According to the man page,
mincore() returns a vector that indicates whether pages of the calling process's virtual memory are resident in core (RAM), and so will not cause a disk access (page fault) if referenced.
But, I'm unclear as to the exact meaning of this -- specifically does RAM refer to the target process's virtual memory, or any processes mapped memory?
Say you have a tmpfs file (tmpfs.file) that is shared between two process's -- process A and process B. Process A maps tmpfs.file, but only accesses page 0. Now process B accesses page 2. And no one accesses page 1. What would be the expected output of mincore as run by process A -- would it be [1,0,0], or [1,0,1]? Technically, page2 exists in process B's RAM, or on the system's RAM, but process A might not have a virtual memory mapping to it, so I'm not sure if this counts... (Tests show it does appear, but I'd like to be sure this is correct).
RAM means physical memory. RAM belongs to the ayarwm, there's no such thing as process A's RAM and process B's RAM.
When the processes map the file, they each have 3 pages of virtual memory allocated to the mapping (I'm assuming the file size is 3 pages).
When A accesses page 0, that page is fetched into RAM. At that time, mincore() should return [1, 0, 0] when run by either process.
When B accesses page 2, that page is fetched into RAM. Assuming nothing causes page 0 to be ejected, mincore() will then return [1, 0, 1] in either process.
It doesn't matter whether the process that calls mincore() has accessed a particular page. If some other process has accessed the page, it will be loaded into RAM, and this process won't get a page fault when trying to access it.
Things can get weird when using virtual filesystems like tmpfs, since the filesystem itself can be stored in RAM -- there's nothing to read in from disk.