Run Linux with bad memory
Long ago in the early days of computers, nothing was reliable. Hard disks would have errors that would come right from the factory. The manufacturer even labeled the drives with the bad parts to be entered so DOS could ignore them. It was not too uncommon to have memory chips go bad from time to time.
In my case I installed a modern system and had two sticks of memory both come up with errors after a burn in. My only guess is that it was a bad batch. Here is how I worked around the bad memory to have a fully operational system.
Memtest86+
This is the defacto standard for a memory testing in the linux world. Just install it and pick it from your boot loader to run
apt-get install memtest86+
I have another page on how to run memtest86+ through a serial console if you have a headless configuration with extlinux.
Be sure to use Error Summary mode to get the correct addresses.
Add a kernel parameter
In our case the bad memory is between 000121640a20 - 00012410abb0. Use the kernel option memmap to define the bad region of memory as reserved.
memmap=0x121640a20$0x12410abb0
| Created: 2024-10-15 | Modified: 2024-10-15 |