- May 22, 2014
- 19
- 0
- 4,520
I have a rather interesting issue. A laptop came to me that would lock up when booting Win7 in normal mode. No BSOD, just hard lock at some point in boot. Safe mode was fine, no errors of not in the event logs. A windows reinstall from factory recovery had been attempted by others. This ran, but the issue persisted after the recovery.
First thought is bad hardware that is not active during safe mode or install, or possibly a driver (unlikely from factory recovery though).
Ran a memtest, and found some memory errors at ~2gb. Problem found, easy solution right? Wrong. Pulled the memory access panel off to find one 2GB stick when the PC has 6GB. Come to find out the other 4GB is soldered to the MOBO. Silly Samsung laptop. It is a nice core i7 laptop but no way to replace the bad ram (and I isolated it to the soldered down chips), by this point several memtest passes have been run, all pointing to memory at ~2gb mark.
That said, in the middle of testing, I decided to run the windows memory diagnostic to see if it gave additional info. The machine pulled a lazurus. That diag found no problems and windows began to boot correctly. I stress tested the machine, checking for overheating, etc. Everything worked fine. Further passes of memtest came up clean.
My suspicion is an intermittently failing chip. I would replace, but obviously cannot.
No, I know linux has a way to pass kernel parameters to ignore RAM addresses. Presumably for similar issues in embedded devices.
Windows has boot options (BURNMEMORY or TRUNCATEMEMORY, vor XP/Vista and above), these only cut off all access above a certain address that I am aware of. Does anyone know of a way in windows to just isolate and not use a specific range of memory addresses? That way if the problem recurs, the laptop is still usable. Truncating is not an option as you would be stuck with <2GB of RAM.
Thanks!
First thought is bad hardware that is not active during safe mode or install, or possibly a driver (unlikely from factory recovery though).
Ran a memtest, and found some memory errors at ~2gb. Problem found, easy solution right? Wrong. Pulled the memory access panel off to find one 2GB stick when the PC has 6GB. Come to find out the other 4GB is soldered to the MOBO. Silly Samsung laptop. It is a nice core i7 laptop but no way to replace the bad ram (and I isolated it to the soldered down chips), by this point several memtest passes have been run, all pointing to memory at ~2gb mark.
That said, in the middle of testing, I decided to run the windows memory diagnostic to see if it gave additional info. The machine pulled a lazurus. That diag found no problems and windows began to boot correctly. I stress tested the machine, checking for overheating, etc. Everything worked fine. Further passes of memtest came up clean.
My suspicion is an intermittently failing chip. I would replace, but obviously cannot.
No, I know linux has a way to pass kernel parameters to ignore RAM addresses. Presumably for similar issues in embedded devices.
Windows has boot options (BURNMEMORY or TRUNCATEMEMORY, vor XP/Vista and above), these only cut off all access above a certain address that I am aware of. Does anyone know of a way in windows to just isolate and not use a specific range of memory addresses? That way if the problem recurs, the laptop is still usable. Truncating is not an option as you would be stuck with <2GB of RAM.
Thanks!
Facebook
Twitter
Instagram