Tinypliny's Journal

12/11/2011 11:10 #55717

Fedora 16: "Soft lockup"
Category: linux

Okay. It just happened AGAIN. I could not even use the REISUB even though I have now enabled it on the system. None of the other methods of system rescue that I listed in my earlier post work with Fedora 16 when it freezes.

This is so insane. I don't understand what is so wrong. NEVER has been an OS so crappy as Fedora 16. I am switching. This is eating into my productivity.

Attached is the text file with the log of the "Oops" errors, which will hopefully show up when (e:paul) sees this message.

This is how it all starts:

Dec 11 10:57:07 why kernel: [  832.169994] BUG: soft lockup - CPU#2 stuck for 23s! [kworker/2:2:520]
Dec 11 10:57:07 why kernel: [  832.169999] Modules linked in: ppdev parport_pc lp parport fuse 8021q fcoe libfcoe libfc scsi_transport_fc scsi_tgt garp stp llc nf_conntrack_ipv4 nf_defrag_ipv4 ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables snd_hda_codec_hdmi snd_hda_codec_realtek snd_hda_intel snd_hda_codec snd_hwdep ir_lirc_codec lirc_dev snd_seq snd_seq_device ir_mce_kbd_decoder rc_rc6_mce ene_ir snd_pcm ir_sony_decoder ir_jvc_decoder snd_timer ir_rc6_decoder ir_rc5_decoder arc4 ir_nec_decoder uvcvideo videodev media rc_core rtl8192ce rtl8192c_common rtlwifi mac80211 wacom cfg80211 iTCO_wdt r8169 snd mii v4l2_compat_ioctl32 sparse_keymap iTCO_vendor_support rfkill jmb38x_ms memstick microcode soundcore snd_page_alloc i2c_i801 joydev uinput sdhci_pci sdhci mmc_core i915 nouveau ttm drm_kms_helper drm i2c_algo_bit i2c_core mxm_wmi wmi video [last unloaded: scsi_wait_scan]
Dec 11 10:57:07 why kernel: [  832.170017] CPU 2 
Dec 11 10:57:07 why kernel: [  832.170017] Modules linked in: ppdev parport_pc lp parport fuse 8021q fcoe libfcoe libfc scsi_transport_fc scsi_tgt garp stp llc nf_conntrack_ipv4 nf_defrag_ipv4 ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables snd_hda_codec_hdmi snd_hda_codec_realtek snd_hda_intel snd_hda_codec snd_hwdep ir_lirc_codec lirc_dev snd_seq snd_seq_device ir_mce_kbd_decoder rc_rc6_mce ene_ir snd_pcm ir_sony_decoder ir_jvc_decoder snd_timer ir_rc6_decoder ir_rc5_decoder arc4 ir_nec_decoder uvcvideo videodev media rc_core rtl8192ce rtl8192c_common rtlwifi mac80211 wacom cfg80211 iTCO_wdt r8169 snd mii v4l2_compat_ioctl32 sparse_keymap iTCO_vendor_support rfkill jmb38x_ms memstick microcode soundcore snd_page_alloc i2c_i801 joydev uinput sdhci_pci sdhci mmc_core i915 nouveau ttm drm_kms_helper drm i2c_algo_bit i2c_core mxm_wmi wmi video [last unloaded: scsi_wait_scan]
Dec 11 10:57:07 why kernel: [  832.170017] 
Dec 11 10:57:07 why kernel: [  832.170017] Pid: 520, comm: kworker/2:2 Not tainted 3.1.4-1.fc16.x86_64 #1 TOSHIBA Satellite P750/PEQAA
Dec 11 10:57:07 why kernel: [  832.170017] RIP: 0010:[]  [] do_raw_spin_lock+0x21/0x25
Dec 11 10:57:07 why kernel: [  832.170017] RSP: 0018:ffff88022fa83e60  EFLAGS: 00000287
Dec 11 10:57:07 why kernel: [  832.170017] RAX: 0000000000003020 RBX: ffff88021847644a RCX: 0000000000006a1d
Dec 11 10:57:07 why kernel: [  832.170017] RDX: 000000000000301f RSI: ffff880218451ff8 RDI: ffff880218451d74
Dec 11 10:57:07 why kernel: [  832.170017] RBP: ffff88022fa83e60 R08: ffffffff813cfb77 R09: 000000007be27d4a
Dec 11 10:57:07 why kernel: [  832.170017] R10: 000000007be27d4a R11: ffffffff81b5d230 R12: ffff88022fa83dd8
Dec 11 10:57:07 why kernel: [  832.170017] R13: ffffffff814be49e R14: ffff88022fa83e60 R15: ffff880218450540
Dec 11 10:57:07 why kernel: [  832.170017] FS:  0000000000000000(0000) GS:ffff88022fa80000(0000) knlGS:0000000000000000
Dec 11 10:57:07 why kernel: [  832.170017] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
Dec 11 10:57:07 why kernel: [  832.170017] CR2: 00007f7e8a6d7000 CR3: 0000000001a05000 CR4: 00000000000406e0
Dec 11 10:57:07 why kernel: [  832.170017] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
Dec 11 10:57:07 why kernel: [  832.170017] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Dec 11 10:57:07 why kernel: [  832.170017] Process kworker/2:2 (pid: 520, threadinfo ffff88021a3a0000, task ffff88021d9ddcc0)
Dec 11 10:57:07 why kernel: [  832.170017] Stack:
Dec 11 10:57:07 why kernel: [  832.170017]  ffff88022fa83e70 ffffffff814b7216 ffff88022fa83e90 ffffffffa0277866
Dec 11 10:57:07 why kernel: [  832.170017]  ffff880218451ff0 ffff880218451ff8 ffff88022fa83ea0 ffffffffa02793cb
Dec 11 10:57:07 why kernel: [  832.170017]  ffff88022fa83ed0 ffffffff8105cdd3 ffffffff81a040b0 ffffffff81a040b0

[txt]1211/SoftBugFedora16.txt[/txt]

Update:
I located this thread on Fedora forum with several people having random freezes on Fedora 16 almost everyday, sometimes multiple times a day, with varied hardware.

In addition, all the system rescue effects including ssh-ing remotely into the system are not working during these freezes. Nothing except a hard boot via the power button rescues the system. This is hardly some random whine isolated just on my laptop.I have added my information to the thread.

There is a bug-report for these odd freezes as well:

One of these freezes now has a bugfix in the updates:

Going to update my system and see if all this happens again.

EDIT 2:
More updates from the FedoraForum: A couple other users think the wireless module is the problem child.

Also, one of them reviewed my logs and said:

You seem to have rtl driver errors and also some apic timer problems.
Try adding "noapic" option to the kernel command line
Edit /boot/grub2/grub.cfg and put "noapic" you'll see some lines like
linux /boot/vmlinuz-3..... buncha options LAN=en_US-UTF-8 noapic

I am going to try that see if it helps.

tinypliny - 12/12/11 18:28
No wonder the wireless performance of Fedora 16 is SO mediocre. Its a driver issue!

tinypliny - 12/12/11 18:27
More insights from the fedora 16 freeze thread seem to indicate that it is a problem with the wireless driver module.

I knew it. That is probably the only thing that is substantially different in F16 compared with F15 (I had compile the driver for the wireless myself in F15) But apparently they have messed up including this driver module in the kernel.

tinypliny - 12/12/11 10:30
Woohoo! Bugfix for this particular freeze!!!!!!!!!!!! :::link:::

tinypliny - 12/11/11 20:30
This thread is filled with people having the same Fedora 16 freezes multiple times a day on a wide variety of hardware: :::link::: It would be interesting to see what we all have in common.

tinypliny - 12/11/11 11:49
The system crashed and I recovered this log from /var/log/messages. There is no PID 520 right now when I do ps -aux
The only things in the 500s range are:

root 500 0.0 0.0 0 0 ? S 11:00 0:00 [kauditd]
root 501 0.0 0.0 17760 1960 ? Ss 11:00 0:00 /sbin/udevd
root 509 0.0 0.0 0 0 ? S 11:00 0:00 [flush-253:1]
root 512 0.0 0.0 23200 908 ? Ss 11:00 0:00 /lib/systemd/systemd-stdout-syslog-bridge
root 523 0.0 0.0 0 0 ? S 11:00 0:01 [kworker/2:2]

Don't PIDs change for processes in every session? Or do they remain stable i.e every process always has the same PID. If they change, there is no way of actually finding out what PID 520 is because the system would have crashed and I would have rebooted it all before accessing /var/log/messages.

Am I missing something here?

uncutsaniflush - 12/11/11 11:31
At the risk of asking an obvious, have you bothered to find out what process 520 is? It might be helpful to know what process is hogging your cpu.

tinypliny - 12/11/11 11:28
I am on Chrome 100% of the time. :/ I write everything nowadays in google docs and use the final revisions to port into M$hit Office and LaTeX.

I could say that I am the ideal consumer for chrome books but then I need the hard drive and the analyses software.

I was happy with fedora 15. I was with it for a month and a half. It did occasionally throw errors but nothing as major as this.

paul - 12/11/11 11:18
I swear its only crashed on me once this whole time, and I think it had to do with my dock becoming disconnected.

I do use chrome but not as much as you.

12/10/2011 22:43 #55694

Justin Bieber
Category: music

What a beautiful voice and wonderful talent

turned into rot.

by fame.

PSL: I actually liked that Christmas song before I saw the video. One word. Ew. I guess Mariah Carey is more pleasant on the radio that on screen. And I don't even want to talk about that little sweet kid who is clearly being overproduced and droned over by pop. What a complete waste of his talents. It's painful to watch.

tinypliny - 12/11/11 11:19
I have to say though west 34th St is pretty magical (like consumerist crazy-buying magical). I went absolutely bonkers in the uniqlo store.

metalpeter - 12/11/11 10:30
It isn't product Placement in a Video the Video is labeled ad.... It is an ad for Macy's and it is his ad...

tinypliny - 12/11/11 09:27
Most music videos seem to require that all females be dressed like hookers

I don't know. Something about not being dressed like a hooker while singing christmas songs is a ton more appealing to me. You are singing a heartfelt sincere song, all the showing off and gestures seem out of place and context.

tinypliny - 12/11/11 09:23
The product placement on that video is atrocious. But then Christmas is clearly about consumerism.

metalpeter - 12/11/11 09:13
She does a great version of this song but when did Justin get added to it... Find the one of her alone it is far better..... She looks great but this out fit doesn't show off her boobs the right way?.... They are so trying to be like the Glee Version of this song .... And they Fail.... Can't say to much cause it is supposed to be an ad for a Charity Thing Macy's does or maybe for them and then you notice the charity still bad though....

12/10/2011 09:38 #55678

When they crossed the himalayas...
Category: the odes

Ha. Yours have just 4 arms?
Ours have 5x more arms and 20x hands.

Because we are better.

Made in China.

paul - 12/10/11 13:15
And this is exactly what scares me about China. That one doesn't even have a lot of arms compared to the crazy one.

12/11/2011 09:15 #55716

Enabling REISUB on Fedora 16
Category: linux

REISUB did not work all these days on Fedora 16 because it was not enabled. UGH.

See if it is enabled with
cat /proc/sys/kernel/sysrq

It will give you either 0 or 1.
If it is 0, enable it with:

sudo su
(enter password)
echo 1 > /proc/sys/kernel/sysrq

Next time your linux machine with Fedora 16 freezes or misbehaves in general, you are set to use the REISUB weapon.

Alt + PRTSC (or the SYSRQ key) +
R
E
I
S
U
B

-> issued in around 6 seconds with a second in between each alphabet.

Other ways to get around freezes and misbehaviour

Ctrl+Alt+Backspace: End the X session

Ctrl+Alt+F1 (or any function key from F2 to F6) followed by
ps aux or
top
find PIDs of processes
kill processes one by one with their PID
kill -9 PID-goes-here

paul - 12/11/11 11:20
Often you can just SSH in from a remote computer to fix GUI issues as well.

12/10/2011 23:42 #55714

Fedora 16 "Oopses"
Category: linux

What irony. Just as I was trying to respond to (e:uncutsaniflush) about how he likes Fedora 16, my system froze up completely. First the mouse was still moving across the screen but I was unable to click anything. Then everything went unresponsive. I had to hard boot and then the whole system crashed. Splendid.

Take a look at the attached /var/log/messages. It found "oopses".
[txt]1211/Oopses.txt[/txt]

Where do I file such crashes and freezes? It is definitely a bug that either Fedora or Chrome people need to sort out. This has never happened in Fedora 15 which almost never crashed. Surely, this is a problem for many people? I have a feeling it is chrome-related because freezes and crashes always happen when I am on Chrome.

PS: (e:Paul): the txt attachments are broken. I tried pasting the log before in the message body and it did not work out. Some quirk made a zillion copies of my post. Every time I would click the edit button and submit some edits, it would make an extra copy and post it.

tinypliny - 12/12/11 10:56
% But of course, you knew all this so you don't need me to tell you.

heh, you overestimate how much I know. the truth is close to nothing. I am just barely getting by with all this linux mania. there is too much to learn and too little time.

tinypliny - 12/12/11 10:55
-- Yum provides readahead lists the following:

systemd-37-3.fc16.x86_64 : A System and Service Manager

-- I disabled selinux long ago because it was annoying me.

-- My chrome has the following extensions:
Autocomplete
Delicious
Lazarus
Link Grabber
Middle Button Scroll
OWA chrome companion
Tab Glutton

None of them are for optimizing browsing. When I disable them, chrome starts even faster and pages load around the same speed.
I will try browsing without them today to test it out. Thanks for the suggestion.

-- How recently was the system updated as recommended?
Last week.
-- Are there packages held back?
hmm? what does that mean?
-- Are there packages from both stable and testing?
only stable.
-- Are there packages (especially kernel modules) compiled locally from 3rd party sources?
No, I didn't install any extra drivers this time.
-- Are there packages installed by force despite depends probs?
No. Whatever whined, I didn't install.

uncutsaniflush - 12/11/11 22:14
Oh yeah, are you doing selinux and something you use needs "readahead"? Sometimes selinux doesn't like readahead. I think I had that prob in Fed 13 or Fed 14. I don't think I fixed it, I just gave up.

I think there is a solution but I don't know what it is.

uncutsaniflush - 12/11/11 22:11
Back in the day when I was just a wee lad and liked breaking betas for fun, the big thing was being able to reproduce crashes and errors on demand to help figure out what is actually going on.

Historically, Browser extensions can be bugbears. I assume that you are using Chrome with extensions enabled. If a browser with extensions crashes or is suspected of causing crashes, usually the first thing to do is disable the extensions and try surfing the same websites that led to the original crash. If stuff crashes, its probably not the extensions; but if it doesn't it might be one of the extensions.

Additionally, in researching bugs and crashes, it is helpful to know how updated and how vanilla the running system is. How recently was the system updated as reccommended? Are there packages held back? Are there packages from both stable and testing? Are there packages (especially kernel modules) compiled locally from 3rd party sources? Are there packages installed by force despite depends probs?

But of course, you knew all this so you don't need me to tell you.

paul - 12/11/11 11:21
I will add the text thing back. I think I overwrite it when some resent updates got merged.

I don't see how the duplicate thing would happen.