I, and a lot of other people, seem to have had many problems with the adi logitech joystick module not correctly detecting the joystick. A little bit of debugging and after contacting the dev, I recieved the following email:
It's most likely the timing in adi_init_digital(). Try removing the line
udelay(-seq[i]*14); /* It looks like mdelay() is off by approx 1.4% */
it's quite likely it'll start working.
SuSE Labs, SuSE CR
This problem seems to fix the recurring bug that I've had for about a year with this joystick, and as this line seems to be a hack, and not needed for this distro anyway, may I suggest it is removed.
This bug has been found on my Logitech Digital 3D, connected by the gameport, and over multiple gentoo-sources versions
Created attachment 78568 [details, diff]
ADI joystick timing problem fix
Have you reported to Vojtech that the fix solves your problem? The patch is not yet in Linus' tree...
Here follows our email exchange. From people on the forums, it seems that this is needed for anyone to get results, on Gentoo, from this driver.
It may well be a distro-specific bug, in which case, I personally feel it makes sense to include it in gentoo-sources and other patchsets.
On Tue, Jan 03, 2006 at 08:36:40PM +0000, Mike Arthur wrote:
> On Tuesday 03 January 2006 19:46, you wrote:
> > It's most likely the timing in adi_init_digital(). Try removing the line
> > udelay(-seq[i]*14); /* It looks like mdelay() is off by approx 1.4% */
> > it's quite likely it'll start working.
> In the most professional way, I love you!
> I can now use my nice Wingman again!
> Is this a proper fix, or a bit of a hack? Is it ok to share the fix on the
> gentoo forums? Should I post this to the LKML?
It's removing a hack needed for some kernels. Removing it is correct,
and finding a real reason for it being necessary on some (quite a large
number of) kernel/hardware combinations is called for.
SuSE Labs, SuSE CR
It won't be a distro-specific thing (the kernel is very independant), it will probably depend on hardware and/or configuration.
We require upstream (i.e. Vojtech) to have accepted a patch before we include it. Please reopen this bug if a conclusion is reached.