Michael Kelly, le lun. 12 mai 2025 08:15:10 +0100, a ecrit: > As part of this research I have also come across another feature that arises > when running low on memory. I have a virtual machine with 2GB of RAM and a > very simple test program to gobble 400M of memory per instance (below). I > run 2 of these and as expected each process has a virtual size of ~550M and > an RSS of 400M. I then run a 3rd instance to consume a further 400M. This > causes memory from the 1st of my test programs to have some of its memory > paged out and its RSS drops to about 150M. > > That all seems quite reasonable until you discover that no memory is > actually being sent to swap as there are no messages sent to the pager.
Ouch, that's not supposed to behave so :) > I was so surprised by this that I actually commented out all of the > code in vm_pageout_scan except for the page balancing but it still did > the same. Have you tried commenting the page balancing? The issue could match the issues I'm still experiencing on buildds where I have just disabled swap. > It is the page balancing that is reducing the resident memory > total for my first process. Each page that is moved gets marked > as VM_PROT_NONE and the task's (pmap) resident_count decremented > accordingly. There doesn't seem to be any equivalent increment > associated with the page that it is moved to. That's odd indeed, do you have the series of calls that leads to the decrement? (e.g. a backtrace) Your test case is valid, it'll be useful to investigate further. Samuel
