__mark_inode_dirty(9)

💅 __mark_inode_dirty(9) 💅

okay so this is a page from the Kernel API Manual, and bestie, it's a kernel-internal one — like, it's the backstage pass, not the general admission ticket. buckle up 🎀

---

✨ NAME ✨

__mark_inode_dirty — the internal function that marks an inode dirty. yeah. that's the whole name. it's giving "unassuming but crucial" 💖

---

🌸 SYNOPSIS 🌸

void __mark_inode_dirty (struct inode *inode , int flags );

---

🦋 ARGUMENTS 🦋

• inode — the inode you wanna mark. that's it. that's the one. 🫶
• flags — okay so this is what kind of dirty we're talking about, e.g. I_DIRTY_SYNC. and yes, it can be a combination of multiple I_DIRTY_* flags — a little flag salad, we love that for us — EXCEPT that I_DIRTY_TIME can't be combined with I_DIRTY_PAGES. no cap, that combo is not allowed 🚫💀

---

⭐ DESCRIPTION ⭐

so the whole job here is: mark an inode as dirty. simple enough, right? but the order matters, and the order is literally the plot:

1. we notify the filesystem first 📣
2. then we update the inode's dirty flags
3. then, if needed, we add the inode to the appropriate dirty list

and here's the thing, bestie — most callers should use mark_inode_dirty or mark_inode_dirty_sync instead of calling this directly. this one's for the internal girlies. you probably don't need to raw-dog this function 😌

CAREFUL! 🔥 we only add the inode to the dirty list if it is hashed or if it refers to a blockdev. unhashed inodes will never be added to the dirty list, even if they get hashed later — because they will have been marked dirty already. so like... the window has passed. it's over. 🙈

in short (and I need you to internalize this): ensure you hash any inodes _before_ you start marking them dirty. order of operations is literally everything here, iykyk 💅

---

and now the blockdev lore, because it's lowkey iconic 🧚

for blockdevs, inode->dirtied_when represents the dirtying time of the block-special inode (/dev/hda1) itself. meanwhile the ->dirtied_when field of the kernel-internal blockdev inode represents the dirtying time of the blockdev's pages. two different times, two different inodes, no confusion intended (there is, in fact, confusion intended, but now you know 🌈).

which is exactly why, for I_DIRTY_PAGES, we always use page->mapping->host — so the page-dirtying time gets recorded in the internal blockdev inode. it's giving "correct accounting or bust" 📊✨

---

👑 SEE ALSO 👑

kernel file ./fs/fs-writeback.c plus this whole squad right here, no gatekeeping:

• inode_io_list_move_locked(9)
• wb_wait_for_completion(9)
• inode_cgwb_move_to_attached(9)
• locked_inode_to_wb_and_lock_list(9)
• inode_to_wb_and_lock_list(9)
• inode_switch_wbs(9)
• cleanup_offline_cgwb(9)
• wbc_attach_and_unlock_inode(9)
• wbc_attach_fdatawrite_inode(9)
• wbc_detach_inode(9)
• wbc_account_cgroup_owner(9)
• wb_split_bdi_pages(9)
• bdi_split_work_to_wbs(9)
• cgroup_writeback_by_id(9)
• cgroup_writeback_umount(9)
• wb_start_background_writeback(9)
• filemap_dontcache_kick_writeback(9)
• writeback_inodes_sb_nr(9)
• writeback_inodes_sb(9)
• try_to_writeback_inodes_sb(9)
• sync_inodes_sb(9)
• write_inode_now(9)
• sync_inode_metadata(9)

---

and that's the page!! 🎀✨💖

signed, sealed, delivered by ./fs — October 2026 — mwah, take care of your inodes bestie 🫶🌸