make_nnode_dirty(9)

💅 make_nnode_dirty — section 9, kernel API, bestie edition 🎀

okay so you found yourself a page from the Kernel API Manual, section 9, and it's all about make_nnode_dirty. ngl this one's a whole little LPT-era deep cut, but we're gonna get through it together, fr fr ✨

---

🌸 NAME 🌸

make_nnode_dirty — the function that goes and finds a nnode, and if it actually finds one, makes it dirty. that's it, that's the whole job 💅

---

⭐ SYNOPSIS ⭐

int make_nnode_dirty (struct ubifs_info *c , int node_num , int lnum ,
offs );

like, yes, it returns an int, and yes, it takes four things. let's meet them 🫶

---

💖 ARGUMENTS 💖

• c — the UBIFS file-system description object. your whole context, your main character struct, the vibe container ✨
• node_num — the nnode number of the nnode you want to make dirty. this is literally which nnode we're talking about 🎀
• lnum — the LEB number where that nnode was written. receipts of where it lives 📍
• offs — the offset where that nnode was written. so precise, we love an exact location 😌

---

🦋 DESCRIPTION 🦋

so, the lore: this function is used by LPT garbage collection. and LPT garbage collection is used only for the "big" LPT model, which is when c->big_lpt == 1 — no cap, that's the condition, that's the gate 💀

the garbage collection itself is honestly so simple it's iconic: it just involves marking all the nodes in the LEB that's being garbage-collected as dirty. that's the move. mark them all 🫶

then the dirty nodes get written at the next commit, and after that, the LEB is free to be reused. slay, recycling 🌈

and the return situation? this function returns 0 on success and a negative error code on failure. short and sweet, periodt ⭐

---

🔥 SEE ALSO 🔥

okay this list is long and honestly kind of iconic, so buckle up bestie, we're naming everyone in the group chat 💖

first up, the kernel file it all lives in: ./fs/ubifs/lpt_commit.c — and then:

first_dirty_cnode(9), next_dirty_cnode(9), get_cnodes_to_commit(9), upd_ltab(9), alloc_lpt_leb(9), layout_cnodes(9), realloc_lpt_leb(9), write_cnodes(9), next_pnode_to_dirty(9), add_pnode_dirt(9), do_make_pnode_dirty(9), make_tree_dirty(9), need_write_all(9), lpt_tgc_start(9), lpt_tgc_end(9), populate_lsave(9), nnode_lookup(9), make_pnode_dirty(9), make_ltab_dirty(9), make_lsave_dirty(9), make_node_dirty(9), get_lpt_node_len(9), get_pad_len(9), get_lpt_node_type(9), is_a_node(9), lpt_gc_lnum(9), lpt_gc(9), ubifs_lpt_start_commit(9), free_obsolete_cnodes(9), ubifs_lpt_end_commit(9), ubifs_lpt_post_commit(9), first_nnode(9), next_nnode(9), ubifs_lpt_free(9), dbg_is_all_ff(9), dbg_is_nnode_dirty(9), dbg_is_pnode_dirty(9), dbg_is_ltab_dirty(9), dbg_is_lsave_dirty(9), dbg_is_node_dirty(9), dbg_check_ltab_lnum(9), dbg_check_ltab(9), dbg_chk_lpt_free_spc(9), dbg_chk_lpt_sz(9), dump_lpt_leb(9), ubifs_dump_lpt_lebs(9), dbg_populate_lsave(9)

like... every single one of them. no skips, no cuts, we keep the whole roster 👑 iykyk

---

ok that's the page!! no hidden extras from me, every flag and every name above is exactly what the manual says — go be nosy in ./fs/ubifs/lpt_commit.c if you want more lore 🧚

---

and that's a wrap, bestie!! this one came from the ./fs/ubifs corner of the kernel, dated October 2026 — drink some water, go touch a filesystem, and remember: mark it dirty, next commit handles the rest 💜✨🍓