pci_ats_queue_depth(9)

📦 version 7.3.1, in nixpkgs unstable

💅 pci_ats_queue_depth(9)

okay bestie, welcome to the Kernel API Manual — yes we're reading kernel docs together, that's the vibe now 🫶 this page is literally called pci_ats_queue_depth, and it's here to answer one (1) very specific question: what's the Invalidate Queue Depth on this PCI device? ✨

🎀 NAME

pci_ats_queue_depth — query the ATS Invalidate Queue Depth

like, it goes and asks the device "hey how deep is your invalidate queue" and reports back 💖

🌸 SYNOPSIS

int pci_ats_queue_depth (struct pci_dev *dev );

one function, one argument, one int coming back at ya. simple. iconic. no cap 💅

🦋 ARGUMENTS

  • dev — the PCI device. that's it, that's the whole argument. the PCI device in question, the main character of this call 👑

✨ DESCRIPTION

Returns the queue depth on success, or negative on failure. so like, positive number = slay, negative number = it's giving error 😭

now here's the part you actually need to know, bestie, because the numbering is lowkey cursed 🙈

The ATS spec uses 0 in the Invalidate Queue Depth field to indicate that the function can accept 32 Invalidate Request. but in this function we use the real values (i.e. 1~32) for the Queue Depth 🔥

and 0 here means the function shares the Queue with other functions (doesn't exclusively own a Queue) — so 0 is not "32 requests" in this context, it's "sharing is caring, no exclusive queue for me" 🫶

⚠️ real talk: don't read the raw spec value and this function's return value the same way, because 0 means a whole different thing in each. the spec-vs-here mismatch is exactly the kind of thing that'll have you debugging for an hour at 2am 💀

🧚 SEE ALSO

Kernel file ./drivers/pci/ats.c — and the whole friend group:

okay that's the whole page, you ate that 🍓 this came out of ./drivers/pci in October 2026 — go forth and query those queue depths, bestie, and remember: 0 is not a number here, it's a lifestyle 💜✨

📦 where this page comes from