Makes sense. Thanks. > On Oct 18, 2020, at 9:13 PM, Ian Lance Taylor <i...@golang.org> wrote: > > On Sun, Oct 18, 2020 at 1:24 PM robert engels <reng...@ix.netcom.com> wrote: >> >> If I set a low process memory limit, and then the OS refuses to grant more >> memory to a malloc request,. will the Go runtime “stop the world” and >> attempt a full garbage collection and try to get the memory from the heap >> and then only crashing the process if it can’t? >> >> I understand that the “pacing” should limit (eliminate?) this situation, but >> I am thinking in cases when the heuristics don’t match the immediate >> allocations it would be beneficial. > > It does not work that way today, no. > > The current Go memory allocator allocates address space and then > expects that address space to be available. Typical GNU/Linux systems > use overcommit, so there is often no system call reporting that memory > could not be obtained; instead, there is a SIGBUS signal when the > program first writes to a page that the kernel can no longer supply. > It might be tricky to safely run a garbage collection and then retry > the write. > > Ian
-- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/golang-nuts/1DA916A6-B365-4C27-9CC8-8C6E155B0CD0%40ix.netcom.com.
