performance hogs? ************************
+CONFIG_AUDIT=y
+CONFIG_AUDITSYSCALL=y
+CONFIG_AUDIT_WATCH=y
+CONFIG_AUDIT_TREE=y

+CONFIG_NETFILTER_XT_TARGET_AUDIT=m

-CONFIG_F2FS_FS_SECURITY=y
-CONFIG_F2FS_CHECK_FS=y

+CONFIG_SECURITY_NETWORK=y

+CONFIG_SECURITY_TOMOYO=y
+CONFIG_SECURITY_TOMOYO_MAX_ACCEPT_ENTRY=2048
+CONFIG_SECURITY_TOMOYO_MAX_AUDIT_LOG=1024
+CONFIG_SECURITY_TOMOYO_POLICY_LOADER="/sbin/tomoyo-init"
+CONFIG_SECURITY_TOMOYO_ACTIVATION_TRIGGER="/usr/lib/systemd/systemd"
+CONFIG_SECURITY_APPARMOR=y
+CONFIG_SECURITY_APPARMOR_BOOTPARAM_VALUE=0
+CONFIG_SECURITY_APPARMOR_HASH=y

+CONFIG_INTEGRITY_AUDIT=y
************************************************

******** other *********************************
-CONFIG_TICK_CPU_ACCOUNTING=y
+CONFIG_IRQ_TIME_ACCOUNTING=y
-CONFIG_NUMA_BALANCING=y
-CONFIG_NUMA_BALANCING_DEFAULT_ENABLED=y
-CONFIG_CGROUP_CPUACCT=y

-CONFIG_AIX_PARTITION=y

+CONFIG_X86_INTEL_USERCOPY=y
+CONFIG_X86_USE_PPRO_CHECKSUM=y
+CONFIG_SMT_NICE=y
-CONFIG_NUMA=y
-CONFIG_AMD_NUMA=y
-CONFIG_X86_64_ACPI_NUMA=y
-CONFIG_NODES_SPAN_OTHER_NODES=y
-CONFIG_NODES_SHIFT=6
-CONFIG_NEED_MULTIPLE_NODES=y
-CONFIG_USE_PERCPU_NUMA_NODE_ID=y
-CONFIG_ACPI_NUMA=y

+CONFIG_RTC_DRV_RX4581=m

-CONFIG_SXGBE_ETH=m
-CONFIG_MFD_RTSX_USB=m
-CONFIG_MFD_RN5T618=m
-CONFIG_MFD_SI476X_CORE=m

+CONFIG_EFI_VARS=m
+CONFIG_EFI_VARS_PSTORE=m

-CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE=0x0
+CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE=0x1

-CONFIG_CRYPTO_SHA1=m
+CONFIG_CRYPTO_SHA1=y

-CONFIG_CRYPTO_DEV_CCP=y
-CONFIG_CRYPTO_DEV_CCP_DD=m
-CONFIG_CRYPTO_DEV_CCP_CRYPTO=m

nous commented on 2015-04-28 10:12

@Thaodan: I used to copy over the ARCH config once every major kernel release and just enable the extra pf-patchset features, i.e. BFS, Hz freq --> 1000, BFQ and TuxOnIce. It's less than a minute's effort and helps to keep the config in order. See also README.maintainers at https://bitbucket.org/nous/linux-pf/src. Cheers!

AnAkkk commented on 2015-04-27 22:51

Well, I won't post the full list of config settings that aren't the same as there are many (can be seen by doing a diff), for example there are the CONFIG_SECURITY_* ones which may lower the performance.

Thaodan commented on 2015-04-27 20:59

It's on my list, but please if you have something to complain give more infos, more detail

AnAkkk commented on 2015-04-27 20:53

Could you please fix the configs to match the official archlinux ones for the non -pf settings, there are still many things that are not in sync. Most of them were modified or removed at some point, and that wasn't changed in -pf.

Thaodan commented on 2015-04-27 11:42

yes this is true, but you could update this too.

gborzi commented on 2015-04-27 11:39

Modules tuxonice_compress, tuxonice_swap and tuxonice_userui are no longer provided by the kernel, so the wiki pages about tuxonice and linux-pf need to be updated, i.e.update the line about MODULES in mkinitcpio.conf.

Thaodan commented on 2015-04-25 16:38

upkg, added script to build all pkgs in a row to the pkg repo

Thaodan commented on 2015-04-15 14:03

upkg, added option to enable batch_mode by src/batch_opts, set CPU to your favorite march and build for it

gborzi commented on 2015-03-28 21:34

Please add WESTMERE to the _CPUSUFFIXES_KBUILD array.

Thaodan commented on 2015-03-26 14:29

sorry I uploaded the wrong source package first, if you watch my git mirror you see that I already updated the pkgbuild. when its out of date please just mark it and dont post ver xx is out, this is redundant and annoing, if you have something to add adout the new ver do it.

vincent commented on 2015-03-26 14:21

v3.19-pf3 is out.

Thaodan commented on 2015-03-26 13:07

upkg, added silvermont support

post-factum commented on 2015-03-01 11:50

v3.19-pf2 is out with updated GCC patch.

https://pf.natalenko.name/forum/index.php?topic=290.0

pavbaranov commented on 2015-02-26 10:46

I don't know why gcc doesn't complains when try to build with wrong flag, but the simpliest solution for Intel's CPU user is add info - something like that: If you are Intel's CPU user and want to optimize your linux-pf kernel, choose MNATIVE flag.
With this flag, kernel will be properly build for sure.

Thaodan commented on 2015-02-26 09:13

But why gcc don't complains when the March flag is wrong?


I'll add the new patch as i already said but at the end of the week.

pavbaranov commented on 2015-02-25 22:06

MNATIVE option is ok and with it we can build proper, optimized kernel for Intel's processors. With "name" of it we cannot, becouse it's for gcc =<4.9.
post-factum - it's not your's work. Of course it's yours decision use new or old graysky's patch, but maintainer linux-pf for Arch should explain us anything that is important or make patch-on-patch :)

post-factum commented on 2015-02-25 20:42

The funny thing is that I'm Arch user as well, but build my kernel with MNATIVE option as I guess it's up to compiler to decide what optimizations to use.

Anyway, if people's lament persists, I may update graysky's patch and break Intel-specific optimizations for older GCC.

I followed versioning voting inspired by Linus, and I guess I should do the same. So, vote please: https://plus.google.com/105187899411277204747/posts/HNaTndmBCyr

pavbaranov commented on 2015-02-25 14:45

No, certainly not. What Natalenko support in his patch it's his choice. But we are in Arch Linux. In Arch Linux we have gcc 4.9+. You are mainteiner of the most important part of system - kernel and offers it with not proper patch. See: https://github.com/graysky2/kernel_gcc_patch/blob/master/outdated_versions/linux-3.15%2B/gcc-4.2%2B/enable_additional_cpu_optimizations_for_gcc_kernel_v3.15%2B.patch for REQUIREMENTS. Since gcc 4.9 names of flags has been changed. It's impossible to build optimized kernel (any binary apps) with wrong compiler's flags. Then, maybe it's possible to build kernel with wrong flags but this kernel might be unstable or doing some things in a wrong way.

Thaodan commented on 2015-02-25 11:57

it if he wan'ts to support gcc 4.9+ it is his problem.
It does build for intel cpus with gcc optimizations just fine.

pavbaranov commented on 2015-02-25 07:59

It's not a problem of Natalenko. His patch works well i.e. in Chakra, KaOS or Debian. Everywhere where GCC=<4,9 is. But not on Arch with GCC 4.9. This is a problem only with PKGBUILD not with patch. Currently the kernel builds fine with GCC optimizations only for AMD processors. For Intel's processors we can build kernel only without GCC optimizations.

Thaodan commented on 2015-02-25 01:07

Please ping Natalenko for this , to update the patch.
I'll add the patch and fix the pkgbuild when I have time.
Currently the kernel builds just fine using the old patch for haswell, sandybridge etc.

pavbaranov commented on 2015-02-25 00:56

Natalenko's patch uses graysky's GCC patch for GCC prior then 4.9. It means, that it use old -march CFF terminology. Just take a look at: https://wiki.archlinux.org/index.php/Repo-ck#Selecting_the_correct_CPU_optimized_package or https://github.com/graysky2/kernel_gcc_patch for new ones and inspect patch-3.19-pf1 for -march (GCC patch started from 2058 lines of Natalenko's patch). For example, for gcc 4.9 and Intel Nehalem platform proper -march flag is nehalem, but in pf patch it's old, but proper for gcc prior then 4.9, corei7. Wrong CFF flags for gcc 4.9 are for IvyBridge, SandyBridge, Haswell... etc.
When PKGBUILD for your's linux-pf is started, we can choose special optimizations for specific processor. But if somebody has got i.e. SandyBridge, Natalenko's patch try to build kernel with -march=corei7-avx. It's unknown flag for GCC 4.9 and kernel will be incorectly built. In extremally causes it will result in kernel panic.
Then it's impossible to build corectly kernel for (i.e.) Sandy Bridge with GCC optimizations in Arch with this PKGBUILD. I think, that your PKGBUILD should use, for example, sed command for correct this.

Thaodan commented on 2015-02-24 23:04

Why please explain? I you have something to add/complain please explain it in detail I can't do anything whithout valid infos.

pavbaranov commented on 2015-02-24 22:56

Natalenko's pf-patch uses old graysky's GCC patch (for GCC prior then 4.9). In Arch we have GCC 4.9 now, then it's impossible to build proper kernel enabling CPU-specific compile option in the Linux kernel for some Intel processors.
The PKGBUILD should changes pf-patch before it's implemented.

Thaodan commented on 2015-02-24 11:53

nous: thanks, I added a workaround from gravsky for this, without dvb it builds just fine now.
Now I just runned qconf and saved the new config. I think I have missed the changes in ARCH config .

nous commented on 2015-02-24 09:53

@Thaodan: About the security stuff, perhaps you accidentally copied over the .config from 3.18*? In the diff @AnAkkk posted seems so.

nous commented on 2015-02-24 09:44

@Tahu363: Device drivers --> <M> Multimedia support --> [*] Digital TV support. You are right, the related config options have changed, I'll update the wiki.

nous commented on 2015-02-24 09:40

@hzy199411 You can do it yourself in the PKGBUILD (change NUMA_off=yes). Very very few of us run multi-cpu (not multi-core) systems and NUMA creates some overhead. This kernel is desktop-oriented, not server...

hzy199411 commented on 2015-02-24 05:52

would you like enable CONFIG_NUMA ? or it will complain "/sys is not mounted" when I run openmpi programs.

Thaodan commented on 2015-02-23 13:02

I'll change the f2fs stuff the security stuff is enabled before me I ask nous why he has enabled it

AnAkkk commented on 2015-02-23 12:49

Well, there aren't that much differences, but there are a few.
https://www.diffchecker.com/ehe0jdbk

CONFIG_SECURITY_* and CONFIG_AUDIT_* are still enabled though, which can impact performance. Are they needed for anything in -pf? This might be why I get less FPS in -pf than in -ck and the default arch kernel.

There are other differences, e.g. CONFIG_F2FS_FS_SECURITY and CONFIG_F2FS_CHECK_FS are not set in -pf, and they're enabled in the default kernel. I use F2FS on two of my computers so this might be useful :)

Thaodan commented on 2015-02-23 12:24

AnAkkk: I looked at this and the config should me like ARCH's except the pf stuff.
@Tahu363: workaround for this copied from -ck look at the PKGBUILD to how this wotks if you want (you don't need to change something).

Tahu363 commented on 2015-02-23 05:51

Does anyone know how to fix the 3.19.1 compile issue?

the arch wiki says "For people who build their own tailored kernels and compilation aborts with with an error about "missing include/config/dvb/*.h files", setting <M> at DVB for Linux at Device Drivers/Multimedia support and leaving everything else out, creates just the necessary dvb.h, which allows the compilation to continue." but menuconfig shows no such option, I believe this information may pertain to lower version kernels.

AnAkkk commented on 2015-02-22 15:34

I think there's more than just tuxonice, aufs3 and ck, some settings seem to have nothing to do with that. I think they were changed in the official config and weren't changed in linux-pf. For example the CONFIG_SECURITY_* ones are different.

Just do a diff between the linux-pf one and https://projects.archlinux.org/svntogit/packages.git/plain/trunk/config.x86_64?h=packages/linux and you'll see :)

Thaodan commented on 2015-02-22 14:51

Anakk: This is intended as taxon ice, aufs3 and ck patchset is applied.
What else do you found?

Thaodan commented on 2015-02-22 14:49

Sorry I I'll fix this in the evening this day.

Tahu363 commented on 2015-02-22 10:36

The install cannot be completed.
Kernel 3.19.1 wont build.
Cannot stat /include/configs/dvb/*h

Anyone have a fix for this? Package half installs and makes my boot hang on a "pending" operation.

Tahu363 commented on 2015-02-22 10:36

The install cannot be completed.
Kernel 3.19.1 wont build.
Cannot stat /include/configs/dvb/*h

Anyone have a fix for this? Package half installs and maked my boot hang on a "pending" operation.

vincent commented on 2015-02-20 19:08

There is an typo in linux.install file (3.18.1 instead of 3.19.1).

AnAkkk commented on 2015-02-20 16:59

Is it normal config/config.x86_64 don't match the stock kernel config ? Doing a diff between the two show many, many differences.

tamjan commented on 2015-02-18 10:00

Do you have TuxOnIce enabled?

tamjan commented on 2015-02-18 09:19

Well, regardless. For instance, these three lines can't be right:
./kernel/power/snapshot.c:2193: pfn = memory_bm_next_pfn(bm);
./kernel/power/snapshot.c:2198: pfn = memory_bm_next_pfn(bm);
./kernel/power/snapshot.c:2301: tmp = pfn_to_page(memory_bm_next_pfn(safe_highmem_bm));

memory_bm_next_pfn requires two arguments. Something is broken. Do you have TuxOnIce enabled?

Thaodan commented on 2015-02-17 21:00

for me it builds just fine what do you changed?

tamjan commented on 2015-02-17 20:30

I get this, more or less: https://paste.softver.org.mk/raw/1c25e3ee2351715a665bc46de1b63dd8

tamjan commented on 2015-02-17 20:28

This doesn't compile for me. I get
error: too few arguments to function ‘memory_bm_next_pfn’, and complaints of multiple definitions of stuff. The TuxOnIce patch seems broken. Anybody else experiencing this? I run 3.17-pf4, which built fine, and I haven't changed my kernel config more than necessary.

Thaodan commented on 2015-02-15 21:28

update is comming, changes aufs3 branch

Thaodan commented on 2015-01-31 23:13

please don't report it out of date if there's no new ver.

nous commented on 2015-01-18 13:31

Also: the conflicts() arrays are not correctly populated; enclose the appended elements in parentheses, otherwise they are appended as raw strings:
conflicts+=(${pkgbase}-${_cpusuffix}) instead of
conflicts+=${pkgbase}-${_cpusuffix}

hzy199411 commented on 2015-01-18 12:32

kmod should in makedepends

nous commented on 2015-01-17 18:50

32-bit binaries also in place.

nous commented on 2015-01-07 19:10

Since Thaodan seems busy at the moment, I've uploaded a new set of precompiled packages at the [pfkernel] repo, with the appropriate iptables NAT modules in place. 64bit packages only, sorry. Version number 3.17.3-2.

nous commented on 2015-01-06 21:13

Please, fix the missing NAT modules issue ASAP.

kris7t commented on 2015-01-06 11:48

3.17-pf2 and 3.17-pf3 seem to build fine by bumping _pfrel to 2 and 3, respectively, and running makepkg -f >>PKGBUILD

I did not try modprobed_db, just used my own .config from 3.17-pf1.

vincent commented on 2014-12-25 12:56

PKGBUILD for fresh 3.17-pf3: http://pastebin.com/SjrC1dKg
Do not forget to change version in linux.install file.

vincent commented on 2014-12-25 11:03

PKGBUILD for fresh 3.17-pf3: http://pastebin.com/SjrC1dKg

vincent commented on 2014-12-09 16:45

When the package will be updated?

jaro3 commented on 2014-11-17 01:18

NAT kernel modules are disabled in the default config, is this on purpose?

fackamato commented on 2014-11-15 12:21

Can you please add a simple if -e or something so this doesn't break the build?

cp: cannot stat ‘include/config/dvb/*.h’: No such file or directory

AnAkkk commented on 2014-11-15 10:17

Where did I say I am flagging anything out of date?
And I know modprobed_db is not part of this kernel, it's just an option and I am pointing out it doesn't work correctly, can be useful for some other people.

Anonymo commented on 2014-11-15 01:22

@AnAkkk Are you flagging out of date, it's not out of date?! modprobed_db is not a part of this kernel. Everything is working well with the included config.

AnAkkk commented on 2014-11-14 21:37

@TheAifam5: it works for me now with the all inclusive one, but not with modprobed_db unfortunately. It looks like modprobed_db misses some modules (or modprobe doesn't load them unless it's early boot), which causes DM_SNAPSHOT, DM_MIRROR, CONFIG_MICROCODE and others to be disabled, even if they're needed :/

Anonymo commented on 2014-11-14 18:27

@Thaodan you doing a great job bro. Love this kernel.

Thaodan commented on 2014-11-14 00:40

Btw: please stay patient, if a new version is released and I didn't updated the package flag the package out of date but please wait until I update the package and don'r post version x.x is released. Give me some time.

Anonymo commented on 2014-11-13 23:27

This package isn't out of date. Check the website!

TheAifam5 commented on 2014-11-11 21:03

"test -e include/generated/autoconf.h -a -e include/config/auto.conf || ( \
echo >&2; \
echo >&2 " ERROR: Kernel configuration is invalid."; \
echo >&2 " include/generated/autoconf.h or include/config/auto.conf are missing.";\
echo >&2 " Run 'make oldconfig && make prepare' on kernel src to fix it."; \
echo >&2 ; \
/bin/false)"

So... this happpends if i install nvidia-pf , but after this error, compilation work

TheAifam5 commented on 2014-11-11 14:32

So.. and i dont know why not working for you.. I selected the same and after this I click ENTER .. ;)

AnAkkk commented on 2014-11-11 13:08

Nevermind, they seem to enabled in the config indeed. I'm not sure why they got disabled when I selected "all-inclusive kernel like stock -ARCH" though.

TheAifam5 commented on 2014-11-11 13:05

@AnAkkk ... for me is ok.. Kernel compiling with CONFIG_MICROCODE.
CONFIG_MICROCODE=m
CONFIG_MICROCODE_INTEL=y
CONFIG_MICROCODE_AMD=y
CONFIG_MICROCODE_OLD_INTERFACE=y

gim commented on 2014-11-10 04:57

AnAkkk, isn't microcode disabled since 3.17 stock kernel?
https://www.archlinux.org/news/changes-to-intel-microcodeupdates/

AnAkkk commented on 2014-11-10 00:20

It says "Hit <N> (or just <ENTER>) to build an all-inclusive kernel like stock -ARCH" but that doesn't seem to be all-inclusive at all, it looks like this is using the config file included, which is far from arch's default, a lot of things are disabled, like CONFIG_MICROCODE, which are pretty important. Is this normal?

AnAkkk commented on 2014-11-09 10:52

I keep getting a missing dvb directory error when the headers package is created, even after enabling Analog TV.

AnAkkk commented on 2014-11-09 10:19

I can compile it fine on one of my computers, but on the other one I keep getting an error when it creates the headers package.
It fails on this line:
cp include/config/dvb/*.h "${pkgdir}/usr/lib/modules/${_kernver}/build/include/config/dvb/"
I don't have any "dvb" directory in include/config.

AnAkkk commented on 2014-11-08 20:37

Ah ok, thanks, I read it but missed this line :)

Thaodan commented on 2014-11-08 20:34

Read the info on source page, pf usally includes last patches.

AnAkkk commented on 2014-11-08 20:28

Yes, I know, but what I mean is that you aren't using the latest kernel sources? Latest stable is 3.17.2, but the PKGBUILD downloads the 3.17 tar.

Thaodan commented on 2014-11-08 20:26

for example pfs ver is:
3.17-pf1
pkgver is:
3.17.1

Thaodan commented on 2014-11-08 20:26

for example pfs ver is:
3.17-pf1
pkgver is:
3.17.1

AnAkkk commented on 2014-11-08 20:20

Are the releases always based on the first major/minor version of a kernel?
For example, it's 3.16, but 3.16.7 is out since a while.

post-factum commented on 2014-11-03 21:25

v3.17-pf1 is out.

https://pf.natalenko.name/forum/index.php?topic=273.0

gim commented on 2014-10-31 15:09

Anonymo, I did the same thing and it works all right for about a week now.

Anonymo commented on 2014-10-31 15:03

Compiling now by changing changing pfrel line to 4 and sha256sum at the bottom. Seems to be working... at the fs/hfs... right now
http://pastebin.com/wx01HBkg

blinxwang commented on 2014-10-25 21:05

pf-kernel 3.16-pf4 is out.

gim commented on 2014-10-17 09:54

It would be great if you implemented PGP dignature checks both for kernel and patch itself in the PKGBUILD.

X-NicON commented on 2014-09-10 17:29

If default config compile
Error:
scripts/link-vmlinux.sh: line 231: cmp: command not found

Depending add "Diffutils"

hzy199411 commented on 2014-09-09 10:11

Do you know how to solve this problems?I followed the Wiki's Setting,But it didn't work.
$ dmesg |grep Tux
[ 0.364051] TuxOnIce 3.3 (http://tuxonice.net)
[ 0.483464] TuxOnIce: Ignoring late initcall, as requested.
[ 2.201971] TuxOnIce: No storage allocators have been registered. Hibernating will be disabled.
[ 2.208399] TuxOnIce: No storage allocators have been registered. Hibernating will be disabled.
[ 2.208402] TuxOnIce: No storage allocators have been registered.

Thaodan commented on 2014-09-07 22:08

updated, however rebuild is not needed as the fix patch was already added by me in the last ver, added request module

post-factum commented on 2014-09-07 21:40

Flagged as out-of-date due to 3.16-pf2 release :).

hzy199411 commented on 2014-09-07 07:58

would you like to add driver support for my "Qualcomm Atheros QCA8171 Gigabit Ethernet Card", you can change "# CONFIG_ALX is not set" to "CONFIG_ALX=m",thank you.

Thaodan commented on 2014-09-06 13:52

issue is fixed the forum post

Thaodan commented on 2014-09-05 14:18

I get the same error however I can't fix it don't know if its my fault pfs fault see: https://pf.natalenko.name/forum/index.php?topic=268.msg1031;boardseen

felixonmars commented on 2014-09-05 14:12

Hrm, I'm getting an error:

ERROR: "resume_file" [kernel/power/tuxonice_file.ko] undefined!
ERROR: "resume_file" [kernel/power/tuxonice_bio.ko] undefined!
scripts/Makefile.modpost:90: recipe for target '__modpost' failed
make[1]: *** [__modpost] Error 1
Makefile:1055: recipe for target 'modules' failed
make: *** [modules] Error 2
make: *** Waiting for unfinished jobs....

Thaodan commented on 2014-09-05 13:48

run kernel config (follow pkgbuild) and set processor type and features to avx2

Anonymo commented on 2014-09-05 13:15

I see the option is in the PKGBUILD to build optimized kernel, I have a Haswell, does it do it automatically or do I have to edit the PKGBUILD somehow?

Thaodan commented on 2014-09-05 08:06

sorry I couldn't update the until yet as my old laptop broke

RibShark commented on 2014-08-24 13:29

3.16-pf1 is out: https://pf.natalenko.name/forum/index.php?topic=267

sir_lucjan commented on 2014-07-19 17:27

abiehf: I have updated PKGBUILD.

http://www.filedropper.com/linux-pf-3153-1srctar



abihf commented on 2014-07-18 03:17

would you like to update this package to 3.15 ?
i tried it my self, but always got "ERROR: A failure occurred in prepare()"

ivenvd commented on 2014-06-27 07:35

v3.15-pf2 is out: https://pf.natalenko.name/forum/index.php?topic=262

mid-kid commented on 2014-06-11 20:15

Please add kmod to makedepends, otherwise it won't compile in a chroot made with devtools, due to the "depmod" command missing at compile time.

mid-kid commented on 2014-06-11 18:24

Please add kmod to makedepends, otherwise it won't compile in a chroot made with devtools, due to the "depmod" command missing at compile time.

nous commented on 2014-06-06 10:03

>>> Updating module dependencies. Please wait ...
depmod: ERROR: could not open directory /lib/modules/3.14.1-pf: No such file or directory
depmod: FATAL: could not search modules: No such file or directory
>>> Generating initial ramdisk, using mkinitcpio. Please wait...
==> Building image from preset: /etc/mkinitcpio.d/linux-pf.preset: 'default'
-> -k /boot/vmlinuz-linux-pf -c /etc/mkinitcpio.conf -g /boot/initramfs-linux-pf.img
==> Starting build: 3.14.0-pf3-pf

Please fix the errors, especially the -pf3-pf.

% file /boot/vmlinuz-linux-pf
/boot/vmlinuz-linux-pf: Linux kernel x86 boot executable bzImage, version 3.14.0-pf3-pf (nous@hyperion) #1 SMP PREEMPT Thu Jun 5 23:56:49 EEST, RO-rootFS, swap_dev 0x2, Normal VGA

PerfectGentleman commented on 2014-04-29 08:16

out-of-date

Thaodan commented on 2014-04-14 04:16

sorry fixed

tkh23 commented on 2014-04-14 04:09

==> Starting package_linux-pf-headers()...
cp: cannot stat ‘drivers/media/usb/sn9c102/*.h’: No such file or directory

pavbaranov commented on 2014-04-13 10:21

First of all, there isn't any patch at: http://pf.natalenko.name/sources/3.14/patch-3.14linux-pf1.xz - 404 Not Found. It should be http://pf.natalenko.name/sources/3.14/patch-3.14-pf1.xz

nous commented on 2014-04-13 09:47

The package version is 3.14.1-pf, but the kernel itself is versioned 3.14.0-pf (which isn't a big deal as far as installation and operation are concerned). Is this intentional?

Also, KERNEL_VERSION in linux.install is at 3.13.1-pf; you might want to uncomment the relevant sed in the PKGBUILD so as to take care of it automagically, since this seems to be easily forgotten.

Thaodan commented on 2014-04-12 15:27

fixed

felixonmars commented on 2014-04-12 04:01

Reversed (or previously applied) patch detected! Skipping patch.
2 out of 2 hunks ignored -- saving rejects to file arch/x86/include/asm/module.h.rej
patching file arch/x86/Kconfig.cpu
Reversed (or previously applied) patch detected! Skipping patch.
9 out of 9 hunks ignored -- saving rejects to file arch/x86/Kconfig.cpu.rej
patching file arch/x86/Makefile
Reversed (or previously applied) patch detected! Skipping patch.
1 out of 1 hunk ignored -- saving rejects to file arch/x86/Makefile.rej
patching file arch/x86/Makefile_32.cpu
Reversed (or previously applied) patch detected! Skipping patch.
2 out of 2 hunks ignored -- saving rejects to file arch/x86/Makefile_32.cpu.rej

ids1024 commented on 2014-04-12 00:09

The cpu optimizations patch is now included in linux-pf.

mid-kid commented on 2014-04-10 21:02

Linux-pf 3.14-pf1 is available.

bruce commented on 2014-03-31 08:11

There's now a 3.13.2 linux-pf patch, just fyi :-)

Thaodan commented on 2014-03-08 10:48

sorry updated

alaviss commented on 2014-03-08 10:40

The kernel version in the install file is still 3.12.4-pf :P

Please update it for depmod to work

nous commented on 2014-03-07 17:38

Ok, I'll rebuild.

Thaodan commented on 2014-03-07 17:34

nous you used an incorrent PKGBUILD wich i correct a second later, sorry but you need to rebuilt the package. All packages in my repo work fine.

nous commented on 2014-03-07 16:58

-core2

I've also received an email from a user (obviously of the repo binaries):

From `pacman -Si linux-pf` I see that you are the packager of this package and (I assume) the maintainer of the unofficial linux-pf repository. If yes, I would like to report that the latest linux-pf-headers package have an invalid symlink which causes the `dkms autoinstall` to fail. The symlink is '/usr/lib/modules/3.13.1-pf/build' and seems to be broken, at least on my system (it points to '../../../src/linux-3.13.1-pf', which doesn't exists). Is this a problem in my system or a mistake in package creation?

Could you confirm it's a package creation issue?

nous commented on 2014-03-07 16:56

-core2

Thaodan commented on 2014-03-07 16:39

which package-suffix do you use?

nous commented on 2014-03-07 16:35

The "build" symlink in /usr/lib/modules/3.13.1-pf is broken, i.e. it points to /usr/src whilst the sources are in /usr/lib/modules. I use the precompiled binaries from my repo, BTW. Could you look at it?

On a sidenote, WHY THE FUCK the sources are placed in a modules dir and not in /usr/src is beyond me. And I'm ranting about the stock -ARCH kernel which started it...

Thaodan commented on 2014-03-05 16:14

updated

alaviss commented on 2014-03-05 02:31

pf-kernel have reached 3.13.1-pf. Please update the package

Thaodan commented on 2014-02-23 17:41

exactly, ask pf for updates. I just maintain the package and add some critical bugfix patches if needed.

clfarron4 commented on 2014-02-23 17:36

@sanerb: The latest pf-kernel patch was written for 3.12.4, so unless there is a patch published for the 3.13 series, I don't think this is going to be updated.

sanerb commented on 2014-02-23 05:45

can we get this updated for 3.13? (current base repo has 3.13.4-1 as of time of writing this)

Thaodan commented on 2014-01-15 19:01

I builds see my repo, I don't know what you have done but do you removed your src dir?

Dr_Drache commented on 2014-01-15 15:02

just to point out, I can't seem to compile.
fremap.c fails patching :
[code]
Hunk #1 FAILED at 207.
1 out of 1 hunk FAILED -- saving rejects to file mm/fremap.c.rej
patching file mm/madvise.c
patching file mm/memory.c
Hunk #1 succeeded at 2777 (offset 32 lines).
Hunk #2 succeeded at 3492 (offset 36 lines).
patching file mm/mmap.c
Hunk #1 succeeded at 251 with fuzz 2 (offset 1 line).
Hunk #2 succeeded at 870 (offset 10 lines).
Hunk #3 succeeded at 1642 (offset 20 lines).
Hunk #4 succeeded at 2433 (offset 21 lines).
Hunk #5 succeeded at 2454 (offset 23 lines).
Hunk #6 succeeded at 2864 (offset 36 lines).
patching file mm/msync.c
patching file mm/nommu.c
patching file fs/file_table.c
patching file fs/inode.c
patching file fs/namespace.c
patching file fs/notify/group.c
patching file fs/notify/mark.c
patching file fs/open.c
patching file fs/splice.c
patching file security/commoncap.c
patching file security/device_cgroup.c
patching file security/security.c


patching file kernel/printk/printk.c
Hunk #1 succeeded at 60 with fuzz 2 (offset 2 lines).
patching file init/Kconfig
Hunk #1 succeeded at 1102 (offset 113 lines).
Hunk #2 succeeded at 1113 (offset 113 lines).
patching file arch/x86/include/asm/module.h
patching file arch/x86/Kconfig.cpu
patching file arch/x86/Makefile
Hunk #1 succeeded at 67 with fuzz 1 (offset 10 lines).
patching file arch/x86/Makefile_32.cpu

=======================================================
=======================================================

(needs IKCONFIG and IKCONFIG_PROC)
(warning: it can take a looong time)


aura >>= Would you like to continue anyway? [Y/n] n
aura >>= Building failed.
[/code]

Thaodan commented on 2013-12-28 17:10

I don't know what you have done but the headers are included in the headers package and not in the kernel package

ilkyest commented on 2013-12-28 16:59

well, I don't know why, or if it's right, but on pkgbuild, on here

# If the following is set, stop right there. We only need the headers for
# dependent drivers' compiling (nvidia, virtualbox etc)

# get kernel version
make prepare

I've unmarked the "make prepare" and, yes, now headers

Thaodan commented on 2013-12-28 14:54

compile as usual

ilkyest commented on 2013-12-28 13:04

how do I compile it with kernel headers?

I need to virtualbox and nvidia driver

Thaodan commented on 2013-12-22 03:19

upkg

mikeroll commented on 2013-11-30 15:47

3.12 is out, please update :)

Thaodan commented on 2013-11-24 23:04

np

aiu commented on 2013-11-24 22:58

Sorry, I flagged it as out-of-date by mistake.

Thaodan commented on 2013-11-07 01:42

@sir: np
@nous: I'm doing on this already.

sir_lucjan commented on 2013-11-06 22:49

@ Thaodan - I'm sorry - I looked only at the version number. My mistake - once again I'm sorry.

sir_lucjan commented on 2013-11-06 22:47

@Thaodan: I'm sorry - I'm sorry - I looked only at the version number.

nous commented on 2013-10-31 18:35

I moved a box from linux-pf to linux-pf-p3 and the respective header packages didn't conflict, i.e. linux-pf-headers-p3 tried to install over linux-pf-headers and failed. Could you fix that? The conflicts() array in package_linux-pf-headers() should be determined more or less like the one in package_linux-pf...

nous commented on 2013-10-29 21:53

justin8: It's NOT an issue. Please, do read the post right on top of Zorael's for a rationale. Just enable DVB, it'll only create dvb.h, nothing else will burden the kernel and everybody will be happy.

justin8 commented on 2013-10-29 04:16

The issue reported by Zorael in January is still present. The build will fail if you do not choose any DVB modules for your kernel...

Thaodan commented on 2013-10-25 02:56

updated, added changes ARCH .config

Thaodan commented on 2013-10-19 18:14

@sir_lucjan: why do you reported it out of date?

Thaodan commented on 2013-10-07 18:24

updated;bcache is now enabled

demon012 commented on 2013-10-02 08:25

FYI Aufs still broken with "aufs loop_backing_file() is not defined". Not your problem though as it is upstreams fault.

Thaodan commented on 2013-10-01 20:34

should be fixed now

walkindude commented on 2013-10-01 19:48

but I always get:
DEPMOD 3.11.2-pf-pf

walkindude commented on 2013-10-01 19:46

but I always get (with P and without):
depmod 3.11.2-pf-pf

Thaodan commented on 2013-10-01 17:17

replaced the P (eg. removing the subver) by the way Arch patches the localversion[1].

[1] https://github.com/Thaodan/thaopkgs/blob/master/linux-pf/PKGBUILD#L324

Thaodan commented on 2013-10-01 15:40

updated aufs is now enabled gain.

demon012 commented on 2013-09-27 06:30

@jakob unfortunately no packages are built for the atom processors but thanks for the link, @Thaodan awesome thanks for the link. Followed you and starred the repo.

Thaodan commented on 2013-09-27 03:56

@demon012 if you want older PKGBUILDS check my git repo, here the 3.10 version:
https://github.com/Thaodan/thaopkgs/blob/b2e1d20d82840ca74619d94e83a65538366355ec/linux-pf/PKGBUILD

jakob commented on 2013-09-26 22:42

@demon012, check the repository: http://dl.dropboxusercontent.com/u/172590784/Linux-pf/x86_64/index.html

demon012 commented on 2013-09-26 21:48

Thaodan, in that case do you happen to have the older 3.10 PKGBUILD handy? I did not keep the older package and now my aufs setup is not working and neither linux-pf or linux-aufs_friendly are working at the moment due to this same bug.

Thaodan commented on 2013-09-26 08:18

I don't used it, at the time I updated the package aufs3 wasn't ready for 3.11

demon012 commented on 2013-09-26 08:12

The AUFS3 patch you are using is causing the error: "aufs au_loopback_init:136: loop_backing_file() is not defined" and it making aufs3 cease to work. Apparently the patch had this function added and then it was for some reason removed: http://www.mail-archive.com/aufs-users@lists.sourceforge.net/msg04337.html .

Need to get that other patch in this package to make it work.

SydMax commented on 2013-09-16 15:39

Hi, Thaodan.
Is there any reason, why you have ignored twice reports on this issue:
"The patchlevel option seems to be the wrong way around - 'p' removes the patch version, <enter> keeps it."
The fix in PKGBUILD is obvious, why not to implement it?

Thaodan commented on 2013-09-09 15:14

updated, aufs3 ins't ready

Thaodan commented on 2013-09-06 22:53

thanks, fixed

post-factum commented on 2013-09-06 21:50

PILEDRIVER target (line 434) has got a typo:

pkgname"${pkgbase}-piledriver"

Should be

pkgname="${pkgbase}-piledriver"

BULLDOZER target has the same issue.

brunogm commented on 2013-07-20 18:05

Hi, problem with CONFIG_MNATIVE=y.

# CONFIG_MEMTEST is not set
# CONFIG_MK8 is not set
# CONFIG_MK10 is not set
# CONFIG_MBARCELONA is not set
# CONFIG_MBOBCAT is not set
# CONFIG_MBULLDOZER is not set
# CONFIG_MPILEDRIVER is not set
# CONFIG_MPSC is not set
# CONFIG_MATOM is not set
# CONFIG_MCORE2 is not set
# CONFIG_MCOREI7 is not set
# CONFIG_MCOREI7AVX is not set
# CONFIG_MCOREAVXI is not set
# CONFIG_MCOREAVX2 is not set
# CONFIG_GENERIC_CPU is not set
CONFIG_MNATIVE=y

brunogm commented on 2013-07-20 17:49

hi,
CONFIG_MNATIVE=y gives build problem at PKGBUILD line 279.

""cp -v .config ${startdir}/config.local
for _cpusuffix_kbuild in ${_CPUSUFFIXES_KBUILD[@]} ; do
_egrepstring="${_egrepstring}M${_cpusuffix_kbuild}=y|"
done
CPU=$(egrep "${_egrepstring}CONFIG_GENERIC_CPU=y|M686=y" ./.config)
CPU=$(sed -e "s/CONFIG_M\(.*\)=y/\1/" <<<$CPU)
CPU=$(sed -e "s/CONFIG_GENERIC_CPU=y/GENERIC/" <<<$CPU)
CPU=$(sed -e "s/^686$/GENERIC/" <<<$CPU)
cp -f .config ${startdir}/config.$CPU-$CARCH

Thaodan commented on 2013-07-18 22:35

updated packages
NOTE: I added patch from https://bugzilla.kernel.org/show_bug.cgi?id=60530 to fix the issue please report issues

J4913 commented on 2013-07-09 19:36

The patchlevel option seems to be the wrong way around - 'p' removes the patch version, <enter> keeps it.

Thaodan commented on 2013-06-16 19:40

With wrong page I mean that this is the page of the nvidia module for linux-pf and not the page of linux-pf.

Thaodan commented on 2013-06-16 10:57

fixed, there was an typo related to piledriver and to generic build.

sinatosk commented on 2013-06-16 10:29

Now it's just saying there is an error...

==> Starting build()...

=======================================================
==> You might be prompted below for some config options
=======================================================

==> Hit <Y> to use your running kernel's config
(needs IKCONFIG and IKCONFIG_PROC)
==> Hit <L> to run 'make localmodconfig'
==> Hit <N> (or just <ENTER>) to build an all-inclusive kernel like stock -ARCH
(warning: it can take a looong time)


==> Using stock ARCH kernel .config (with BFS, BFQ and TuxOnIce enabled).

==> Kernel configuration options before build:
<M> make menuconfig (console menu)
<N> make nconfig (newer alternative to menuconfig)
<G> make gconfig (needs gtk)
<X> make xconfig (needs qt)
<O> make oldconfig
<L> make localyesconfig
<ENTER> to skip configuration and use stock -ARCH defaults
X
HOSTCC scripts/basic/fixdep
CHECK qt
SHIPPED scripts/kconfig/zconf.tab.c
MOC scripts/kconfig/qconf.moc
HOSTCC scripts/kconfig/conf.o
SHIPPED scripts/kconfig/zconf.lex.c
SHIPPED scripts/kconfig/zconf.hash.c
HOSTCXX scripts/kconfig/qconf.o
HOSTCC scripts/kconfig/zconf.tab.o
HOSTLD scripts/kconfig/qconf
scripts/kconfig/qconf Kconfig
Fontconfig warning: "/etc/fonts/conf.d/50-user.conf", line 14: reading configurations from ~/.fonts.conf is deprecated.
#
# configuration written to .config
#
‘.config’ -> ‘/home/username/pkg/linux-pf/config.local’
==> ERROR: A failure occurred in build().
Aborting...

sinatosk commented on 2013-06-16 10:27

forgot to mention... this is 3.9.4-4 I'm talking about

sinatosk commented on 2013-06-16 10:24

Press enter the first time, 'X' the second time to set the CPU to 'piledriver'. Save the config and quit that, then error related to 'cp' command about invalid option 'c'

==> Starting build()...

=======================================================
==> You might be prompted below for some config options
=======================================================

==> Hit <Y> to use your running kernel's config
(needs IKCONFIG and IKCONFIG_PROC)
==> Hit <L> to run 'make localmodconfig'
==> Hit <N> (or just <ENTER>) to build an all-inclusive kernel like stock -ARCH
(warning: it can take a looong time)


==> Using stock ARCH kernel .config (with BFS, BFQ and TuxOnIce enabled).

==> Kernel configuration options before build:
<M> make menuconfig (console menu)
<N> make nconfig (newer alternative to menuconfig)
<G> make gconfig (needs gtk)
<X> make xconfig (needs qt)
<O> make oldconfig
<L> make localyesconfig
<ENTER> to skip configuration and use stock -ARCH defaults
X
HOSTCC scripts/basic/fixdep
CHECK qt
SHIPPED scripts/kconfig/zconf.hash.c
HOSTCC scripts/kconfig/conf.o
SHIPPED scripts/kconfig/zconf.lex.c
SHIPPED scripts/kconfig/zconf.tab.c
MOC scripts/kconfig/qconf.moc
HOSTCC scripts/kconfig/zconf.tab.o
HOSTCXX scripts/kconfig/qconf.o
HOSTLD scripts/kconfig/qconf
scripts/kconfig/qconf Kconfig
Fontconfig warning: "/etc/fonts/conf.d/50-user.conf", line 14: reading configurations from ~/.fonts.conf is deprecated.
#
# configuration written to .config
#
‘.config’ -> ‘/home/halional/pkg/linux-pf/config.local’
cp: invalid option -- 'c'
Try 'cp --help' for more information.
==> ERROR: A failure occurred in build().
Aborting...

Thaodan commented on 2013-06-16 10:04

fixed

sinatosk commented on 2013-06-16 09:57

after pressing enter twice I see this

==> Starting build()...

=======================================================
==> You might be prompted below for some config options
=======================================================

==> Hit <Y> to use your running kernel's config
(needs IKCONFIG and IKCONFIG_PROC)
==> Hit <L> to run 'make localmodconfig'
==> Hit <N> (or just <ENTER>) to build an all-inclusive kernel like stock -ARCH
(warning: it can take a looong time)


==> Using stock ARCH kernel .config (with BFS, BFQ and TuxOnIce enabled).

==> Kernel configuration options before build:
<M> make menuconfig (console menu)
<N> make nconfig (newer alternative to menuconfig)
<G> make gconfig (needs gtk)
<X> make xconfig (needs qt)
<O> make oldconfig
<L> make localyesconfig
<ENTER> to skip configuration and use stock -ARCH defaults

‘.config’ -> ‘/home/username/pkg/linux-pf/config.local’
cp: invalid option -- 'c'
Try 'cp --help' for more information.
==> ERROR: A failure occurred in build().
Aborting...

Thaodan commented on 2013-06-16 09:41

fixed

Anonymous comment on 2013-06-16 04:50

I got the same issue "PKGBUILD: line 273: 0: command not found" with a custom config.

Looking at the PKGBUILD, there's something screwy going on in the relevant section - it's trying to figure out what CPU you're compiling for and failing miserably. Several variables appear to be misnamed, etc. For the time being, here's a workaround:

Comment out line 273 and add CPU="GENERIC" (or something else if you're not compiling for the generic CPU target) just before the cp command around 283.

felixonmars commented on 2013-06-15 14:20

PKGBUILD: line 273: 0: command not found

I'm using stock settings :(

Thaodan commented on 2013-06-15 11:39

should be fixed.

Anonymous comment on 2013-06-15 05:21

I can't set k10

Anonymous comment on 2013-06-14 10:48

I already downloaded 3.9.4 from repo and updated with "pacman -U"
But now it is ok with "pacman -S" too. Thanks.

nous commented on 2013-06-14 10:07

firebbs: you probably caught the repo at updating the packages to 3.9.4. Please try again.

Thaodan commented on 2013-06-13 20:25

I think you mean nous repo? If yes ask him in his thread.
If not: there's no generic linux-pf package in my repo.

Anonymous comment on 2013-06-13 20:19

#pacman -Syu --ignore filesystem,bash

error: failed retrieving file 'linux-pf-3.9.3-1-x86_64.pkg.tar.xz' from dl.dropbox.com : The requested URL returned error: 404 NOT FOUND
warning: failed to retrieve some files
error: failed retrieving file 'linux-pf-headers-3.9.3-1-x86_64.pkg.tar.xz' from dl.dropbox.com : The requested URL returned error: 404 NOT FOUND

Thaodan commented on 2013-06-10 19:21

Updated PKGBUILD, aufs3 is enabled again (url to gitrepo was wrong).

SydMax commented on 2013-06-05 14:54

It seems that handlers of <P> or <Enter> answers in PKGBUILD is mixed up and are in contradiction with descriptions.

SydMax commented on 2013-06-05 14:52

It seems that handlers of <P> or <Enter> answers in PKGBUILD is mixed up and are in contradiction with descriptions.

Thaodan commented on 2013-05-28 20:58

fixed

mikeroll commented on 2013-05-28 20:45

Someone forgot to add MK10=y to egrep in line 266. Please fix.
Otherwise build fails for K10 cpus.

Thaodan commented on 2013-05-24 11:50

The $_kernelname var is set in line 12, to pf. There must be something in your scripts.

nous commented on 2013-05-24 11:20

It must be something in my 32bit chroot then. I'm using distcc to shorten compile time and there's a strong warning in the wiki against distcc/multilib: https://wiki.archlinux.org/index.php/Distcc#Multilib_GCC_Method_.28Not_Recommended.29

Anyway, I've pointed at the issue because linux-pf used to compile/pack correctly in the chroot a few revisions ago.

Thaodan commented on 2013-05-23 21:35

Using 32bit makepkg.conf and gcc-multilib

nous commented on 2013-05-23 21:12

I must pin this down. Are you building in an i686 chroot or natively?

Thaodan commented on 2013-05-23 20:37

As you see it doesn't.

nous commented on 2013-05-23 20:34

Found an easy one: when copy/pasting from linux-ARCH you _must_ change ${_kernelname:--ARCH} to ${_kernelname:--pf}.

Thaodan commented on 2013-05-23 20:28

It's your script or system. I builded it just fine without any issues and with the correct path see: http://dl.dropboxusercontent.com/u/172590784/Linux-pf/i686/linux-pf-3.9.3-1-i686.pkg.tar.xz

Thaodan commented on 2013-05-23 18:31

I'm searching it, but as the settting of EXTRAVERSION isn't depent on $CARCH I don't see it. You may search on your own.

nous commented on 2013-05-23 18:17

I have an i686 server box running 3.8.2-3-pf and updated to 3.9.2-pf but not rebooted yet. The new kernel modules and headers are installed in /usr/lib/modules/3.9.0-ARCH and /usr/src/linux-3.9.0-ARCH and the running kernel's version is 3.8.0, instead of 3.8.2-pf. So, there's definitely something broken in the PKGBUILD.

Thaodan commented on 2013-05-23 18:02

Hmm, It should I don't see anything in the PKGBUILD.

nous commented on 2013-05-23 17:54

It might be my repo build script, but whereas the x86_64 packages correctly show install paths suffixed with 3.9.3-pf (i.e. in /usr/src and /usr/lib/modules), the i686 ones show instead 3.9.0 (without -pf). The returned kernel version (uname -r) might also be wrong in i686, but I haven't booted one to check.

Anonymous comment on 2013-05-23 11:00

This package mentions AUFS3 in the title, but if I'm not mistaken the patches/includes are missing. Just FYI http://aufs.sourceforge.net/ currently shows patches for 3.9; it would be great to see these re-added if it isn't too much trouble. My hope is for the ability to use Docker (http://www.docker.io/) which unfortunately requires AUFS for now...

Thaodan commented on 2013-05-22 22:25

updated, added patch to enanble cpufreq/intelp with ivybridge cpus.

trizen commented on 2013-05-09 19:01

'bc' package should be included in makedepends array, otherwise:

CC kernel/itimer.o
HZFILE kernel/hz.bc
BC kernel/timeconst.h
/bin/sh: bc: command not found
make[1]: *** [kernel/timeconst.h] Error 127
make: *** [kernel] Error 2
==> ERROR: A failure occurred in build().
Aborting...

Thaodan commented on 2013-05-04 23:39

fixed

walkindude commented on 2013-05-04 21:10

wrong sum for pf patch

walkindude commented on 2013-05-04 21:08

wrong sum for patch...

nous commented on 2013-04-12 06:14

Sorry to be such a noisy chap, but I think you should put the PKGBUILD section between lines 150-289 into build(), i.e. from
if [[ "$_BATCH" != "y" ]];
to
export CPU

Rationale: if someones wants to retry compiling with 'makepkg -e', so as to not unpack and patch again the source tree, then he (or she) won't be presented with the configuration/customization options. I use this option extensively either manually or scripted, especially for building the optimized repository binaries.

nous commented on 2013-04-10 17:34

You forgot again to update the configs, the build script stops at the same points.

% diff -u config.x86_64 config.local
--- config.x86_64 2013-03-16 22:39:26.000000000 +0200
+++ config.local 2013-04-10 20:32:01.343274958 +0300
@@ -1,6 +1,6 @@
-# Linux/x86_64 3.8.1-pf Kernel Configuration
+# Linux/x86_64 3.8.2-pf Kernel Configuration
@@ -303,8 +303,6 @@
CONFIG_CFQ_GROUP_IOSCHED=y
CONFIG_IOSCHED_BFQ=y
CONFIG_CGROUP_BFQIO=y
-# CONFIG_DEFAULT_DEADLINE is not set
-# CONFIG_DEFAULT_CFQ is not set
CONFIG_DEFAULT_BFQ=y
# CONFIG_DEFAULT_NOOP is not set
CONFIG_DEFAULT_IOSCHED="bfq"
@@ -341,6 +339,7 @@
CONFIG_NO_BOOTMEM=y
# CONFIG_MEMTEST is not set
# CONFIG_MK8 is not set
+# CONFIG_MK10 is not set
# CONFIG_MBARCELONA is not set
# CONFIG_MBOBCAT is not set
# CONFIG_MBULLDOZER is not set
@@ -5287,6 +5286,8 @@
# CONFIG_EDD_OFF is not set
CONFIG_FIRMWARE_MEMMAP=y
CONFIG_EFI_VARS=y
+CONFIG_EFI_VARS_PSTORE=y
+# CONFIG_EFI_VARS_PSTORE_DEFAULT_DISABLE is not set
CONFIG_DELL_RBU=m

Thaodan commented on 2013-04-08 09:52

yes

lockheed commented on 2013-04-08 09:46

Does that version finally include appropriate /include/linux/version.h file to enable installation of nvidia-pf package?

nous commented on 2013-04-08 07:07

I also found tremendously helpful to regularly commit to the bitbucket repo, so I could easily revert mistakes or use a past version which worked better in my laptop.

Thaodan commented on 2013-04-08 07:03

Ah something that I forgot :)

nous commented on 2013-04-08 07:01

With every upstream kernel update new options are introduced, which are not always 'default'; others are changed by some patches. You can see which ones if you press <ENTER> at the second prompt (kernel configuration options before build). In my box it stops at the CPU selection and at EFI vars.

nous commented on 2013-04-08 07:00

With every upstream kernel new options are introduced, which are not always 'default'; others are changed by some patches. You can see which ones if you press <ENTER> at the second prompt (kernel configuration options before build). In my box it stops at the CPU selection and at EFI vars.

Thaodan commented on 2013-04-08 06:48

I used the old config whats bad with them?

nous commented on 2013-04-08 06:45

You should 'make nconfig' (or any *config you prefer) through the PKGBUILD in both i686 and x86_64 and rename the resulting config.local into config and config.x86_64 respectively, then regenerate the sha256sums. I know it's trivial, but my repo build scripts find unresolved configuration options and stop prompting for input.

Thaodan commented on 2013-04-07 22:14

Updated the package added prepare() removed the two drm-i915 patches.

Thaodan commented on 2013-04-07 22:05

Updated the package added prepare() removed the two drm-i915 packages.

Thaodan commented on 2013-04-07 21:55

I'll update the PKGBUILD yet but the packages will not come until the next day or later.

walkindude commented on 2013-04-07 21:52

3.8.2 is out
http://pf.natalenko.name/forum/index.php?topic=202.0

nous commented on 2013-03-30 19:54

Due to increasingly pressing real-life obligations, my free time is less than required to provide quality maintainance of such a demanding package. It's been a wonderful ride, thanks for your support and kind words. Good luck, Thaodan!

nous commented on 2013-03-30 19:47

Due to increasingly real-life demands, my free time is less than required to provide quality maintainance of such a demanding package. It's been a wonderful ride, thanks for your support and kind words. Good luck, Thaodan!

Thaodan commented on 2013-03-28 02:43

Updated my builds for sandbridge and ivvy bridge (NUMA off, removed minor version string):
https://docs.google.com/folder/d/0B_FQtIHEHNn8d3ktYVpDcFYzdEU/edit?usp=sharing

nous commented on 2013-03-20 09:07

Ok.

Anonymous comment on 2013-03-20 09:04

Works fine (all other 3.8 version I "had cannot stat ‘/tmp/yaourt-tmp-indie/aur-linux-pf/src/linux.preset’: No such file or directory"... and I have an x64 CPU ;))

Thaodan commented on 2013-03-20 08:51

Nous: I suggest to add the option to disable NUMA it the beginning of the PKGBUILD instead adding it in build()

jnbek commented on 2013-03-19 05:13

Build works fine for me, perhaps y'all are using the 32 bit build and need to join the 21st century already? ;-)

mytbk commented on 2013-03-19 03:27

@indie: Me,too. I think the submitter should try building a package to ensure the PKGBUILD is okey before submitting.

mytbk commented on 2013-03-19 02:59

@indie: Me,too. I think the submitter should try building a package to ensure the PKGBUILD is okey before submitting.

indie commented on 2013-03-19 02:52

install: cannot stat ‘/tmp/yaourt-tmp-indie/aur-linux-pf/src/linux.preset’: No such file or directory

Thaodan commented on 2013-03-18 19:40

Just add the functionality from linux-ck like I did (this disables NUMA by default if not enabled):
http://pastebin.com/6fZE1eJU (i915 currently not applied),

nous commented on 2013-03-18 16:37

I'll post a PKGBUILD with the i915 patches ASAP, as that bug's a showstopper. The CPU optimization patches will follow soonest if not with the former. I've also been considering the feasibility of NUMA in linux-pf's scope. Is there ANY one needing it?

Thaodan commented on 2013-03-18 07:55

My GDocs folder with my builds optimized for sandy- and ivvy bridge (comming soon): https://docs.google.com/folder/d/0B_FQtIHEHNn8d3ktYVpDcFYzdEU/edit
NUMA is disabled in this build.

Thaodan commented on 2013-03-18 07:49

You may use the updated gcc patch from linux-ck?
http://repo-ck.com/source/gcc_patch/kernel-38-gcc47-1.patch.gz

Thaodan commented on 2013-03-18 03:37

You may add patches from FS #34327?

Thaodan commented on 2013-03-18 03:33

I get this when patching aufs:
patching file fs/file_table.c
patching file fs/inode.c
patching file fs/splice.c
patching file include/linux/fs.h
Hunk #1 succeeded at 2560 (offset 9 lines).
patching file include/linux/splice.h
patching file fs/Kconfig
patching file fs/Makefile
patching file include/uapi/linux/Kbuild
patching file drivers/block/loop.c
patching file fs/aufs/f_op.c
patching file fs/aufs/loop.c
patching file fs/aufs/loop.h
patching file fs/aufs/super.c
patching file include/linux/fs.h
patching file include/linux/loop.h
patching file fs/buffer.c
patching file fs/proc/nommu.c
patching file fs/proc/task_mmu.c
patching file fs/proc/task_nommu.c
patching file include/linux/mm_types.h
patching file kernel/fork.c
patching file mm/memory.c
Hunk #1 succeeded at 3409 (offset 36 lines).
patching file mm/mmap.c
Hunk #1 succeeded at 243 with fuzz 2 (offset 1 line).
Hunk #2 succeeded at 828 (offset 10 lines).
Hunk #3 succeeded at 2396 (offset 48 lines).
Hunk #4 succeeded at 2419 (offset 50 lines).
Hunk #5 succeeded at 2831 (offset 63 lines).
patching file mm/nommu.c
patching file fs/file_table.c
patching file fs/inode.c
patching file fs/namespace.c
Hunk #2 succeeded at 1432 (offset 7 lines).
patching file fs/notify/group.c
patching file fs/notify/mark.c
patching file fs/open.c
patching file fs/splice.c
patching file security/commoncap.c
patching file security/device_cgroup.c
patching file security/security.c
patching file kernel/printk.c
Hunk #1 succeeded at 61 (offset 3 lines).
patching file arch/x86/include/asm/module.h
Hunk #1 succeeded at 15 (offset -2 lines).
Hunk #2 succeeded at 41 (offset -2 lines).
patching file arch/x86/Kconfig.cpu
Hunk #1 succeeded at 139 (offset -8 lines).
Hunk #2 succeeded at 147 (offset -8 lines).
Hunk #3 succeeded at 155 (offset -8 lines).
Hunk #4 succeeded at 280 (offset -8 lines).
Hunk #5 succeeded at 288 (offset -8 lines).
Hunk #6 FAILED at 375.
Hunk #7 succeeded at 394 (offset -32 lines).
Hunk #8 succeeded at 426 (offset -32 lines).
1 out of 8 hunks FAILED -- saving rejects to file arch/x86/Kconfig.cpu.rej
patching file arch/x86/Makefile
Hunk #1 succeeded at 58 (offset 7 lines).
patching file arch/x86/Makefile_32.cpu
Hunk #1 succeeded at 24 (offset -1 lines).
Hunk #2 succeeded at 36 (offset -1 lines).
PS: You may add patches from FS #34327?

nous commented on 2013-03-17 20:25

felixonmars: Are you using any kind of compiler cache, like ccache? It's certainly not linux-pf related.

mytbk: Thanks, it slipped from the -ARCH portion; I'll rename the linux-pf.{preset|install} files to just linux.* soon, as the relevant section uses $pkgbase anyway.

First generic i686 and x86_64 3.8-pf series are already in the pfkernel repo, optimized builds added every hour or so. Please report any problems here, especially tuxonice related.

mytbk commented on 2013-03-17 04:21

and also 'linux-pf.install' is written as 'linux.install'.

mytbk commented on 2013-03-17 03:49

I think the string 'linux.preset' in this PKGBUILD should be 'linux-pf.preset'.

mytbk commented on 2013-03-17 03:46

I think the string 'linux.preset' in this PKGBUILD should be 'linux-pf.preset'.

felixonmars commented on 2013-03-17 03:30

I'm getting a strange error:
unrecognized ELF data encoding 0: drivers/atm/eni.o
make[2]: *** [drivers/atm/eni.o] Error 1

nous commented on 2013-03-06 07:27

No idea but I'm guessing postfactum is awaiting all patches to sync with 3.8.

Thaodan commented on 2013-03-06 00:41

Any succes torwards linux 3.7.10?

nous commented on 2013-03-01 22:35

I just tried both URLs and the one without /pub/ fails.

% wget -Y off ftp://www.kernel.org/linux/kernel/v3.x/linux-3.7.tar.xz
Resolving www.kernel.org (www.kernel.org)... 149.20.4.69, 198.145.20.140
Connecting to www.kernel.org (www.kernel.org)|149.20.4.69|:21... connected.
Logging in as anonymous ... Logged in!
==> SYST ... done. ==> PWD ... done.
==> TYPE I ... done. ==> CWD (1) /linux/kernel/v3.x ...
No such directory ‘linux/kernel/v3.x’.

The one with /pub/ works.

Anonymous comment on 2013-03-01 17:08

Hi,

Can't install here, default link broken to download Vanilla kernel at this time...
To succeed I have replaced it manually by this one : ftp://www.kernel.org/linux/kernel/v3.x/linux-3.7.tar.xz
in PKGBUILD

compiling now so not a big deal.

BR,

nous commented on 2013-02-16 23:57

3.7.5-2 is for i686 users only: -1's config was vanilla -ARCH without any -pf settings... sorry.

nous commented on 2013-02-16 23:55

3.7.5-2 is of i686 users only: -1's config was vanilla -ARCH without any -pf settings.

nous commented on 2013-02-04 13:07

3.7.4-2's only change is a move from ZCACHE to ZCACHE2 so as I won't forget to implement it later. Do not bother updating.

jnbek commented on 2013-01-27 05:07

is there not a post on the forums for the purpose of discussing this kernel and it's package? Can we please move these discussions here: http://is.gd/4hDDzF as opposed to here. This comment system is to report build errors and to give thanks to the maintainer for being the kick ass mofo to be making this kernel so easily available in the first place... Feature requests and other assorted 'zOMFG!!! I NEEDZ AnaLogtV!!!!' can be discussed in the Forum post.

nous commented on 2013-01-16 21:14

Zorael: This has been discussed time and again (see older comments here); I synchronize my PKGBUILD with the stock ARCH, therefore the default linux-pf contains DVB. Just enable Analog TV under Device Drivers -> Multimedia without anything else to have dvb.h created.

Creating an exception for this is not an option as I will most certainly forget to re-create it at the next update.

Anonymous comment on 2013-01-16 20:36

Fails to build if you opted to not compile any DVB drivers;

==> Starting package_linux-pf-headers()...
cp: cannot stat ‘include/config/dvb/*.h’: No such file or directory
==> ERROR: A failure occurred in package_linux-pf-headers().

One dirty workaround is to just append || true to the offending command (line ~557). Would you mind adding that? Or maybe you'd prefer some more elegant logic around that and the other these-files-should-be-there fixes.

felixonmars commented on 2013-01-12 15:25

@nous
Sorry I've found out that the problem was caused by testing/file 5.12 (which is not present now).

nous commented on 2013-01-09 19:17

I admit I hadn't tested, so I went ahead and compiled it. It works ok. Are you using any pacman wrapper?

felixonmars commented on 2013-01-09 12:09

I'm getting the following error when upgrading from 3.7.1-3 to 3.7.1-4:

(1/2) upgrading linux-pf
[------------------------------------------------------] 100%
>>> Updating module dependencies. Please wait ...
>>> Generating initial ramdisk, using mkinitcpio. Please wait...
==> Building image from preset: 'default'
-> -k /boot/vmlinuz-linux-pf -c /etc/mkinitcpio.conf -g /boot/initramfs-linux-
pf.img
==> ERROR: invalid kernel specified: `/boot/vmlinuz-linux-pf'
==> Building image from preset: 'fallback'
-> -k /boot/vmlinuz-linux-pf -c /etc/mkinitcpio.conf -g /boot/initramfs-linux-
pf-fallback.img -S autodetect
==> ERROR: invalid kernel specified: `/boot/vmlinuz-linux-pf'
error: command failed to execute correctly

nous commented on 2013-01-08 16:14

klazutin: https://bbs.archlinux.org/viewtopic.php?pid=1212358#p1212358 :)

Anonymous comment on 2013-01-08 13:48

>To compile nvidia, until they sniff the change in 3.7: sudo ln -s /usr/src/linux-$(uname -r)/include/generated/uapi/linux/version.h /usr/src/linux-$(uname -r)/include/linux/version.h

Just wanted to note that this is also required to compile VMWare modules. Took me a while to figure out.

nous commented on 2013-01-07 09:38

You should report this to the BFS developers, it must be very hardware-specific. I've tested connecting the charger in two laptops with no issues.

Anonymous comment on 2013-01-05 14:23

i have kernel panic, when plugin charger (trobles with BFS). if i change bfs to noop - work fine.

jnbek commented on 2013-01-04 15:46

I got the prompt for _SUBLEVEL and it worked like peaches and creme, thanks again for being so proactive @nous :D

jnbek commented on 2013-01-01 00:40

@nous: I really appreciate your quick action to suggestions and queries. I will test the _SUBLEVEL prompt here in a while when I have a chance to upgrade. Thanks again for your hard work in acquiescing to my nonsensical whims :)

nous commented on 2012-12-31 23:38

To compile nvidia, until they sniff the change in 3.7: sudo ln -s /usr/src/linux-$(uname -r)/include/generated/uapi/linux/version.h /usr/src/linux-$(uname -r)/include/linux/version.h

As for the sublevel prompt, it would only appear when a non-generic CPU was selected; fixed in 3.7.1-3 (hopefully).

Happy 2013!

jnbek commented on 2012-12-31 15:21

yea, I been meaning to mention this for a few days, and keep forgetting for one reason or a thousand. I have yet to be prompted by the PKGBUILD if I want to set _SUBLEVEL=y or not.. I go through the questions, and it just goes straight to the build... defaulting to no sublevel... Not sire what the issue is, I've just gone and set it to Y manually in the PKGBUILD to accomodate my sometimes seedy desires. But now that I remember, I can now live my life to it's fullest knowing that I have finally reported my _SUBLEVEL issue here :D

jahiy commented on 2012-12-31 11:59

Please set _SUBLEVEL to "y"(default), or nvidia cannot be complied correctly without SUBLEVEL!!

nous commented on 2012-12-22 11:52

"Core2 is one that the kernel devs include in the source."

Thank you.

graysky commented on 2012-12-22 11:01

@nous - Core2 vs generic was only 88 ms faster. Core2 is one that the kernel devs include in the source. If you are rejecting the others based on your argument about practical significance, then why are you offering any optimized builds? Based on these data, they all offer, in you option, no practical significance. Do you see the disconnect here?

nous commented on 2012-12-22 10:32

@Thaodan: The gcc-4.7 patch is there, the optimizations are enabled but I won't pass them onto the package name beyond the in-kernel maximum (i.e. core2 and k8). To put an end to this, there's already provision for extreme optimization enthusiasts in the PKGBUILD: uncomment the 'export KCFLAGS="-march=native -Ofast"' line just before make bzImage and one gets the fastest kernel in the universe one's CPU can run (I do, for my core2 laptop). That been said, I won't add optimizations into the PKGBUILD other than those from the upstream kernel.

@graysky: The same obvious fact is stated in lkml; statistically significant does not mean practically significant. As I've written elsewhere, I believe the kernel developers know better than me when to include a CPU optimization into the kernel configuration. Please, if it's not a great hassle, run a comparison between core2 and avx-i kernels.

We've spent more time on this topic than all our CPU's combined would've saved running coreavx kernels for several thousand years.

nous commented on 2012-12-22 09:28

@Thaodan: The gcc-4.7 patch is there, the optimizations are enabled but I won't pass them onto the package name beyond the in-kernel maximum (i.e. core2 and k8). To put an end to this, there's already provision for extreme optimization enthusiasts in the PKGBUILD: uncomment the 'export KCFLAGS="-march=native -Ofast"' line just before make bzImage and one gets the fastest kernel in the universe one's CPU can run (I do, for my core2 laptop). That been said, I won't add optimizations into the PKGBUILD other than those from the upstream kernel.

@graysky: The same obvious fact is stated in lkml; statistically significant does not mean practically significant. As I've written elsewhere, I believe the kernel developers know better than me when to include a CPU optimization into the kernel configuration. Please, if it's not a great hassle, run a comparison between core2 and avx-i kernels.

We've spent more time on this topic than all our CPU's combined would've saved running coreavx kernels for a several thousand years.

Thaodan commented on 2012-12-21 23:51

BTW: nvidia-pf detecs all builds that I added fine without any issues.

Thaodan commented on 2012-12-21 23:50

nous: I don't say build all aviable optimisations, I only say inclue it in the PKGBUILD nothing more.

jnbek commented on 2012-12-21 15:01

egads, aur spam!!! lol, woke up this morning to an inbox full of notices from this pkg... :D Love a well used package.

graysky commented on 2012-12-21 14:25

@nous - You're missing the point. Upstream already includes optimization options in the nconfig that give the same order of decrease (hundreds of milliseconds) vs. generic. The options the patch introduces just expands the existing menu. The real point is that there are statistically significant gains by using them -- at least on the machines I looked at... for more, see https://lkml.org/lkml/2012/12/9/47 and the reply.

nous commented on 2012-12-21 13:39

Thaodan: Good to know, as I've included in the PKGBUILD the option to follow the 3-number versioning in 3.7 (right after the CPU-naming selection); it leaves SUBLEVEL at the original upstream number with no further tweaking.

nous commented on 2012-12-21 13:33

About the proposed optimization schemes and the benchmarks graysky provided: the slowest tested CPU vs GENERIC gives a
echo '(191.9256-191.8459)/191.8459*100'|bc -l
.04154375986143045000
WOW! 0.0415% gain in performance! 79.7 timeless and torturing milliseconds after 191 seconds of wait. And, mind you, that gain is against the GENERIC kernel. Now, if graysky is kind enough to compare a core2 kernel vs an avx-i, I bet my sideburns that the measured gain won't be greater than 1 millisecond, if that. It's just nonsensical and futile; optimize your userspace, not the kernel. Again, see http://tinyurl.com/cp7wjv5 and read the comments [0].

Thaodan: thanks for your work but I won't spend 20 hours (40 minutes * 15 CPUs * 2 arches) compiling 30 different kernels which run barely 1/10th of a second faster in benchmarks than linux-pf (and 1/1000th second faster than linux-pf-[core2|k8]. Not to mention the nvidia-pf mess that'll ensue.

graysky: I've queued the watchdog patch for inclusion, thanks.

[0] Read http://tinyurl.com/cp7wjv5. It was the last thing I ever did on Gentoo before moving to Arch.

Thaodan commented on 2012-12-21 12:38

nous: i fixed it by setting SUBLEVEL in /usr/src/linux-pf-3.7/Makefile to "1"

nous commented on 2012-12-21 11:59

Thaodan: I do.
% cat /usr/lib/modules/3.6-pf/build/include/linux/version.h
#define LINUX_VERSION_CODE 198144
#define KERNEL_VERSION(a,b,c) (((a) << 16) + ((b) << 8) + (c))
% cat /usr/lib/modules/3.7-pf/build/include/linux/version.h
cat: /usr/lib/modules/3.7-pf/build/include/linux/version.h: No such file or directory
% cat /usr/lib/modules/3.7.0-1-ARCH/build/include/linux/version.h
cat: /usr/lib/modules/3.7.0-1-ARCH/build/include/linux/version.h: No such file or directory

It would seem version.h is missing from 3.7. A simple copy from a *-headers-3.6 will temporarily resolve this.

Thaodan commented on 2012-12-21 05:25

Has someone an idea why drivers like nvidia don't build if SUBLEVEL in /usr/src/linux-pf-3.7 is empty?

Thaodan commented on 2012-12-21 03:40

My modifed PKGBUILD with proper pkgnames and description for optimized builds, I addded grayskys docs and disable numa (if not enabled) and I the upstream fix for watchdog regression as suggest: https://docs.google.com/open?id=0B_FQtIHEHNn8UTBQempFcl9jb1E

Thaodan commented on 2012-12-21 03:20

My modifed PKGBUILD with proper pkgnames and description for optimized builds, I addded grayskys docs and disable numa (if not enabled) and I the upstream fix for watchdog regression as suggest: https://docs.google.com/open?id=0B_FQtIHEHNn8TWxVZmNpYk9Ka0k

graysky commented on 2012-12-20 22:32

/type on/typo/

walkindude commented on 2012-12-20 22:30

all builds now, thanks)

walkindude commented on 2012-12-20 22:30

all builds now, thanks)

graysky commented on 2012-12-20 22:27

...also, there is a type on in the kernel-33-gcc47-0 patch which I emailed the author about but haven't heard back from him. See: http://repo-ck.com/source/gcc_patch/kernel-37-gcc47-1.patch.gz

graysky commented on 2012-12-20 22:26

@nous - Does the pf set contain this fix? If not you might wanna add it.

https://projects.archlinux.org/svntogit/packages.git/commit/trunk?h=packages/linux&id=5e9d4a043761d12d2dfbc57ae8a753dd2f419fed

nous commented on 2012-12-20 22:20

Fixed, thanks. The repo is being populated right now.

cian1500ww commented on 2012-12-20 20:38

@walkindude

Yep, you're right, the PKGBUILD for 3.7 in Arch testing doesn't have mtd in the for loop either.

walkindude commented on 2012-12-20 20:10

@cian1500ww
because of

for i in acpi asm-generic config crypto drm generated linux math-emu \
media mtd net pcmcia scsi sound trace video xen; do
cp -a include/${i} "${pkgdir}/usr/src/linux-${_kernver}/include/"
done

in PKGBUILD

i just deleted mtd and everything worked..

by the way, I got this when I wanted to build modules for virtualbox:

DKMS make.log for vboxhost-4.2.4 for kernel 3.7-pf (x86_64)
Fri Dec 21 00:07:23 MSK 2012
make: Entering directory `/usr/src/linux-3.7-pf'
LD /var/lib/dkms/vboxhost/4.2.4/build/built-in.o
LD /var/lib/dkms/vboxhost/4.2.4/build/vboxdrv/built-in.o
CC [M] /var/lib/dkms/vboxhost/4.2.4/build/vboxdrv/linux/SUPDrv-linux.o
In file included from /var/lib/dkms/vboxhost/4.2.4/build/vboxdrv/include/iprt/types.h:129:0,
from /var/lib/dkms/vboxhost/4.2.4/build/vboxdrv/include/VBox/types.h:30,
from /var/lib/dkms/vboxhost/4.2.4/build/vboxdrv/linux/../SUPDrvInternal.h:35,
from /var/lib/dkms/vboxhost/4.2.4/build/vboxdrv/linux/SUPDrv-linux.c:31:
include/linux/types.h:5:30: fatal error: uapi/linux/types.h: No such file or directory
compilation terminated.
make[2]: *** [/var/lib/dkms/vboxhost/4.2.4/build/vboxdrv/linux/SUPDrv-linux.o] Error 1
make[1]: *** [/var/lib/dkms/vboxhost/4.2.4/build/vboxdrv] Error 2
make: *** [_module_/var/lib/dkms/vboxhost/4.2.4/build] Error 2
make: Leaving directory `/usr/src/linux-3.7-pf'

cian1500ww commented on 2012-12-20 19:23

I'm getting the following error when packing linux-pf-headers: "cp: cannot stat ‘include/mtd’: No such file or directory"

I'm using the default config.

Thaodan commented on 2012-12-19 18:08

Anyone has an idea how is the state of tuxonice for linux-3.7?

jnbek commented on 2012-12-18 22:40

Muchas Gracias @jakob

jakob commented on 2012-12-18 07:01

@jnbek: pfkernel is the repo maintained by nous. No idea who set up the other one or what's the benefit.

jnbek commented on 2012-12-18 03:40

which of the 'unofficial user repos' mirror this pkg? pflinux or linux-pf ? seems there are two...

Decepteiskon commented on 2012-12-17 16:08

@jnbek Thanks, it worked.

jnbek commented on 2012-12-16 19:00


@nous: w00t!!! thanks :)

@Decepteiskon, remove it, and it's corresponding md5sum (4th line from the bottom iirc, starts with d) from the PKGBUILD, this patch never actually applied, well at least for quite a while. The maintainer has stated that this will be removed when the 3.7 build is released.

Decepteiskon commented on 2012-12-16 18:39

Can't compile, error:
curl: (22) The requested URL returned error: 404 Not Found
...given when downloading interactive-governor.tar.gz

nous commented on 2012-12-15 18:03

Ok then, the customer is always right! I'll give the 2/3-digit kernel version option in 3.7-pf.

jnbek commented on 2012-12-14 20:31

When I did it last night, using yaourt it removed the old 3.6-pf and installed the 3.6.11-pf just fine. And there's no 3.6-pf dir in /lib/modules seems fine.

nous commented on 2012-12-14 19:14

@jnbek: A simple versioning preference option and an "if-fi" around the SUBLEVEL sed would be enough for that, but I don't know what happens if one compiles a 3-digit linux-pf and at the next update one forgets to select again a 3-digit version: does pacman remove the old modulesdir from /usr/lib/modules or it ends up accomodating both 3.7.1-pf and 3.7-pf? In the pacman manpage it says -U is "remove-then-add", so I'm guessing we're OK but what happens with 2-digit kernver packages downloaded from the repo and installed with -S over a previous 3-digit version?

jnbek commented on 2012-12-14 18:49

@ nous, I totally understand the rationale, I guess I came across a little on teh smartass side of the discussion initially. Perhaps, a possible solution would be a -fullversion package for those that don't mind the nvidia/vbox compilings, and then this one keeping the 2 digit versioning for the folks who don't wish to compile. I mean, it is just a one line difference in the PKGBUILD, that way you could do some A/B testing, if you will, to see which of the two get's the most attention, then y'all can at that point decide which of the two seems best. Just a thought.

nous commented on 2012-12-14 18:07

Incidentally, I've been having this discussion with post-factum (the patchset author) a couple of days now. Here's an excerpt from my rationale on why we should stick to 2-digit versioning:

"[...] if we want to show the patchlevel and increment it with every release, then I'm OK with that, but there is an unbeatable advantage in leaving the patchlevel out of the kernel version: binary modules don't need rebuilding with every update and that was the way of linux-ARCH a few months back, until Tobias decided he didn't mind rebuilding nvidia and virtualbox with every kernel update."

Most of the boxes I own or run sport an nvidia card and I hate it when I update the kernel and Xorg won't start at the next reboot due to a simple bump from 3.6.1 to 3.6.2.

It's just unproductive.

J4913 commented on 2012-12-14 18:06

Yeah, I was aware of that. I only have PHC (which seems to be working with '3.6.11-pf' without a recompile, somehow) and VirtualBox (which is taken care of by the DKMS service). Either way, I needed to do it to test something in CrashPlan - it seems to fail to use inotify if the kernel version is only <major>.<minor>, for some crazy reason.

jnbek commented on 2012-12-14 17:51

@ J4913 just keep in mind, nous and the AUR maintainers have set it that way to prevent external module problems, so, you will likely need to rebuild any RAID, nVidia/ATI, Wireless, etc modules to accommodate the 3 level kernel value. I personally don't use any of that stuff so it's not an issue for me, I prefer a proper smelling uname -r as well, but if you find yourself having to spend 937 years rebuilding modules, you may want to relent as run it as offered.

J4913 commented on 2012-12-14 17:37

Actually, no, you're entirely correct: that was what I wanted. Thanks.

J4913 commented on 2012-12-14 09:29

Actually, what I wanted was the same version name as the stock kernel (which is just the version without '-pf', right?).

nous commented on 2012-12-14 06:40

The pf-patchset includes more patches than just BFS/CK. Post-factum is waiting for them to sync with upstream 3.7 before releasing 3.7-pf, unless some delay more than a few days in which case they're temporarily omitted from the patchset.

jnbek commented on 2012-12-14 04:07

-ck kernel just released 3.7.1, as an fyi

jnbek commented on 2012-12-14 03:47

@ J4913 Remove this line ~237:

sed -ri 's|^(SUBLEVEL =).*|\1|' Makefile

to keep the 3.6.11-pf style uname -r

J4913 commented on 2012-12-14 00:38

Is there an easy way to change the kernel name? I changed _kernelname and fixed _pfpatchname, but it still calls itself '3.6-pf' (output of 'uname -r').

nous commented on 2012-12-13 17:57

About time. Just comment it out, it didn't apply for months now anyway. I'll remove it with 3.7.1-pf.

J4913 commented on 2012-12-13 11:22

https://gist.github.com/gists/1172524/download seems to be deleted.

nous commented on 2012-12-09 16:45

Pastebin the error messages somewhere.

Anonymous comment on 2012-12-09 13:08

Can't use non-standard config. Localmod/menuconfig - no matter.

Burning_aces commented on 2012-11-30 02:28

Just did then, building package now, will edit if it doesn't work.
Thanks Nous

nous commented on 2012-11-29 07:00

Have you tried checking "Digital TV support"?

Burning_aces commented on 2012-11-29 04:47

That isn't an option for me
http://i.imgur.com/CEQEC.png

nous commented on 2012-11-13 11:15

The config options have changed slightly, but I believe all one needs is:
Device Drivers ---> Multimedia Support ---> [*] DVB/ATSC adapters

Burning_aces commented on 2012-11-13 02:14

yeah, unless I did it wrong, that still didnt work...
I can post the .config I used?

nous commented on 2012-11-12 12:30

Press "All comments" button and see the solution below. It's been discussed many times before.

Burning_aces commented on 2012-11-12 07:38

==> Starting package_linux-pf-headers()...
cp: cannot stat âinclude/config/dvb/*.hâ: No such file or directory
==> ERROR: A failure occurred in package_linux-pf-headers().

Keep getting this at the end of the makepkg.

nous commented on 2012-11-07 11:32

@lockheed: My bad, I commented-out "depends=('linux-pf')" in package_linux_headers() and forgot the "&& depends=${depends}-${cpuopt}" directive below. Please use -dd in pacman, I'm uploading a new PKGBUILD.

lockheed commented on 2012-11-07 10:44

I can't upgrade linux kernel any more, because it has a vicious circle dependency:

warning: cannot resolve "-core2", a dependency of "linux-pf-headers-core2"
:: The following package cannot be upgraded due to unresolvable dependencies:
linux-pf-headers-core2

Do you want to skip the above package for this upgrade? [y/N]
error: failed to prepare transaction (could not satisfy dependencies)
:: linux-pf-headers-core2: requires -core2

Thaodan commented on 2012-11-07 08:28

@nous: exept that it dosn't inclueds a propper conflicts array its not.

nous commented on 2012-11-07 07:44

It doesn't. It purposedly seds all newer AMD and Intel CPUs into k8 and core2 respectively. As I said before, it's non-sensical to create so many different packages for every single CPU out there. Furthermore, your submitted PKGBUILD besides being from an earlier (pre 3.6.8-pf) version, doesn't deal with a very important aspect: the conflicts() array, which is already a mess.

Thanks for your interest and your work, but I won't implement it. Besides, if you really want the ultimately optimized kernel, I have provided the proper key in the PKGBUILD many months ago: uncomment the 'export KCFLAGS="-march=native -Ofast"' line just before the final make. Then see http://termite.apcdn.com/full/84760.jpg for a definitive explanation of my rationale.

Thaodan commented on 2012-11-07 02:46

your package detects false package names for the processor that were added with the 'http://www.linuxforge.net/linux/kernel/kernel-33-gcc47-0.patch' patch.
So I updated my modifed PKGBUILD according to your latest changes:
http://pastebin.com/fYSd7hgT

Thaodan commented on 2012-11-07 02:46

your package detects false package names for the processor that were added with the 'http://www.linuxforge.net/linux/kernel/kernel-33-gcc47-0.patch' patch.
I updated my modifed PKGBUILD according to your latest changes:
http://pastebin.com/fYSd7hgT

Thaodan commented on 2012-11-07 00:34

@nous: thanks that you've inclued the patch to set the right -march -mtune for past intel core2 processors!

nous commented on 2012-10-30 23:03

@travnick: I downloaded virtualbox-modules-bfs (https://aur.archlinux.org/packages.php?ID=62966), ran 'sed -i s/bfs/pf/g PKGBUILD', renamed virtualbox-modules-bfs.install into virtualbox-modules-pf.install and I got a perfect virtualbox-modules-pf package (https://aur.archlinux.org/packages.php?ID=64162). But I won't maintain it as I don't use virtualbox. I will compile it though and put it in [pfkernel] if someone keeps it up-to-date.

@Thaodan: That's a negative. Virtualbox might be picky, but nvidia-pf (which I use) works without recompiling throughout patchlevels. I really can't see _one_ reason for a user to know the exact patchlevel of a running kernel, other than bug reporting, in which case it's easily obtainable by other means.

Thaodan commented on 2012-10-30 22:27

@Martdadinata666: as for me just cause uname shoud report the right kernel release.

travnick commented on 2012-10-30 20:26

As I wrote:


When kernel updates, old modules (eg virtualbox) still are there and make conflicts, they are not loadable
"ERROR: could not insert 'vboxdrv': Exec format error
ERROR: could not insert 'vboxnetflt': Exec format error
ERROR: could not insert 'vboxnetadp': Exec format error"

But you cannot simply create new ones with dkms (virtualbox-hook), because the old one exist in kernel modules folder ...
"Module vboxhost/4.2.2 already installed on kernel 3.6-pf/x86_64"
So you must uninstall (remove vbox modules) and do dkms remove to remove old builds from dkms.


But maybe there is need to create vbox package dedicated to linux-pf

martadinata666 commented on 2012-10-30 14:15

hmm, why you need to name this versionas 3.6.4? as long as im using this kernel, if you have modules like vbox, you just need to rebuild this modules if this kernel update to major version 3.5 to 3.6 or when virtualbox upgrade...

Thaodan commented on 2012-10-30 13:01

Ah ok thanks. A suggestion please rename the kernel release (the from uname) to the release number of the based kernel release, actual: 3.6.4.

nous commented on 2012-10-30 10:25

The pf-kernel patchset does not closely follow the kernel.org, as it increments with updates from various patches, but usually includes the latest stable linux kernel. 3.6.6-pf apparently includes 3.6.4 from kernel.org. See http://freecode.com/projects/pf-kernel.

martadinata666 commented on 2012-10-29 23:31

so afaik, this have its own patch package which u can use just on bare linux 3.6 not 3.6.x :)

Thaodan commented on 2012-10-29 21:38

So its not 3.6.4?

Det commented on 2012-10-29 17:46

Yes.

Thaodan commented on 2012-10-29 13:57

One question: Is it right that the kernel version is 3.6?

Thaodan commented on 2012-10-29 13:13

@nous I don't that there shoud be packages for all these cpu familys, its just nice too have, if you need i can upload builds for corei7avx and coreavxi

nous commented on 2012-10-29 12:59

@claudiaj: I've not encountered the zram issue myself, on several computers I run this kernel. The reports you point to indicate it's not linux-pf related though; therefore, we should expect a fix from upstream.

nous commented on 2012-10-29 07:40

I'm building 3.6.6-pf for i686 now. The problem with dropbox is solved, and right now the daemon is uploading 3.6.5-2/64-bit; 3.6.6 for x86_64 will follow before the end of the day.

@Thaodan: it's not the menuconfig I'm concerned about, it's the PKGBUILD which should reflect the extra CPUs and the extra time, cpu-power and webspace needed to build and accommodate the additional packages. Which, I still insist, are not worth building as there's no gain whatsoever in performance kernel-wise.

That being said, I might add the patch to an upcoming release, but I'll have to decide whether or not let the PKGBUILD know about i7*.

Anonymous comment on 2012-10-28 23:40

@jakob: I had no intention to come off as rude. My main concern is zram, I just noted that after nous has stated 'Should be OK now', http://dl.dropbox.com/u/11734958/ still 404's, indicating that not all is gravy just yet.

I have this unfortunate tendency to spend days fine tuning menuconfig getting rid of stuff I'll never use and adding things I might think I'd like to see. Just my luck would have it though, there's always an update ready by the time I finish :P

Thaodan commented on 2012-10-28 23:33

@nou: this patch dosn't modifies the kernel code it just adds the option to set the right -march and -mtune for cpus that are never than year 2007 in menuconfig.

jakob commented on 2012-10-28 23:07

Hey claudiaj, this service by nous is an exceptional contribution that he has been committing since a long time now! I think we should be a bit more patient if there are computer problems on his side! It's always possible to just build the kernel, the PKGBUILDs are there, if you're in such need to have it now!

Anonymous comment on 2012-10-28 22:40

I'm getting an insane amount of zram errors in dmesg

[305870.803049] zram: Error allocating memory for compressed page: 121722, size=4118
[305870.803089] Write-error on swap-device (254:0:973776)

Seems related to this
http://code.google.com/p/compcache/issues/detail?id=102

Still getting 404 on the dropbox repo. How long do you imagine 'OK now' means working again?

nous commented on 2012-10-28 20:50

@dkasak: I upgraded my main dropbox box (which is headless)to systemd and lost all connectivity to it, as setting a static IP with systemd requires user intervention. Should be OK now.

nous commented on 2012-10-28 20:07

@Thaodan: thanks for your work, but I intend to follow the upstream kernel optimization options.
Allow me to elaborate:
1. I believe Linus and the rest of the kernel developers know their job better than me.
2. Extreme kernel optimizations offer very little performance-wise; it's far more beneficial to optimize in user-space. Please, read this post of mine about it: http://ubuntuforums.org/showpost.php?p=1202195&postcount=86
3. By large, newest CPUs offer only a few additional instruction sets, mostly multimedia-related, like Advanced Vector Extensions; the AES-NI instruction set is really not worth to mention, unless one AES encrypts-decrypts all day long. Therefore, a new CPU choice usually enters the kernel only after a major architecture evolution.

dkasak commented on 2012-10-28 11:04

I've been getting this error for a few days when trying to install from the unofficial repo: error: failed retrieving file 'linux-pf-core2-3.6.5-2-x86_64.pkg.tar.xz' from dl.dropbox.com : The requested URL returned error: 404 NOT FOUND.

Thaodan commented on 2012-10-28 02:47

Forgot to mod everything in the PKGBUILD to be conform with the changes:
http://pastebin.com/RHM2hXME

Thaodan commented on 2012-10-28 02:19

If forgot to say that it only affects Kconfig/menuconfig.

Thaodan commented on 2012-10-28 02:16

Added patch to include -march and -mtune flags for corei7, corei7-avx core-avxi and core-avx2 etc that were intrued in gcc 4.6 and 4.7 use it if you need:
http://pastebin.com/KnZzgCNT

Patch:
--- linux-pf/PKGBUILD 2012-10-26 11:34:57.000000000 +0200
+++ PKGBUILD 2012-10-28 03:03:53.340907261 +0100
@@ -33,7 +33,9 @@
"interactive-governor.tar.gz::https://gist.github.com/gists/1172524/download"
'change-default-console-loglevel.patch'
'dcache.patch'
- "${_pfpatchhome}${_pfpatchname}.bz2") # the -pf patchset
+ "${_pfpatchhome}${_pfpatchname}.bz2" # the -pf patchset
+ "http://www.linuxforge.net/linux/kernel/kernel-33-gcc47-0.patch"
+)

_aufs3git="git://aufs.git.sourceforge.net/gitroot/aufs/aufs3-standalone.git"
_aufs3name=aufs3
@@ -100,6 +102,8 @@
msg "Patching dcache.c"
patch -Np1 -i "${srcdir}/dcache.patch"

+ # added gcc 4.7 support for Kconfig and menuconfig
+ patch -Np1 -i "${srcdir}/kernel-33-gcc47-0.patch"
if [ "$CARCH" = "x86_64" ]; then
cat "${startdir}/config.x86_64" >| .config
else
@@ -188,7 +192,7 @@
;;
esac
cp -v .config ${startdir}/config.local
- CPU=`egrep "MK8=y|MCORE2=y|MPSC=y|MATOM=y|MPENTIUMII=y|MPENTIUMIII=y|MPENTIUMM=y|MPENTIUM4=y|MK7=y|CONFIG_GENERIC_CPU=y|M686=y" ./.config`
+ CPU=`egrep "MK8=y|MCORE2=y|MPSC=y|MATOM=y|MPENTIUMII=y|MPENTIUMIII=y|MPENTIUMM=y|MPENTIUM4=y|MK7=y|CONFIG_GENERIC_CPU=y|M686=y|MBARCELONA=y|MBOBCAT=y|MBULLDOZER=y|MPILEDRIVER=y|MCOREI7=y|MCOREI7AVX=y|MCOREAVXI=y|MCOREAVX2=y" ./.config`
CPU=`sed -e "s/CONFIG_M\(.*\)=y/\1/" <<<$CPU`
CPU=`sed -e "s/CONFIG_GENERIC_CPU=y/GENERIC/" <<<$CPU`
CPU=`sed -e "s/^686$/GENERIC/" <<<$CPU`
@@ -571,4 +575,6 @@
'd537a358c9f8b9cbcad34ba99f88af070f54d99c66d3e5d3d4734576ed2fb6e1'
'b9d79ca33b0b51ff4f6976b7cd6dbb0b624ebf4fbf440222217f8ffc50445de4'
'c49eca2478b7d911ee95a61c3f9080082b982ba4de34b9acff9b1ff2d5224957'
- 'f9cf30022dcb86a70891c5c1f64713fa4da108beec72e5e264942c674268a6b6')
+ 'f9cf30022dcb86a70891c5c1f64713fa4da108beec72e5e264942c674268a6b6'
+ '8f2dad5b89dac8c76c3dd3ad902928d6032ec2c6ada41d873d752bad5586b154')

nous commented on 2012-10-25 19:36

That exactly was the purpose of removing the patchlevel number from the kernel version: to let binary blobs work without the need to recompile throughout all a given minor version series (e.g. 3.6.[1-N]-pf). The nvidia-pf module for one works as intended since 3.6.1-pf, at least on my computers. I wasn't aware of the virtualbox situation... I'll test it with the next -pf release and if your report is confirmed, I'll either consider changing the versioning scheme or creating an AUR package for virtualbox-pf-modules.

At any rate, if you can and want create it yourself I'll happily compile it and include it in the [pfkernel] repo, just like nvidia-pf.

travnick commented on 2012-10-25 09:42

Shouldn't this package create folders in /usr/lib/modiles like: 3.6.3-1-pf instead of 3.6-pf?

When kernel updates, old modules (eg virtualbox) still are there and make conflicts, they are not loadable
"ERROR: could not insert 'vboxdrv': Exec format error
ERROR: could not insert 'vboxnetflt': Exec format error
ERROR: could not insert 'vboxnetadp': Exec format error"

But you cannot simply create new ones with dkms (virtualbox-hook), because the old one exist in kernel modules folder ...
"Module vboxhost/4.2.2 already installed on kernel 3.6-pf/x86_64"

nous commented on 2012-10-22 00:15

It's a known and discussed issue (search previous comments); just enable DVB without any drivers.

neitrino commented on 2012-10-21 18:29

Hmm..
1. pacman -Syu
2. yaourt -S linux-rt
Localmod

and see:

DEPMOD 3.6-pf

-> Compressing package...
==> Starting package_linux-pf-headers()...
cp: cannot stat ‘include/config/dvb/*.h’: No such file or directory
==> ERROR: A failure occurred in package_linux-pf-headers().

nous commented on 2012-10-20 07:13

Why don't you just put "echo cfq >| /sys/block/sda/queue/scheduler" in /etc/rc.local? CFQ is a quite large chunk of code to build into kernel, and some people (me included) use linux-pf on memory-restrained boxes too.

That said though, it would make more sense to make to make NO-OP the default I/O scheduler and leave all other modularized, but we're addressing the silent majority here who just want their kernels to boot.

nous commented on 2012-10-20 07:12

Why don't you just put "echo cfq >| /sys/block/sda/queue/scheduler" in /etc/rc.local? CFQ is a quite large chunk of code to build into kernel, and some people (me included) use linux-pf on memory-restrained boxes.

That said though, it would make more sense to make to make no-op the default I/O scheduler and leave all other modularized, but we're addressing the silent majority here who just want their kernels to boot.

3V0LU710N commented on 2012-10-19 13:18

@nous: I've found if I want to use CFQ instead of BFQ as my main disk scheduler, I need to put in the mkinitcpio.conf modules section "cfq_iosched". Shouldn't it be better to have CFQ compiled as a "built-in" option in the kernel (like linux-ck does with CFQ and BFQ), so, we just needed to put "elevator=cfq" in the grub cmdline?

Cheers and keep up the good work...

nous commented on 2012-10-04 23:45

Aufs3 isn't yet ready for 3.6; therefore it's disabled on configs.

martadinata666 commented on 2012-10-04 12:20

i already got to 3.6

you can edit the pkgbuild to patch the pf-patch and the console-level, and logo-linux patch only...
Build success here, in my try... using another patch like i2c, channel-negative-one-maxim patch will get reject :)

PLum commented on 2012-10-04 10:57

When You find some time please update to new 3.6.1-pf :-), thx

nous commented on 2012-09-20 05:21

I think I also had some section mismatches but building continued. All binaries are available at the pfkernel repo.

martadinata666 commented on 2012-09-17 10:08

well pretty weird

....
LD arch/x86/boot/setup.elf
OBJCOPY arch/x86/boot/setup.bin
BUILD arch/x86/boot/bzImage
Setup is 17040 bytes (padded to 17408 bytes).
System is 2749 kB
CRC 4470713b
Kernel: arch/x86/boot/bzImage is ready (#1)
WARNING: modpost: Found 3 section mismatch(es).
To see full details build your kernel with:
'make CONFIG_DEBUG_SECTION_MISMATCH=y'
make[1]: *** [__modpost] Error 1
make: *** [modules] Error 2
==> ERROR: A failure occurred in build().
Aborting...

J4913 commented on 2012-09-17 09:03

Fixed in the update, thanks.

nous commented on 2012-09-17 07:09

3.4.5-1 is up for grabs.

post-factum commented on 2012-09-17 04:48

Try to change

depmod -b "$pkgdir" -F System.map "$_kernver"

to

depmod -b "$pkgdir/usr" -F System.map "$_kernver"

J4913 commented on 2012-09-16 19:21

I'm getting the same error that wind1r commented about:

INSTALL sound/soundcore.ko
DEPMOD 3.5-pf
ERROR: could not open directory /home/j/linux-pf/pkg/linux-pf/lib/modules/3.5-pf: No such file or directory
FATAL: could not search modules: No such file or directory
==> ERROR: A failure occurred in package_linux-pf().
Aborting...

And similarly, the modules are in usr/lib/ (as, it seems, they should be). I don't know enough about depmod and stuff to know what other information might be useful.

post-factum commented on 2012-09-16 12:37

@leodag: do not hurry nous — he won't release updated PKGBUILD without me :).

leodag commented on 2012-09-14 23:42

upstream's 3.5.4 was released :)

nous commented on 2012-09-04 20:44

Wtf... Tobias introduced new thingamajigs into his PKGBUILD.

walkindude commented on 2012-09-04 20:12

%.*}.3-3 ?

walkindude commented on 2012-09-04 20:11

%.*}.3-3 ?

nous commented on 2012-08-18 19:15

Sorry guys, it was a patch I forgot to remove from 3.5.1...

martadinata666 commented on 2012-08-18 17:33

@leodag i compiled by comment the amifritz section,,, i also get that error

leodag commented on 2012-08-18 17:14

When installing linux-pf 3.5.2-1, I get this:

<many, many patches>
==> Patching for restless souls
patching file net/wireless/chan.c
Hunk #1 succeeded at 84 (offset 35 lines).
Hunk #2 succeeded at 136 (offset 57 lines).
patching file kernel/printk.c
Hunk #1 succeeded at 61 (offset 3 lines).
patching file drivers/isdn/hardware/mISDN/avmfritz.c
Reversed (or previously applied) patch detected! Skipping patch.
3 out of 3 hunks ignored -- saving rejects to file drivers/isdn/hardware/mISDN/avmfritz.c.rej
==> ERROR: A failure occurred in build().
Aborting...
The build failed.

I'm using packer as a pacman wrapper, but almost always building linux-pf with it works.

Anonymous comment on 2012-07-18 23:47

Wow, worked with the latest update. Sorry.

nous commented on 2012-07-18 20:42

Strange, are you using a pacman/makepkg wrapper to build?

Anonymous comment on 2012-07-18 03:57

I getting error:

INSTALL /var/abs/local/linux-pf/pkg/linux-pf/lib/firmware/yamaha/ds1e_ctrl.fw
DEPMOD 3.4-pf
ERROR: could not open directory /var/abs/local/linux-pf/pkg/linux-pf/lib/modules/3.4-pf: No such file or directory
FATAL: could not search modules: No such file or directory

While:

ls /var/abs/local/linux-pf/pkg/linux-pf/usr/lib/modules/3.4-pf
extramodules kernel modules.builtin modules.order

nous commented on 2012-06-22 05:55

I'll test on my personal box, along with the alleged TRANSPARENT_HUGEPAGE-breaks-TOI issue. If it works and does not break other things, I'll apply it in the PKGBUILD.

Huulivoide commented on 2012-06-21 19:51

It would seem it is needed to run 'make headers_install'
for aufs3 headers to be installed correctly. Simple cp
will result in non-working unusable headers. Currently
aufs3-util package cannot be built with linux-pf due to
this very reason, even after using the CPPFLAGS hack
provided in comments. Please take a look at this mailinglist
conversation, expecially the 6th message.
http://comments.gmane.org/gmane.linux.file-systems.aufs.user/3241

Anonymous comment on 2012-06-19 09:43

My internet connection now is not cool enough to download and compile kernel, but about a year ago i used kernel without initrd and modules and it was booting faster.
I have 8gb of RAM.

nous commented on 2012-06-18 18:20

On my laptop zram boot parameter is 'zram.num_devices=2', not zram.zram_num_devices.

nous commented on 2012-06-18 17:54

If you're low on physical RAM, you might want to append the keywords 'zcache' and 'zram.zram_num_devices=N' to the kernel boot parameters, where N your CPU cores, in order to activate zcache and allocate all memory to the CPUs for zram. Also, install zramswap from the AUR and add zramswap to /etc/rc.conf DAEMONS array.

The results are impressive.

nous commented on 2012-06-18 17:34

Vekt: you realize that you won't probably get a measurable speed-up, don't you? Please, apply the following patch to PKGBUILD and test your new kernel with bootchart. If there's significant increase, I'll implement it (and probably build one kernel for my laptop too).

--- PKGBUILD 2012-06-18 01:06:09.000000000 +0300
+++ PKGBUILD.koko 2012-06-18 20:29:22.762549123 +0300
@@ -183,2 +183,3 @@
echo " <O> make oldconfig"
+ echo " <L> make localyesonfig"
echo " <ENTER> to skip configuration and start compiling"
@@ -196,2 +197,4 @@
;;
+ l) make localyesconfig
+ ;;
default)

Anonymous comment on 2012-06-18 09:45

Please, add a posibility to compile it with "make localyesconfig". I wouldn't like to have any modules in my system in order to make it boot faster, but there is no possibility to do "make localyesconfig"...
Thank you.

J4913 commented on 2012-06-15 19:14

Ah, great. Thanks.

nous commented on 2012-06-15 19:00

Not strange at all, see https://bbs.archlinux.org/viewtopic.php?pid=1116326#p1116326 and update to 3.4.1-4.

J4913 commented on 2012-06-15 11:29

I got an e-mail about a comment here, but it seems to have been removed. I'm on the 3.4 stock kernel now, and it's working fine - strange.

nous commented on 2012-06-14 16:33

Google Earth runs fine here on 3.4-pf 64bit.

J4913 commented on 2012-06-14 10:58

I seem suddenly to have lost the ability to run 32-bit programs with this (on a 64-bit system), both with the repo versions and when I compile myself. The stock kernel works fine, but it's not on 3.4 yet, so I don't know if it's 3.4 or just pf. Any ideas?

nous commented on 2012-06-12 20:48

When asked by the PKGBUILD, make menuconfig, go to "Power management and ACPI options" and deselect "Enhanced Hibernation (TuxOnIce)".

Burning_aces commented on 2012-06-12 17:21

@nous: Sorry, I meant how do I build the kernel without Tuxonice

nous commented on 2012-06-12 15:33

Burning_aces: see http://chakra-linux.org/wiki/index.php/Tuxonice#Disabling_tuxonice

Burning_aces commented on 2012-06-12 14:44

How do I disable TOI?

nous commented on 2012-06-12 05:55

Disable zcache and zram (they're in staging), as they currently break TOI.

Burning_aces commented on 2012-06-11 22:04

I keep getting the following error:

CC kernel/power/tuxonice_compress.o
kernel/power/tuxonice_compress.c: In function ‘toi_compress_write_page’:
kernel/power/tuxonice_compress.c:183:25: error: too few arguments to function ‘memory_bm_test_bit_index’
In file included from kernel/power/tuxonice.h:24:0,
from kernel/power/tuxonice_compress.c:19:
kernel/power/power.h:339:12: note: declared here
kernel/power/tuxonice_compress.c:183:25: warning: left-hand operand of comma expression has no effect [-Wunused-value]
make[2]: *** [kernel/power/tuxonice_compress.o] Error 1
make[1]: *** [kernel/power] Error 2
make: *** [kernel] Error 2
==> ERROR: A failure occurred in build().
Aborting...

nous commented on 2012-06-10 16:20

Erm... it should be 'hg update 40', not 'revert'. Glad it worked otherwise though.

Anonymous comment on 2012-06-10 14:48

wow, it compiled like a charm!
ty, nous!

Anonymous comment on 2012-06-10 13:43

@nous
unfortunatly:
$ hg revert 40
40: no such file in rev 015486e5e37d

trying to compile with "_minor=3 \n _pfrel=6"
Thank you

nous commented on 2012-06-10 11:43

Scratch that, it won't probably work as it's a minor version bump, not just a patchlevel one.

Do the following:
$ hg clone https://bitbucket.org/nous/linux-pf
$ cd linux-pf
$ hg revert 40

And you'll have 3.3.6-pf ready for compilation.

nous commented on 2012-06-10 11:18

Just download this tarball and edit the PKGBUILD as follows:
_minor=3
_pfrel=6

and build with 'makepkg --skipinteg'

Anonymous comment on 2012-06-10 11:05

Hi, nous!
Unfortunatly I updated to 3.4 from 3.3 mindlessly this time, and now i'm having some heavy mindfuck with radeon video.

how could i get a previos version of your excelent package for Archlinux?
It would be awesome, if you share the PKGBUILD for v.3.3!

Hope for you reply!
Thanks

nous commented on 2012-06-05 11:15

You might succeed compiling nvidia-pf by editing its PKGBUILD and setting good and bad kernver.

lockheed commented on 2012-06-05 11:11

I just upgraded linux-pf before reading this thread and before discovering nvidia-pf is not compatible with it.
I have not a very nice feeling that once I reboot, I am screwed.:)

nous commented on 2012-06-05 11:06

3.4.1-1 is up but not for grabs (unless you build yourself and don't use tuxonice); it croaks when compiling tuxonice_compress.o, so you shouldn't remove 3.3.6 just yet. Please wait for 3.4.1-2.

nous commented on 2012-06-05 11:05

3.4.1 is up but not for grabs; it croaks when compiling tuxonice_compress.o, so you shouldn't remove 3.3.6 just yet.

nous commented on 2012-06-05 07:12

3.4.1 is up for grabs.

nous commented on 2012-06-01 12:16

The current version is 3.3.6; please update your database. What's more, p4 somehow fails to build with my repobuild script on i686. You'll have to use either p3 or the generic one, sorry.

cuihao commented on 2012-06-01 11:54

error: failed retrieving file 'linux-pf-p4-3.3.4-1-i686.pkg.tar.xz' from dl.dropbox.com : The requested URL returned error: 404

post-factum commented on 2012-04-27 16:31

You are welcome to visit pf-kernel community forum: http://pf.natalenko.name/forum/

msx commented on 2012-04-16 03:46

Hi all, adding this repo to your /etc/pacman.conf will let you download an already compiled linux-pf kernel that matches your CPU architecture:

[pfkernel]
# Generic and optimized binaries of the ARCH kernel patched with BFS, TuxOnIce, BFQ, Aufs3
# linux-pf, kernel26-pf, gdm-old, nvidia-pf, nvidia-96xx, xchat-greek, arora-git
Server = http://dl.dropbox.com/u/11734958/$arch

nous commented on 2012-04-13 20:49

I could, but it's not practical and difficult PKGBUILD-wise, as it would require several changes. It's much easier for you to configure and compile the package once, and then use the "Hit <Y> to use your running kernel's config" PKGBUILD option at every subsequent compile (that's how I do it for my laptop tailored linux-pf).

To considerably reduce compile time, you could install modprobed_db from the AUR (http://aur.archlinux.org/packages.php?ID=41689) and use the "Hit <L> to run 'make localmodconfig'" option the _first_ time you compile your custom kernel.

shadowed87 commented on 2012-04-13 14:03

hello, thanks for this awesome pkg...a question: is it possible to add in the repository a kernel with these config (for laptops)

Notebook / mobile device:
300 Hz
Tickless
Full preemption
CPU frequency scaling, ondemand default

taken here: http://ck.kolivas.org/patches/bfs/bfs-configuration-faq.txt

nous commented on 2012-04-09 05:47

ericab: In both AUR and repo, nvidia-pf is updated; please retry.

Anonymous comment on 2012-04-09 01:53

arg;

==> Finished making: linux-pf 3.3.1-2 (Sun Apr 8 18:08:53 PDT 2012)

==> Continue installing linux-pf ? [Y/n]
==> [v]iew package contents [c]heck package with namcap
==> ---------------------------------------------------
==>

loading packages...
resolving dependencies...
looking for inter-conflicts...
error: failed to prepare transaction (could not satisfy dependencies)
:: nvidia-pf: requires linux-pf<3.3

nous commented on 2012-04-07 23:01

Latest linux-pf and nvidia-pf for all CPUs are up for grabs at [pfkernel]

nous commented on 2012-04-06 05:30

That version is too old, try 'pacman -Syw linux-pf-atom'

gaougalos commented on 2012-04-05 10:58

Is there a problem on this package (linux-pf-atom-3.2.7-2-i686.pkg.tar.xz) on your repo?

$ pacman -Sw linux-pf-atom
resolving dependencies...

Targets (1): linux-pf-atom-3.2.7-2

Total Download Size: 39.68 MiB

Proceed with download? [Y/n] y
:: Retrieving packages from pfkernel...
error: failed retrieving file 'linux-pf-atom-3.2.7-2-i686.pkg.tar.xz' from dl.dropbox.com : The requested URL returned error: 404
warning: failed to retrieve some files from pfkernel
error: failed to commit transaction (download library error)
Errors occurred, no packages were upgraded.

Max-P commented on 2012-04-05 00:28

@gaougalos

Your error have nothing to do with this package. linux-pf is not in the repo, you can't just install it with pacman. The package you are trying to install come from a [pfkernel] repo YOU added. I just built it, and it works fine...

gaougalos commented on 2012-04-04 21:42

$ pacman -Sw linux-pf-atom
resolving dependencies...

Targets (1): linux-pf-atom-3.2.7-2

Total Download Size: 39.68 MiB

Proceed with download? [Y/n] y
:: Retrieving packages from pfkernel...
error: failed retrieving file 'linux-pf-atom-3.2.7-2-i686.pkg.tar.xz' from dl.dropbox.com : The requested URL returned error: 404
warning: failed to retrieve some files from pfkernel
error: failed to commit transaction (download library error)
Errors occurred, no packages were upgraded.

nous commented on 2012-04-03 21:04

Sorry guys, 3.2.7-4 was a working draft for 3.3 and was uploaded by mistake. 3.2.7-3 still is the latest and greatest.

ryosei commented on 2012-04-03 12:48

and here...

hermes14 commented on 2012-04-03 12:48

And here...

lockheed commented on 2012-03-31 15:49

And here.

Anonymous comment on 2012-03-31 15:39

The same problem here:

==> Starting package_linux-pf-headers()...
cp: cannot stat `include/aufs3': No such file or directory
==> ERROR: A failure occurred in package_linux-pf-headers().
Aborting...

Anonymous comment on 2012-03-31 04:30

I have a problem with 3.2.7-4 and stock arch config

==> Starting package_linux-pf-headers()...
cp: cannot stat `include/aufs3': No such file or directory
==> ERROR: A failure occurred in package_linux-pf-headers().
Aborting...
==> ERROR: Makepkg was unable to build linux-pf.

nous commented on 2012-03-31 00:10

You can't, it's obsoleted. The latest binaries are at 3.2.7-3. Try 'pacman -Sy linux-pf-core2'. Alternatively, download from the AUR and edit $pkgrel in the PKGBUILD.

Anonymous comment on 2012-03-30 12:35

Where can we get linux-pf-core2-3.2.7-2-i686.pkg.tar.xz from?

sudo pacman -S linux-pf-core2
resolving dependencies...
looking for inter-conflicts...

Targets (1): linux-pf-core2-3.2.7-2

Total Download Size: 39.66 MiB
Total Installed Size: 56.85 MiB

Proceed with installation? [Y/n] y
:: Retrieving packages from pfkernel...
error: failed retrieving file 'linux-pf-core2-3.2.7-2-i686.pkg.tar.xz' from dl.dropbox.com : The requested URL returned error: 404
warning: failed to retrieve some files from pfkernel
error: failed to commit transaction (download library error)
Errors occurred, no packages were upgraded.

Anonymous comment on 2012-03-25 11:49

That explains it. I have a single-core CPU.

nous commented on 2012-03-16 13:34

I took the time to compile a wiki page for linux-pf: https://wiki.archlinux.org/index.php/Linux-pf

nous commented on 2012-03-12 22:40

It works. TuxOnIce people, add this to your /etc/hibernate/tuxonice.conf: "ProcSetting no_multithreaded_io 1".

post-factum commented on 2012-03-12 21:22

Possible temporary fix: http://lists.tuxonice.net/pipermail/tuxonice-devel/2012-March/007186.html
Give it a try.

post-factum commented on 2012-03-08 23:23

Here is my call for help with TOI+BFS issue: http://lists.tuxonice.net/pipermail/tuxonice-devel/2012-March/007183.html

Soul_Est commented on 2012-03-06 17:53

Just saw on the homepage that the pf-kernel patch has been updated to 3.2.6.

nous commented on 2012-03-03 18:48

TOI scans the partitions for a hibernation image, so it should even work without any directives.

technic93 commented on 2012-03-02 17:33

I have resume=swap:/dev/sdxx instead if resume=/dev/sdxx, could this cause the error?

Anonymous comment on 2012-03-02 14:58

What exactly is error -22 by the way?

I have the "resume" hook enabled and "lzo" in the MODULES list of mkinitcpio.conf.
I am using pm-utils with the UPower dbus scripts.
I have a file in /etc/pm/sleep.d that sets /sys/power/tuxonice/compression/algorithm to "lzo".
I have another file in /etc/pm/config.d that sets the variable "SLEEP_MODULE" to "tuxonice".
I also passed "resume=/dev/sdxx" to my kernel, with "sdxx" being my swap partition.
Hope that helps.

nous commented on 2012-03-01 06:00

Can't confirm. I built TOI into the kernel and it fails with -22.

technic93 commented on 2012-02-29 14:14

Good news, I'll try it too.

Anonymous comment on 2012-02-29 13:35

Good news: I hibernated then blacklisted lzo with the kernel command line. The system then fails to resume with an error "failed to initialize compressor".
That means the lzo compressor is indeed being used.

Anonymous comment on 2012-02-29 09:13

@technic93:

I am not completely sure if lzo is used by TuxOnIce or not, but both /sys/power/tuxonice/compression/algorithm and the /etc/hibernate config file are set to use lzo. Plus hibernation is quite faster than when using the built-in function.

I will test further to see if lzo is being used or not.

technic93 commented on 2012-02-29 07:17

Hi deathfantasy,
is LZO actually used by tuxonice?

Anonymous comment on 2012-02-29 06:46

nous: I built LZO as a module.

nous commented on 2012-02-27 17:55

deathfantasy: Please elaborate: did you also build the compressor in-kernel? Which one, lzo or lzf?

Anonymous comment on 2012-02-27 12:48

I built TuxOnIce into the kernel instead of as a module, and it works for me.

nous commented on 2012-02-24 21:42

I've personally tested lzf and it fails too.

technic93 commented on 2012-02-24 15:39

I guess we could add lzf module to the config.

technic93 commented on 2012-02-23 20:12

thank you, I've not found this in wiki.

nous commented on 2012-02-23 16:02

technic93: Please see my comment on 13 Nov 2011 13:59:23. You must add your allocator of choice to /etc/mkinitcpio.conf, MODULES array [e.g. toi_swap, toi_compress (which still fails to resume with lzo; compressor must be disabled) and optionally toi_userui].

technic93 commented on 2012-02-23 15:37

Hi,
Why neither CONFIG_TOI_FILE nor CONFIG_TOI_SWAP is on by default?

nous commented on 2012-02-07 17:46

It's not PKGBUILD related; it downloads ok here, but I use wget instead of curl at the DLAGENTS array in /etc/makepkg.conf:

DLAGENTS=('ftp::/usr/bin/wget -c --passive-ftp -t 3 --waitretry=3 -O %o %u'
'http::/usr/bin/wget -c -t 3 --waitretry=3 -O %o %u'
'https::/usr/bin/wget -c -t 3 --waitretry=3 --no-check-certificate -O %o %u'
'rsync::/usr/bin/rsync -z %u %o'
'scp::/usr/bin/scp -C %u %o')

The relevant option in wget for https is '--no-check-certificate'.

leafonsword commented on 2012-02-07 12:36

-> 正在下载 interactive-governor.tar.gz...
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
0 0 0 0 0 0 0 0 --:--:-- 0:00:01 --:--:-- 0curl: (77) error setting certificate verify locations:
CAfile: /etc/ssl/certs/ca-certificates.crt
CApath: none

==> 错误: 无法下载 interactive-governor.tar.gz
正在放弃...

nous commented on 2012-01-27 09:04

You need to update the nvidia-pf package first, the one from the AUR is fine.

Anonymous comment on 2012-01-27 01:54

thanks @nous, @crush
that issue is fixed now, but now a new one !

see it here: http://pastebin.com/Z3hJQ0vR

nous commented on 2012-01-26 20:32

I had fixed it quite a while ago (see https://bitbucket.org/nous/linux-pf/changeset/200b757b9b9c), but an older version of the PKGBUILD somehow slipped in. Uploaded fixed version again.

liviucmg commented on 2012-01-26 19:46

@ericab I'm getting the same thing. I suppose the default kernel doesn't come with the config options. You can use <N> without a problem, thought.

Anonymous comment on 2012-01-26 18:58

anyone able to help with this ??
---------------------------------


=======================================================
==> You might be prompted below for some config options
=======================================================

==> Hit <Y> to use your running kernel's config
(needs IKCONFIG and IKCONFIG_PROC)
==> Hit <L> to run 'make localmodconfig'
==> Hit <N> (or just <ENTER>) to build an all-inclusive kernel like stock -ARCH
(warning: it can take a looong time)

Y
==> running 'sudo modprobe configs'
==> ERROR: A failure occurred in build().
Aborting...
==> ERROR: Makepkg was unable to build linux-pf.
==> Restart building linux-pf ? [y/N]
==> ---------------------------------
==>

nous commented on 2012-01-24 14:47

Thanks for the feedback on compiling.
Your second couple of lines should be reversed; I'll put it in the next update.

Anonymous comment on 2012-01-24 12:41

Had the same error as user crush. Just fetched the package again and it compiles now.

Also a note on the PKGBUILD: makepkg automatically decompresses every file in ${source}.
The following lines can be changed:
_pfpatchname="patch-${_basekernel}.${_pfrel}${_kernelname}.bz2" ->
_pfpatchname="patch-${_basekernel}.${_pfrel}${_kernelname}"

${_pfpatchhome}${_pfpatchname}.bz2) # the -pf patchset ->
${_pfpatchhome}${_pfpatchname}) # the -pf patchset

bzip2 -dc ${srcdir}/${_pfpatchname} | patch -Np1 ->
patch -Np1 -i ${_pfpatchname}
It doesn't really make any real difference though.

Thanks for the package !

nous commented on 2012-01-24 11:34

Beats the crap out of me. This morning I tested it and worked OK, then it broke. I took the CVE patch from linux.org and it worked again...

liviucmg commented on 2012-01-24 09:19

I'm getting the following error:

$ aurget -S linux-pf
[...]
==> Fetching and applying aufs3 patches
[...]
patching file security/security.c
patching file fs/proc/base.c
Hunk #1 succeeded at 194 (offset -4 lines).
Hunk #2 succeeded at 205 (offset -4 lines).
Hunk #3 succeeded at 214 (offset -4 lines).
Hunk #4 FAILED at 699.
Hunk #5 succeeded at 821 (offset 64 lines).
Hunk #6 succeeded at 830 (offset 64 lines).
Hunk #7 succeeded at 862 (offset 64 lines).
Hunk #8 succeeded at 882 (offset 64 lines).
1 out of 8 hunks FAILED -- saving rejects to file fs/proc/base.c.rej
==> ERROR: A failure occurred in build().
Aborting...

It worked fine just yesterday before the update. Any ideas?

nous commented on 2012-01-24 07:17

hyperion:[nous]:/tmp% ./mempodipper
===============================
= Mempodipper =
= by zx2c4 =
= Jan 21, 2012 =
===============================

[+] Opening socketpair.
[+] Executing child from child fork.
[+] Waiting for transferred fd in parent.
[+] Opening parent mem /proc/23140/mem in child.
[+] Sending fd 5 to parent.
[+] Received fd at 5.
[+] Assigning fd 5 to stderr.
[+] Reading gpasswd for exit@plt.
[+] Resolved exit@plt to 0x4018a0.
[+] Calculating gpasswd padding.
[+] Seeking to offset 0x401890.
[+] Executing gpasswd with shellcode.
sh-4.2#


Please, everyone UPDATE to 3.2.1-5

http://blog.zx2c4.com/749

nous commented on 2012-01-24 07:14

Please, everyone UPDATE to 3.2.1-5!

http://blog.zx2c4.com/749

nous commented on 2012-01-22 21:36

@vadik I'm waiting for linux-ARCH to change that before I do it too.

Anonymous comment on 2012-01-22 15:21

kmod replaces module-init-tools.

http://www.archlinux.org/news/kmod-replaces-module-init-tools/

Anonymous comment on 2012-01-20 09:22

Thank you Nous.
After updating works .

nous commented on 2012-01-19 22:02

It was not the headers after all. Somehow, /usr/src/linux-3.2-pf/vmlinux was getting stripped. I recompiled a generic x86_64 linux-pf and now it works (I had the same problem here compiling nvidia) - virtualbox compiles OK too. You can download it from the repo, the optimized binaries are on their way.

Anonymous comment on 2012-01-19 18:44

The problem may be in the headers.
Just to test I made a simple cp -rv /usr/src/linux-3.2.1-1-ARCH /usr/src/linux-3.2-pf and he compiled but the module did not work .

nous commented on 2012-01-19 18:33

Googling around, I saw some talk (mainly by archers) with the same problem, with 3.1.9-ARCH (3.2 too), related to libarchive. I'll investigate more; in the meanwhile, if anyone finds a solution please post.

nous commented on 2012-01-19 18:31

Googling around, I saw some talk (mainly by archers) with the same problem, with 3.1.9 and 3.2, related to libarchive. I'll investigate more; in the meanwhile, if anyone finds a solution please post.

post-factum commented on 2012-01-19 18:20

The issue with TOI as module is fixed by Nigel but not pushed to git repo yet. I expect it to be included in 3.2.2-pf.

@lgf: I've got no idea :/. Probably, some problem with headers?

Anonymous comment on 2012-01-19 17:25

After upgrading to 3.2 and have got an error when compiling the modules VirtualBox and catalyst, in both cases is the same
(FATAL: vmlinux is truncated. Sechdrs [i]. Sh_offset = 18446744073709551615> sizeof (* hrd) = 64)
if it goes to boot the default kernel even this also being the 3.2 modules compiled without problem.

post-factum commented on 2012-01-19 00:33

I'll mail him.

nous commented on 2012-01-18 23:11

Somehow, the freeze_kernel_threads and in_suspend symbols are not passed to the kernel image, with tuxonice as all-modules. It's not that important to devote time to it. Maybe just email Nigel?

post-factum commented on 2012-01-18 22:21

If there are any ideas how to fix it, you are welcome.

nous commented on 2012-01-18 11:29

"<*> Enhanced Hibernation (TuxOnIce)" seems to do the trick. The rest of the submenu are again modules.

nous commented on 2012-01-18 07:09

The error is not PKGBUILD related; it only appears with tuxonice configured as modules. I just tested a different config with built-in tuxonice and it compiled OK.

nous commented on 2012-01-18 07:09

The error is not PKGBUILD related; it only appears with tuxonice configured as modules. I just tested a different config with built-in tuxonice and it compiled OK.

nous commented on 2012-01-18 07:08

The error is not PKGBUILD related; it only appears with tuxonice configured as modules. I just tested a different config with built-in tuxonice and it compiled OK.

nous commented on 2012-01-18 07:01

The error is not PKGBUILD related; it only appears with tuxonice configured as modules. I just tested a different config with built-in tuxonice and it compiled OK.

hermes14 commented on 2012-01-17 23:18

Same error here

nixtrian commented on 2012-01-17 20:22

build halts with:

BUILD arch/x86/boot/bzImage
Setup is 17036 bytes (padded to 17408 bytes).
System is 2521 kB
CRC e0604fdd
Kernel: arch/x86/boot/bzImage is ready (#1)
ERROR: "freeze_kernel_threads" [kernel/power/tuxonice_core.ko] undefined!
ERROR: "in_suspend" [kernel/power/tuxonice_core.ko] undefined!
WARNING: modpost: Found 9 section mismatch(es).
To see full details build your kernel with:
'make CONFIG_DEBUG_SECTION_MISMATCH=y'

nous commented on 2012-01-17 18:15

Updated and uploaded (untested, as usual :)
Will start batch building now, expect the repos to be populated by tommorow morning EET.

nixtrian commented on 2012-01-17 16:03

3.2 released http://pf.natalenko.name/sources/3.2/

Rolinh commented on 2012-01-09 08:25

Don't flag it out of date because it is not the same version as mainline.
The latest pf-kernel patch is 3.1.6 at this point, as you can see here: http://pf.natalenko.name/sources/3.1/

nous commented on 2011-11-28 22:14

LinuxIMQ is back in. People who want it will have to compile it themselves though...

nous commented on 2011-11-28 21:47

LinuxIMQ is back in. If you want it though, you'll have to configure it in yourselves.

nous commented on 2011-11-21 21:53

Cheers, bardo. I've used a few diff tools from time to time but none quite impressed me. Although I must admit I'm not their most expert user. I'll have a look at meld too.

bardo commented on 2011-11-21 19:09

@nous: might not be the holy grail, but I suggest using meld when you need to compare a newer PKGBUILD to your customizations, it saves a lot of headaches. You can also try vimdiff if you are more frugal or dislike gtk. Or just "svn diff" the latest official build with the version you synchronized to, discover what changed since then and decide what to integrate in your package. HTH :)

nous commented on 2011-11-20 20:32

@kalpik: The ones from /lib/modules/3.1-pf/kernel/kernel/power [_bio, _compress (lzo still fails here), _core, _file/_swap and _userui]
I believe that specifying one (e.g. tuxonice_swap) which depends on _bio and _core would do, but haven't tested as I don't use an initrd.

@myra: That's about how I had it till recently:
[[ -e include/config/dvb/ ]] && cp include/config/dvb/*.h ${pkgdir}/usr/src/linux-${_kernver}/include/config/dvb/
The thing is, I _must_ synchronize with the official linux-ARCH; their config, patches/fixes and packaging on every minor bump. It's very tedious and tiring for a maintainer to compare line-by-line two huge PKGBUILDs and merge them into one. This time I forgot to add the aforementioned line, which resulted in your error (mine too!) But as I explained, I decided to leave it as it was. You see, for people running custom/tailored kernels, selecting <M> for DVB for Linux at "Device Drivers/Multimedia support" and leaving everything else out, creates just the necessary dvb.h without any further demands.

Anonymous comment on 2011-11-20 19:28

nous:

About the dvb headers, would it not be as easy to add this to the PKGBUILD

# http://bugs.archlinux.org/task/11194
mkdir -p "${pkgdir}/usr/src/linux-${_kernver}/include/config/dvb/"
if [ -a include/config/dvb/*.h ] ; then
cp include/config/dvb/*.h "${pkgdir}/usr/src/linux-${_kernver}/include/config/dvb/*.h
fi

Seems as though that would save building the modules. I haven't installed my build yet cause I don't want to rebuild nvidia until tonight, but the packaging succeeds this way.

Myra

kalpik commented on 2011-11-20 04:26

"tuxonice is no longer built-in; You MUST add the relevant modules in your /etc/mkinitcpio.conf MODULES array"

What modules do I need to add?

nous commented on 2011-11-19 19:49

Since the headers package section is taken as-is from linux-ARCH, which has almost all options 'moduled', it expects the dvb headers in place. I too have had this error in my tailored laptop config and circumvented it simply by enabling as module the dvb submenu in drivers/multimedia.

lucak3 commented on 2011-11-19 11:58

Sorry for deleted message, i found the real problem:
- _kernver too has KARCH problem

lucak3 commented on 2011-11-19 11:55

Another one, for my package (nvidia-pf):
- "Unable to determine the target kernel version." returned :(

lucak3 commented on 2011-11-19 11:50

There are some problems in package_linux-pf-headers():
- Missing include/config/dvb/*.h files (i did build again to confirm it wasn't a temp issue)
- Used KARCH variable defined only in package_linux-pf()

nous commented on 2011-11-17 22:43

Thanks to Julian who caught an incomplete path in the aufs3 branch, 3.1.1-7 fetches the aforementioned patch too.

nous commented on 2011-11-15 21:04

My bad, guys. See https://bbs.archlinux.org/viewtopic.php?pid=1017107#p1017107

I also reverted to the good ol' 3.X-pf versioning scheme for modules, as having to rebuild nvidia or other binary blobs when pf is incremented is redundant.

ryosei commented on 2011-11-15 20:00

I've got the same issue and I'm not using yaourt...
I do "makepkg -i" get an error then I do
"pacman -U -- -3.1.1-5-x86_64.pkg.tar.xz linux-pf-headers-3.1.1-5-x86_64.pkg.tar.xz"
and i get an error again...
it seems to me that the name of the package don't get catch by the first packaging operation makepkg does...
I don't know much about split packages so I can't tell what is the cause but surely something seems wrong...

Anonymous comment on 2011-11-15 19:34

@nous,
how can i get this package installed then ? ive got the same problem as gfang

nous commented on 2011-11-15 13:52

As of 3.1.1-5, linux-pf is a split package: linux-pf and linux-pf-headers.

@gfang It's yaourt's issue; perhaps it's got something to do with the oodoo PKGBUILD does to get past the AUR parser...

Anonymous comment on 2011-11-15 00:39

Compile completes successfully but error on install...
"error: missing package name in /tmp/yaourt-tmp-root/PKGDEST.I3H/-3.1.1-5-i686.pkg.tar.xz"
"error: '/tmp/yaourt-tmp-root/PKGDEST.I3H/-3.1.1-5-i686.pkg.tar.xz': invalid or corrupted package"

Reproduced on both i686 and x86_64 machines.

nous commented on 2011-11-14 10:16

@Cdh The kernel CPU configuration menu doesn't offer i[357] as an option. Therefore, I won't force things unnecessarily. However, you can set your KCFLAGS and KCPPFLAGS per your heart's desire at the 'make' directive in the PKGBUILD (and cross your fingers :) See https://bbs.archlinux.org/viewtopic.php?pid=977942#p977942 for an explanation as to why I don't.

@shahid and strephil: I'll let you know, thanks for stepping up.

haagch commented on 2011-11-13 23:57

I guess core2 should work for i3/i5/i7 first and second generation, but will there be separate packages?

krevedko commented on 2011-11-13 19:59

noun, you can review my candidature for new maintainer.
I have several self-written pkgbuilds here.

nous commented on 2011-11-13 17:52

The interactive governor patch breaks compilation - disabled.

Anonymous comment on 2011-11-13 17:18

Thanx!

nous commented on 2011-11-13 13:59

The interactive CPU governor is available as of 3.1.1-3! AUFS3 has not yet sync'ed with 3.1; this means that my precompiled repo packages which are being furiously compiled right now, will not sport it. Those who need it, will have to makepkg themselves. Also: tuxonice is no longer built-in; You MUST add the relevant modules in your /etc/mkinitcpio.conf MODULES array or your boxes won't resume.

nous commented on 2011-11-13 13:47

The interactive CPU governor is available as of 3.1.1-3! AUFS3 has not yet sync'ed with 3.1; this means that my precompiled repo packages which are being furiously compiled right now, will not sport it.

strephil commented on 2011-11-13 13:32

I know bash and Arch well and I could manage to maintain PKGBUILD.
I although have some experience in kernel compilation.

nous commented on 2011-11-12 22:31

Hi people. I created this package (starting with kernel26-pf) because I used both BFS and tuxonice, patching tediously my kernel, before I found out that post-factum had actually set up a git repository doing the same thing the right and elegant way. That was when I decided to provide a package here, to let other Archers benefit from his effort. That being said and as "real life" obligations keep squeezing my free time to extinction, I've decided to seek a devoted and capable maintainer for this package. I still use linux-pf exclusively on all my boxes and will do so, therefore I need someone to take good care of it. At the very least, one should have:
1. A firm grasp of bash shell scripting (for handling the PKGBUILD)
2. Adequate experience in linux generally and in Arch linux specifically
3. Some knowledge of kernel compilation tasks (e.g. patching, resolving gcc errors, understanding kernel configuration options etc)
I'll keep providing precompiled binaries in my repo, as building them is scripted and trivial once the PKGBUILD is updated and the 2 kernel configs from the stock -ARCH kernel are adapted to -pf every time the minor version number changes. I'll also help the new maintainer with the whys and wtfs of the PKGBUILD.

nous commented on 2011-11-12 22:24

AUFS3 is not yet ported to 3.1, thus the error. I'm sorry I uploaded an untested PKGBUILD, but my free time has become really scarce... I've uploaded a new one, with sanity check on aufs3 git. When aufs-3.1 becomes available, it should be pulled automagically.

nous commented on 2011-11-12 21:18

AUFS3 is not yet ported to 3.1, thus the error. I'm sorry I uploaded an untested PKGBUILD, but my free time has become really scarce... I've uploaded a new one, with sanity check on aufs3 git. When aufs-3.1 becomes available, it should be pulled automagically.

xF0E commented on 2011-11-12 14:26

ty guys, solved by removing aufs patch.

post-factum commented on 2011-11-12 14:01

@xF0E: or you may try to change this string in PKGBUILD:

git checkout origin/aufs3.0

to

git checkout origin/aufs3.1

and try to build again. It seems, that nous has forgotten about it.

post-factum commented on 2011-11-12 13:56

@xF0E: that function is defined by aufs3-loopback.patch file, you may checkout aufs3.0 and check yourself.

post-factum commented on 2011-11-12 13:55

@xF0E: loop_real_file function in drivers/block/loop.c is related to aufs and appears ONLY in aufs. So, you've patched pf-kernel with aufs before build. Please, revert aufs or disable it in PKGBUILD and try again.

xF0E commented on 2011-11-12 13:35

@post-factum

ty for your comment:

just some screens above LD error i found this:

drivers/block/loop.c: In function ‘loop_real_file’:
drivers/block/loop.c:661:32: error: ‘const struct super_operations’ has no member named ‘real_loop’
drivers/block/loop.c:662:33: error: ‘const struct super_operations’ has no member named ‘real_loop’
make[2]: *** [drivers/block/loop.o] Error 1
make[1]: *** [drivers/block] Error 2
make: *** [drivers] Error 2
make: *** Waiting for unfinished jobs....

this is not aufs relataed, since i don't checked aufs support before compilation start.

post-factum commented on 2011-11-12 12:22

@xF0E: try to find errors above.

post-factum commented on 2011-11-12 12:21

Yep, aufs seems to be broken.

xF0E commented on 2011-11-12 12:21

unfortunately i get this error on compiling the source every time i try to build the kernel:
LD net/built-in.o
==> ERROR: A failure occurred in build().
Aborting...
==> ERROR: Makepkg was unable to build linux-pf

prMoriarty commented on 2011-11-12 11:53

fs/aufs/super.c:758:2: error: unknown field ‘real_loop’ specified in initializer

prMoriarty commented on 2011-11-12 11:39

fs/aufs/super.c:758:2: error: unknown field ‘real_loop’ specified in initializer

nous commented on 2011-11-06 19:06

You are 100% right but I'm still using this on my laptop for its working lzo, so I'll allow myself a small "maintainer's prerogative"...

schivmeister commented on 2011-11-06 10:33

Please request a merge with linux-pf, else let's help remove packages with older naming schemes since they disturb the consistency of the repos and the distribution in general. The working state of a specific version is independent of the naming scheme.

nous commented on 2011-10-28 20:04

Thanks for pointing out, added git to makedepends.

prMoriarty commented on 2011-10-28 18:47

add to dependencies: "git"

nous commented on 2011-10-27 14:11

Please, be more specific: which patch?

The patches that could break the process are the aufs3 ones from git. They already are covered with the || operator, as you described (as a matter of fact, the moment I uploaded the package at least one wouldn't apply cleanly and its exit status was caught by that ||).

dflt commented on 2011-10-27 12:11

I'm getting build error from this pkg. It's because some patches apply partialy and exiting with status diffrent then 0. So you need to add at the end of every line that contains patch sth like this: || echo asdf, so the whole line should look like this:

patch -p1 blablabla ||echo asdf

(echo returns 0 as default and || means OR so if patch does not run succesly bash executes echo and echo returns 0 and everyone is happy)

nous commented on 2011-10-26 11:17

Gr1N: just hit <ENTER> at all prompts. But then again, you'll get the generic pf kernel which is also available at the repo and that kinda beats the purpose of self-compiling.

Anonymous comment on 2011-10-26 09:30

Before starting compile I see message with three paragraphs. Tell me please which one to choose?

post-factum commented on 2011-10-24 16:53

pf-kernel website is available via IPv6.

nous commented on 2011-10-23 11:44

Thanks for noticing. Moved to aufs3 from git.

Dayita commented on 2011-10-22 21:24

Chakra repo with aufs patches is dead for now. I think, links to aufs.sourceforge.net will be more reliable.

nous commented on 2011-10-19 06:12

Yes. LZO compression with tuxonice broke at pf4. For people with less swap space than RAM (like myself) LZO's a must. Others should move to linux-pf anyway. To absolutely use the last kernel26-pf4, you can download the package tarball, change pf3 to pf4 in the PKGBUILD and build it with 'makepkg --skipinteg'.

shadowed87 commented on 2011-10-19 05:15

is it there any reason not to use 2.6.39-pf4 instead of pf3 ??

nous commented on 2011-10-08 17:53

I'm waiting for linux-pf to start working with lzo'ed tuxonice. Once that happens, I'll request deletion and transfer of comments/votes.

falconindy commented on 2011-10-08 17:29

Can this be deleted?

Anonymous comment on 2011-09-24 22:03

change link for:
http://kernelorg.mirrors.tds.net/pub/linux/kernel/v3.0/linux-3.0.tar.bz2

thanks.

nous commented on 2011-08-31 17:50

3.0.3-pf incorporates the new aufs3 tree and the interactive cpufreq governor.

nous commented on 2011-08-29 09:04

Yeah, same here with lzo and lzf. See https://bbs.archlinux.org/viewtopic.php?pid=977921#p977921 and https://bbs.archlinux.org/viewtopic.php?pid=983362#p983362

aarone commented on 2011-08-29 06:23

I have an i686 machine (Atom N450, specifically), and I get -22 errors when trying to resume using lzo compression. Lzf also doesn't work (as some have reported it does) -- it yields an error code of -5 with a message about "decompression yielded XXXX bytes instead of 4096" (where XXXX is a number slightly less than 4096). This is with the latest version (3.0.2) of linux-pf.

The last kernel version where suspend/resume consistently works for me is 2.6.37. 2.6.38 and .39 die silently on resume (some fiddling with transparent hugepages once made the problem seem to go away, I can't remember the details). And the 3.0 series has the -22 error.

nous commented on 2011-08-26 16:22

I just read a couple of threads over at the Gentoo forums and it would seem that the -22 decompression error only concerns x86_64 kernels. Can anyone on i686 confirm that? I only have access to 64bit boxes and they all fail. Also, someone reported that non-BFS kernels don't get that error...

nous commented on 2011-08-26 16:16

I just read a couple of threads over at the Gentoo forums and it would seem that the -22 decompression error only concerns i686 kernels. Can anyone confirm that? I only have access to 64bit boxes and they all fail. Also, someone reported that non-BFS kernels don't get that error...

nous commented on 2011-08-24 09:00

@skydrome Fixed, thanks. It seems that dropbox gets confused when more than a few computers are linked.

skydrome commented on 2011-08-23 20:08

in your repo, the packages are all current, but the db still shows old versions.
Specifically:
the -k7, -k8, and -p4 builds

nous commented on 2011-08-22 19:13

Please, see the post below about the default linux-pf I/O scheduler (BFQ) and post some feedback. I'm not satisfied with its performance, but I'd want to hear from others.

https://bbs.archlinux.org/viewtopic.php?pid=980278#p980278

nous commented on 2011-08-22 17:44

Please, do not flag as out-of-date before checking the project page. 3.0.3 is incorporated into linux-pf-3.0.2

nous commented on 2011-08-21 09:39

@jgottula Either dropbox has just changed something or I did something wrong. In any case, please fall back to the original URL which I just tested and verified to work.

[pfkernel]
Server = http://dl.dropbox.com/u/11734958/$arch

In case you're trying to access the repo through a browser, dropbox it doesn't allow directory listing. One can only download specific files with a full url (e.g. http://dl.dropbox.com/u/11734958/x86_64/linux-pf-core2-3.0.2-1-x86_64.pkg.tar.xz). In the past few days I also run a cronjob on my main server that generates an index.html for all public directories, thus allowing people to browse the repo by appending index.html to the URL: http://dl.dropbox.com/u/11734958/index.html

jgottula commented on 2011-08-21 01:38

Sorry to bother, but the Dropbox URL has been giving a 404 for the past day or two (the tiny.cc one works). Also, and I don't know if this is related to the Dropbox being inaccessible, when I attempt to pacman -S linux-pf using the tiny.cc URL, the database downloads successfully but I get that the package is corrupted (it downloads within a second, so it's probably pulling down a 404 page or something).

nous commented on 2011-08-19 06:09

@Phate666 See https://bbs.archlinux.org/viewtopic.php?pid=976753#p976753 where xclaude reported working compressed hibernation with lzf.

Anonymous comment on 2011-08-19 05:36

ohhh yes .. sorry ... downloading...

post-factum commented on 2011-08-19 05:10

@willianholtz: see my previous comment. 3.0.2-pf contains 3.0.3.

Anonymous comment on 2011-08-19 04:03

waiting for 3.0.3....

Phate666 commented on 2011-08-18 22:13

tuxonice still broken

nous commented on 2011-08-18 10:54

Updated. LZF is enabled as module in this version, as some people have reported success at resuming with it.

post-factum commented on 2011-08-17 21:14

3.0.2-pf is out. Note, that it contains Linux 3.0.3!!!

nous commented on 2011-08-15 12:11

The repo is now populated with the precompiled generic and optimized linux-pf packages. Also with a new, easier to remember hostname alias.

[pfkernel]
Server = http://tiny.cc/linux-pf-$arch
or
Server = http://dl.dropbox.com/u/11734958/$arch

People who use hibernation should either disable tuxonice compression or switch to swsusp.

nous commented on 2011-08-15 11:32

The repo is now populated with the precompiled generic and optimized linux-pf packages.

People who use hibernation should either disable tuxonice compression or switch to swsusp.

jakob commented on 2011-08-13 21:37

kalpik: I have the same error: -22, although I’ve built-in lzo now instead of module

post-factum commented on 2011-08-13 19:50

@nous: I've given up using TuxOnIce, because suddenly built-in hibernation works well for me.

kalpik commented on 2011-08-13 16:46

So, tuxonice still not functional? Error -22?

nous commented on 2011-08-13 16:45

Could you be more specific? What's the error? Anyway, I'm not going to break the PKGBUILD yet for pacman3 in favor of pacman4.

Huulivoide commented on 2011-08-13 15:51

Doesent currently work with testing
version of pacman4, it doesent aplly
any of the patches. its the noextraxt
if clause that breaks it ^^

nous commented on 2011-08-13 14:20

I deleted my previous uneducated post about lzf. It seems that lzo is the only available/working (de)compressor for tuxonice.

@kalpik: My old Pentium 3 at home is working overtime to build the i686 packages (6 hours for each one, 4 so far; generic, atom, k8 and core2). I'm building the generic x86_64 one on my laptop. You'll have to wait 'til Aug 16 for the rest as I don't want to stress my laptop in this heat.

kalpik commented on 2011-08-13 13:32

When are you planning to update your repo? :)

nous commented on 2011-08-13 13:27

I googled a little and found that lzf compresses almost as good as lzo, only faster. I'm switching the built-in compressor of the included configs to lzf and leaving lzo as module. People who use tuxonice and experience the -22 error in resume, should change lzo to lzf in /etc/hibernate/tuxonice.conf; all others should add 'lzo' to MODULES array of /etc/mkinitcpio.conf.

nous commented on 2011-08-13 13:02

It's built-in in my laptop config. Does yours resume OK?

post-factum commented on 2011-08-13 12:55

Probably, the trouble is in lzo module? Try to compile it as built-in, not a module.

nous commented on 2011-08-13 12:01

Yeah, I was just informing the users. For the time being, I've changed the compressor in /etc/hibernate/tuxonice.conf from 'lzo' to 'none'.

post-factum commented on 2011-08-12 19:26

Concerning TuxOnIce it's better to interact with Nigel. I've just merged it, nothing more…

nous commented on 2011-08-12 14:29

Fixed. BTW, tuxonice resume still croaks with error -22...

lucak3 commented on 2011-08-12 09:39

In the package renaming options, you left the old kernel26-pf name:
# Give option to rename package according to CPU
echo
echo "=============================================================="
msg "Hit <ENTER>: to create a package named kernel26-pf"

nous commented on 2011-08-11 22:22

It boots fine here on my Vostro with SLUB.

nous commented on 2011-08-11 21:40

Hey people! The new and shiny linux-pf is waiting. Check http://aur.archlinux.org/packages.php?ID=50956. If some wish to stick with kernel26-pf for a little longer, it's recommended to use 2.6.39-pf3 in order to have a working tux-on-ice.

This package will eventually be deleted. Thank you all.

nous commented on 2011-08-11 21:39

Hey people! The new and shiny linux-pf is waiting. Check http://aur.archlinux.org/packages.php?ID=50956. Is some wish to stick to kernel26-pf for a little longer, it's recommended to use 2.6.39-pf3 in order to have a working tux-on-ice.

This package will eventually be deleted. Thank you all.

post-factum commented on 2011-08-11 21:25

I've already caught 1 panic with SLOB. Hope, SLUB won't panic.

nous commented on 2011-08-11 20:05

Please test and report either here or at https://bbs.archlinux.org/viewtopic.php?id=103462

lucak3 commented on 2011-08-11 14:24

http://ck-hack.blogspot.com/2011/08/300-ck1-and-bfs-for-300.html -> Let's hope pf gets out soon :)

post-factum commented on 2011-08-04 21:23

We've moved to new server. I hope, you haven't noticed that :). In case of some troubles mail me, please.

post-factum commented on 2011-08-04 21:23

We've moved to new server. I hope, you haven't noticed that :). In case of some troubles mail me, please.

nous commented on 2011-07-30 20:49

This might be my second-to-last post here, as kernel26 is being dropped in favor of linux 3.0. The versioning pattern of the respective linux-pf has been finalized to linux-pf 3.x.y, as requested by Oleksandr, where y is the actual pf version. This also makes my life easier, as it allows for updates not related to the pf-patchset to be reflected on the pkgrel number and give timely notifications to users.

Awaiting for Con Kolivas to publish BFS for 3.0, linux-pf stays at 3.0.0 and builds a stock ARCH kernel. In the meantime, the "old" kernel26-pf has been updated not to conflict with linux-headers, as those who have the pfkernel repo might have noticed.

The new linux-pf resides at http://aur.archlinux.org/packages.php?ID=50956 and my last post here will announce the transition to it.

nous commented on 2011-07-30 20:44

OK, the versioning pattern has been finalized to linux-pf 3.x.y, as requested by Oleksandr, where y is the actual pf version. This also makes my life easier, as it allows for updates not related to the pf-patchset to be reflected on the pkgrel number and give timely notifications to users.

Awaiting for Con Kolivas to publish BFS for 3.0, linux-pf stays at 3.0.0 and builds a stock ARCH kernel. In the meantime, the "old" kernel26-pf has been updated not to conflict with linux-headers, as those who have the pfkernel repo might have noticed.

Anonymous comment on 2011-07-28 14:58

I also use a custom config, but I have removed mkinitcpio from the PKGBUILD and the install script. I just thought it would be nice to do it "properly" without editing the PKGBUILD. I haven't thought of the possibility of just ignoring the warnings. Thank you for the hint.

nous commented on 2011-07-28 10:08

@euri I use a custom config for my laptop without initramfs. Just build as usual, remove the 'initrd=' line from grub and ignore the mkinitcpio warnings. Just make sure you have compiled-in your hard disk controller and root filesystem driver.

Anonymous comment on 2011-07-28 07:41

Is it possible to add an option for a kernel without initramfs?

post-factum commented on 2011-07-27 21:44

Unfortunately, 3.0-pf will be released not so soon as expected. -ck as well as BFQ isn't ready and won't be for a week or two.

Also, consider changing version numbers. I'll use such a scheme:

3.x.y-pf

where x equals to current mainline kernel version (e.g., 0) and y equals to pf-kernel release (e.g., 1). So, the first version will be 3.0.1-pf.

post-factum commented on 2011-07-27 21:44

Unfortunately, 3.0-pf will be released not so soon as expected. -ck isn't ready and won't be for a week or to.

Also, consider changing version numbers. I'll use such a scheme:

3.x.y-pf

where x equals to current mainline kernel version (e.g., 0) and y equals to pf-kernel release (e.g., 1). So, the first version will be 3.0.1-pf.

nous commented on 2011-07-27 10:06

I modified the PKGBUILD to not provide kernel26-headers so as to avoid conflict with the mainstream linux kernel package. Awaiting the release of the pf-patchset for linux-pf-3.0, I'm rebuilding all the 64-bit and the generic i686 binaries of the pfkernel repo.

jgottula commented on 2011-07-26 21:37

That sounds reasonable to me. I'm also looking at the linux-mainline package at the moment because it's also installed on my system, and it definitely provides headers in /usr/src/, but it just omits the whole kernel26-headers bit, and it never seemed to have a problem installing with linux-headers present.

nous commented on 2011-07-26 21:18

The linux-pf package installs its headers in /usr/src/linux-pf and they are available to any (nvidia comes to mind) driver that might ask for them. No conflict with linux-arch.

nous commented on 2011-07-26 21:15

The new linux-headers provides, replaces AND conflicts with kernel26-headers. Thus, it's impossible to force pacman to install any other kernel that also provides kernel26-headers. It makes no sense to make linux-pf provide 'linux-pf-headers' (which package would ask for it anyway?). I'm thinking of removing that 'provides' altogether, certainly before the pf patchset for 3.0 becomes available. Any objections?

jgottula commented on 2011-07-26 10:07

I don't think it is strictly because of providing linux-headers, because kernel26-pf (the old version) provided kernel26-headers and that was never a conflict; rather, I think it's the conflicts=('kernel26-headers') bit that is meant to rule out mixing 2.6.x headers with 3.x headers, but which mistakenly overlooks the fact that the new linux-headers package provides kernel26-headers presumably as a backwards compatibility thing... someone want to back me up on this?

sva_h4cky0 commented on 2011-07-26 09:50

@ jgottula
bcoz linux-pf provide linux-header

jgottula commented on 2011-07-26 09:12

I'm also getting the following conflict when I try to install linux-pf-core2 alongside Arch's regular linux 3.0-1 and linux-headers 3.0-1 packages:

linux-pf-core2 and linux-headers are in conflict (kernel26-headers). Remove linux-headers? [y/N]

I seem to recall that kernel26-pf-core2 didn't mind kernel26-headers being installed concurrently. Perhaps this is cropping up because linux-headers provides kernel26-headers, with which this package has an explicit conflict?

nous commented on 2011-07-26 08:33

Right! I had already downloaded the kernel tarball and makepkg skipped that step...

jgottula commented on 2011-07-26 06:57

The PKGBUILD appears to be trying to grab linux-3.0.tar.bz2 from ftp://ftp.kernel.org/pub/linux/kernel/v2.6/, whereas it should really be looking in /v3.0/. The build was failing for me until I made that amendment in my local PKGBUILD file.

nous commented on 2011-07-24 13:19

Same archbbs project page: https://bbs.archlinux.org/viewtopic.php?id=103462
Freshmeat changelog: http://freshmeat.net/projects/pf-patchset

nous commented on 2011-07-24 13:15

At http://aur.archlinux.org/packages.php?ID=50956 you'll find the new linux-pf package. I'm waiting for the assorted patches to be prepared. For now it just builds a stock ARCH kernel.

nous commented on 2011-07-24 13:09

It won't build until postfactum releases the appropriate pf1 patch for kernel3.

nous commented on 2011-07-24 13:08

Well, it builds now but uses the stock ARCH config.

nous commented on 2011-07-24 12:03

It won't build until postfactum releases the appropriate pf1 patch for kernel3.

nous commented on 2011-07-15 18:53

Again, the pf patchset contains the latest kernel patchlevel with pf-patchset ≥ patchlevel. When in doubt check the related kernel26-pf page at https://bbs.archlinux.org/viewtopic.php?id=103462 or the changelog at freshmeat: http://freshmeat.net/projects/pf-patchset

jakob commented on 2011-07-15 17:31

though I wasn’t quite right, as the answer is located 8 lines below my last citation:
it says:
“Included patches:

[m] mainline update;”

So this is, where .3 is actually coming in…

Anonymous comment on 2011-07-15 11:52

jakob: Thanks for your answer.

jakob commented on 2011-07-15 11:04

marceliq: Have a look at http://pf.natalenko.name/:

“Latest patch 2.6.39-pf4 (11.07.2011), applies to bare 2.6.39 kernel with no stable patches”

So, pf4 IS the current version. But beware: tuxonice, as noted already by Phate666 (and I can confirm that) doesn’t work with pf4. pf3 works, however.

Anonymous comment on 2011-07-15 11:01

Hi. Is it posible to update this to a newest kernel version 2.6.39.3? Thanks.

nous commented on 2011-07-14 11:11

OK. I modified the PKGBUILD to provide one last option regarding the package name. Users that compile kernel26-pf themselves will have the option to have a package name that reflects or not their processor. That option affects ONLY the name of the package, not the actual package optimization (if any).

jakob commented on 2011-07-13 22:05

I removed the [pfkernel] repo very consciously since I had the impression of dropbox being rather slow and additionally found a localmodconfig kernel more attractive. True, kernel compilation takes some time, but mostly, I consider it fun and since my thinkpad is running so smooth in all regards, I enjoy this last resort of tinkering around currently. Instead of putting in another query whether to name the pkg appropriately or not, it may be better to comment out lines 202-237 of the current PKGBDUILD. We won’t have as neatly named packages anymore then, but that shouldn’t be a too big caveat. Are there other opinions on this?

nous commented on 2011-07-13 20:31

If you add the [pfkernel] repo in your pacman.conf, you'll be given the option to use the updated binaries whenever available, as I build optimized packages for all major processors with every update. I could add an option to PKGBUILD to control the package name for those who compile it themselves, I just don't like too many questions and options. Tell me what you think.

jakob commented on 2011-07-13 17:13

What I found a problem for me is the automatic renaming of the pkg (in my case to kernel26-pf-core2): thus, an update isnt recognized as there’s no such pkg.

nous commented on 2011-07-13 12:16

Your kernel or repo's? If the latter, tell me your cpu/architecture to give you a link for a pf3. Due to real-life delays, I haven't upgraded myself to pf4 (some maintainer, huh?)...

Phate666 commented on 2011-07-13 09:56

TuxOnIce stopped working with pf4. After waking up from suspend to disk, image reading crashes with an IO error

GI_Jack commented on 2011-07-10 22:08

yeah its fixed it wasn't -pf it was my chroot. thanks

GI_Jack commented on 2011-07-10 22:08

yeah its figured out now, wasn't kernel26, it was my little chroot. fixed/

nous commented on 2011-07-06 09:13

"mkinitcpio -p kernel26-pf"

GI_Jack commented on 2011-07-06 07:41

with makeinitcpio kernel chokes with
Parsing hook [sata]
>_/lib/initcpio/functions: line 54: /dev/fd/62: No such file or directory

GI_Jack commented on 2011-07-06 07:17

with makeinitcpio kernel chokes with
Parsing hook [sata]
>_/lib/initcpio/functions: line 54: /dev/fd/62: No such file or directory

nous commented on 2011-06-26 21:10

Done. You still need to manually copy your old .config (or zcat from /proc/config.gz) into the source tree.

Anonymous comment on 2011-06-26 19:00

Would be good to include a 'make oldconfig' option in the build menu.

Cape commented on 2011-06-26 12:56

Recompiling the kernel solved the problem... it appears that i somehow configured the kernel without the "-pf" suffix.

Cape commented on 2011-06-26 12:51

@nous: Even the output of uname -a returns "Linux myhost 2.6.39 #1 SMP PREEMPT...". I somehow configured this kernel without the trailing -pf!

@lucak3: thanks for the help

nous commented on 2011-06-25 12:24

@Cape I can't reproduce the bug. Are you using makepkg or some wrapper like yaourt/clyde? Running 'make kernelrelease' inside $builddir/src/linux-2.6.39 should return '2.6.39-pf', otherwise something's broken on your side.

lucak3 commented on 2011-06-25 07:57

It would be better to report nvidia-pf problems on its AUR page :)
BTW, if you are referring to this:
------------------------------------------------------------------------------------------
NVIDIA: calling KBUILD...
test -e include/generated/autoconf.h -a -e include/config/auto.conf || ( \
echo; \
echo " ERROR: Kernel configuration is invalid."; \
echo " include/generated/autoconf.h or include/config/auto.conf are missing.";\
echo " Run 'make oldconfig && make prepare' on kernel src to fix it."; \
echo; \
/bin/false)
------------------------------------------------------------------------------------------
then i can assure you there is no problem: it's a known KBUILD harmless bug, it just prints what it would do, easily understandable with a bit of sh knowledge.
But of course you can check if it really picked up the correct headers: just check some lines under that and you should see:
cc -Wp,-MD,/home/luca/nvidia-pf/src/NVIDIA-Linux-x86-275.09.07/kernel/.nv.o.d -nostdinc -isystem /usr/lib/gcc/i686-pc-linux-gnu/4.6.0/include -I/usr/src/linux-2.6.39-pf/arch/x86/include
As you can note, if the last part with -I has -pf in the path, you're good.

Cape commented on 2011-06-24 12:00

@Nous: I'm building my own on i686. I don't have kernel-headers installed but:

cape ~ $ pacman -Qo /usr/src/linux-2.6.39*/* 2>/dev/null
/usr/src/linux-2.6.39-ARCH/vmlinux is owned by kernel26 2.6.39.1-1
/usr/src/linux-2.6.39/Kconfig is owned by kernel26-pf-core2 2.6.39-pf2
/usr/src/linux-2.6.39/linux-2.6.39 is owned by kernel26-pf-core2 2.6.39-pf2
/usr/src/linux-2.6.39/Makefile is owned by kernel26-pf-core2 2.6.39-pf2
/usr/src/linux-2.6.39/Module.symvers is owned by kernel26-pf-core2 2.6.39-pf2
/usr/src/linux-2.6.39/vmlinux is owned by kernel26-pf-core2 2.6.39-pf2

And modules are also stored in '/lib/modules/2.6.39' (and the 'build' symlink correctly points to '/usr/src/linux-2.6.39/')

Plus, when i build nvidia-pf, the installer keeps telling me that the sources for the current kernel are not correctly configured...

PS: i've used the Qt frontend to configure the kernel + localmodconfig and modprobed_db.

nous commented on 2011-06-23 14:43

@Cape Are you using a repo package or your own? Which version? i686 or x86_64 arch? In all my boxes I get (depending on the CPU) the following:

% pacman -Qo /usr/src/linux-2.6.39*/* 2>/dev/null
/usr/src/linux-2.6.39-ARCH/Kconfig is owned by kernel26-headers 2.6.39.1-1
/usr/src/linux-2.6.39-ARCH/Makefile is owned by kernel26-headers 2.6.39.1-1
/usr/src/linux-2.6.39-ARCH/Module.symvers is owned by kernel26-headers 2.6.39.1-1
/usr/src/linux-2.6.39-ARCH/vmlinux is owned by kernel26 2.6.39.1-1
/usr/src/linux-2.6.39-pf/Kconfig is owned by kernel26-pf-k8 2.6.39-pf2
/usr/src/linux-2.6.39-pf/Makefile is owned by kernel26-pf-k8 2.6.39-pf2
/usr/src/linux-2.6.39-pf/Module.symvers is owned by kernel26-pf-k8 2.6.39-pf2
/usr/src/linux-2.6.39-pf/linux-2.6.39-pf is owned by kernel26-pf-k8 2.6.39-pf2
/usr/src/linux-2.6.39-pf/vmlinux is owned by kernel26-pf-k8 2.6.39-pf2

The 'build' symlink in /lib/modules/2.6.39-pf/ is correct too. Are you sure it's not nvidia-pf's fault?
To have a generic package name remove lines 196-241.

Cape commented on 2011-06-22 20:29

I've noticed that sources from this build are stored in "/usr/src/linux-2.6.39/" causing nvidia-pf to complain about not finding the kernel (unless i modify the PKGBUILD).
Shouldn't it be "/usr/src/linux-2.6.39-pf/"?
Also, i don't want my package to be called "kernel26-pf-core2" so if i comment out everything from line 196 to line 235, will i be fine?
Thanks in advance

nous commented on 2011-06-04 21:47

Thanks, jakob.
@antiwinhacker Unfortunately, dropbox doesn't allow normal browsing, you have to put it somewhere in /etc/pacman.conf or /etc/pacman.d/unofficial (and include it from /etc/pacman.conf).

jakob commented on 2011-06-04 19:58

Just an awesome package! Almost easier and certainly faster than downloading the package each time from dropbox. Thanks alot!

Anonymous comment on 2011-06-04 18:04

I just installed your package, thanks for the great work! I have a AMD Athlon X2 250 CPU. Your dropbox repository is not reachable, it is moving?

jakob commented on 2011-06-03 01:46

Ah yeah.

Just did that: http://cxg.de/_4f746d.htm

However, I deleted my aufs* files in ~/tmp/pkgs/sources, downloaded fresh copies et voila: it works: and patch application works fine :) I seemed to have broken soures or somethig like that

Thanks for your hint!

nous commented on 2011-06-02 19:47

@jakob: please, try downloading the tarball with 'wget -Y off http://aur.archlinux.org/packages/kernel26-pf/kernel26-pf.tar.gz' and report back. The filesizes should be (see also http://chakra-project.org/sources/kernel26/patches/39/aufs2.1/):

orion:[nous]:/tmp/kernel26-pf% ls -l
total 660
-rw-r--r-- 1 nous users 17622 Jun 1 19:28 PKGBUILD
-rw-r--r-- 1 nous users 2727 May 19 09:46 aufs2-base.patch
-rw-r--r-- 1 nous users 976 May 19 09:46 aufs2-kbuild.patch
-rw-r--r-- 1 nous users 7781 May 19 09:46 aufs2-standalone.patch

jakob commented on 2011-06-02 12:59

I can confirm slane’s issue. After makepkg -g >> PKGBUILD, I have the following sums which don’t fail:
http://cxg.de/_4f8961.htm

Unfortunately, the patches don’t apply and building breaks with applying aufs2-patches.

nous commented on 2011-06-02 08:33

I can't confirm. It works OK, I even tried it manually with wget and sha256sum and the sums check out. Are you using a proxy?

Anonymous comment on 2011-06-02 08:16

Hello,
==> Validating source files with sha256sums...
aufs2-base.patch ... FAILED
aufs2-kbuild.patch ... FAILED
aufs2-standalone.patch ... FAILED

post-factum commented on 2011-05-31 19:00

That's it.

post-factum commented on 2011-05-23 21:31

Well, guys, I guess, Nigel is busy at the moment and still cannot update TuxOnIce against 2.6.39 tree. But other things are already merged, so you can give them test by cloning pf-kernel from my github account, compiling it and reporting. I'd really appreciate it.

One more thing =). I'll immediately release 2.6.39-pf1 as soon as all components arrive.

nous commented on 2011-05-23 21:20

There's no download link on the project's page. Usually, Oleksandr flags as out-of-date when he's ready.

smoge commented on 2011-05-23 17:49

.39 is ready? https://github.com/pfactum/pf-kernel/commits/pf-2.6.39

nous commented on 2011-04-23 18:53

But anyway, one can safely disable that patch, it's only for those uneasy spirits that delve into the dark arts of WEP cracking...

nous commented on 2011-04-23 18:52

Funny, it works from here...

% wget -Y off http://patches.aircrack-ng.org/channel-negative-one-maxim.patch
--2011-04-23 21:50:19-- http://patches.aircrack-ng.org/channel-negative-one-maxim.patch
Resolving patches.aircrack-ng.org (patches.aircrack-ng.org)... 213.186.33.2, 2001:41d0:1:1b00:213:186:33:2
Connecting to patches.aircrack-ng.org (patches.aircrack-ng.org)|213.186.33.2|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 1021 [text/plain]
Saving to: “channel-negative-one-maxim.patch”

100%[=======================================================================>] 1,021 --.-K/s in 0s

2011-04-23 21:50:19 (117 MB/s) - “channel-negative-one-maxim.patch” saved [1021/1021]

akurei commented on 2011-04-23 15:53

-> Downloading channel-negative-one-maxim.patch...
--2011-04-23 17:45:23-- http://patches.aircrack-ng.org/channel-negative-one-maxim.patch
Resolving patches.aircrack-ng.org... 2001:41d0:1:1b00:213:186:33:2, 213.186.33.2
Connecting to patches.aircrack-ng.org|2001:41d0:1:1b00:213:186:33:2|:80... connected.
HTTP request sent, awaiting response... Read error (Connection reset by peer) in headers.
Retrying.

My patch:

URL: http://paste.pocoo.org/show/FSt2NWwhNurdktQG8wNL/
RAW: http://paste.pocoo.org/raw/FSt2NWwhNurdktQG8wNL/

akurei commented on 2011-04-23 15:46

-> Downloading channel-negative-one-maxim.patch...
--2011-04-23 17:45:23-- http://patches.aircrack-ng.org/channel-negative-one-maxim.patch
Resolving patches.aircrack-ng.org... 2001:41d0:1:1b00:213:186:33:2, 213.186.33.2
Connecting to patches.aircrack-ng.org|2001:41d0:1:1b00:213:186:33:2|:80... connected.
HTTP request sent, awaiting response... Read error (Connection reset by peer) in headers.
Retrying.

post-factum commented on 2011-04-22 19:25

-pf6 with 38.4 included is out.

nous commented on 2011-04-22 10:17

Thanks, fixed.

nous commented on 2011-04-22 09:13

Thanks, fixed.

nous commented on 2011-04-22 08:48

Thanks, fixed.

Anonymous comment on 2011-04-22 07:56

there is a small typo in the PKGBUILD, nconfig is selectable via "c" key not "n".

post-factum commented on 2011-04-21 18:49

-pf5 brings BFSv401.

post-factum commented on 2011-04-15 17:50

-pf4 is out. It contains 38.3, which reverts buggy commit that caused troubles with hibernation, so Anvin's patch is also removed.

nous commented on 2011-04-13 16:01

A trouble-free kernel, for a change...

nous commented on 2011-04-11 19:26

Great. Updated and started building.

post-factum commented on 2011-04-11 15:53

-pf3 is out. -ck3 and hibernation regression fix have been merged.

Anonymous comment on 2011-04-05 13:24

Just a feedback: it works like a charm now! :)

nous commented on 2011-04-05 11:32

post-factum: Correct, the patch is just a temporary fix.

freakazoid88: It's my laptop tailored config, I'm sorry it slipped through. Fixed now.

bicer: The aufs2 patches are fetched from the chakra repo. I chose this approach to let makepkg warn me for updates 'cause I rarely check for them myself. The patches were updated a little after the last PKGBUILD was uploaded. Also fixed, one can always override the sums with 'makepkg --skipinteg'.

Anonymous comment on 2011-04-03 22:28

Nous could you please confirm SUMs of aufs2-base.patch, aufs2-kbuild.patch, and aufs2-standalone.patch are correct. They're failing verification.
Thank You.

Anonymous comment on 2011-04-03 15:17

The file config.x86_64 looks quite strange, it's not set at the default values o.O Can anyone confirm that?

post-factum commented on 2011-04-03 10:06

I guess, cr4-restore.patch is not final version of patch, so I'll recommend using it, but still waiting for 2.6.38.3.

Thanks.

nous commented on 2011-04-02 22:42

New pf2.5 PKGBUILD with cr4-restore.patch fixes resume. The repo is not updated, sorry. Those who need hibernate/resume functionality, please download and compile.

nous commented on 2011-04-02 20:43

post-factum: Since this is is critical, and because I'll be far from computers for the next 2 days, how about upgrading 2.5 to pf3 now so I put my dropbox server to work, populating the repo with the prebuilts? BFS 372 performs significantly better than 363.

nous commented on 2011-04-02 20:36

Included patch from http://marc.info/?l=linux-kernel&m=130158291603814&w=4

Tested to work, resume continues now.

post-factum commented on 2011-04-02 20:32

OK, thanks for info!

Anonymous comment on 2011-04-02 19:37

Resume issue: http://marc.info/?l=linux-kernel&m=130151206329088&w=4
I guess we need to wait for 38.3, TOI or BFS seems not to be buggy as for 38.1 (pf1) it works, I currently run 372-test2 patch on top of pf1...

post-factum commented on 2011-04-02 13:24

No idea :/. Try to disable BFS, then BFQ. If not — TuxOnIce is buggy.

nous commented on 2011-04-02 12:52

Grrrr, still hangs on resume.

post-factum commented on 2011-04-02 11:13

Ok, I've released -pf2.5. Please, test it and report here or to me using email.

nous commented on 2011-04-02 09:59

Why don't you upload it as patch-2.6.38-pf2.5.bz2 for the rest to test, and make it pf3 in due time?

post-factum commented on 2011-04-02 09:37

I'm testing new BFS version right now.

[pf@eternity]:[~][0]% dmesg | grep BFS
[ 1.364306] BFS CPU scheduler v0.372 by Con Kolivas.


[pf@eternity]:[~][0]% uname -a
Linux eternity 2.6.38-pf2.5 #1 SMP PREEMPT Sat Apr 2 11:16:23 EEST 2011 i686 Intel(R) Pentium(R) Dual CPU T2330 @ 1.60GHz GenuineIntel GNU/Linux

post-factum commented on 2011-04-02 07:48

Well, I suspect some deadlock to be present in one of pf2 patches. I guess, this is BFS problem, and I'm waiting for next BFS release.

nous commented on 2011-03-31 16:05

Easiest way to downgrade: change _pfrel=pf2 to pf1 in PKGBUILD and build with 'makepkg --skipinteg'

nous commented on 2011-03-31 09:15

As Kirugs has reported, -pf2 is broken for me too. It fails to resume on 2 desktops and 1 laptop...

nous commented on 2011-03-30 14:32

Kirugs: obviously, as -pf1 works, your setup is ok and you wait for the next tuxonice update. Since hibernation is a necessity for my laptop, I had to hold back updates more than a couple of times in the past.

shadowed87: sha256sums updated, but I'm not rebuilding the repo packages just for aufs - I doubt anyone uses it anyway :)

shadowed87 commented on 2011-03-30 08:50

aufs has been updated on chakra, sha256 are different ;)

Anonymous comment on 2011-03-30 06:14

pf2 is broken for me. Hibernate doesn't work, going sleep works, but resuming ends with blinking caps lock, nothing in logs. I tried pfkernel repo and situation is the same, except that it gave me no blinking caps.
pf1 works brilliant, so I'm reverting to that.

nous commented on 2011-03-29 09:00

http://freshmeat.net/projects/pf-patchset

shadowed87 commented on 2011-03-29 08:58

anyone knows where to find a pf changelog? i really can't find it..

shadowed87 commented on 2011-03-29 08:56

anyone knows where to find a pf changelog? i really can't find it..

nous commented on 2011-03-29 04:35

hellnest: Silly me, I saw that error in my laptop and I just enabled dvb in menuconfig... Meh, I'm either getting lazier or dumber. Fixed and cheers.

hellnest commented on 2011-03-28 21:55

@nous
- cp include/config/dvb/*.h ${pkgdir}/usr/src/linux-${_kernver}/include/config/dvb/
+ [[ -e include/config/dvb/ ]] && cp include/config/dvb/*.h ${pkgdir}/usr/src/linux-${_kernver}/include/config/dvb/

Like your comment on -ck page. You must modify PKGBUILD to make it success. My build now success thanks once again :)

hellnest commented on 2011-03-28 19:44

Ok thanks for your help, :) I didn't read a whole comments sorry ^^

nous commented on 2011-03-28 19:42

hellnest: Downgrade to make 3.81-5 from [core], I had similar errors (fackamato also reported it, see comments) and I currently have it pinned with IgnorePkg.

hellnest commented on 2011-03-28 19:17

I need some enlightment here... >.<" 10times try compile it without success

make[1]: *** No rule to make target `/home/lee/AUR/kernel26-pf/pkg/lib/firmware/./', needed by `/home/lee/AUR/kernel26-pf/pkg/lib/firmware/atmsar11.fw'. Stop.
make: *** [_modinst_post] Error 2

nous commented on 2011-03-27 12:13

litemotiv I was hit by the error you reported; googling it I found that it's caused by binutils-2.21

nous commented on 2011-03-27 00:33

shadowed87 yup, but it's gonna take a couple of days... My 32bit box is slow.

shadowed87 commented on 2011-03-26 23:53

wonderful...thank you as usual...
one morequestion...is it possible to add core2 version to i686 repository?

nous commented on 2011-03-26 22:55

Thanks everyone for the feedback.
Missing /lib/modules/2.6.38-pf/build symlink fixed.
Forgotten version 37 chakra patches updated to 38.

CheariX, as Alger noted, if you run optimized kernel26-pf you won't see the updates until my prebuilt optimized packages hit the pfkernel repo.

post-factum commented on 2011-03-26 20:36

I'm not going to include BFS 363-test2 into pf-kernel until it become stable.

shadowed87 commented on 2011-03-26 16:35

@Alger
I had the same problem...executing sudo ln -s /usr/src/linux-2.6.38-pf/ /lib/modules/2.6.38-pf/build all was fine with nvidia drivers

Anonymous comment on 2011-03-26 15:51

@CheariX Yaourt can't detect because you have other name package installed like kernel26-pf-core2 so it's differ from this as far as i think this is the reason

CheariX commented on 2011-03-26 15:17

Why can't yaourt detect new version?
This problem was from 2.6.37-pf2 up to 2.6.37-pf6 and now with 2.6.38, yaourt still thinks kernel26-pf is up to date...

shadowed87 commented on 2011-03-26 14:46

why not these patches?
http://chakra-project.org/sources/kernel26/patches/38
instead of
http://chakra-project.org/sources/kernel26/patches/37

Anonymous comment on 2011-03-26 14:28

aren't we need a simlink /lib/modules/build/2.6.38-pf/build to /usr/src/linux-2.6.38-pf or something like this , because i can't build ati properity driver now

nous commented on 2011-03-26 13:47

@graysky - Thanks for the heads-up, but as far as BFS is concerned, I'm just following the pfkernel patchset, whose author anyway is a close follower of CK's blog and I suspect he'll implement this promising patch very soon.

graysky commented on 2011-03-26 13:37

Nous - you might wanna add this patch to ck1: http://ck-hack.blogspot.com/2011/03/bfs-and-optimising-global-scheduler-for.html

nous commented on 2011-03-26 13:17

Alright, alright. Four kernel configuration options for all tastes.

Anonymous comment on 2011-03-26 08:21

I suggest write this to PKGBUILD:

msg "Run make menuconfig(console), gconfig(GTK based) or xconfig(Qt based) before build? (m/g/x/N)"
read answer
case "$answer" in
m | M) make menuconfig
cp -v .config ${startdir}/config.local
;;
g | M) make gconfig
cp -v .config ${startdir}/config.local
;;
x | X) make xcongig
cp -v .config ${startdir}/config.local
;;
default)
;;
esac

nous commented on 2011-03-26 02:18

It's very easy to put that in the PKGBUILD, but it's just another configuration option; I've also been asked to use xconfig or gconfig instead of menuconfig. It would be rather annoying for most people to have the PKGBUILD asking a ton of questions before actually building the package, while one could just issue 'sed -i s/menuconfig/nconfig/ PKGBUILD' before makepkg.

misc commented on 2011-03-26 01:27

It would be great if an option for nconfig were to be included in this PKGBUILD.

nous commented on 2011-03-19 11:33

If the stock -ARCH kernel has the same issue, then something might wrong with your build tools. I really can't imagine where you should start, but you could grep the main kernel Makefile for 'export' directives (I see make, awk, perl, gcc among others; definitely more in submakefiles) and reinstall them. It's really weird, have you tested on another box?

Cape commented on 2011-03-18 22:35

Nope... same output, but still i'm unable to enter that submenu. The same happens in /usr/src/linux-2.6.37-ARCH

cape ~/Packages/kernel26-pf/src/linux-2.6.37 $ make menuconfig
HOSTCC scripts/basic/fixdep
HOSTCC scripts/basic/docproc
HOSTCC scripts/kconfig/conf.o
HOSTCC scripts/kconfig/kxgettext.o
HOSTCC scripts/kconfig/lxdialog/checklist.o
HOSTCC scripts/kconfig/lxdialog/inputbox.o
HOSTCC scripts/kconfig/lxdialog/menubox.o
HOSTCC scripts/kconfig/lxdialog/textbox.o
HOSTCC scripts/kconfig/lxdialog/util.o
HOSTCC scripts/kconfig/lxdialog/yesno.o
HOSTCC scripts/kconfig/mconf.o
SHIPPED scripts/kconfig/zconf.tab.c
SHIPPED scripts/kconfig/lex.zconf.c
SHIPPED scripts/kconfig/zconf.hash.c
HOSTCC scripts/kconfig/zconf.tab.o
HOSTLD scripts/kconfig/mconf
scripts/kconfig/mconf Kconfig
#
# configuration written to .config
#


*** End of the configuration.
*** Execute 'make' to start the build or try 'make help'.

nous commented on 2011-03-18 21:37

I just tested the PKGBUILD again on an i686 box without problems. Please do the following and see if you get any errors.
1. Download the kernel26-pf tarball, untar and run makepkg
2. At the '==> Run make menuconfig before build? (y/N)' prompt press CTRL-C
3. cd into src/linux-2.6.37 and run 'make menuconfig', check the processor option and exit menuconfig

My output is the following:
================================
proxy:[nous]:/temp/pkgbuild/kernel26-pf/src/linux-2.6.37% make menuconfig
HOSTCC scripts/basic/fixdep
HOSTCC scripts/basic/docproc
HOSTCC scripts/kconfig/conf.o
HOSTCC scripts/kconfig/kxgettext.o
HOSTCC scripts/kconfig/lxdialog/checklist.o
HOSTCC scripts/kconfig/lxdialog/inputbox.o
HOSTCC scripts/kconfig/lxdialog/menubox.o
HOSTCC scripts/kconfig/lxdialog/textbox.o
HOSTCC scripts/kconfig/lxdialog/util.o
HOSTCC scripts/kconfig/lxdialog/yesno.o
HOSTCC scripts/kconfig/mconf.o
SHIPPED scripts/kconfig/zconf.tab.c
SHIPPED scripts/kconfig/lex.zconf.c
SHIPPED scripts/kconfig/zconf.hash.c
HOSTCC scripts/kconfig/zconf.tab.o
HOSTLD scripts/kconfig/mconf
scripts/kconfig/mconf Kconfig
#
# configuration written to .config
#


*** End of the configuration.
*** Execute 'make' to start the build or try 'make help'.
=====================================

What happens if you run 'sudo make menuconfig' in /usr/src/linux-2.6.38-ARCH?

Cape commented on 2011-03-18 13:07

I'm on i686 and i'm using makepkg. Of course, i answer "y" when asked to run menuconfig. The problem is that "Processor type and features ---> Processor family" is "locked", i can't enter the submenu...
It's the same if choose localmodconfig, vanilla or current kernel config. I also noticed that other options are locked.

Cape commented on 2011-03-18 13:06

I'm on i686 and i'm using makepkg. Of course, i answer "y" when asked to run menuconfig. The problem is that "Processor type and features ---> Processor family" is "locked", i can't edit it...
It's the same if choose localmodconfig, vanilla or current kernel config. I also noticed that other options are locked.

Cape commented on 2011-03-18 13:06

I'm on i686 and i'm using makepkg. Of course, i answer "y" when asked to run menuconfig. The problem is that "Processor type and features ---> Processor family" is "locked", i can't edit it...
It's the same if choose localmodconfig, vanilla or current kernel config. I also noticed that other options are locked.

Cape commented on 2011-03-18 13:00

I'm on i686 and i'm using makepkg. Of course, i answer "y" when asked to run menuconfig. The problem is that "Processor type and features ---> Processor family" is "locked", i can't edit it...
It's the same if choose localmodconfig, vanilla or current kernel config. I also noticed that other options are locked.

nous commented on 2011-03-16 22:06

Cape: It's not normal. I assume you follow "Processor type and features ---> Processor family", right? Can you be more specific? (i686 or x86_64? makepkg or yaourt/clyde? etc)

I also assume of course that you answer 'y' to "==> Run make menuconfig before build? (y/N)"

Cape commented on 2011-03-16 20:19

Maybe i'm doing something wrong, but i'm not able to choose my cpu family in menuconfig.
It's the same if i choose locamodconfig, running kernel or stock... I just can't edit that entry. Is it normal?

Anonymous comment on 2011-03-15 17:07

Thank you Christos. I will try the prebuild version from your repo first. :-)

nous commented on 2011-03-15 16:59

litemotiv: that option is for users that already have a tailored kernel for their systems and don't want to compile the entire kernel tree. Using that option entails manual configuring of the uncovered options the pf-patchset provides, i.e. what you were asked and it's normal. Had you answered 'yes' to 'make menuconfig' question before, you'd be directed to the more user-friendly menu configuration. If you used that option on, say, the stock -ARCH kernel you'd have no gain whatsoever, since that's a generic kernel. You'd have to deselect the unneeded options/modules one by one, but that's usually a one-time chore. Google for 'kernel configure' for more.

As for the compilation error, I haven't bumped into it, ever. I compile 6 i686 and 5 x86_64 pf-kernels for my unofficial repo and I've never seen that error. What's worse, that's an assembler error and I hate x86 assembly. Have you tried a clean compile? A different binutils version?

Anonymous comment on 2011-03-14 17:24

Thanks for this, nice installer too. :-)

Two questions:

I chose the current-running-kernel config option, this then gave a lot of the PF-specific options a default value of No (example: BFQ [N/y/m/]). Is this fixable, or is it just the way it works?

Secondly, i'm getting a compilation error:

AS arch/x86/kernel/entry_64.o
arch/x86/kernel/entry_64.S: Assembler messages:
arch/x86/kernel/entry_64.S:1531: Error: .size expression does not evaluate to a constant

Is this just a temporary source problem that i have to wait a bit for to get fixed?

Thanks again,
-lite

nous commented on 2011-03-09 16:52

pf6 is 2.6.37.3 bump

nous commented on 2011-03-01 23:21

Changelog of pf4 to pf5: BFQ updated to version 2.

nous commented on 2011-02-10 17:02

Indeed, I had uploaded a previous source tarball without the relevant patch line. Nice catch.

floft commented on 2011-02-10 01:33

I installed this kernel for the b44 fix, but I still got a kernel panic when trying to connect via ethernet. I looked at the swiotlb.c file and the patch wasn't being applied, so I modified the PKGBUILD to apply the included patch and now it works. I'm not sure if this is as expected or not...

shadowed87 commented on 2011-02-08 17:48

tnx man, you're a great mantainer

nous commented on 2011-02-02 20:52

A patch to fix the b44 kernel panic is included. Darn, .37 is not the best kernel I've run.

nous commented on 2011-01-26 21:12

@shadowed87 I just put the libata-alignment.patch in. I'm rebuilding the 64bit packages right now; those unaffected by the bug, do not need to upgrade.

shadowed87 commented on 2011-01-26 18:59

the problem for me is that after udev 165 upgrade some days ago, sometimes there is a kernel panic on boot, this happens with all kernels but not all computers..so I downgrade temporarily udev...

here is the bug report:
https://bugs.archlinux.org/task/22343

and now they have patched all archlinux kernel for this bug...so it's a shame not to have the fix in this package...

nous commented on 2011-01-26 17:43

The ARCH patchset is mostly the archlinux bootlogos and a few other minor patches most of which conflict with BFS/BFQ/TuxOnIce. It's just not worth isolating the clean ones (and not wise either, considering they probably need the dirty ones).

nous commented on 2011-01-26 17:07

The ARCH patchset is mostly the archlinux bootlogos and a few other minor patches most of which conflict with BFS/BFQ/TuxOnIce. It's just not worth isolating the clean ones (and not wise, considering they probably need the dirty ones).

shadowed87 commented on 2011-01-26 11:53

why don't you add arch patches?

nous commented on 2011-01-20 17:17

New prebuilt packages for all needs!
x86_64 packages for: Intel Pentium4/D/Nocona Xeon - Core2 - Atom, AMD Athlon64 K8, Generic x86_64
i686 packages for : Intel Pentium3 - Pentium4 - -Pentium-M - Atom, AMD K7, Generic i686

[pfkernel]
Server = http://dl.dropbox.com/u/11734958/$arch

nous commented on 2011-01-17 12:04

@elmariachi I think that should be reported at http://aur.archlinux.org/packages.php?ID=41689

Anonymous comment on 2011-01-17 10:45

==> Attempting to run /usr/bin/reload_database from modprobe_db...
WARNING: Module led_class not found.
Aborting...

wasn't this fixed already?

nous commented on 2011-01-11 22:31

A quick fix for the nvidia-pf without the need for reinstallation:
# cd /lib/modules/2.6.37-pf
# rm -f source build
# ln -sf /usr/src/linux-2.6.37-pf build

You *may* need to force the next update to overwrite the link.

nous commented on 2011-01-11 22:03

@Cape please, see my previous comment as it applies to your condition too. The oversight I mentioned was related also to not copying all the necessary header files correctly :/
It was the same reason as yours that lead me to discover and fix it. I should update the pf#, but it won't follow pfkernel's own version and I don't like decimals (if you see the PKGBUILD, I cannot just update -pf2 to -pf2-1, as the $pkgrel variable (which controls the revision) is reserved for the pf version itself and no more dashes are allowed. So, please, reinstall, either from the repo (if on x86_64, including nvidia-pf) or from the AUR, they have been fixed. The i686 packages build in a slow P3 and will be available shortly.

nous commented on 2011-01-11 21:54

@Cape please, see my previous comment as it applies to your condition too. The oversight I mentioned was related also to not copying all the necessary header files correctly :/
It was the same reason as yours that lead me to discover and fix it. I should update the pf#, but it won't follow pfkernel's own version and I don't like decimals (if you see the PKGBUILD, I cannot just update -pf2 to -pf2-1, as the $pkgrel variable (which controls the revision) is reserved for the pf version it self. So, please, reinstall, either from the repo (if on x86_64, including nvidia-pf) or from the AUR, they have been fixed. The i686 packages build in a slow P3 and will be available shortly.

Cape commented on 2011-01-11 20:18

I'm trying to build nvidia-pf, but it complains about "determining the kernel version".
The nvidia-pf PKGBUILD points the nvidia installer to "/lib/modules/2.6.37-pf/build" witch doesn't exist.
The stock arch kernel instead, contains "/lib/modules/2.6.36-ARCH/build" as a link to "/usr/src/2.6.36-ARCH/".
Should i create that link manually? Or did i messed up something during the kernel configuration?

nous commented on 2011-01-10 10:10

To all users of kernel26-pf: I had an oversight in the .37 PKGBUILD that left inside the firmware files, so if pacman complained about conflicting files in /lib/firmware during the installation, please delete the kernel26-pf package from /var/cache/pacman/packages (if you used the prebuilt repo package), uninstall and re-install both kernel26-pf and linux-firmware so you don't have any conflicting files in future updates.

nous commented on 2011-01-08 17:34

For those who don't tweak much their kernels or are bored of compiling, assorted kernel26-packages can be found at the repo below.

[pfkernel]
# kernel26-pf, kernel26-pf-p3, kernel26-pf-k8, kernel26-pf-core2, nvidia-pf, squid3, arora-git
Server = http://dl.dropbox.com/u/11734958/$arch

nous commented on 2011-01-08 17:00

For those who don't tweak much their kernels or are bored of compiling, assorted kernel26-packages can be found at the repo below.

[pfkernel]
# kernel26-pf, kernel26-pf-p3, kernel26-pf-k8, kernel26-pf-core2, nvidia-pf, squid3, arora-git
Server = http://dl.dropbox.com/u/11734958/$arch

nous commented on 2011-01-08 16:27

I've built successfully the generic x86_64 version (check the pfkernel repo: search for unofficial user repositories in archwiki) and I'm building right now the optimized 64-bit ones and the generic i686 version without errors. Wrapper scripts like yaourt are known to have problems with the AUR; please use 'aurdownload kernel26-pf' from aurscripts and just 'makepkg -i' inside the kernel26-pf directory.

Anonymous comment on 2011-01-08 15:46

Doesn't build for me:


HOSTCC scripts/basic/fixdep
HOSTCC scripts/basic/docproc
HOSTCC scripts/kconfig/conf.o
HOSTCC scripts/kconfig/kxgettext.o
SHIPPED scripts/kconfig/zconf.tab.c
SHIPPED scripts/kconfig/lex.zconf.c
SHIPPED scripts/kconfig/zconf.hash.c
HOSTCC scripts/kconfig/zconf.tab.o
HOSTLD scripts/kconfig/conf
scripts/kconfig/conf --silentoldconfig Kconfig
fs/Kconfig:194: can't open file "fs/aufs/Kconfig"
make[2]: *** [silentoldconfig] Error 1
make[1]: *** [silentoldconfig] Error 2
make: *** No rule to make target `include/config/auto.conf', needed by `include/config/kernel.release'. Stop.
Aborting...
==> ERROR: Makepkg was unable to build kernel26-pf.
==> Restart building kernel26-pf ? [y/N]
==> ------------------------------------
==>
==> WARNING: Following packages have not been installed:
kernel26-pf


This could be the problem:


==> Applying pf-kernel patch
bzip2: skipping "/tmp/yaourt-tmp-root/aur-kernel26-pf/src/patch-2.6.37-pf2.bz2": not a regular file
==> Applying aufs2 and squashfs-lzma patches
patching file fs/namei.c
patching file fs/splice.c
patching file include/linux/namei.h
patching file include/linux/splice.h
patching file fs/Kconfig
patching file fs/Makefile
patching file include/linux/Kbuild
patching file fs/file_table.c
patching file fs/inode.c
patching file fs/namei.c
patching file fs/namespace.c
patching file fs/notify/group.c
patching file fs/notify/mark.c
patching file fs/open.c
patching file fs/splice.c
patching file security/commoncap.c
patching file security/device_cgroup.c
patching file security/security.c
bzip2: skipping "/tmp/yaourt-tmp-root/aur-kernel26-pf/src/aufs2.1-2010-12-20.patch.bz2": not a regular file

krevedko commented on 2011-01-08 13:23

i have a problem when compiling 2.6.37-pf2:
$ makepkg
....
LD [M] sound/usb/usx2y/snd-usb-usx2y.ko
IHEX firmware/acenic/tg2.bin
IHEX firmware/acenic/tg1.bin
....
IHEX firmware/cpia2/stv0672_vp4.bin
Преждевременный выход...

nous commented on 2011-01-08 10:24

LZO compression is implemented now into mainline kernel hibernation and it should be as fast as tuxonice, but I still like the latter's framebuffer interface.

post-factum commented on 2011-01-08 09:46

BTW, I've deactivated tuxonice for my laptop, because hibernation works better without it.

nous commented on 2011-01-08 08:28

Grrrrreat! I'd hate to lose tuxonice.

post-factum commented on 2011-01-07 23:36

It was Nigel's mistake with tuxonice, pf-kernel has been updated against new tuxonice branch. pf2 is out.

nous commented on 2011-01-07 17:13

To Oleksandr: I use tuxonice extensively, as it behaves especially well on my vostro 1500 and thinkpad r50e laptops. It behaved equally well on my desktops at home and at work until 2.6.36, which is when it started failing to suspend. That said, I would regret to see it go but I barely have time to maintain this package, not to mention messing with reversing patchsets. The kernel26-ck maintainer (http://aur.archlinux.org/packages.php?ID=32877) doesn't even bother to include tuxonice and the kernel26-ice one (http://aur.archlinux.org/packages.php?ID=15224) has kept BFS at v0.357.
I'll keep maintaining this package, but I'll certainly keep 2.6.36-pf6 on my laptops until it's too outdated or this is resolved.

To all: package updated to 2.6.37-pf1, you'll need to say "y" to "make menuconfig" until I find some time to go through the new kernel config. The aufs2 and squashfs-lzma patches too will be temporarily be removed until I find the aforementioned time (or someone else compiles a patchset for 2.6.37-pf1).

post-factum commented on 2011-01-07 15:18

Any ideas about tuxonice? See pf-kernel site for details.

nous commented on 2011-01-07 10:40

Please, please, please, see http://pf.natalenko.name/ before flagging as out-of-date.

Anonymous comment on 2011-01-04 15:06

Thanks for this. Knocks at least 2W off my power consumption. Boots faster too!

nous commented on 2011-01-01 22:30

Thy will is done. :)

graysky commented on 2011-01-01 15:21

@lucak3 - thanks for the kind words - glad you're finding that script useful.
@nous - you should know that according to Steven (the author of the streamline script), calling it directly isn't right. He recommends calling it make localmodconfig instead. Also know that for 2.26.37, there is a fixed version of his script.

See http://git390.marist.edu/cgi-bin/gitweb.cgi?p=linux-2.6.git;a=commitdiff;h=8ef17fa2ef8be74b946e725c2afb0e2a54981da1;hp=20d190473328b90755eb2434cf2d26b73a53ef23

If you look in my PKGBUILD for kernel26-ck I pull the git version and replace the 2.6.32.x version with it, then call the make localmodconfig.

lucak3 commented on 2011-01-01 15:17

I found interesting and useful the use of modprobed_db script in AUR (by graysky) https://aur.archlinux.org/packages.php?ID=41689.
It would be a good idea to add a few lines to the PKGBUILD to use that before streamline_config.pl, as in kernel26-bfs.
This reduces to these lines
---------------------------
msg "If you have modprobe_db installed, running reload_database now"
if [ -e /usr/bin/reload_database ]; then
/usr/bin/reload_database
fi
---------------------------
as you can see here, https://aur.archlinux.org/packages/kernel26-bfs/kernel26-bfs/PKGBUILD

nous commented on 2011-01-01 09:25

Hrmpf... I re-introduced the aufs2 and squashfs-lzma patch into the package, but I inadvertedly applied it against the vanilla kernel tree... so it wiped completely the pf patch. Apologies, it should be fixed now.

MaJia321 commented on 2011-01-01 08:15

Why doesn't config.x86_64 have CONFIG_SCHED_BFS=y ?
Thanks

nous commented on 2010-12-30 00:41

Re-introduced aufs2 and squashfs-lzma patches.

Anonymous comment on 2010-12-06 22:02

I am a robot. This is not an official message. AUR guidelines suggest to not include binaries. Examples:
kernel26-pf/logo_linux_clut224.ppm.bz2
kernel26-pf/logo_linux_mono.pbm.bz2
kernel26-pf/logo_linux_vga16.ppm.bz2
Feel free to disregard this as you would any other comment. This robot will not post here again.

nous commented on 2010-11-29 15:37

Links updated.

post-factum commented on 2010-11-29 14:18

Please, update links to pf-kernel site. I've changed its domain name to pf.natalenko.name. Old links still works via redirect.

LookTJ commented on 2010-11-24 00:44

it has mainline update.

LookTJ commented on 2010-11-23 22:38

nous, sorry about that, won't happen again from me.

nous commented on 2010-11-23 12:39

FFS, I've said it so many times, kernel26-pf follows http://pf-kernel.org.ua, NOT http://kernel.org!

DO NOT FLAG AS OUT-OF-DATE.

nous commented on 2010-11-21 18:57

mpsnets I've never used it and my time is really sparse.... Why don't you copy an existing PKGBUILD of, say, -kernel26-ck and create a new package, as lucak3 did with nvidia-pf?

Anonymous comment on 2010-11-19 12:46

Please
You need to create the package vhba-module-pf!

graysky commented on 2010-11-18 22:23

Plus, see what ck has to say about it: http://ck-hack.blogspot.com/2010/11/create-task-groups-by-tty-comment.html

Con Kolivas wrote:

I've had every man and his dog either drop into IRC or email me asking me what my thoughts are on the grouping tasks by tty layer patch discussed here: Phoronix link and slashdot. I guess people didn't understand my 2.6.36-ck1 announcement clearly enough, so I'll quote it again here:

Those following the development of the patches for interactivity at massive
load, I have COMPLETELY DROPPED them as they introduce regressions at normal
workloads, and I cannot under any circumstances approve changes to improve
behaviour at ridiculous workloads which affect regular ones. I still see
precisely zero point at optimising for absurd workloads. Proving how many
un-niced jobs you can throw at your kernel compiles is not a measure of one's
prowess. It is just a mindless test.

Remember, I already had developed a hierarchical tree-based penalty patch for BFS and blogged about it here. I can do it in a 10 line patch for BFS, but it introduced regressions, which is why I dropped it (see earlier blog entry here: further-updates-on-hierarchical-tree).

Again, I can't for the life of me see why you'd optimise for make -j64 on a quad core machine. It is one workload, unique to people who compile all the time, but done in a way you wouldn't normally do it anyway. It is not going to magically make anything else better. If for some god-forsaken reason you wanted to do that, you could already do that with nice, or even better, by running it SCHED_IDLEPRIO.

nice -19 make -j 64 blahblah
or
schedtool -D -e make -j64 blahblah

It's not really that hard folks...

And if you really really really still want the feature for BFS, the patch that does the hierarchical tree based penalty is rolled into a bigger patch (so a lot more than just the 10 lines I mentioned) that can also group threads and enable/disable the features and it's still here:

bfs357-penalise_fork_depth_account_threads.patch

It is worth noting also that the mainline approach costs you in throughput, whereas this patch is virtually free.

EDIT: I forgot to mention that for YEARS now I've been using my "toolsched" wrapper scripts that do this automatically. See toolsched for the scripts. Make always starts as SCHED_IDLEPRIO for me at home.

nous commented on 2010-11-18 21:41

From what I read, the patch groups tty-originated tasks, which has little to do with the desktop:
http://marc.info/?l=linux-kernel&m=128993894920527&w=2

Besides: 1st, the patch doesn't apply cleanly and I don't have the time to fix it and 2nd, nice(1) behaves very well with BFS.

nous commented on 2010-11-18 21:35

From what I read, the patch groups tty-originated tasks from the rest, which has little to to with the desktop:
http://marc.info/?l=linux-kernel&m=128993894920527&w=2

Besides: 1st, the patch doesn't apply cleanly and I don't have the time to fix it and 2nd, nice(1) behaves very well with BFS.

nous commented on 2010-11-18 21:11

mpsnets: the patch is conflicting with BFS.

Anonymous comment on 2010-11-18 19:16

This thinking of putting the patch on autogroup-pf?

See
http://aur.archlinux.org/packages.php?ID=43689

nous commented on 2010-11-04 20:27

@fackamato: It's make's fault (and it's random). Downgrade to make from [core].

nous commented on 2010-11-04 20:12

@fackamato: It's make's fault (and it's random). Downgrade to make from [core].

fackamato commented on 2010-11-04 19:09

Hm. Am I doing something wrong? It won't work for me. This is the error:


IHEX2FW firmware/keyspan_pda/keyspan_pda.fw
IHEX2FW firmware/keyspan_pda/xircom_pgs.fw
IHEX firmware/cpia2/stv0672_vp4.bin
H16TOFW firmware/vicam/firmware.fw
Aborting...

I've deleted the src dir but it's the same.

nous commented on 2010-10-31 22:09

Fixed, thanks.

Anonymous comment on 2010-10-31 14:26

@nous - Could you take a look at "config.x86_64". Some things like BFQ are not being set correctly. Is this just me?

CONFIG_IOSCHED_BFQ=y | # CONFIG_IOSCHED_BFQ is not set
CONFIG_CGROUP_BFQIO=y | # CONFIG_DEFAULT_DEADLINE is not set
CONFIG_DEFAULT_BFQ=y | CONFIG_DEFAULT_CFQ=y
CONFIG_DEFAULT_IOSCHED="bfq" | CONFIG_DEFAULT_IOSCHED="cfq"

nous commented on 2010-10-30 15:03

@sht0rm -ck2 is included in -pf1, check line 1483 of kernel/sched_bfs.c, "BUG_ON(rq != this_rq());" is removed.

BTW, I'm not russian/ukrainian :p

Anonymous comment on 2010-10-30 11:07

Можете включить patch -ck2 или он уже включен?
Являюсь владельцем intel 4965, и вайфай заглючил.

Can you include an -ck2 patch in your kernel, or it already included?
I am an owner of this wifi chip
0c:00.0 Network controller: Intel Corporation PRO/Wireless 4965 AG or AGN [Kedron] Network Connection (rev 61)
in my laptop, and I have a problems with 2.6.26-pf1.

http://ck-hack.blogspot.com/2010/10/2636-ck2.html

nous commented on 2010-10-29 19:22

LinuxIMQ hasn't released a patch for 2.6.36, so the author of pfkernel ditched it. I also ditched aufs2 and squashfs-lzma, because I couldn't find patches to apply cleanly in 2.6.36. I'm building generic and optimized binaries right now to upload at the repo.

drexya commented on 2010-10-27 22:24

The Con Kolivas patchset is very impressive, x264 encoding (on the same clip) with stock kernel = ~ 15fps, with this kernel = ~ 20fps !

nous commented on 2010-10-22 15:09

Please, do not flag as out-of-date before checking the pf-kernel page. The author is waiting for IMQ to sync with 2.6.36 before releasing -pf1.

[EDIT] Please, again, DO NOT FLAG AS OUT-OF-DATE! This package follows http://pf-kernel.org.ua - NOT http://kernel.org.

nous commented on 2010-10-22 11:00

Please, do not flag as out-of-date before checking the pf-kernel page. The author is waiting for IMQ to sync with 2.6.36 before releasing -pf1.

nous commented on 2010-10-12 08:08

Thanks for the suggestion, I replaced the relevant section from your package.

graysky commented on 2010-10-11 22:10

@nous - you might wanna modify your make localmodconfig line to call the script directly as that make lmc never worked right. See my PKGBUILD in the kernel26-ck package for more.

nous commented on 2010-10-11 18:24

@graysky I vaguely remember that despite my MAKEFLAGS="-j2" in /etc/makepkg.conf, makepkg only used 1 core as I could see it in the xfce4 CPU monitor. Those lines intended to fix that and were forgotten since. Nice catch!

graysky commented on 2010-10-11 15:34

Nice package, thanks for maintaining it. Aren't the following lines redundant?

# Make some good use of MAKEFLAGS
MAKEFLAGS=`grep -v '#' /etc/makepkg.conf | grep MAKEFLAGS= | sed s/MAKEFLAGS=// | sed s/\"//g`
# get kernel version
make ${MAKEFLAGS} prepare

nous commented on 2010-10-11 08:01

I'm sorry, I uploaded a test PKGBUILD, it's fixed now.

gaougalos commented on 2010-10-11 07:42

scripts/kconfig/conf -s arch/x86/Kconfig
***
*** You have not yet configured your kernel!
*** (missing kernel config file ".config")
***
*** Please run some configurator (e.g. "make oldconfig" or
*** "make menuconfig" or "make xconfig").
***
make[2]: *** [silentoldconfig] Error 1
make[1]: *** [silentoldconfig] Error 2
make: *** No rule to make target `include/config/auto.conf', needed by `include/config/kernel.release'. Stop.
Aborting...
[christos@christos-arch kernel26-pf]$ cd

nous commented on 2010-10-11 06:26

http://ck-hack.blogspot.com/2010/10/further-updates-on-hierarchical-tree.html
Here CK says he has received a few negative reports about the fork depth patch and will not merge it into the main BFS patch set. For the time being, I'll keep it in the PKGBUILD since it works oh-so-nice for me, but if anyone reports bad behavior I'll remove it.

nous commented on 2010-10-10 07:32

Alright, I was curious about the fork_depth patch so I applied it. I was so impressed by the results, that I updated the PKGBUILD. I haven't changed the version number though, so people will have to explicitly install from the AUR.

I started 'grep -ir nous /' and fired up a dozen of the heaviest apps I have installed in my laptop one after the other, switching between them, clicking here and there, playing video and audio. Not a single frame or sound skip, no click action delayed. Superb.

nous commented on 2010-10-10 07:18

Alright, I was curious about the fork_depth patch so I applied it. I was so impressed by the results, that I updated the PKGBUILD. I haven't changed the version number though, so people will have to explicitly install from the AUR.

nous commented on 2010-10-09 12:23

@jens It's my understanding that those patches are highly experimental, but Con Kolivas is quickly intergrating them into the main -ck patchset, given enough positive feedback. I don't want to mess with them right now since my systems are responsive enough as they are already and I don't want ranting from people who use kernel26-pf and experience Bad Things. The interbench results though that CK posted are so impressive that I think the patch will be possibly merged in the next BFS version.

jens commented on 2010-10-09 09:39

http://ck-hack.blogspot.com/2010/10/interactive-at-any-load.html

Sorry, I´m just not sure, are the above patches already included in this package?

Thanks

nous commented on 2010-10-01 20:37

@sht0rm I can't confirm, it boots fine in my 5 boxes.

Anonymous comment on 2010-09-30 12:45

in last build i have an error
"waiting for udev uevents to be processed [BUSY]"
and system stay to long on this message.

nous commented on 2010-09-29 16:28

@mpsnets Great! Could you post your configs (/etc/hibernate/* and /etc/mkinitcpio.conf) to the forum kernel26-pf thread? Pankajmore reported trouble resuming... https://bbs.archlinux.org/viewtopic.php?id=103462&p=4

nous commented on 2010-09-29 16:27

@mpsnets Great! Could you post your configs (/etc/hibernate/* and /etc/mkinitcpio.conf) to the forum kernel26-pf thread? Pankajmore reported trouble resuming...

Anonymous comment on 2010-09-28 20:04

Here it works perfectly. It's the best I've found so far.
It works perfectly and suspend to RAM Suspend to disk.
I have a Core2Duo 4GB RAM ArchLinux x86_64.

pankajmore commented on 2010-09-28 19:10

this kernel is awesome , the system feels very snappy , but sadly hibernation doesn't work , after hibernation , during "..kernel and process data ..." the caps lock led continuously blinks and i have to manually power down by pressing the power button , kernel26-ice works perfectly , so tuxonice works fine on my hardware , does anybody else has issues with hibernation??

nous commented on 2010-09-28 07:04

@kazuo It's already there! Make sure you don't download some older version through a proxy.

kazuo commented on 2010-09-27 20:17

And nous, please what about adding a '-k' to the 'bzip2 -d ${startdir}/logo_linux_*.bz2' so the logo files are not deleted after every makepkg?

lucak3 commented on 2010-09-27 15:07

pf10 patchset out

Anonymous comment on 2010-09-23 20:19

sorry, fixed.

I uninstalled everything and installed again.

Anonymous comment on 2010-09-23 19:26

I finished upgrading this package and when I run "yaourt-Syu - aur" keeps appearing on the list of packages to update

==> Package upgrade only (new release):
aur/kernel26-pf pf7 pf7-2.6.35 -> pf8

--

# uname -rv
2.6.35-pf #1 SMP PREEMPT Thu Sep 23 15:13:36 BRT 2010

nous commented on 2010-09-22 18:30

The 2.6.35.5-ARCH is basically the mainstream .5 patch, plus aufs2 and the archlogos. The pf patch already contains the .5 patch. I'm adding the archlogos now.

Anonymous comment on 2010-09-22 12:37

Why do not you use the patch ArchLinux?

"ftp://ftp.archlinux.org/other/kernel26/patch-2.6.35.5-1-ARCH.bz2"

nous commented on 2010-09-21 18:06

The fixed kernel packages have been uploaded. 64-bit users, please update your kernels.

nous commented on 2010-09-19 19:13

The fixed kernel packages are being uploaded right now. 64-bit users, please update your kernels.

nous commented on 2010-09-19 11:32

@harryNID Touched!

Anonymous comment on 2010-09-19 09:36

aufs2-sqlzma.patch.bz2 1970-Jan-01 04:13:08 107.3K application/x-bzip
kernel26.preset 1970-Jan-01 04:13:08 0.4K text/plain

These files were created a year before I was born? You might want to touch them. Anyway, what's up with these timestamps? Looks kinda suspicious!

nous commented on 2010-09-18 13:47

After the local root exploit that affects 64-bit kernels (described at http://sota.gen.nz/compat2/ and confirmed to work in my 64-bit boxes), I updated the PKGBUILD.

Until further notice, PLEASE DO NOT USE THE PREBUILT PACKAGES FROM THE X86_64 REPO, unless you're on a single-user machine and you really know your stuff.

nous commented on 2010-09-17 21:11

The repo has moved to dropbox, please update your unofficial mirrorlist.

[pfkernel]
Server = http://dl.dropbox.com/u/11734958/i686 or
Server = http://dl.dropbox.com/u/11734958/x86_64

In x86_64 you can find core2 and k8 optimized packages. I use a single PKGBUILD for all versions to make my life easier, so make sure you uninstall any existing kernel26-pf package before you install the -core2 or -k8, otherwise you'll have lots of conflicting files.

nous commented on 2010-09-17 11:03

I was just informed the dns record of the repo was deleted. I'm setting up another, I'll update the information as soon as I'm done.

nous commented on 2010-09-17 11:02

I was just informed the dns record of the repo is down permanently. I'm setting up another, I'll update the information as soon as I'm done.

gaougalos commented on 2010-09-17 10:37

error: failed retrieving file 'pfkernel.db.tar.gz' from pfkernel.gtklocker.com : No address record
error: failed to update pfkernel (No address record)

nous commented on 2010-09-04 22:02

We now have an unofficial repository for pf-kernel and nvidia-pf, thanks to gtklocker.

[pfkernel]
Server = http://pfkernel.gtklocker.com/i686 or
Server = http://pfkernel.gtklocker.com/x86_64

On x86_64 you can find core2 and k8 optimized packages. I use a single PKGBUILD for all versions to make my life easier, so make sure you uninstall any existing kernel26-pf package before you install the -core2 or -k8, otherwise you'll have lots of conflicting files.

nous commented on 2010-08-30 17:52

Since the updates come out more often than some anticipate, it's highly recommended to have a kernel config tailored for your boxes. Use hwd or hwdetect to show you your hardware modules and build the disk controller (and tuxonice and lzo) in-kernel. Everything else can be modules and loaded at startup with /etc/rc.conf. Initrd is not needed this way and you'll reduce kernel build time to a small fraction. Keep at least one alternative kernel (e.g. the stock arch kernel) as a backup.

nous commented on 2010-08-30 17:40

Since the updates come out more often than some anticipate, it's highly recommended to have a kernel config tailored for your boxes. Use hwd or hwdetect to show you your hardware modules and build the disk controller (and tuxonice and lzo) in-kernel. Everything else can be modules and loaded at startup with /etc/rc.conf. Initrd is not needed this way and you'll reduce kernel build time to a small fraction. Keep the stock arch kernel as a backup.

nous commented on 2010-08-30 14:33

Since the updates come out more often than some anticipate, it's highly recommended to have a kernel config tailored for your boxes. Use hwd or hwdetect to show you your hardware modules and build the disk controller in-kernel. Everything else can be modules and loaded at startup with /etc/rc.conf. Initrd is not needed this way and you'll reduce kernel build time to a small fraction. Keep the stock arch kernel as a backup.

nous commented on 2010-08-30 14:29

It's highly recommended to have a kernel config tailored for your boxes. Use hwd or hwdetect to show you your hardware modules and build the disk controller in-kernel. Everything else can be modules and loaded at startup with /etc/rc.conf. Initrd is not needed this way and you'll reduce kernel build time to a small fraction. Keep the stock arch kernel as a backup.

nous commented on 2010-08-30 09:49

@lucak3 I'll try :) But I can't possibly compile and update the prebuilt packages every single time...

lucak3 commented on 2010-08-30 08:22

pf7 patchset out: are you planning to update every time they release one?
It seems this patchset updates very frequently...

lucak3 commented on 2010-08-30 08:21

pf7 patchset out

nous commented on 2010-08-29 20:38

@MaJia321 GOT IT! It's make's fault. Downgrading to 3.81-5 from [core] solves the problem.

drexya commented on 2010-08-29 18:43

Thanks for the prebuilt packages.

nous commented on 2010-08-28 20:17

AMD Athlon64 K8 optimized package:
http://www.4shared.com/file/bP9-ipTS/kernel26-pf-2635-pf6-AMD64_K8p.html
sha256sum: d525b8abbcba2201696fbe8a3a1b544b20ceb731f45a28a84d5c8c2c24dc4a74

Intel Core2 optimized package:
http://www.4shared.com/file/c_PVGzPs/kernel26-pf-2635-pf6-Core2-64p.html
sha256sum: 3eef5f67963113f925e3a228515c49bb4f05ba5aa991d3295f4a8cec6d985398

Generic x86_64 package:
http://www.4shared.com/file/u3yzd1tl/kernel26-pf-2635-pf6-x86_64pkg.html
sha256sum: c057d32fb0c782d1988a7ee98369247c1248d922245b6c37bda7094d2676a8c8

Generic i686 package:
http://www.4shared.com/file/VohYWrQw/kernel26-pf-2635-pf6-i686pkgta.html
sha256sum: 4661d12c3a13d20939e4c78acde34c9a2c33e5bab842bfacdf1f36e7b1fe8cd4

nous commented on 2010-08-28 13:06

Generic i686 package:
http://www.4shared.com/file/VohYWrQw/kernel26-pf-2635-pf6-i686pkgta.html
md5sum: 3025644b7b262808af9f6ac0e93b2972
sha256sum: 4661d12c3a13d20939e4c78acde34c9a2c33e5bab842bfacdf1f36e7b1fe8cd4

AMD Athlon64 K8 optimized package:
http://www.4shared.com/file/H8CPRlA1/kernel26-pf-2635-pf5-x86_64-K8.html
md5sum: 557e787c201013afb1497bc9699c42b1
sha256sum: 210ad1330d78ffd987a428fc236adaabf594aaf4da599391b14acce5d50dc2ef

Generic x86_64 package:
http://www.4shared.com/file/z9LIAhiX/kernel26-pf-2635-pf5-x86_64pkg.html
md5sum: a1a809bef6c45d38a35d9376acb5c82a
sha256sum: c7ac47a673820098541db3215dcd1aaa53a15f03253980b73b33c4bec25fafd5

Intel Core2 optimized package:
http://www.4shared.com/file/NRG0MuDe/kernel26-pf-2635-pf4-x86_64-CO.html
md5sum: 09f548e160188d58a3ff916842b5ec14
sha256sum: b86b0c53c5d60c02a5021534cb234f22da189501e8336f53eeda1e4d60accc70

nous commented on 2010-08-28 13:04

Intel Core2 optimized package:
http://www.4shared.com/file/NRG0MuDe/kernel26-pf-2635-pf4-x86_64-CO.html
md5sum: 09f548e160188d58a3ff916842b5ec14
sha256sum: b86b0c53c5d60c02a5021534cb234f22da189501e8336f53eeda1e4d60accc70

AMD Athlon64 K8 optimized package:
http://www.4shared.com/file/H8CPRlA1/kernel26-pf-2635-pf5-x86_64-K8.html
md5sum: 557e787c201013afb1497bc9699c42b1
sha256sum: 210ad1330d78ffd987a428fc236adaabf594aaf4da599391b14acce5d50dc2ef

Generic x86_64 package:
http://www.4shared.com/file/z9LIAhiX/kernel26-pf-2635-pf5-x86_64pkg.html
md5sum: a1a809bef6c45d38a35d9376acb5c82a
sha256sum: c7ac47a673820098541db3215dcd1aaa53a15f03253980b73b33c4bec25fafd5

Generic i686 package:
http://www.4shared.com/file/VohYWrQw/kernel26-pf-2635-pf6-i686pkgta.html
md5sum: 3025644b7b262808af9f6ac0e93b2972
sha256sum: 4661d12c3a13d20939e4c78acde34c9a2c33e5bab842bfacdf1f36e7b1fe8cd4

lucak3 commented on 2010-08-27 19:30

pf6 patchset out (for 2.6.35.4)

nous commented on 2010-08-27 16:21

Prebuilt packages to be found at http://www.4shared.com/account/dir/vB9rGCLh/

Main differences from stock archlinux kernel:
- BFS as default task scheduler
- BFQ as default I/O scheduler
- CONFIG_HZ_1000
- TuxOnIce built-in
- SquashFS-LZMA and AUFS2

nous commented on 2010-08-27 13:55

Prebuilt packages to be found at http://www.4shared.com/account/dir/vB9rGCLh/

As of now BFQ is the default I/O scheduler, as is CONFIG_HZ_1000.

nous commented on 2010-08-27 13:54

Prebuilt packages to be found at http://www.4shared.com/account/dir/vB9rGCLh/

As of now, BFQ is the default I/O scheduler, as is CONFIG_HZ_1000.

Anonymous comment on 2010-08-27 07:40

Nice.
Привет землякам.

MaJia321 commented on 2010-08-27 01:54

@nous Thanks ;)

nous commented on 2010-08-26 09:52

@MaJia321 I had that same error a couple of times, couldn't pin it down. Try building straight with makepkg instead of using wrappers or download one of the prebuilt packages.

nous commented on 2010-08-26 09:50

Intel Core2 optimized package:
http://www.4shared.com/file/NRG0MuDe/kernel26-pf-2635-pf4-x86_64-CO.html
md5sum: 09f548e160188d58a3ff916842b5ec14
sha256sum: b86b0c53c5d60c02a5021534cb234f22da189501e8336f53eeda1e4d60accc70

AMD Athlon64 K8 optimized package:
http://www.4shared.com/file/H8CPRlA1/kernel26-pf-2635-pf5-x86_64-K8.html
md5sum: 557e787c201013afb1497bc9699c42b1
sha256sum: 210ad1330d78ffd987a428fc236adaabf594aaf4da599391b14acce5d50dc2ef

Generic x86_64 package:
http://www.4shared.com/file/z9LIAhiX/kernel26-pf-2635-pf5-x86_64pkg.html]kernel26-pf-2.6.35-pf5-x86_64.pkg.tar.xz
md5sum: a1a809bef6c45d38a35d9376acb5c82a
sha256sum: c7ac47a673820098541db3215dcd1aaa53a15f03253980b73b33c4bec25fafd5

Generic i686 package:
http://www.4shared.com/file/RV72ITnt/kernel26-pf-2635-pf4-i686pkgta.html
md5sum: e0f5ee2d177192308189c2f8a8373228
sha256sum: 56b4530320e6abdf8ffa41f96fda88ae01078226474c765d09233793f2c197e9

nous commented on 2010-08-26 05:35

Intel Core2 optimized package:
http://www.4shared.com/file/NRG0MuDe/kernel26-pf-2635-pf4-x86_64-CO.html
md5sum: 09f548e160188d58a3ff916842b5ec14
sha256sum: b86b0c53c5d60c02a5021534cb234f22da189501e8336f53eeda1e4d60accc70

AMD Athlon64 K8 optimized package:
[url]http://www.4shared.com/file/H8CPRlA1/kernel26-pf-2635-pf5-x86_64-K8.html[/url]
md5sum: 557e787c201013afb1497bc9699c42b1
sha256sum: 210ad1330d78ffd987a428fc236adaabf594aaf4da599391b14acce5d50dc2ef

Generic x86_64 package:
http://www.4shared.com/file/Xx-l4mZi/kernel26-pf-2635-pf4-x86_64pkg.html
md5sum: 41f059473df339b68281e66b91b1d4f5
sha256sum: 3d52d16fdc8c6a9d2a0e7a1b0c702495be91d2b96bb3219690b1ce3b83de3184

Generic i686 package:
http://www.4shared.com/file/RV72ITnt/kernel26-pf-2635-pf4-i686pkgta.html
md5sum: e0f5ee2d177192308189c2f8a8373228
sha256sum: 56b4530320e6abdf8ffa41f96fda88ae01078226474c765d09233793f2c197e9

MaJia321 commented on 2010-08-26 01:46

Sorry for the error :
$....
INSTALL sound/usb/usx2y/snd-usb-usx2y.ko
MKDIR /var/abs/local/kernel26-pf/pkg/lib/firmware/acenic/
make[1]: *** No rule to make target `/var/abs/local/kernel26-pf/pkg/lib/firmware/./', needed by `/var/abs/local/kernel26-pf/pkg/lib/fi.fw'. Stop.
make[1]: *** Waiting for unfinished jobs....
MKDIR /var/abs/local/kernel26-pf/pkg/lib/firmware/adaptec/
make: *** [_modinst_post] Error 2
Aborting...

Thanks.

nous commented on 2010-08-25 20:46

@mpsnets please, read my first comment, the pf patch includes the latest kernel patchlevel, which as of now is .3

Anonymous comment on 2010-08-25 20:43

@mpsnets: The version of the kernel built, using this package is indeed 2.6.35.3. See for yourself: http://pf-kernel.org.ua/

Anonymous comment on 2010-08-25 16:23

kernel 2.6.35 out

2.6.35.3 relesed !

nous commented on 2010-08-25 15:08

Intel Core2 optimized package:
http://www.4shared.com/file/NRG0MuDe/kernel26-pf-2635-pf4-x86_64-CO.html
md5sum: 09f548e160188d58a3ff916842b5ec14
sha256sum: b86b0c53c5d60c02a5021534cb234f22da189501e8336f53eeda1e4d60accc70

Generic x86_64 package:
http://www.4shared.com/file/Xx-l4mZi/kernel26-pf-2635-pf4-x86_64pkg.html
md5sum: 41f059473df339b68281e66b91b1d4f5
sha256sum: 3d52d16fdc8c6a9d2a0e7a1b0c702495be91d2b96bb3219690b1ce3b83de3184

Generic i686 package:
http://www.4shared.com/file/RV72ITnt/kernel26-pf-2635-pf4-i686pkgta.html
md5sum: e0f5ee2d177192308189c2f8a8373228
sha256sum: 56b4530320e6abdf8ffa41f96fda88ae01078226474c765d09233793f2c197e9

nous commented on 2010-08-25 15:05

Intel Core2 optimized package:
http://www.4shared.com/file/NRG0MuDe/kernel26-pf-2635-pf4-x86_64-CO.html
md5sum: 09f548e160188d58a3ff916842b5ec14
sha256sum: b86b0c53c5d60c02a5021534cb234f22da189501e8336f53eeda1e4d60accc70

Generic x86_64 package:
http://www.4shared.com/file/Xx-l4mZi/kernel26-pf-2635-pf4-x86_64pkg.html
md5sum: 41f059473df339b68281e66b91b1d4f5
sha256sum: 3d52d16fdc8c6a9d2a0e7a1b0c702495be91d2b96bb3219690b1ce3b83de3184

Generic i686 package:
http://www.4shared.com/file/RV72ITnt/kernel26-pf-2635-pf4-i686pkgta.html]kernel26-pf-2.6.35-pf4-i686.pkg.tar.gz
md5sum: e0f5ee2d177192308189c2f8a8373228
sha256sum: 56b4530320e6abdf8ffa41f96fda88ae01078226474c765d09233793f2c197e9

nous commented on 2010-08-25 15:05

Intel Core2 optimized package:
http://www.4shared.com/file/NRG0MuDe/kernel26-pf-2635-pf4-x86_64-CO.html
md5sum: 09f548e160188d58a3ff916842b5ec14
sha256sum: b86b0c53c5d60c02a5021534cb234f22da189501e8336f53eeda1e4d60accc70

Generic x86_64 package:
http://www.4shared.com/file/Xx-l4mZi/kernel26-pf-2635-pf4-x86_64pkg.html
md5sum: 41f059473df339b68281e66b91b1d4f5
sha256sum: 3d52d16fdc8c6a9d2a0e7a1b0c702495be91d2b96bb3219690b1ce3b83de3184

Generic i686 package:
http://www.4shared.com/file/RV72ITnt/kernel26-pf-2635-pf4-i686pkgta.html]kernel26-pf-2.6.35-pf4-i686.pkg.tar.gz
md5sum: e0f5ee2d177192308189c2f8a8373228
sha256sum: 56b4530320e6abdf8ffa41f96fda88ae01078226474c765d09233793f2c197e9

Anonymous comment on 2010-08-25 14:13

pf5 out! Seeing that it's automatically added via the "special file" (direct link to newest version from hp) perhaps the out of date flag was wrong of me, if so sorry -- mind you that the description is out of date tho!

nous commented on 2010-08-25 11:02

Intel Core2 optimized package:
http://www.4shared.com/file/NRG0MuDe/kernel26-pf-2635-pf4-x86_64-CO.html
md5sum: 09f548e160188d58a3ff916842b5ec14
sha256sum: b86b0c53c5d60c02a5021534cb234f22da189501e8336f53eeda1e4d60accc70

Generic x86_64 package:
http://www.4shared.com/file/Xx-l4mZi/kernel26-pf-2635-pf4-x86_64pkg.html
md5sum: 41f059473df339b68281e66b91b1d4f5
sha256sum: 3d52d16fdc8c6a9d2a0e7a1b0c702495be91d2b96bb3219690b1ce3b83de3184

nous commented on 2010-08-25 09:27

Generic x86_64 package:
http://www.4shared.com/file/Xx-l4mZi/kernel26-pf-2635-pf4-x86_64pkg.html
md5sum: 41f059473df339b68281e66b91b1d4f5
sha256sum: 3d52d16fdc8c6a9d2a0e7a1b0c702495be91d2b96bb3219690b1ce3b83de3184

nous commented on 2010-08-24 08:55

The latest pf4 patch updates to kernel 2.6.35.3. The PKGBUILD has been tested in 2 Arch64 machines and the original config.x86_64 has been modified to include -ck1 by default and tuxonice as modules. Please, submit a good 32bit config.
I haven't tested tuxonice hibernation yet, make sure you read http://wiki.archlinux.org/index.php/Tuxonice first before you hibernate in the middle of your work!
If you run a vanilla/custom kernel, you can use its own config - you'll be asked so before build.

nous commented on 2010-08-24 08:54

The latest pf4 patch updates to kernel 2.6.35.3. The PKGBUILD has been tested in 2 Arch64 machines and the original config.x86_64 has been modified to include -ck1 by default and tuxonice as modules. Please, submit a good 32bit config.
I haven't tested tuxonice hibernation yet, make sure you read http://wiki.archlinux.org/index.php/Tuxonice first before you hibernate in the middle of your work!