On 3/26/20 2:56 PM, Mike Kravetz wrote: > Perhaps it would be best to check hugepages_supported() when parsing > hugetlb command line options. If not enabled, throw an error. This > will be much easier to do after moving all command line parsing to > arch independent code.
Yeah, that sounds sane. > Is that a sufficient way to address this concern? I think it is a good > change in any case. (Thanks to Kirill for pointing this out.) So, it turns out the x86 huge page enumeration is totally buggered. X86_FEATURE_PSE is actually meaningless on 64-bit (and 32-bit PAE). All CPUs architecturally support 2MB pages regardless of X86_FEATURE_PSE and the state of CR4.PSE. So, on x86_64 at least, hugepages_supported() should *always* return 1. 1GB page support can continue to be dependent on X86_FEATURE_GBPAGES.
