Bug#394971: [powerpc64] load the fan control modules
On Thu, Nov 23, 2006 at 01:51:59PM +0100, Frans Pop wrote:
> The provided patch is broken: standard error is redirected to a file
> named "1" instead of file handle 1.
> Suggest to use the following syntax instead:
> modprobe -q <module> || true
Ok, will be done, i wonder why you didn't start by telling us this instead of
reverting the patch ?
Furthermore, the patch was in the BTS since almost a month, without any
comment on your part, and at the latest, i would have expected from you to
investigate it and comment when you did the 1.42 upload.
> Also, I wonder if all listed modules really need to be modprobed
> individually: modprobe will after all load modules that other modules
> depend on automatically.
Well, this is not how things are, after investigation and discussion with
Benjamin Herrenschmidt, the powerpc/powermac linux kernel maintainer. The
current way of doing this is the best compromise for within the 2.6.18 and
etch timeframe.
> I would expect that i2c-powermac and windfarm_core can safely be dropped.
And you would expect wrong. When are you going to stop to try to second guess
the work and investigation i do, and try by all mean to show me as incapable ?
They cannot be dropped, especially i2c-powermac cannot be dropped, since it
will not be pulled in by the other modules. There are numerous comments about
this fact on both debian-kernel and debian-powerpc, as well as numerous irc
conversations on debian-kernel, where you are also present.
So, either you chose to get involved, and try to stay informed about the
different powerpc developments, or you don't, but then it is only fair to ask
you to trust my knowledge and competence, as well as the interaction with the
upstream linux/powerpc community such as Benjamin Herrenschmidt. Your current
attitude is quite insulting to them and me, please modify it.
> So please only modprobe those modules that are strictly needed to enable
> the fan control.
Well, given that your premises are wrong, this conclusion is obviously not
valid.
> Finally, why was S50directfb-linux-powerpc added in the Makefile? This
> seems unrelated to this patch.
It was not. Or should not have been. If it was, it is uniquely a result of a
bad manipulation caused by me not having the svn commit right, and you are as
thus solely to blame for it.
But seriously, find attached my svn diff, there is no trace of a
S50directfb-linux-powerpc in my diff, if it was there, it most probably did
come from an older commit. I did in fact do an apt-get source of rootskel, and
then applied the svn diff output, so i have no idea where this
S50directfb-linux-powerpc file came from. I guess it came from earlier
experiments done while me and attilio and a few others tried to investigate
the g-i breakage.
> Colin suggested that when we apply this patch in the installer, a similar
> patch should added in initramfs-tools to make sure that the modules are
> loaded too on the installed system. Has that already been coordinated?
A similar patch for therm_pm72 is already in the installer, and i have
regularly been pushing maks to apply the whole stuff, but without success upto
now.
We even have some material from benjamin herrenschmidt for doing much more
fine-grained checking in the initramfs-tools, based on actual model
information and a static mapping from them to the actual needed modules.
But in the context of the d-i, this is hardly worth it, which is why i didn't
go into such a detailed and more error prone patch.
> P.S. The main reason (beside the fact that the patch was broken) that I
> reverted the upload was that normal NMU procedure was not followed:
Indeed, it was not an NMU. I am the defacto d-i powerpc porter, even though
manipulated me into resigning back in march while my mother was dying, for
which i don't think i will ever forgive you, unless you are very sincere when
you present your excuses for such unpardonable behavior. Your inability to
find anyone able to replace me and do the work that needs dooing, and this
more than 6 months after the fact, fully supports me in this.
I reject your or anyone's right to expulse me from d-i, i am doing the powerpc
port work, so i am the lead powerpc porter, that you want it or not. Debian
has always been a community where the only thing that counted was the work you
did, and as thus it is my full right to do this upload, despite your
dillusions of power or whatever.
Furthermore, by uploading 1.42 without even investigating the fancontrol
patch, you furthermore failed in your responsabilities, and furthermore lose
all right to revert that maintainer upload.
> - an NMUer never adds himself to Uploaders
Indeed, which is why it was not an NMU.
> - an NMU uses a different version number from normal uploads
Indeed, but as it was a maintainer upload.
> - bug closure was missing
Indeed, since the bug was holding the patch, and since you are refusing me the
right to commit to the svn repo normally, the patch should not be closed
before the patch is commited.
Maybe i should have opened a second patch though, against debian-installer,
including the patch of what i uploaded. I will do so in the future, but
seriously you should consider giving me the commit access back, which at this
point is the most reasonable thing to do, would avoid any further such mess,
and in general help us all put this lamentable dispute behind us and work
together in a positive way. But as always, the decision is solely in your own
hands.
> - there was no mail to the open BR with a final patch based on the debdiff
> for the upload
See above. But as this is excess work imposed to me by you unfairly removing
my svn commit access, maybe the burden to maintaining the svn consistency
should be your responsability, don't you think ? If you actually have to do
the work, maybe you will soon get bored of this childishness, and behave in
this issue again as a mature and adult human being. I certainly hope so. 6
months i long enough, and there is no sane reason for keeping this up anymore,
apart from petty vengeance, and every day you persist in this will only weaken
your position and strengthen mine, especially seeing at how you take the users
hostage of this petty dispute.
Friendly,
Sven Luther
Index: debian/control
===================================================================
--- debian/control (revision 42811)
+++ debian/control (working copy)
@@ -2,7 +2,7 @@
Section: debian-installer
Priority: standard
Maintainer: Debian Install System Team <debian-boot@lists.debian.org>
-Uploaders: Petter Reinholdtsen <pere@debian.org>, Joey Hess <joeyh@debian.org>, Matt Kraai <kraai@debian.org>, Colin Watson <cjwatson@debian.org>, Kenshi Muto <kmuto@debian.org>, Bastian Blank <waldi@debian.org>, Bdale Garbee <bdale@gag.com>, Frans Pop <fjp@debian.org>
+Uploaders: Petter Reinholdtsen <pere@debian.org>, Joey Hess <joeyh@debian.org>, Matt Kraai <kraai@debian.org>, Colin Watson <cjwatson@debian.org>, Kenshi Muto <kmuto@debian.org>, Bastian Blank <waldi@debian.org>, Bdale Garbee <bdale@gag.com>, Frans Pop <fjp@debian.org>, Sven Luther <luther@debian.org>
Build-Depends: debhelper (>= 4.2), dpkg-dev (>= 1.7.0), libklibc-dev (>= 1.4.29), klibc-utils (>= 1.4.29)
XS-Vcs-Svn: svn://svn.debian.org/d-i/trunk/packages/rootskel
Index: debian/changelog
===================================================================
--- debian/changelog (revision 42811)
+++ debian/changelog (working copy)
@@ -1,3 +1,14 @@
+<<<<<<< .mine
+rootskel (1.43) unstable; urgency=low
+
+ [ Sven Luther ]
+ * Added S05fancontrol-linux-powerpc, in order to actually load the
+ fancontrol modules, in order to not have G5 apple box go into aircraft
+ noise level a few minutes after the start of the installation.
+
+ -- Sven Luther <luther@debian.org> Sun, 19 Nov 2006 10:59:02 +0100
+
+=======
rootskel (1.44) unstable; urgency=low
* Revert to previous known good version as upload was against agreement,
@@ -14,6 +25,7 @@
-- Sven Luther <luther@debian.org> Sun, 19 Nov 2006 10:59:02 +0100
+>>>>>>> .r42811
rootskel (1.42) unstable; urgency=low
[ Sylvain Ferriol ]
Index: src/lib/debian-installer-startup.d/S05fancontrol-linux-powerpc
===================================================================
--- src/lib/debian-installer-startup.d/S05fancontrol-linux-powerpc (revision 0)
+++ src/lib/debian-installer-startup.d/S05fancontrol-linux-powerpc (revision 0)
@@ -0,0 +1,13 @@
+# Load fan control modules, to stop the fans to go into aircraft-db levels
+modprobe i2c-powermac >/dev/null 2>1 || true
+modprobe therm_pm72 >/dev/null 2>1 || true
+modprobe windfarm_core >/dev/null 2>1 || true
+modprobe windfarm_cpufreq_clamp >/dev/null 2>1 || true
+modprobe windfarm_lm75_sensor >/dev/null 2>1 || true
+modprobe windfarm_max6690_sensor >/dev/null 2>1 || true
+modprobe windfarm_pid >/dev/null 2>1 || true
+modprobe windfarm_pm81 >/dev/null 2>1 || true
+modprobe windfarm_pm91 >/dev/null 2>1 || true
+modprobe windfarm_smu_sat >/dev/null 2>1 || true
+modprobe windfarm_pm112 >/dev/null 2>1 || true
+modprobe windfarm_smu_sensors >/dev/null 2>1 || true
Index: src/lib/debian-installer-startup.d/Makefile
===================================================================
--- src/lib/debian-installer-startup.d/Makefile (revision 42811)
+++ src/lib/debian-installer-startup.d/Makefile (working copy)
@@ -32,6 +32,7 @@
ifeq ($(DEB_HOST_ARCH_CPU),powerpc)
files += \
+ S05fancontrol-linux-powerpc \
S45keyboard-linux-powerpc
endif
Reply to: