Hello It's the first bug I submit and I hope I am doing it right. I have a Compal FL90 laptop which as an Intel Core 2 Duo T9300 @ 2,50GHz with 4 GB RAM I have it since 2008 and I never managed to run properly any Linux distribution on it due to the fact that the computer freezes unless the mouse is moved or a key is pressed. I have found a workaround time ago which is adding "nolapic_timer clocksource=jiffies" to the startup command line in GRUB. It affected other computer models in the past. It also affects booting as sometimes the computer will freeze as well. I don't know if the startup command line affects battery life but I think it drains it faster This has been present on all version through time and I think it's a similar issue like this "https://bugzilla.kernel.org/show_bug.cgi?id=21032" which affects other processors. I expected someone else would have the same problem and submit the bug, but since it's been going for such a long time I decided to submit it myself. Please let me know if you need any info. I am not an expert on Linux but I am sure I can help. Thanks
(In reply to Alberto Castillo from comment #0) > Hello > > It's the first bug I submit and I hope I am doing it right. > > I have a Compal FL90 laptop which as an Intel Core 2 Duo T9300 @ 2,50GHz > with 4 GB RAM > > I have it since 2008 and I never managed to run properly any Linux > distribution on it due to the fact that the computer freezes unless the > mouse is moved or a key is pressed. > please be clearer about "computer freezes", how do you know your computer is freezed if you don't move the mouse/keyboard? > I have found a workaround time ago which is adding "nolapic_timer > clocksource=jiffies" to the startup command line in GRUB. It affected other > computer models in the past. It also affects booting as sometimes the > computer will freeze as well. > so the computer boots well without those parameters?
Hello It's pretty difficult to explain. It's easy to know that the computer is frozen when it's doing some kind of activity, such as a filecopy, download, or even when installing a Linux distribution. The hard drive light will go off and will start blinking and the computer will start reacting as soon a key is pressed or the mouse is moving. In many cases the computer won't even boot with those parameters, althought in most cases it does. I can find more cases of this for you if you need. It's a difficult issue to explain. Thank you!
please attach the dmesg output after boot. please attach the output of "cat /proc/interrupts"
Oki, I will need to put this on hold. I am in Denver at the moment and I didn't bring my laptop with me. As soon as I go back I will do this
I just returned to Spain. I will do this and report back
Hello I attached the needed attachments. I am sorry I took long to submit them. I was away
Created attachment 188851 [details] dmesg dmesh output
Created attachment 188861 [details] interrupts "cat /proc interrupts" output
Both files have been created without using the "nolapic_timer clocksource=jiffies" startup flag
Should I do something else? Thanks!
Hello?
does the problem still exist if you boot with "idle=poll"?
Hello Sorry for the delay Apparently it works I have modified /etc/default/grub file and set: GRUB_CMDLINE_LINUX_DEFAULT="quiet splash idle=poll" Apparently it works well. Thanks
I noticed you are using quite old kernel 3.13, please download the latest kernel at https://www.kernel.org and test if this issue still exist, thanks.
could you also attach your dmidecode(you might need to install this tool) result and cat /proc/cpuinfo?
I am unable to install a new kernel. I will get the dmiecode data
(In reply to Alberto Castillo from comment #2) > Hello > > It's pretty difficult to explain. It's easy to know that the computer is > frozen when it's doing some kind of activity, such as a filecopy, download, > or even when installing a Linux distribution. > The hard drive light will go off and will start blinking and the computer > will start reacting as soon a key is pressed or the mouse is moving. > > In many cases the computer won't even boot with those parameters, althought > in most cases it does. > > I can find more cases of this for you if you need. It's a difficult issue to > explain. > > Thank you! According to this description, it looks like the system has been fallen auto-asleep if you did not press any key? After un-freeze, can you provide your dmesg? Since there are quite many fix after 3.13, I think running latest kernel version would be helpful.
Does intel_idle.max_cstate=1 appended in grub work for you?
Yes intel_idle.max_cstate=1 works for me, however not that well I couldn't run the latest kernel yet to see if it worked, sorry. My distro doesn't have it.
Do you need anything else? I doubt changes are on the latest kernel as these issues have been going on since I tried Linux on this machine (2008) Thanks
Please provide your dmesg as Comment 17. Compiling the kernel is not difficult IMO, you can download the source code from kernel.org, and it would not take too much time to compile(I think at most 1 hour for compiling it)
ping
Alberto, please check if latest kernel work for you, because there are too many changes in the linux day after day, and many bugs have been fixed(maybe also introduced during the linux lifecycle) if you still encounter this issue on new kernel, please feel free to open it.
Hi The laptop broke one year ago and I repaired it 2 weeks ago. I tried a newer version of the kernel included on Ubuntu MATE 16.04 LTS. I still had the same issue. What else could I try? Thanks
please attach the output of "cat /proc/cpuinfo"
(In reply to Alberto Castillo from comment #24) > Hi > > The laptop broke one year ago and I repaired it 2 weeks ago. I tried a newer > version of the kernel included on Ubuntu MATE 16.04 LTS. I still had the > same issue. > > What else could I try? > > Thanks please provide your dmesg
Created attachment 256731 [details] cpuinfo cpuinfo.txt
Created attachment 256733 [details] dmesg dmesg output
Created attachment 256735 [details] uname uname output
Thank you
Hello Is there anything else you need? Thanks
please attach the lspci output
#define INTEL_FAM6_CORE2_PENRYN 0x17 this is a pretty old platform that intel_idle driver does not support [ 5.807750] intel_idle: does not run on family 6 model 23 please attach the output of "grep . /sys/devices/system/cpu/cpuidle/*" to confirm it is the acpi cpuidle driver running on your platform. please attach the acpidump as well. please also confirm the problem still exists with the latest BIOS version available. TBH, given this is a really old processor and it is probably because the processor c-state is broken on this specific model, I'm not sure if we can find necessary support for this model any more...
Created attachment 257021 [details] lspci output lspci output
Created attachment 257023 [details] output of "grep . /sys/devices/system/cpu/cpuidle/*" grep . /sys/devices/system/cpu/cpuidle/*
Hello acpidump gets an empty file, this is what shows on the terminal when the command is ran ubuntu-mate@ubuntu-mate:~/Desktop$ acpidump Could not open table file: /sys/firmware/acpi/tables/APIC2 Could not get ACPI tables, AE_ACCESS
All files were uploaded I don't know, but this has always happened since I got the device in 2008. Every Linux distribution always had this problem since then. The only way to solve the issues were to add to boot command line in GRUB: nolapic_timer clocksource=jiffies Or any of these: idle=poll intel_idle.max_cstate=1 However, the one that gets the best results is "nolapic_timer clocksource=jiffies" not sure why The others still cause some glitches. Thanks
Forgot to mention that yes, I am using the latest BIOS version available (1.18)
(In reply to Alberto Castillo from comment #37) > All files were uploaded > > I don't know, but this has always happened since I got the device in 2008. > Every Linux distribution always had this problem since then. > > The only way to solve the issues were to add to boot command line in GRUB: > nolapic_timer clocksource=jiffies > > Or any of these: > idle=poll > intel_idle.max_cstate=1 > I'm confused, you're using acpi_idle driver rather than intel_idle driver on this platform, but why intel_idle.max_cstate=1 makes difference...
Hi Thanks for you response. Igot intel_idle.max_cstate=1 from another bug repoted by someone else. It did work for me but not as good as nolapic_timer clocksource=jiffies These 2 didn't work as good: idle=poll intel_idle.max_cstate=1
As you are using acpi idle driver, the intel_idle driver will not load, so intel_idle.max_cstate would not take effect in your case. As nolapic_timer works for you, I assumed this should be a problem in apic driver. Could you please help test on latest upstream kernel? I remembered that the apic maintainer has pushed some modification recently.