[coreboot] Super I/Os Winbond W83627HF and Nuvoton NCT5572D detected as NCT6775F (was: Ivy Bridge desktop support)

[Paul Menzel]

Am Dienstag, den 02.04.2013, 12:12 -0500 schrieb Aaron Durbin:
> On Tue, Apr 2, 2013 at 11:56 AM, Nico Huber <nico.h at gmx.de> wrote:

> > Am 02.04.2013 11:12, schrieb Paul Menzel:
> >> Am Sonntag, den 31.03.2013, 21:10 +0200 schrieb Paul Menzel:
> >>
> >>> Am Samstag, den 30.03.2013, 19:57 +0100 schrieb Nico Huber:
> >>>
> >>>> I'm looking forward to write a port for my Ivy Bridge desktop system based
> >>>> on this board: Intel DH77EB (codenamed Eb Lake, [1]). It comes with the H77
> >>>> platform controller hub (PCH) and a Winbond W83677HG super I/O chip (build
> >>>> by Nuvoton as NCT6775F, [2]). I'm using it with a Core i3-3225 processor [3].
> >>>
> >>> awesome news! Best luck with this!
> >>>
> >>>> I've got the datasheet for the super I/O, so writing support for it should
> >>>> be no problem.
> >>>
> >>> The ASRock E350M1 [4] also uses the Nuvoton NCT6775F. Currently the code
> >>> from the Winbond W83627HF [5].
> >>>
> >>>     device pci 14.3 on # LPC        0x439d
> >>>         chip superio/winbond/w83627hf
> >>>
> >>> Serial works for me. The only problem that might be related from any
> >>> incompatibilities is, that a PS/2 keyboard connected to the shared PS/2
> >>> port does not work correctly, meaning, entering a character the
> >>> characters is displayed several times more. Entering »n«, I get
> >>> »nnnnnnnn« for example. Alvaro, having a ASRock E350M1/USB3, also uses a
> >>> PS/2 keyboard *but* has no problems with this at all. So I have no idea
> >>> if this is related to the Super I/O or not.
> > Sounds weird. But something I can't test / run into: My board doesn't even
> > have a PS/2 port.
> >
> >> Please note that Martin Roth pointed out to me, that when looking at the
> >> the ASRock E350M1, one actually finds that the Nuvoton NCT5572D is used.
> >> Both superiotool and Linux’ w83627ehf are incorrect.
> >>
> >>         superiotool r4.0-3888-gce9f8be
> >>         Found Nuvoton NCT6775F (B) (id=0xb473) at 0x2e
> >>
> >> And if you know what you have to look for, you also find that this was
> >> known all time along [8].
> >>
> >>         The board has a socketed SPI flash BIOS chip and a serial port
> >>         header. The SIO is Nuvoton NCT5572D. Using coreboot's existing
> >>         Winbond w83627hf is a good enough match to get the serial port
> >>         and keyboard working.
> > Uh, does this mean that both chips share the same ID? I get the same output
> > from superiotool:
> >     Found Nuvoton NCT6775F (B) (id=0xb473) at 0x2e
> > My chip is labeled W83677HG and I vaguely remember that I saw on a web page
> > that this is a NCT6775F. But I can't find the page anymore :(
> > Also, W83677HG sounds more like NCT6775F than NCT5572D.
> 
> They are probably the same chip. http://en.wikipedia.org/wiki/Nuvoton
> Nuvoton was spun off from Winbond a few years ago. It's all just
> branding and part number differences, I suspect.

According to the LM Sensors developers replied that all three are
variants and should be compatible, but cannot be told apart [1].

        > > > Should the Linux message be changed to the following?
        > > > 
        > > >         w83627ehf: Found NCT6775F (or NCT5572D) chip at 0x290
        > > > 
        > > > Is it planned to add support for that the NCT5572D?
        > > > 
        > > I don't think so. Many of the superio chips have different labels - we would
        > > end up spending a lot of time trying to keep the labels up to date.
        > > Besides, that chip can also be labeled as NCT6772F or NCT6771F.
        > > So it should read
        > > 
        > >     w83627ehf: Found NCT6775F, or NCT6771F, or NCT6772F, or NCT5572D chip at 0x290
        > > 
        > > which would get a bit awkward. And then there is its original name, W83677HG-I,
        > > which we would presumably have to add as well.
        > 
        > "NCT6775F or compatible" is one way to name it, that would make
        > everyone happy I think.
        > 
        Ok, I'll add something like that to the nct6775 driver. Turns out the other
        supported chips (NCT6776, NCT6779) have similar variants, so "or compatible"
        applies to all those chips. One interesting detail is that the variants
        support fewer sensors/fans according to the data sheets, but there is no way
        to distinguish the chips from each other.

Should we adapt superiotool to also print »NCT6775F or compatible«?


Thanks,

Paul


[1] http://www.spinics.net/lists/lm-sensors/msg38272.html
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part
URL: <http://www.coreboot.org/pipermail/coreboot/attachments/20130403/e321379f/attachment.sig>