i found out after a couple of tests that it was caused by core 3. when i used it as a triple core with cores 1, 2 and 4 it appeared stable. but it scared me out of using more than the default 2 cores anyway, i'm not interested in an unstable system. i overclocked the cores one at a time up to 4ghz (black edition, unlocked multiplier) and none of them would fail ~15mins of mprime stress test (without changing any voltages)... at 4.1ghz the entire system would crash. debian terminal was the only sign of any instability. i intended to test it further, increasing vcore and northbridge voltage, but lost interest in core unlocking after the error flood (normally i underclock).
]]>I’m having the same problem with an AMD FX-6300 CPU.
I’m fine for now with linux-lts since I discovered that there are VirtualBox kernel modules for the LTS kernel as well.
]]>Mai 06 21:49:22 archpc kernel: [Hardware Error]: MC4 Error (node 0): L3 data cache ECC error.
Mai 06 21:49:22 archpc kernel: [Hardware Error]: Error Status: Corrected error, no action required.
Mai 06 21:49:23 archpc kernel: [Hardware Error]: CPU:0 (15:2:0) MC4_STATUS[Over|CE|MiscV|-|AddrV|-|-|CECC]: 0xdd0144e5001c011b
Mai 06 21:49:23 archpc kernel: [Hardware Error]: MC4_ADDR: 0x00000000a6bf2b84
Mai 06 21:49:23 archpc kernel: [Hardware Error]: cache level: L3/GEN, tx: GEN, mem-tx: RD
This appears every five minutes on a 3.14.x kernel, but way more on a newer kernel so that the computer is not usable any more … my KDE session freezes and the virtual console gets flooded with the above error message.
Can these error messages be suppressed?
]]>@ kokoko3k
You are probably right but my programming skills are close to zero. I will have a look at your link though
You may also have some luck improving cooling, increasing Vcore (risk of hardware damage, blah blah blah) or reducing clock speed.
]]>Anyway, if no dev will help you with a fast reply, you will need to bisect the kernel to, at least, pointing the kernel developers to the problematic commit.
]]>Also, you can find the different loglevel values on the following page: https://www.kernel.org/doc/Documentatio … meters.txt
PS: By the way, since when did you start noticing these errors?
]]>