Seems fine to me. Does Linux have a default high value equivalent to "no limit?" I'd be slightly worried that applications would be unprepared to handle extremely large values but I imagine INT32_MAX is unlikely to be a problem.

This revision is now accepted and ready to land.May 30 2022, 5:29 PM

In D35351#801575, @emaste wrote:

Seems fine to me. Does Linux have a default high value equivalent to "no limit?" I'd be slightly worried that applications would be unprepared to handle extremely large values but I imagine INT32_MAX is unlikely to be a problem.

Linux limits it to 65530 due to Linux coredump method restrictions, described here https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/include/linux/mm.h?h=v5.17
I don't think the number matters here, we can return USHORT_MAX-5 as in the Linux, also Illumos also limits it to some high value

Closed by commit rG7e2a45111937: linprocfs: Add /proc/vm/max_map_count (authored by dchagin). · Explain WhyMay 31 2022, 9:00 AM

This revision was automatically updated to reflect the committed changes.

dchagin added a commit: rG7e2a45111937: linprocfs: Add /proc/vm/max_map_count.

dchagin added a commit: rGb3c809b96f77: linprocfs: Add /proc/vm/max_map_count.Jun 17 2022, 7:47 PM