PageWriteWrapper: Failed to write page X: General system error

[dalmemail]

The log gets full of this error when the VMs end their execution. This seems unrelated to the problems of getting the VMs run on QEMU.
This happens probably when delete_area() gets called. Should our caches be temporary caches?
Why are pages written to disk when an area is deleted?

[waddlesplash]

This happens if the areas are memory-mapped files on disk with changes set to write back. (It also may happen in the case of swap usage, but I think you should have everything locked into memory for the VM so that shouldn't matter?)

[waddlesplash]

I don't know if the fix is the right one here. Non-swappable caches shouldn't try to use the PageWriteWrapper at all. Some other setting may be wrong.