In [1] Bingbu report an error that vmap_udmabuf invoke failed due to use vmap_pfn and blocked by !pfn_valid.
Due to misunderstand it, I change vmap_pfn check of !pfn_valid to pfn_valid, then send [2], then Christoph point that: vmap_pfn is entirely for memory not backed by pages or folios, i.e. PCIe BARs and similar memory. This must not be mixed with proper folio backed memory. So, I given a misuse of vmap_pfn. But to fix vmap_udmabuf, and consider HVO effect, I offer a RFC patchset [3], which simple copy vmap_pfn to vmap_udmabuf, other implement folio range based vmap. But in [3], Muchun point that I misunderstand HVO, which do not released any page struct pointer in vmemmap, only change this VA's PTE point to hugetlb's **HEAD** page frame, and release remainned **tail** page frame. So any page struct pointer do exist, and folio_page, folio_pfn or other api still can work. By this, we can fix this error simpliy: Patch 1 revert this vmap_pfn misuse patch. Patch 2 fix missed offset page set. [1] https://lore.kernel.org/all/9172a601-c360-0d5b-ba1b-33deba430...@linux.intel.com/ [2] https://lore.kernel.org/all/20250312061513.1126496-1-l...@vivo.com/ [3] https://lore.kernel.org/all/20250327092922.536-1-l...@vivo.com/ Huan Yang (2): Revert "udmabuf: fix vmap_udmabuf error page set" udmabuf: fix vmap missed offset page drivers/dma-buf/Kconfig | 1 - drivers/dma-buf/udmabuf.c | 23 ++++++++--------------- 2 files changed, 8 insertions(+), 16 deletions(-) base-commit: b425262c07a6a643ebeed91046e161e20b944164 -- 2.48.1
