| Message ID | 20200320164905.21722-6-jgg@ziepe.ca (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [hmm,1/6] mm/hmm: remove pgmap checking for devmap pages | expand |
On Sat, Mar 21, 2020 at 09:43:17AM +0100, Christoph Hellwig wrote: > On Fri, Mar 20, 2020 at 01:49:04PM -0300, Jason Gunthorpe wrote: > > From: Jason Gunthorpe <jgg@mellanox.com> > > > > This code can be compiled when CONFIG_TRANSPARENT_HUGEPAGE is off, so > > remove the ifdef. > > It can compile, but will the compiler optimize it away? Yes, the enclosing conditional: if (pmd_devmap(pmd) || pmd_trans_huge(pmd)) { is statically false if !CONFIG_TRANSPARENT_HUGEPAGE This is proven today, as the fallback stub is a function prototype with no implementation: -int hmm_vma_handle_pmd(struct mm_walk *walk, unsigned long addr, - unsigned long end, uint64_t *pfns, pmd_t pmd); If the compiler wasn't optimizing the above branch we'd get link failures. Jason
diff --git a/mm/hmm.c b/mm/hmm.c index 7f77fb6e35cf78..a09b4908e9c81a 100644 --- a/mm/hmm.c +++ b/mm/hmm.c @@ -192,7 +192,6 @@ static inline uint64_t pmd_to_hmm_pfn_flags(struct hmm_range *range, pmd_t pmd) range->flags[HMM_PFN_VALID]; } -#ifdef CONFIG_TRANSPARENT_HUGEPAGE static int hmm_vma_handle_pmd(struct mm_walk *walk, unsigned long addr, unsigned long end, uint64_t *pfns, pmd_t pmd) { @@ -215,11 +214,6 @@ static int hmm_vma_handle_pmd(struct mm_walk *walk, unsigned long addr, hmm_vma_walk->last = end; return 0; } -#else /* CONFIG_TRANSPARENT_HUGEPAGE */ -/* stub to allow the code below to compile */ -int hmm_vma_handle_pmd(struct mm_walk *walk, unsigned long addr, - unsigned long end, uint64_t *pfns, pmd_t pmd); -#endif /* CONFIG_TRANSPARENT_HUGEPAGE */ static inline bool hmm_is_device_private_entry(struct hmm_range *range, swp_entry_t entry)