Personal knowledge base of fixes, workarounds, and notes for problems I've run into — mostly with old, unsupported, or niche hardware on Linux (Nobara / Fedora). When I solve something, I dump notes here so the next person hitting the same error message gets a result.
Each subdirectory covers one problem: what it was, what didn't work, what did.
8bitdo-ultimate-2-wireless — 8BitDo Ultimate 2 Wireless: 2.4GHz dongle ignored by games
The controller works fine over Bluetooth and Steam Big Picture sees it, but games don't pick it up. Root cause: the dongle's default firmware mode exposes the controller as a composite HID device (gamepad + keyboard + mouse), which breaks SDL2's GameController classification. Fix: hold B while powering on to boot the controller into D-input mode.
Stack: 8BitDo Ultimate 2 Wireless + 2.4GHz dongle, Nobara 43, kernel 6.17.x, SDL2-based games, Unity-on-Proton.
Likely search hits: 8BitDo Ultimate 2 Wireless not working in games, dongle USB ID 2dc8:310b vs 2dc8:6012, composite HID gamepad, SDL2 GameController classification, Steam Input dongle Linux.
hp560sfp-firmware-update — Flashing HP 560SFP+ firmware on non-ProLiant hardware
HPE's firmware update tool refuses to run on consumer motherboards, but the firmware files inside the package fully support the 560SFP+. Ancient EEPROM firmware on used cards (e.g. 0x800007C7) can hang POST/GRUB when SFP+ modules are linked at boot. Fix: force-install the RPM, symlink libpci.so.3, unbind the ixgbe driver, run .setup interactively (no flags) — this bypasses the HPSUM platform check.
Stack: HP 560SFP+ (Intel 82599ES, PCI 8086:10fb / 103c:17d3), ixgbe, Linux x86_64.
Likely search hits: HP 560SFP+ firmware update Linux, HPE firmware-nic-intel-1.27.30, Possible execution of smart component from an older unsupported HPSUM release, NIC firmware flash without ProLiant, Adapter initialization failed during NVM update.
ixgbe-third-party-sfp — Third-party SFP+ modules on Intel ixgbe NICs
Three loosely related issues running non-Intel SFP+ transceivers on ixgbe-driven NICs (Intel X520-DA1, HP 560SFP+, etc.): (1) ixgbe rejects unrecognized SFP+ modules by default — fix with allow_unsupported_sfp=1; (2) some copper RJ45 transceivers (e.g. RealHD) spoof optical DDM data, so ethtool -m reports nonsense laser/power values, while honest modules like the MikroTik S+RJ10 report -inf dBm zeros — only temperature and voltage are reliable across modules; (3) RJ45 SFP+ modules run hot in dead-air bracket pockets and need airflow. For temperature monitoring see the CC-SFP-module-sensor daemon — module-agnostic, writes hwmon-format files for CoolerControl.
Stack: Intel X520-DA1 / HP 560SFP+ (Intel 82599ES) or any ixgbe-driven NIC, MikroTik S+RJ10 / RealHD SFP+ → RJ45 transceivers, Nobara / Fedora.
Likely search hits: ixgbe allow_unsupported_sfp, MikroTik S+RJ10, RealHD SFP+ RJ45 10GBase-T, SFP+ DDM spoofed laser power, copper SFP+ thermal throttle, CoolerControl SFP+ custom file sensor, ethtool -m bogus optical readings.
Hardware-specific Linux problems tend to have their solutions buried in a forum post, a kernel mailing list thread, or a five-year-old GitHub issue. This is my own breadcrumb trail back to the working answer, made public so others can find it too.
Notes are written from a Nobara Linux 43 (Fedora-based) workstation but most fixes apply to any modern Fedora / RHEL-family system, and many to Linux in general.
Issues and PRs welcome — corrections, better workarounds, or related fixes to add. If you hit one of these problems on a different distro / kernel / hardware revision and the fix needed tweaking, a note about that is useful.
MIT — use, modify, redistribute freely. No warranty.