dma_get_seg...ary_nr_pages(9)

💖 dma_get_seg...ary_nr_pages(9) 💖

okay so bestie, we're in the Kernel API Manual today and the star of the show is dma_get_seg...ary_nr_pages(9) (the full name got cut off in the header bc it's long, we'll get to it, don't worry 🫶). it's giving "small function, big responsibility" and honestly? iconic. let's get into it ✨

---

🌸 NAME 🌸

dma_get_seg_boundary_nr_pages — returns the segment boundary in "page" units, no cap 💅

---

✨ SYNOPSIS ✨

unsigned long dma_get_seg_boundary_nr_pages (struct device *dev ,
       unsigned int page_shift );

literally just two args. we love a low-maintenance signature 🎀

---

🦋 ARGUMENTS 🦋

• dev — the device you're querying the boundary for, iykyk 🌈
• page_shift — the ilog of the IOMMU page size. yes, that's what it says, that's the vibe ⭐

---

💜 DESCRIPTION 💜

okay so what it actually does: it returns the segment boundary in IOMMU page units for the device you passed in. and heads up — those IOMMU page units may be different from the CPU page size, so don't go assuming they're the same, that's the whole plot twist 🔥

if dev is NULL, a boundary of U32_MAX is assumed. and the docs are clear about why: this case is just for non-DMA API callers. so like, not your average DMA situation, bestie 🙈

---

🧚 SEE ALSO 🧚

• Kernel file ./include/linux/dma-mapping.h 👑
• dma_use_iova(9)
• dma_unmap_sgtable(9)
• dma_sync_sgtable_for_cpu(9)
• dma_sync_sgtable_for_device(9)

go read those next, they're all in the same friend group fr fr 🍓

---

that's the whole page, periodt 💀✨ this little legend lives over in ./include/linux and the page is dated October 2026 — mwah, go query some boundaries responsibly, love u bye 🎀💖🌸