Bug#345067: Draft of documenting the ide-generic problem
On Tue, Mar 07, 2006 at 06:43:04PM +0100, Jonas Smedegaard wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Greetings, fellow hackers!
>
> I have startet a wiki page to document the ide-generic problem with the
> Linux kernel (and thus the ramdisk tools as well). The page is here:
> http://wiki.debian.org/LinuxKernelIdeProblem
>
> The intend is to isolate the cause of all this and fix it at the root
> rather than maintaining workarounds in the ramdisk tools (causing us in
> the end to bother the technical comittee with silly non-core problems).
>
> Please go correct all my hilarious mistakes, either directly (if you
> are a WikiFan) or by commenting in this thread - then I'll pass on your
> comments on the wiki page.
So much lost time, just because you are unable to look at the problem.
Please revert the broken patch that caused this problem in the first place.
The first of your unaccuracies :
The above combination is currently unsupported with yaird: Machines using a
problematic IDE chipset cannot be assured to work without ide-generic, so
yaird fails to build the image when it cannot include that module.
This is pure bullshit, the kernel has worked fine with yaird for months on the
pegaos, before you and erik implemented the buggy workaround of forcing the use
of ide-generic. So, since you make claims of this being a problem, i would
really like you to give a single example of this causing a problem on non-x86,
and more exactly on powerpc.
The second inaccuracy is :
A patch is proposed for yaird that disables the ide-generic force-loading
for powerpc.
The proposed patch does indeed backout the original ide-generic force-loading
patch.
It is tested to work with Pegasos using official Debian Linux kernels, but
may silently break in other scenarios.
Ok. I claim that it cannot break other scenarios, first, be it only that the
debian kernel worked on all involved hardware today, and doesn't even build
the ide-generic module. This allone is proof enough that the ide-generic can
in no way be actually needed on powerpc. It may be used possibly, and the user
is free to add it to its /etc/modules in this case, but there is absolutely no
chance of it breaking in a way you fear. I have tried to examine the cases
involved in erkelenz with you, but then and now again, you have absolutely no
interest in looking at it logically, and show us even a remote case where this
could break. So this sentence is wrong.
Furthermore, i am calling you an hypocrit, since you are being over-cautious
here, but absolutely did nothing to plainly test the patch which caused the
brokeness in the first place.
Then again :
If the guess in the yaird internal notes is correct, then powerpc succeeding
without ide-generic is due to other IDE modules being builtin, causing
ide-core to be builtin as well.
So, yaird used to work on powerpc before you broke it, because the ide modules
where builtin. This again is patently false, since powerpc has had modular ide
modules from the start, except the powermac-ide stuff, but then generic x86
stuff is not recomended to run on mac hardware anyway. Furthermore, all
powerpc 2.6 kernels had modular ide drivers for everything except pmac-ide,
for the reasons explained above.
So, i guess your text is just an assembly of innacuries and lies, coming
either of malice or plain ignorance, in order for you to justify your abysmal
packaging skills and ignoring this bug report.
Please stop loosing anyones time, and try to be a bit more acurate next time,
and don't pretend to know more about powerpc kernels than i do, please.
Sven Luther
Reply to: