qman_create_cgr(9)

📦 version 7.3.1, in nixpkgs unstable

💅 qman_create_cgr — the Kernel API Manual, bestie edition (section 9, obviously)

okay so picture this: you're deep in the Linux kernel API manual, section 9, and you've stumbled onto qman_create_cgr(9) — yes that's the real name, that's the page, we're doing this 🎀

✨ NAME ✨

qman_create_cgr — literally just register a congestion group object. that's the whole vibe. 🦋

🌸 SYNOPSIS 🌸

int qman_create_cgr (struct qman_cgr *cgr , u32 flags , struct
qm_mcc_initcgr *opts );

she takes three things and returns an int, periodt 💅

🎀 ARGUMENTS 🎀

  • cgr — the 'cgr' object itself, with its fields already filled in. like, you gotta bring your own homework here, she's not doing it for you 📝
  • flags — the QMAN_CGR_FLAG_* values. pick your flags, bestie ✨
  • opts — optional state of CGR settings. optional!! key word 💖

🔥 DESCRIPTION 🔥

So what does she actually do? She registers this object for receiving congestion entry/exit callbacks — and the way she does it is on the portal affine to the cpu portal on which this API is executed. iykyk 👑

If opts is NULL then only the callback (cgr->cb) function is registered. that's it. that's the whole registration. lowkey minimal 🫶

But if flags contains QMAN_CGR_FLAG_USE_INIT, then an init hw command will be used rather than a modify hw hardware. and yes, the init hw command will reset any unspecified parameters — so if you weren't paying attention to the ones you left out, they're getting reset, no cap 😭 (the modify hw hardware, for the record, only modifies the specified parameters — much more chill about it).

fr fr read that flag twice before you set it. resetting unspecified parameters is exactly as much of a moment as it sounds 💀

🌈 SEE ALSO 🌈

like, the whole friend group showed up to this party 🧚💜

and that's the page!! mwah 💋 sign-off time: this one lives in ./include/soc/fsl, dated October 2026, and it answers to qman_create_cgr(9) — go forth and register responsibly, bestie 🍓⭐

📦 where this page comes from