Re: [PATCH v4 3/3] x86/vmemmap: Handle unpopulated sub-pmd ranges
From: David Hildenbrand <hidden>
Date: 2021-03-05 18:27:59
Also in:
lkml
From: David Hildenbrand <hidden>
Date: 2021-03-05 18:27:59
Also in:
lkml
On 04.03.21 18:08, Dave Hansen wrote:
On 3/4/21 9:02 AM, Dave Hansen wrote:quoted
quoted
+#define PAGE_UNUSED 0xFD +/* + * The unused vmemmap range, which was not yet memset(PAGE_UNUSED) ranges + * from unused_pmd_start to next PMD_SIZE boundary. + */ +static unsigned long unused_pmd_start __meminitdata;This whole 'unused_pmd_start' thing was unmentioned in the changelog.One tiny suggestion: *Sometimes* for these optimizations, it's easiest to write the code up without it in one patch, then add the optimization in the next patch. It makes it 100% clear what is part of the "core" algorithm and what is pure optimization. I don't know if it will work here, but it might be worth taking a look.
For this reason the s390x part by me (see patch description) was two separate patches. Maybe it also makes sense to split it up here. -- Thanks, David / dhildenb