[LAU] RME multiface detected as digiface

[Tito Latini]

On Sat, Mar 14, 2015 at 02:18:19AM +0100, Peter P. wrote:
> Dear Tito,
> 
> * Tito Latini <tito.01beta at gmail.com> [2015-03-13 18:16]:
> > On Fri, Mar 13, 2015 at 12:13:27PM +0100, Peter P. wrote:
> > > Dear list,
> > > 
> > > the following problem is giving me headache since a very long time:
> > > 
> > > My RME Multiface connected via the ExpressCard to my Lenovo X201s
> > > running Debian/GNU testing is 20% of the time detected as Digiface when
> > > inserting the card into the computer. Dmesg tells me that
> > > 	Hammerfall-DSP: Digiface found
> > > 
> > > I have to eject the card multiple times and power cycle the Multiface
> > > until it is eventually randomly detected as the multiface it is.
> > > 
> > > Does anyone have any idea why this could happen? Does anyone have
> > > similar experiences?
> > 
> > The code in `hdsp_get_iobox_version' (sound/pci/rme9652/hdsp.c) seems
> > to fail with the ExpressCard (perhaps it is related to HDSP_SHORT_WAIT).
> > 
> > If you have/use only the Multiface, you could try to replace this
> > function with
> > 
> > static int hdsp_get_iobox_version (struct hdsp *hdsp)
> > {
> >         hdsp->io_type = Multiface;
> > 	snd_printk("Hammerfall-DSP: Multiface found\n");
> >         return 0;
> > }
> > 
> > and recompile the module.
> 
> Wow, thanks! This sounds like a good idea. I am using a Debian stock
> kernel. Do you have a good advice on how to recompile a single module
> only? I browsed the net a bit but some info there is either old or not
> clear.

I'm using a personal patched and boosted slackware but,
if you have a Makefile in

  ls /lib/modules/`uname -r`/build

the faster way to recompile your module is

cd YOUR_LINUX_SRC/sound/pci/rme9652
cp hdsp.c hdsp.c.orig

### change the function in hdsp.c ###

make -C /lib/modules/`uname -r`/build M=$PWD
make -C /lib/modules/`uname -r`/build M=$PWD modules_install

More info in "linux/Documentation/kbuild/modules.txt".