cpuid_leaf_0x2(9)

💅 cpuid_leaf_0x2(9) 💅

okay so you're in the Kernel API Manual, bestie, and this is the page for cpuid_leaf_0x2(9). It's giving "sanitize the CPUID(0x2) register output before anyone gets hurt" realness, and honestly? iconic. ✨

🌸 NAME 🌸

cpuid_leaf_0x2 — Return sanitized CPUID(0x2) register output

✨ SYNOPSIS ✨

void cpuid_leaf_0x2 (union leaf_0x2_regs *regs );

🎀 ARGUMENTS 🎀

• regs — Output parameter, bestie. This is where the CPUID(0x2) output gets stored. 🫶

💖 DESCRIPTION 💖

Query CPUID(0x2) and store its output in regs. Then — and this is the important glow-up — if the hardware hands back any invalid 1-byte descriptor, this function force sets it to zero (that's the NULL cache/TLB descriptor) before returning it to the caller. Sanitized. Clean. No cursed descriptor energy making it through. 💅

Use for_each_cpuid_0x2_desc to iterate over the register output in parsed form. That's your bestie for walking the results without doing raw register brain surgery. 🧚

Lowkey crucial: invalid 1-byte descriptors do not stay invalid — they get zeroed out as the NULL cache/TLB descriptor before the caller sees them. fr fr.

🌈 SEE ALSO 🌈

• Kernel file: ./arch/x86/include/asm/cpuid/api.h
• for_each_cpuid_0x2_desc(9)
• cpuid_subleaf(9)
• cpuid_leaf(9)
• cpuid_leaf_raw(9)
• cpuid_subleaf_n(9)
• cpuid_subleaf_n_raw(9)
• cpuid_subleaf_count(9)

mwah 💖 — from the source/version at ./arch/x86/include/asm/cpuid, October 2026, with cpuid_leaf_0x2(9) signing off 💅