I tested this patch on the nigeria-sp7 in the Kitchener office, and it resolves an exception that originally took place during the loader stage.

Remove contraction and break sentences across lines.

In D57242#1317770, @bz wrote:

This isn't quite correct yet (though probably not wrong).

mt7921 has an entry for a different vendor as well these days:

{ PCI_DEVICE(PCI_VENDOR_ID_ITTIM, 0x7922),
        .driver_data = (kernel_ulong_t)MT7922_FIRMWARE_WM },

Not sure anymore if the script doesn't check or if it simply didn't barf and it goes unnoticed as it is the same device ID as for mediatek?

Another reason we need a newer fwget (I should get back and rewrite my lua take I) to avoid adding different files for vendors but rather have the logical unit be a driver or similar. *sigh*

I would suggest, for now, to manually add the extra file to fwget and not try to automate this just yet. Maybe a printf saying to check for other vendor IDs in the driver or to list them at the end?

grep -r PCI_VENDOR_ID sys/contrib/dev/mediatek/mt76/ | grep -v PCI_VENDOR_ID_MEDIATEK
sys/contrib/dev/mediatek/mt76/mt7921/pci.c:     { PCI_DEVICE(PCI_VENDOR_ID_ITTIM, 0x7922),

In D57140#1319211, @kib wrote:

By consistency I mean that blindly ignoring some CPU IDs might result in some devices not working. For instance, there might be interrupts hard-coded to be routed through specific APICs, which would break.

We already have the knob to disable SMP, and knobs to disable specific LAPICs. I am not sure that blindly adding the cap on the OS-assigned CPUIDs it good idea, at least on x86. Although I probably think that the knob is fine as far as people who use it know the consequences. Might be a wording in smp(4) should be added to explain why the knob is there but broken.

Removed realtek changes.

In D57242#1311904, @bz wrote:

Not saying that I'll take them but where did you get the two Realtek driver files from?
The ones I am aware off are different, that's why I am asking.

Changed potentially ambiguous KASSERTs into MPASS

Removed unnecessary line wrap

Rebased onto main

Added suggested edit to comment

Added missing diff context

• Shortened error messages
• Added entry to exterr_cat_filenames.h
• Addressed potential memleak introduced in linker_load_module() line 2315

Changed ENOENT errors to ENOEXEC for more semantic correctness in the case where the file exists but is unusable due to version incompatibility.