hwlat_mode_write(9)

💖 NAME 💖

Okay bestie, welcome to hwlat_mode_write(9) — yes, section 9, straight out of the Kernel API Manual — and this page is literally the “Write function for "mode" entry.” We're writing to the hardware latency detector's "mode" knob, and it's giving very specific thread-vibe energy. 💅

✨ SYNOPSIS ✨

ssize_t hwlat_mode_write (struct file *filp , const char __user *ubuf ,
size_t cnt , loff_t *ppos );

🫶 ARGUMENTS 🫶

• filp — the active open file structure, aka the file's main character at this exact moment.
• ubuf — the user buffer that contains the value to write. This is where the tea lives.
• cnt — the maximum number of bytes to write to "file" — yes, that quoted "file" is exactly the situation, no cap.
• ppos — the current position in file. Tracking where we are, bestie.

🌸 DESCRIPTION 🌸

So this function provides a write implementation for the "mode" interface to the hardware latency detector. And hwlatd has different operation modes, so this is where you pick the whole vibe:

• "none" — sets the allowed cpumask for a single hwlatd thread at the startup and lets the scheduler handle the migration. Like, the scheduler is driving, periodt.
• "round-robin" — this is the default mode, and it's giving rotation. A single hwlatd thread runs and migrates among the allowed CPUs in a round-robin fashion. One thread, many CPUs, iconic behavior.
• "per-cpu" — creates one hwlatd thread per allowed CPU. One thread for every allowed CPU, matching energy all around. 🦋

🎀 SEE ALSO 🎀

• Kernel file ./kernel/trace/trace_hwlat.c — get_sample(9), init_tracefs(9)

Source/version: ./kernel/trace · October 2026 — that’s the whole tea, now go write that "mode" entry 💅✨🍓