Zen-Kernel Discussion/Support Thread - Part 4

[cach0rr0]

cach0rr0, you tried it with

pci=nomsi

or

pci=nommconf

that i have not. After I made that post I went and updated my BIOS, which went smoothly enough - it was very old, but the idea of updating a BIOS always scares me a bit. So I had been avoiding it, but well, turns out it was a few *years* old. Neat! So that's done.

Many thanks for those tidbits, if the bios update did not sort it, I will try those (time to google around for an explanation of each! - ah wait, first hit on google explains it , cool, so that's what MSI stands for)

[mbar]

Happens for me on .34-zen1:

Code: Select all

  LD      drivers/built-in.o
  LD      vmlinux.o
  MODPOST vmlinux.o
WARNING: modpost: Found 1 section mismatch(es).
To see full details build your kernel with:
'make CONFIG_DEBUG_SECTION_MISMATCH=y'
  GEN     .version
  CHK     include/generated/compile.h
  UPD     include/generated/compile.h
  CC      init/version.o
  LD      init/built-in.o
  LD      .tmp_vmlinux1
net/built-in.o: In function `br_multicast_stop':
(.text+0xb00ec): undefined reference to `rcu_barrier_bh'
make: *** [.tmp_vmlinux1] Błąd 1

any hints? .config is the same as was for .33-zen3


EDIT: CONFIG_BRIDGE_IGMP_SNOOPING is the culprit, foud that in "part 3" thread.

[aTan]

Is there a custom flags patch for 2.6.34? The old one doesn't fit.
patching file Makefile
Hunk #1 FAILED at 324.
Hunk #2 succeeded at 349 with fuzz 2 (offset 8 lines).
Hunk #3 succeeded at 612 (offset 10 lines).
1 out of 3 hunks FAILED -- saving rejects to file Makefile.rej
patching file init/Kconfig
Hunk #1 succeeded at 927 with fuzz 2 (offset 196 lines).

UPD: Looks like 2.6.34 already has this option. Hmm.

[Colt45]

I built a .34-zen1 kernel and it panics on boot because it cant find root. It appears to throw up 3 or 4 backtraces before that point, but they are going by so fast I have no way of reading them.
I used make oldconfig on my .33-zen3 config. Any ideas?

[cheater1034]

UPD: Looks like 2.6.34 already has this option. Hmm.

Yes

Post a .config and a emerge --info and what FS are you using as root, etc?

[Colt45]

Post a .config and a emerge --info and what FS are you using as root, etc?

http://sarah.colt45.ws/emerge.xps.log
http://sarah/config-2.6.34-zen1
Im using btrfs.

[kernelOfTruth]

using 2.6.35-rc4-zen1 now

SLQB seems to work well (again), too

besides slqb and CONFIG_CRYPTO_PCRYPT afaik I'm currently not using anything exotic

reiser4 (as a module) and LZF/LZO support compiled fine - thanks !

[ponciarello]

I've spotted the start of project zendroid

great news for my tattoo tnx again also for this

[broch]

2.6.35-rc4-zen1

compiles just fine but two things seems to be broken:
1) ath9k
with dhcpcd timing out (confirmed on vanilla 2.6.35-rc4)
2) tuxonice
when trying tohibernate system hangs at "Doing atomic copy/restore" . Hybernate is nor reliable but issue is related to intel video problem (well known): system hibernates/resumes just fine but freezes after a while with intel video errors. I was hoping that 2.6.35-rc will fix this, unfortunately I can't test this as it seems that tuxonice is broken.

This is netbook with Atom n450/video Intel GMA 3150/2GB RAM
I am using the same config file as with 2.6.34-zen1 where ath9k works.

thanks

note:
I see that unstable zen is adding mysterious + at the end of the kernel name. This is the case for two different (no cpopied config) systems. ?

[kernelOfTruth]

2.6.35-rc4-zen1

compiles just fine but two things seems to be broken:
1) ath9k
with dhcpcd timing out (confirmed on vanilla 2.6.35-rc4)
2) tuxonice
when trying tohibernate system hangs at "Doing atomic copy/restore" . Hybernate is nor reliable but issue is related to intel video problem (well known): system hibernates/resumes just fine but freezes after a while with intel video errors. I was hoping that 2.6.35-rc will fix this, unfortunately I can't test this as it seems that tuxonice is broken.

This is netbook with Atom n450/video Intel GMA 3150/2GB RAM
I am using the same config file as with 2.6.34-zen1 where ath9k works.

thanks

note:
I see that unstable zen is adding mysterious + at the end of the kernel name. This is the case for two different (no cpopied config) systems. ?

I get this too - I thought it was an issue caused by genkernel

[Waninkoko]

2.6.35-rc4-zen1

compiles just fine but two things seems to be broken:
1) ath9k
with dhcpcd timing out (confirmed on vanilla 2.6.35-rc4)
2) tuxonice
when trying tohibernate system hangs at "Doing atomic copy/restore" . Hybernate is nor reliable but issue is related to intel video problem (well known): system hibernates/resumes just fine but freezes after a while with intel video errors. I was hoping that 2.6.35-rc will fix this, unfortunately I can't test this as it seems that tuxonice is broken.

This is netbook with Atom n450/video Intel GMA 3150/2GB RAM
I am using the same config file as with 2.6.34-zen1 where ath9k works.

thanks

note:
I see that unstable zen is adding mysterious + at the end of the kernel name. This is the case for two different (no cpopied config) systems. ?

I get this too - I thought it was an issue caused by genkernel

What u guys mean? Where that's happening?

[kernelOfTruth]

2.6.35-rc4-zen1

compiles just fine but two things seems to be broken:
1) ath9k
with dhcpcd timing out (confirmed on vanilla 2.6.35-rc4)
2) tuxonice
when trying tohibernate system hangs at "Doing atomic copy/restore" . Hybernate is nor reliable but issue is related to intel video problem (well known): system hibernates/resumes just fine but freezes after a while with intel video errors. I was hoping that 2.6.35-rc will fix this, unfortunately I can't test this as it seems that tuxonice is broken.

This is netbook with Atom n450/video Intel GMA 3150/2GB RAM
I am using the same config file as with 2.6.34-zen1 where ath9k works.

thanks

note:
I see that unstable zen is adding mysterious + at the end of the kernel name. This is the case for two different (no cpopied config) systems. ?

I get this too - I thought it was an issue caused by genkernel

What u guys mean? Where that's happening?

cat /usr/src/linux/Makefile
VERSION = 2
PATCHLEVEL = 6
SUBLEVEL = 35
EXTRAVERSION = -rc4-zen1
NAME = Weed overdose

ls -l /lib/modules/
total 8
drwxr-xr-x 4 root root 4096 Jul 5 21:42 2.6.35-rc4-zen1
drwxrwx--- 3 root root 4096 Jul 5 21:44 2.6.35-rc4-zen1+

the "+" gets added automagically with these releases:

du -hs /lib/modules/2.6.35-rc4-zen1*
5.0M /lib/modules/2.6.35-rc4-zen1
38M /lib/modules/2.6.35-rc4-zen1+

in the folder without "+" there's only the fglrx and virtualbox module

ls -l /lib/modules/2.6.35-rc4-zen1/
total 112
drwxr-xr-x 2 root root 4096 Jul 5 21:42 misc
-rw-r--r-- 1 root root 8375 Jul 5 21:42 modules.alias
-rw-r--r-- 1 root root 7021 Jul 5 21:42 modules.alias.bin
-rw-r--r-- 1 root root 69 Jul 5 21:42 modules.ccwmap
-rw-r--r-- 1 root root 105 Jul 5 21:42 modules.dep
-rw-r--r-- 1 root root 312 Jul 5 21:42 modules.dep.bin
-rw-r--r-- 1 root root 52 Jul 5 21:42 modules.devname
-rw-r--r-- 1 root root 73 Jul 5 21:42 modules.ieee1394map
-rw-r--r-- 1 root root 141 Jul 5 21:42 modules.inputmap
-rw-r--r-- 1 root root 81 Jul 5 21:42 modules.isapnpmap
-rw-r--r-- 1 root root 74 Jul 5 21:42 modules.ofmap
-rw-r--r-- 1 root root 15569 Jul 5 21:42 modules.pcimap
-rw-r--r-- 1 root root 43 Jul 5 21:42 modules.seriomap
-rw-r--r-- 1 root root 131 Jul 5 21:42 modules.softdep
-rw-r--r-- 1 root root 10501 Jul 5 21:42 modules.symbols
-rw-r--r-- 1 root root 11712 Jul 5 21:42 modules.symbols.bin
-rw-r--r-- 1 root root 189 Jul 5 21:42 modules.usbmap
drwxr-xr-x 2 root root 4096 Jul 5 21:40 video

[eusanpe]

I cannot mount btrfs on 2.6.34-zen1. It starts with /etc/mtab not updateable, then cannot create lock file /etc/mtab~2379 no space left on device
and then alot of no space left on device errors. My subvolumes never mounted. I updated btrfs-progs to btrfs-progs-9999 and still no luck. I
rebooted back to 2.6.33-zen2 and everything works again.

[DigitalCorpus]

run btrfsck on that partition. I have a funny feeling that you have corruption. Sounds similar to what I experienced a few months ago.

There is new btrfs code in your new kernel and it might be hanging up on the corruption that .33 doesn't know about due to the lack of refinements. I have 4 btfr partitions over 430 GB and I have about ~1329GB on them without any current problems running the latest git copy of zen-stable.

[Waninkoko]

2.6.35-rc4-zen1

compiles just fine but two things seems to be broken:
1) ath9k
with dhcpcd timing out (confirmed on vanilla 2.6.35-rc4)
2) tuxonice
when trying tohibernate system hangs at "Doing atomic copy/restore" . Hybernate is nor reliable but issue is related to intel video problem (well known): system hibernates/resumes just fine but freezes after a while with intel video errors. I was hoping that 2.6.35-rc will fix this, unfortunately I can't test this as it seems that tuxonice is broken.

This is netbook with Atom n450/video Intel GMA 3150/2GB RAM
I am using the same config file as with 2.6.34-zen1 where ath9k works.

thanks

note:
I see that unstable zen is adding mysterious + at the end of the kernel name. This is the case for two different (no cpopied config) systems. ?

I get this too - I thought it was an issue caused by genkernel

What u guys mean? Where that's happening?

cat /usr/src/linux/Makefile
VERSION = 2
PATCHLEVEL = 6
SUBLEVEL = 35
EXTRAVERSION = -rc4-zen1
NAME = Weed overdose

ls -l /lib/modules/
total 8
drwxr-xr-x 4 root root 4096 Jul 5 21:42 2.6.35-rc4-zen1
drwxrwx--- 3 root root 4096 Jul 5 21:44 2.6.35-rc4-zen1+

the "+" gets added automagically with these releases:

du -hs /lib/modules/2.6.35-rc4-zen1*
5.0M /lib/modules/2.6.35-rc4-zen1
38M /lib/modules/2.6.35-rc4-zen1+

in the folder without "+" there's only the fglrx and virtualbox module

ls -l /lib/modules/2.6.35-rc4-zen1/
total 112
drwxr-xr-x 2 root root 4096 Jul 5 21:42 misc
-rw-r--r-- 1 root root 8375 Jul 5 21:42 modules.alias
-rw-r--r-- 1 root root 7021 Jul 5 21:42 modules.alias.bin
-rw-r--r-- 1 root root 69 Jul 5 21:42 modules.ccwmap
-rw-r--r-- 1 root root 105 Jul 5 21:42 modules.dep
-rw-r--r-- 1 root root 312 Jul 5 21:42 modules.dep.bin
-rw-r--r-- 1 root root 52 Jul 5 21:42 modules.devname
-rw-r--r-- 1 root root 73 Jul 5 21:42 modules.ieee1394map
-rw-r--r-- 1 root root 141 Jul 5 21:42 modules.inputmap
-rw-r--r-- 1 root root 81 Jul 5 21:42 modules.isapnpmap
-rw-r--r-- 1 root root 74 Jul 5 21:42 modules.ofmap
-rw-r--r-- 1 root root 15569 Jul 5 21:42 modules.pcimap
-rw-r--r-- 1 root root 43 Jul 5 21:42 modules.seriomap
-rw-r--r-- 1 root root 131 Jul 5 21:42 modules.softdep
-rw-r--r-- 1 root root 10501 Jul 5 21:42 modules.symbols
-rw-r--r-- 1 root root 11712 Jul 5 21:42 modules.symbols.bin
-rw-r--r-- 1 root root 189 Jul 5 21:42 modules.usbmap
drwxr-xr-x 2 root root 4096 Jul 5 21:40 video

I have been checking everything and I think it has something to do with .35-rc4. The script that generates include/linux/version.h is adding that '+' in purpose, so it's not a typo or something else.

[kernelOfTruth]

I have been checking everything and I think it has something to do with .35-rc4. The script that generates include/linux/version.h is adding that '+' in purpose, so it's not a typo or something else.

thanks for the clarification

an annoyance of that whole thing however is that I need to create symbolic links from the "+" folder to the folder without "+" when installing kernel modules (e.g. ati-drivers, virtualbox-modules) and depmod afterwards since it doesn't find the modules automatically

[Waninkoko]

Everything was working fine for me when I tested 2.6.35-rc4-zen1.

[tranquilcool]

I have been checking everything and I think it has something to do with .35-rc4. The script that generates include/linux/version.h is adding that '+' in purpose, so it's not a typo or something else.

thanks for the clarification

an annoyance of that whole thing however is that I need to create symbolic links from the "+" folder to the folder without "+" when installing kernel modules (e.g. ati-drivers, virtualbox-modules) and depmod afterwards since it doesn't the modules automatically

why create a link.
just add + in the Makefile( es. EXTRAVERSION = -rc5+), install kernel modules and then remove it.

[Ant P.]

What git revision do I need to revert to to make my computers' off switches work again? (currently on b33c0e23)

[kernelOfTruth]

Everything was working fine for me when I tested 2.6.35-rc4-zen1.

strange

here you go - another "victim": http://lkml.org/lkml/2010/7/16/290

I have been checking everything and I think it has something to do with .35-rc4. The script that generates include/linux/version.h is adding that '+' in purpose, so it's not a typo or something else.

thanks for the clarification

an annoyance of that whole thing however is that I need to create symbolic links from the "+" folder to the folder without "+" when installing kernel modules (e.g. ati-drivers, virtualbox-modules) and depmod afterwards since it doesn't the modules automatically

why create a link.
just add + in the Makefile( es. EXTRAVERSION = -rc5+), install kernel modules and then remove it.

I'll give that a try - thanks !



could you guys please pull in the latest fixes in zen-upstream (>rc5) ?

being able to test the latest rc-releases in zen on a regular basis would be nice so we could find/track down potential issues with zen addons and upstream


Thanks !


if you don't have the time:

how could I do this with git commands locally ?

[kernelOfTruth]

thanks a lot !

[Ant P.]

I'm getting this compile failure when I try to compile with Classic RCU enabled (Tree RCU works fine):

Code: Select all

  HOSTCC  scripts/mod/file2alias.o
In file included from include/linux/module.h:18,
                 from include/linux/crypto.h:21,
                 from arch/x86/kernel/asm-offsets_64.c:8,
                 from arch/x86/kernel/asm-offsets.c:4:
include/linux/tracepoint.h: In function 'tracepoint_synchronize_unregister':
include/linux/tracepoint.h:80: error: implicit declaration of function 'synchronize_sched'
make[1]: *** [arch/x86/kernel/asm-offsets.s] Error 1

[kernelOfTruth]

thanks for bfq and the recent pulls in zen-upstream - not being in the master-tree it's however not accessible yet by simply pulling stuff ...

update:

seems that too much changed in 2.6.35-rc* compared to 2.6.34:

--
CC drivers/ata/libata-acpi.o
CC mm/init-mm.o
CC drivers/acpi/dock.o
In file included from block/bfq-iosched.c:121:0:
block/bfq-ioc.c: In function ‘cic_free_func’:
block/bfq-ioc.c:58:2: error: ‘struct cfq_io_context’ has no member named ‘dead_key’
block/bfq-ioc.c:61:45: error: ‘struct cfq_io_context’ has no member named ‘dead_key’
block/bfq-ioc.c: In function ‘__bfq_exit_single_io_context’:
block/bfq-ioc.c:102:5: error: ‘struct cfq_io_context’ has no member named ‘dead_key’
block/bfq-iosched.c: In function ‘__bfq_dispatch_requests’:
block/bfq-iosched.c16: warning: passing argument 1 of ‘atomic_inc’ from incompatible pointer type
/usr/src/linux-2.6.35-rc5-zen1_bfq/arch/x86/include/asm/atomic.h:93:20: note: expected ‘struct atomic_t *’ but argument is of type ‘struct atomic_long_t *’
CC mm/bounce.o
make[1]: *** [block/bfq-iosched.o] Error 1
make: *** [block] Error 2
--

[mantoo]

After I pulled today, compiling the kernel fails:

Code: Select all

linux # make && make modules && make modules_install
  CHK     include/linux/version.h
  CHK     include/generated/utsrelease.h
  CALL    scripts/checksyscalls.sh
  CHK     include/generated/compile.h
  CC      kernel/softirq.o
kernel/softirq.c: In function ‘run_ksoftirqd’:
kernel/softirq.c:719: error: implicit declaration of function ‘rcu_note_context_switch’
make[1]: *** [kernel/softirq.o] Error 1
make: *** [kernel] Error 2

something went wrong with http://git.zen-kernel.org/?p=kernel/zen ... 207d22d4fa ?

[Colt45]

After I pulled today, compiling the kernel fails:

Code: Select all

linux # make && make modules && make modules_install
  CHK     include/linux/version.h
  CHK     include/generated/utsrelease.h
  CALL    scripts/checksyscalls.sh
  CHK     include/generated/compile.h
  CC      kernel/softirq.o
kernel/softirq.c: In function ‘run_ksoftirqd’:
kernel/softirq.c:719: error: implicit declaration of function ‘rcu_note_context_switch’
make[1]: *** [kernel/softirq.o] Error 1
make: *** [kernel] Error 2

something went wrong with http://git.zen-kernel.org/?p=kernel/zen ... 207d22d4fa ?

Yeah. I backed that out and it compiles. Have not tested if it boots yet.

...it fails to boot. :/