qed_dbg_mcp_trace_dump(9)

📦 version 7.3.1, in nixpkgs unstable

💅 qed_dbg_mcp_trace_dump — NAME ✨

okay so bestie, this is qed_dbg_mcp_trace_dump from section 9 of the Kernel API Manual, and it literally does what it says on the tin: it performs an mcp trace and writes the results into the specified buffer 🎀 no cap, that's the whole vibe.

🦋 SYNOPSIS 🦋

enum dbg_status qed_dbg_mcp_trace_dump (struct qed_hwfn *p_hwfn , struct
qed_ptt *p_ptt , u32 *dump_buf , u32 buf_size_in_dwords , u32
*num_dumped_dwords );

like, yes, it's a whole mouthful, but she ate 💅

🌸 ARGUMENTS 🌸

here's the rundown of everyone showing up to the function party:

  • p_hwfn — the HW device data. main character energy, honestly 👑
  • p_ptt — the Ptt window used for writing the registers. she's the one doing the actual writing, iykyk 🖊️
  • dump_buf — pointer to write the mcp trace data into. this is where all the good stuff lands 🎯
  • buf_size_in_dwords — the size of the specified buffer, measured in dwords. yes bestie, dwords, not bytes, keep up 😌
  • num_dumped_dwords — (OUT) number of dumped dwords. she's an out-param, so she comes back out with the tea 💖

🔥 RETURN 🔥

so like, you get an error if any one of these things is being a little hater:

  • The version wasn't set. 🙈
  • The specified buffer is too small. 😭 (give her room to breathe fr fr)
  • The trace data in MCP scratchpad contain an invalid signature. 💀
  • The bundle ID in NVRAM is invalid. 🚫
  • The trace meta data cannot be found (in NVRAM or image file). 🕵️
  • The trace meta data cannot be read (from NVRAM or image file). 😵

Otherwise, returns ok. slay 💜

🎀 SEE ALSO 🎀

first up, the kernel file, right here:

./drivers/net/ethernet/qlogic/qed/qed_dbg_hsi.h

and then the whole friend group, no one gets left out bestie:

✨ and that's the whole page, bestie ✨

this lil masterpiece lives in ./drivers/net/ethernet/qlogic/qed, dated October 2026, and it's literally qed_dbg_mcp_trace_dump(9) — periodt 💅🫶

📦 where this page comes from