Re: Swap partition priority

On 8/18/26 18:03, Joshua Armstrong wrote:
> If there’s no problem with adding an optional parameter to the struct that swapon(2) passes to the kernel

Actually I was thinking about something much simpler and very basic.

unknown_sysctl=0: default behaviour
unknown_sysctl=1: use the devices in order of "swapon" (when the first 
is full, move to the second and so on).

This would not require any modifications to swapon(2), but only to 
swp_pager_getswapspace (besides adding the sysctl, obviously).
Right?

At least this could be a first step and allow for testing.



> IMO. We’d have to be really really careful about testing it under all sorts of conditions because blocking is
 > disallowed in some parts of this code path and it’s really easy
 > withblock I/O to accidentally cause a block.

How would the new strategy differ from the old one in this regard?



  bye & Thanks
	av.