Hi Knuth,
but Grub 2 is necessary because the driver in the kernel obviously
fails to initialise something, and so Grub2 is needed to get the console in
to 1920x1080 before the kernel is loaded... unless there is an equivalent
of the "gfxpayload=keep" setting in Grub2.
Thanks! Wasn't aware of this yet. How does this work though? Because
ususally the kernel does everything on it's own? I see 2 possibilities:
I *think* ( and this is just guesswork ) that the current ( 2.6.35+ )
driver simply fails to initialise the chip properly, and therefore there is a
register or maybe several which are just left alone completely by the driver.
Grub presumably sets the card up using BIOS calls, which know how to
initialise things, and using gfxpayload = keep means that the chip is left in
this state when the kernel starts to boot, meaning that "whatever isn't
initialised" is initialised by Grub. Note though that I also think that
getting 1920x1080 to work in 2.6.35-rc6 was a bit of a fluke really, the
2.6.36 driver does not yet work at all :(.
> Unfortunately if you don't have the HD panel, then at the minute I'm not
> sure of a solution.
I saw such a lot of noise on the ml already about this panel (but did
not follow it in detail)...
Is the grub2 gfxpayload=keep "trick" working? And then it is "just" the
Xorg driver?
Maybe someone else on this list could respond to that (if he/she has a
working (but probably "bodgy" - I like that word (see my PS) solution?
Maybe someone who has a working dual boot/RAID set-up working with the
Intel card on an Arrandale can resond to that! There seems to be a way to get
pre 2.6.35 kernels working with Grub 1 ( legacy ) but personally I wanted grub
2 ( and actually wanted to push things along so that at some point it should
just work out of the box! )
> constantly try to prod Jesse at Intel and feed him
How do you do that (to join you in your efforts :)?
Would you be on IRC somewhere? .. e.g. the intel channel:
irc.freenode.net: #intel-gfx?
I don't have much time and so haven't tried that approach, though I'm
sure that they are working on this because I would guess that many Arrandale +
Ironlake machines are affected. I've just been using my bug report at
https://bugs.freedesktop.org/show_bug.cgi?id=29141, which seems to have turned
in to an "it just doesn't work" thread :(. There are probably enough of my
random ramblings on there already though to be honest, but if anyone who has
the knowledge and time wants to look through it all then there may be
something in there that is useful. Progress is being made, but it's slow :(. I
would also ask that anyone who does read through it all take in to account
that I really don't know what I'm doing :D.
/// GRUB2 ///
c) For as the hardware of the Sony is concerned:
Is there somewhere a /complete/ grub2 config? (like what other commands
other than the gfxpayload=keep?)
I think that the only important things are:
Within the main config:
if loadfont /usr/share/grub/unicode.pf2 ; then
set gfxmode=1920x1080
insmod gfxterm
insmod vbe
if terminal_output gfxterm ; then true ; else
terminal gfxterm
fi
fi
And within a menu entry:
set gfxpayload = keep
linux<kernel> <root=... etc> i8042.nopnp i8042.reset i8042.nomux i8042.noloop
text
Obviously "text" shouldn't be needed but is currently because the "fix"
requires a delay between loading i915 and starting X. You also need to prevent
the i915, nouveau and uvesafb modules from loading a boot by blacklisting it
in whatever way your distro does it. ( Some may be ok but I haven't had a
chance to try the different combos. )
/// KERNEL ///
e) For as the hardware of the Sony is concerned:
Is there somewhere a /complete/ kernel config?
I've stuck my .config file in my www.voip-x.co.uk/files/adam directory -
but this is only one that works, not the cleanest or the most complete!
f) For as the hardware of the Sony is concerned:
Is there somewhere a /complete/ list of the kernel-patches you are using?
The patch file for the 2.6.35-rc6 source is already in there. There is
another important fix which I added to stop the hda-intel driver from trying
to initialise the NVidia sound chip when the card is powered down, and it also
incorporates my version of Norbert's sony-laptop module.
g) For as the hardware of the Sony is concerned:
Is there somewhere a /complete/ list of the Xorg package versions (as
the intel driver usually needs specific versions of e.g. mesa...) you
are using?
I don't have that, but to be honest I don't think that it is massively
critical since it seems to work with both the Ubuntu standard repository
versions as well as the code from xorg-edgers.
h) For as the hardware of the Sony is concerned:
Is there somewhere a /complete/ list of the patches necessary for any of
these Xorg packages (e.g. the intel graphic driver I suppose?)?
I didn't use any.
i) For as the hardware of the Sony is concerned:
Is there somewhere your /complete/ Xorg config file (or are you using
the auto-generated one)?
Its auto generated. The old kernel boot trick with the NVidia drivers
needs the NVidia created config.
/// MISC ///
j) For as the hardware of the Sony is concerned:
Is there somewhere a /complete/ list of software (additional to the
patched kernel, grub2 and the Xorg specific packages) you are using?
Don't think that it's relevent, though I'm using KDE 4.5 now.
l) Your BIOS is (still) unhacked (not unlocked), right?
Correct.
m) ( I've got rid of the need for the printing of the dots by the way! )
Where would I find an explanation on how this is done (or it is fixed in
one of the patches?)
The version that's up there now already has this fix. I found that the
driver wasn't waiting long enough for the transcoder to disable after clearing
the enable bit.
P.S.: Funnily the word bodge was in none of my dictionaries ;)
("a bodge": A clumsy or inelegant job, usually a temporary repair.)
Hehe.... probably a "Britishism" :) It very much works for this though!
Adam.
