Run-time huge page configuration is broken

[oandras]

salt-formula-linux/linux/system/hugepages.sls

Line 32 in 1ba71ad

# Make hugepages available right away with a temporary systctl write

The code tries to set the desired number of huge pages by updating vm.nr_hugepages via sysctl. This sysctl manipulates the default sized hugepages only, so it's wrong to call this in a loop iterating over all huge page sizes.

https://www.kernel.org/doc/Documentation/vm/hugetlbpage.txt:

When multiple huge page sizes are supported, /proc/sys/vm/nr_hugepages
indicates the current number of pre-allocated huge pages of the default size.

Huge pages of a given size can be manipulated via /sys/kernel/mm/hugepages during run-time. It is also a question if the allocation of 1GB huge pages can be modified during run-time on the x86-64 architecture.

[oandras]

I created a patch https://github.com/oandras/salt-formula-linux/tree/fix_runtime_hugepage_allocation

Did not have the opportunity to test it yet.