[SOLVED] dell optiplex 7010 pagefault pf-sources 6.9.7

[sjbiaga]

Hi,

I have an Dell OptiPlex 7010 Corei5-3470S for 9 months now, and since 4 months ago it started to give pagefaults.
I used pf-sources from the start, but I kept updating it regularily. I even switched to gentoo-sources, to no avail. (It hanged once or twice after I bought it - second hand, new SSD.)

It just hangs, even every half an hour: image freezes, the red light at the optical USB mouse turns off, sometimes the sound goes on in a loop, sometimes it doesn't resume from hibernation; twice it corrupted the / filesystem that it didn't boot; once I had to manually answer Y/N to fsck.ext4.

journalctl shows the 'BUG' with a full debug stack trace, but not always. Sometimes it even shows me on what CPU it occurred. Once it hanged when I booted with Gentoo Live USB, while I was 'fsck.ext4'-ing the partitions on the SSD.

In the kernel cmdline I appended the parameters: "acpi_osi=Linux processor.max_cstate=2".

What could have changed to cause these hangs?

Thanks,
Sebastian

[Hu]

Welcome to the forums. As a first guess, this sounds like a hardware fault. Have you run a memory test on it to confirm that its RAM is stable? Bad RAM can cause random memory corruption, which can cause a variety of weird and inconsistent failures.

[sjbiaga]

Yes, the memory test from the Gentoo Live USB and it finished OK.

[Hu]

How many passes did the memory test run before you stopped it?

[sjbiaga]

All of the passes, I didn't stop it. I inserted the log with the BUG.

[sjbiaga]

I have followed your suggestion and did a memtest twice and it failed twice! Do you think it's just memory failure (in which case I'll have to just change it) or the memory circuits?
It fails after 94% at about 52k errors, so the circuits work 94% of the time, after which I guess it's just bad memory.

[Hu]

If the fault address is consistent, that usually implicates the RAM stick which backs that address. If the fault address is inconsistent, or you have large numbers of faults, that usually implicates other components, such as excessively aggressive RAM timings. RAM needs a certain amount of time to complete each operation before it can produce the correct value. The exact time requirement depends on the quality of the manufacturing. Most firmware makes it easy to configure the RAM controller to use more or less aggressive timing, and if you use too short a timing, you get unstable results.

[sjbiaga]

It's been a week of uptime after the memory DIMM has been replaced and it's Ok now. Thanks.