[Bug 233863] Various PowerMac G5 models may require kern.smp.disabled=1 and must set usefdt=1 which causes net interface reorder

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=233863

--- Comment #43 from Mark Millard <marklmi26-fbsd at yahoo.com> ---
(In reply to Justin Hibbits from comment #40)

Continuity and common structure certainly are
considerations.

For reference, I updated my code to dump out
the HID0, HID1, HID4, and HDI5 values for the
ap's, dumping values from both before FreeBSD
changes them and after they are changed. I was
curious. The result was:

HIOR(311) before: fff00000, 0 after
before: HID0 fc1c000000000000
after:  HID0 0151108100000000
before: HDI1                0
after:  HDI1 fd3c200000000000
before: HID4 10000000000, HID5 0
after:  HID4  1000000000, HID5 0

So differences in:
(Hopefully I matched everything up correctly)

HID0: one_ppc, do_single, isync_sc, ser-gp, reserved bits 4:5,
deep nap, doze, nhr, ext_tb_en, reserved bit 24, en_attn.

HID1: bht_pm, en_ls, en_cc, en_ic, pf_mode, en_if_cach,
en_ic_rec, ic_pe.

HID4: rm_ci, en_sp_dtw.

HID5: none.

I'll note that en_ic is described with:

QUOTE
Enable instruction cache (must be ‘1’ for proper functioning).
END QUOTE

(Not that there is a description of what the improper functioning
of the 970MP would be.) This was the wording I was most
worried about.

[There are some other bits that have differences from "preferred
state", not that I know the importance of them.]

[slbtrap and handle_kernel_slb_spill are using mftb() and so
ext_tb_en being different is in use before HID0 is updated.]

-- 
You are receiving this mail because:
You are the assignee for the bug.