You are not logged in.
Same problem here on a Toshiba Z30-B laptop (i7). The freeze only appears when the internal display is in use.
Offline
I think I found the reason for system freezing with 4.16*(at least in my case)! It is the new default setting of SATA Active Link Power Management - Power management - ArchWiki. I set it back to max_performance (the default setting is med_power_with_dipm since 4.16* kernels) and I haven't had a single freeze in two days with the newest kernel! It would be nice to know if ti is the chipset itself (Intel i5 Broadwell) that can't handle power saving or it is the Samsung 850 Evo SSD inside my laptop that doesn't like it and if the problem can be fixed somehow as the difference in power consumption is significant! When power saving on Powertop reports power consumption of less than 3.5W when idle and when off - about 5.5W!!! About 2W difference.. Any ideas guys?
Offline
Samsung SSDs are notorious for issues with the SATA Link power management, it's definitely the SSD that doesn't handle it.
Online
The problem still persists with 4.16.11 on my side
@stoychod : I will still try the solution you indicated, with the 4.16.12 kernel.
I will keep you informed
Offline
The problem is solved on my side since kernel 4.16.12 and the power management tip.
No crash since yesterday.
But I do not know which one solved it...
Offline
The problem is solved on my side since kernel 4.16.12 and the power management tip.
No crash since yesterday.
But I do not know which one solved it...
Just for you, I switched back to the default setting (I am on the same kernel) and it took about 15 min to get another crash... So I am 99.99% sure it is the new default ALPM setting(at least for me)... On top of that I recall some time ago (probably at least 2 years , had just bought the laptop and installed Arch) I had exactly the same crashes. It was driving me crazy until I figured it was some power tweak from TLP. At the time I didn't bother figuring out wich one exactly and just uninstalled it. Now I realize it must've been setting ALPM to min_power (med_power_with_dipm wasn't available ta the time as fa ras I remember)! By the way, I have been testing ALPM set to medium_power for a couple of days and no crashes so far! So you can give it a try (to get at lest some power saving). Would you mind sharing what HDD/SSD you have on your machine, by the way? As V1del suggests it is probably my SSD not being able to handle SATA power saving so I am curious about yours....
@V1del Thanks a lot, mate , good to know that...
Offline
Thank you for the informations and for making the manipulation
HDD infos :
Model Family: Toshiba 2.5" HDD MQ01ABD...
Device Model: TOSHIBA MQ01ABD100
Firmware Version: AX1P4M
User Capacity: 1,000,204,886,016 bytes [1.00 TB]
Sector Size: 512 bytes logical/physical
Rotation Rate: 5400 rpm
ATA Version is: ATA8-ACS (minor revision not indicated)
SATA Version is: SATA 2.6, 3.0 Gb/s (current: 3.0 Gb/s)
Offline
I'd like to verify that the ALPM setting really is the culprit, but as of 4.16.11-1, the freeze is really hard to reproduce (but it's still there).
- If anyone has a tip on how to provoke ALPM-related errors, then please leave a note. My naive approach to loop "read/write --> drop caches --> sleep a few secs" did not do the trick.
- My system also has a Samsung 850 EVO SSD, so that fits the picture. However, I cannot understand that the system is 100% stable when running on AC power && using external display && wired network (I double checked that link_power_management_policy is med_power_with_dipm when on AC power).
Last edited by azara (2018-05-31 18:46:58)
Offline
FWIW med_power_with_dpm is definitely not the default without any other configuration. It is the default of newer versions of TLP, but not from kernel config itself.
Apparently not, see below
Last edited by V1del (2018-05-31 19:34:16)
Online
No freezes so far on 4.16.12-1
Update: yes an hour of work and it froze again.
Last edited by novica (2018-05-31 19:33:31)
Offline
FWIW med_power_with_dpm is definitely not the default without any other configuration.
Hm, I think it is (running linux 4.16.11-1):
$ zgrep LPM /proc/config.gz
CONFIG_SATA_MOBILE_LPM_POLICY=3
"3" maps to "Medium power with Device Initiated PM enabled" acc. to kernel sources.
Am I missing something?
Offline
Ouh, it is! I just double checked quickly on my desktop didn't know they would do that kind of distinction, sorry for the noise.
Online
V1del wrote:FWIW med_power_with_dpm is definitely not the default without any other configuration.
Hm, I think it is (running linux 4.16.11-1):
$ zgrep LPM /proc/config.gz CONFIG_SATA_MOBILE_LPM_POLICY=3
"3" maps to "Medium power with Device Initiated PM enabled" acc. to kernel sources.
Am I missing something?
Also from ArchWiki:
Since Linux 4.15 there is a new setting called med_power_with_dipm that matches the behaviour of Windows IRST driver settings and should not cause data loss with recent SSD/HDD drives. The power saving can be significant, ranging from 1.0 to 1.5 Watts (when idle).It will become a default setting for Intel based laptops in Linux 4.16 [5]
As I mentioned earlier I've got two Intel based laptops and can confirm that without any configuration form my side the old one defaults to min_power and the new one to med_power_with_dipm and both have the:
CONFIG_SATA_MOBILE_LPM_POLICY=3
in their kernel config!
I rarely use my laptop connected to AC so I don't know if that makes any difference but I can give it a try and see if the system still crashes...
Offline
As I mentioned earlier I've got two Intel based laptops and can confirm that without any configuration form my side the old one defaults to min_power and the new one to med_power_with_dipm and both have the:
CONFIG_SATA_MOBILE_LPM_POLICY=3
in their kernel config!
As you noted the following if must evaluate to true for mobile_lpm_policy so for the old laptop it would indicate it does not assuming both are using 4.15+
if ((hpriv->flags & AHCI_HFLAG_IS_MOBILE) &&
mobile_lpm_policy >= ATA_LPM_UNKNOWN &&
mobile_lpm_policy <= ATA_LPM_MIN_POWER)
ap->target_lpm_policy = mobile_lpm_policy;
Offline
Also seem to have the same symptoms on a Lenovo Thinkpad 11e (1st gen). Originally installed Arch on it a few days ago and have had three freezes so far.
Update: no freezes since installing linux-lts kernel a few days ago (not long enough to be sure yet, though).
Another update: Got a freeze with linux-lts, now trying stoychod's fix.
Last edited by Charn Xuli (2018-06-11 11:36:00)
Offline
Got another freeze after stoychod's fix. Will now cease using Arch Linux on this machine, so that I can detect whether other Linux distros have a similar problem.
Now believe the freezing was caused by a faulty SATA hard drive.
Last edited by Charn Xuli (2018-06-17 18:42:24)
Offline
Definitely no more freeze here with ALPM set to max_performance. Thanks stoychod for the hint! My next SSD won't have a SAMSUNG tag for sure...
Last edited by azara (2018-06-13 19:33:37)
Offline
Have anyone tried the 4.17 kernel yet? Does it have the same issue?
Offline
Also seem to have the same symptoms on a Lenovo Thinkpad 11e (1st gen).
Thinkpad 11e has a Intel Celeron N2940 Processor = Baytrail. There is a well known bug. Baytrail doesn't handle c-states correctly. See:
No new thing under the sun
Offline
Thinkpad 11e has a Intel Celeron N2940 Processor = Baytrail. There is a well known bug.
Thanks, I obviously wasn't aware of this.
I tried installing several other linux distros on the SATA hard drive and got freezes so then tried installing Arch on a USB flash drive. After debugging the touchpad, sound and dhcpd (mostly using questions from this forum and modprobe blacklisting), everything's working perfectly (with a reasonable amount of testing time behind it, although a freeze is still possible I suppose).
Now trying the suggested fix.
Update: As identified by HaCeMei, the Lenovo Thinkpad 11e (1st gen) freezing was caused by the Baytrail bug and seemingly wasn't related to this kernel update.
Last edited by Charn Xuli (2018-07-04 12:34:13)
Offline
Definitely no more freeze here with ALPM set to max_performance.
No freezes for me as well with ALPM set to max_performance.
Offline
Hi
Same problem 4.16.8-1-ck-broadwell
This has been happening for a few days, especially when I'm browsing with firefox
I have the same problem with my T450, it still happens even with the latest kernel: 4.18.12-arch1-1-ARCH.
Does anyone knows if it's related to tpm? Because I see some TPM error messages in dmesg, my T450 is broadwell as well.
Last edited by malcontent (2018-10-13 20:21:40)
Offline
I've tried disabling TPM on the BIOS, and while the dmesg error is gone, I still got a freeze after a week of using my laptop.
Offline
Just switched /sys/class/scsi_host/host*/link_power_management_policy to max_performance, we'll see how it goes.
Offline