WarheadsSE wrote:Arktronic wrote:How does this fault manifest itself? I'm running 188.8.131.52 using telzey's merged patchset and haven't noticed any instability.
Press it, hard, with memory & CPU.. compile a kernel on it, with a decent sized swap file. It's got to be long running, heavy resource-hitting. Then it will just *huurrrrk* stop dead, without a fault.
That's not what I saw in my testing with various kernel compilations, but I don't have any trouble believing that the problem exists. Bummer!
There is code in the newer kernels specifically related to handling disk cache blocks that don't get released properly. That could easily cause a soft-lock like you're seeing.
I never fully stress-tested the patchset that I made based on the Silverstone 184.108.40.206 kernel, but if anyone would like it, I can make it available.
For myself, because of the recent availability of cheap sata-based kirkwood machines I'm switching to that platform instead.