More issues with hdsp and 1.0.7 and 1.0.8 drivers

More issues with hdsp and 1.0.7 and 1.0.8 drivers

[Ivica Ico Bukvic]

Hi all,

Ever since I finally got to solve my pcmcia cardbus problem, I've been using 
the 1.0.5 driver for a while and hdsp+multiface worked more or less fine.

By more or less, I mean that one problem persisted: the hdspmixer's levels 
were way off and it exhibited the following problems:

1) the monitoring of analog inputs only worked for the first 7 inputs which 
accurately showed signal levels below -105 dB, which should be as expected.
2) the middle row, bottom row, as well as main outs showed some levels but the 
peaks were extremely latent and often hung below the actual levels which 
obviously made no sense. Furthermore, the green level bars were also very 
inert and bottom row almost never showed them at all

Recently I read somewhere discussion that some of these problems have been 
addressed so I resorted to installing the latest drivers and here are the 
problems I found:

1.0.8 driver install completely from scratch using sources has the following 
problem:

1) hdsp card gets detected and everything goes fine until the firmware needs 
to be loaded.
2) at that point if I run hdsploader I get the following undocumented error:

hdsploader - firmware loader for RME Hammerfall DSP cards
Looking for HDSP + Multiface or Digiface cards :
Card 0 : VIA 8235 with ALC250 at 0x1000, irq 22
Card 1 : RME Hammerfall DSP at 0x20000000, irq 17
Upload firmware for card hw:1
Hwdep ioctl error on card hw:1 : Invalid argument.
Card 2 : Virtual MIDI Card 1

No other errors are noted in syslog or /var/log/messages

I tried recompiling twice and I also made sure that no stale rpm's were around 
or old sources. I am using 2.6.7 custom kernel (slightly modified sds version 
linked to from the thac's rpms site http://rpm.nyvalls.se -- modifications 
should have nothing to do with audio though).

==========================

Then I tried installing 1.0.7 from scratch and that one went fine and 
hdsploader worked just fine. However, when I started hdspmixer, it had 
exhibited the following behavior:

1) only first 7 inputs (instead of 8) were still monitored but all of them 
were at -6dB meaning most of the bars were showing continuous levels of input 
even though there was nothing coming into the card which leads me to believe 
that their values were simply offset since the 1.0.5 version.
2) middle row now had very active peaks which were accurately reflecting level 
peaks but the levels were somewhat messed up (on the screenshot I am sending 
you link below you'll notice that the 2nd channels shows nothing even though 
a stereo file is being played and there are obvious peaks present, while the 
1st channel flickered a lot, it was way greater than it's supposed level, and 
it took it a long time to dissipate once the signal ceased)
3) bottom rows apparently worked ok

I am also including a link to the screenshot for your convenience:
http://meowing.ccm.uc.edu/~ico/Linux/hdsp.jpg

NB:

1) soundcard works ok in Windows so hw is not the issue.
2) sound used for the screenshot is dc-offset-free normalized sound that 
should not clip
3) system worked fine as pointed before on 1.0.5 except for similar issues 
with hdspmixer

==========================

Just for fun I also tried to figure out where the source of the problem  was 
regarding the first issue (hdsploader failing), so I did some tests (please 
note that each of the situations were tested both with sources and rpm's 
where available):

1) when I installed all 1.0.7 packages everything worked ok (except hdspmixer 
as described above)
2) when I installed all 1.0.8 packages hdsploader failed
3) when I installed all 1.0.7 drivers with *only* alsa-lib being 1.0.8, 
everything was still fine.
4) when I installed all 1.0.7 drivers with *only* alsa-lib and hdsploader 
being 1.0.8, everything was still fine
5) when I installed 1.0.8 drivers in addition to 1.0.8 libs (everything else, 
tools, oss, utils 'n' stuff remained 1.0.7) the hdsploader failed.

This leads me to believe that the problem associated with hdsploader failing 
on 1.0.8 has to do specifically with the hdsp driver as that is the only 
thing that changed between points 3, 4, and 5 (ok, obviously my other sound 
driver for the built-in card also changed, as well as the driver for virmidi 
that I use but I seriously doubt that they had anything to do with it).

For the record, I use heavily updated mdk 10.0 community, aforementioned 2.6.7 
kernel with patches, Xorg 6.7 on a mobile  AMD64 notebook running 32-bit 
version of Linux, and here's my (ugly) /etc/modprobe.conf:

alias eth0 via-rhine
alias wlan0 ndiswrapper
alias char-major-116-* snd
alias char-major-14-* soundcore
alias snd-card-0 snd-via82xx
options snd-via82xx index=0
#alias snd-card-1 snd-hdsp
options snd-hdsp index=1
alias snd-card-2 snd-virmidi
options snd-virmidi index=2
alias sound-slot-0 snd-card-0
alias sound-service-0-0 snd-mixer-oss
alias sound-service-0-1 snd-seq-oss
alias sound-service-0-3 snd-pcm-oss
alias sound-service-0-8 snd-seq-oss
alias sound-service-0-12 snd-pcm-oss
#alias sound-slot-1 snd-hdsp
#alias sound-service-1-0 snd-mixer-oss
#alias sound-service-1-1 snd-seq-oss
#alias sound-service-1-3 snd-pcm-oss
#alias sound-service-1-8 snd-seq-oss
#alias sound-service-1-12 snd-pcm-oss
alias sound-slot-2 snd-card-2
alias sound-service-1-1 snd-seq-oss
alias sound-service-1-8 snd-seq-oss
install processor /sbin/modprobe powernow-k8; /sbin/modprobe --ignore-install 
processor
remove processor /sbin/modprobe -r --ignore-remove processor && /sbin/modprobe 
-r powernow-k8
install scsi_hostadapter /sbin/modprobe imm; /sbin/modprobe ppa; /bin/true
#install snd-hdsp /sbin/modprobe --first-time --ignore-install snd-hdsp && 
{ /sbin/modprobe snd-pcm-oss; /bin/true; }
install snd-via82xx /sbin/modprobe --first-time --ignore-install snd-via82xx 
&& { /sbin/modprobe snd-pcm-oss; /bin/true; }
install snd-hdsp /sbin/modprobe --ignore-install snd-hdsp && 
{ /usr/bin/autohdsploader; }
install usb-interface /sbin/modprobe usb-uhci; /sbin/modprobe 
ehci-hcd; /bin/true
remove snd-hdsp { /sbin/modprobe -r snd-pcm-oss; } ; /sbin/modprobe -r 
--first-time --ignore-remove snd-hdsp
remove snd-via82xx { /sbin/modprobe -r snd-pcm-oss; } ; /sbin/modprobe -r 
--first-time --ignore-remove snd-via82xx

(I know that the aliases are not any more necessary, but I've been too lazy to 
deal with that part as of yet).

Any help in this matter would be greatly appreciated!

Best wishes,

Ico


-------------------------------------------------------
The SF.Net email is sponsored by: Beat the post-holiday blues
Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt

Re: More issues with hdsp and 1.0.7 and 1.0.8 drivers

[Takashi Iwai]

At Thu, 13 Jan 2005 22:16:56 -0500,
Ivica Ico Bukvic wrote:
> 
> 1.0.8 driver install completely from scratch using sources has the following 
> problem:
> 
> 1) hdsp card gets detected and everything goes fine until the firmware needs 
> to be loaded.
> 2) at that point if I run hdsploader I get the following undocumented error:

You need no longer to call hdsploader.  The hdsp driver uses hotplug
firmware loader if available.  So, installating firmware files to the
proper directory would suffice.  (Note that the directory is different
from the path for hdsploader.)

Did you install alsa-firmware 1.0.8, too?


Takashi


-------------------------------------------------------
The SF.Net email is sponsored by: Beat the post-holiday blues
Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt

RE: More issues with hdsp and 1.0.7 and 1.0.8 drivers

[Ivica Ico Bukvic]

> At Thu, 13 Jan 2005 22:16:56 -0500,
> Ivica Ico Bukvic wrote:
> >
> > 1.0.8 driver install completely from scratch using sources has the
> following
> > problem:
> >
> > 1) hdsp card gets detected and everything goes fine until the firmware
> needs
> > to be loaded.
> > 2) at that point if I run hdsploader I get the following undocumented
> error:
> 
> You need no longer to call hdsploader.  The hdsp driver uses hotplug
> firmware loader if available.  So, installating firmware files to the
> proper directory would suffice.  (Note that the directory is different
> from the path for hdsploader.)
> 
> Did you install alsa-firmware 1.0.8, too?

I tried to and made no difference. For more info please see my other
e-mails. If the hotplug is supposed to do that, it certainly fails on my
system as I have to either explicitly tell it to do it or if the snd-hdsp is
being probed for the first time, I added the hdsploader to the "install"
clause so that it is triggered with modprobing of the snd-hdsp module...

Of course, this only works on the 1.0.7 driver as 1.0.8 fails with the
aforementioned error. Just to make things as clear as possible, when using
1.0.8 driver if I do not do anything, the firmware never gets loaded and the
red HOST light never turns off on the multiface suggesting that nothing got
to the card.

Best wishes,

Ico 

-- 
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.300 / Virus Database: 265.6.11 - Release Date: 1/12/2005
 



-------------------------------------------------------
The SF.Net email is sponsored by: Beat the post-holiday blues
Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt

Re: More issues with hdsp and 1.0.7 and 1.0.8 drivers

[Takashi Iwai]

At Fri, 14 Jan 2005 14:32:10 -0500,
Ivica Ico Bukvic wrote:
> 
> > At Thu, 13 Jan 2005 22:16:56 -0500,
> > Ivica Ico Bukvic wrote:
> > >
> > > 1.0.8 driver install completely from scratch using sources has the
> > following
> > > problem:
> > >
> > > 1) hdsp card gets detected and everything goes fine until the firmware
> > needs
> > > to be loaded.
> > > 2) at that point if I run hdsploader I get the following undocumented
> > error:
> > 
> > You need no longer to call hdsploader.  The hdsp driver uses hotplug
> > firmware loader if available.  So, installating firmware files to the
> > proper directory would suffice.  (Note that the directory is different
> > from the path for hdsploader.)
> > 
> > Did you install alsa-firmware 1.0.8, too?
> 
> I tried to and made no difference. For more info please see my other
> e-mails. If the hotplug is supposed to do that, it certainly fails on my
> system as I have to either explicitly tell it to do it or if the snd-hdsp is
> being probed for the first time, I added the hdsploader to the "install"
> clause so that it is triggered with modprobing of the snd-hdsp module...

Invoking hdsploader does nothing with the driver with hotplug-fw
loader support.  It's simply ignored.

> Of course, this only works on the 1.0.7 driver as 1.0.8 fails with the
> aforementioned error. Just to make things as clear as possible, when using
> 1.0.8 driver if I do not do anything, the firmware never gets loaded and the
> red HOST light never turns off on the multiface suggesting that nothing got
> to the card.

Did you get any kernel messages after loading snd-hdsp?

Anyway, please check the following:

- Whether your kernel supports CONFIG_FW_LOADER.
  See /proc/config.gz (if any) or /usr/src/linux/.config file.

- If fw-loader is supported, the firmware files must be installed
  properly.  Check /usr/lib/hotplug/firmware directory whether
  multiface*.bin files exist.

- Remove install in /etc/modprobe.conf to be sure.

If they are OK and the system still doesn't work, something wrong in
the driver or in the hotplug.


As a last resort, you can disable the fw-loader support in hdsp
driver.  Remove the check in alsa-driver/pci/rme9652/hdsp.c

	#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0)

so that the following is enabled always.

	#define HDSP_USE_HWDEP_LOADER

(don't forget to remove the corresponding #endif, too)


Takashi


-------------------------------------------------------
The SF.Net email is sponsored by: Beat the post-holiday blues
Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt

re: More issues with hdsp and 1.0.7 and 1.0.8 drivers

[Ivica Ico Bukvic]

Oops, I forgot to add something to the previous report. This is just to teach 
me that no matter how meticulous I try to be in reporting bugs, I always fail 
to include something important :-)

At any rate, what I forgot to mention is that even if I use 1.0.8 version of 
hdspmixer with 1.0.8 libs and hdsploader (while drivers and the rest are 
1.0.7), the same issue with hdspmixer still persists which further reinforces 
the claim that something is wrong with the 1.0.8 hdsp driver.

Best wishes,

Ico


-------------------------------------------------------
The SF.Net email is sponsored by: Beat the post-holiday blues
Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt

re: More issues with hdsp and 1.0.7 and 1.0.8 drivers

[Ivica Ico Bukvic]

And yet another thing.

It seems that the memory allocation issue that plagued really old drivers is 
still not completely gone. After disconnecting and reconnecting the hdsp 
after some time (and obviously allowing the RAM to get fragmented) I get 
this:

RME Hammerfall DSP: no buffers available
RME Hammerfall DSP: probe of 0000:02:00.0 failed with error -12

I guess this has something to do with RAM being fragmented. Please correct me 
if I am wrong...

Best wishes,

Ico


-------------------------------------------------------
The SF.Net email is sponsored by: Beat the post-holiday blues
Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt

Re: More issues with hdsp and 1.0.7 and 1.0.8 drivers

[Ivica Ico Bukvic]

On Thursday 13 January 2005 22:44, Ivica Ico Bukvic wrote:
> RME Hammerfall DSP: no buffers available
> RME Hammerfall DSP: probe of 0000:02:00.0 failed with error -12
>
> I guess this has something to do with RAM being fragmented. Please correct
> me if I am wrong...

Never mind on that one. It seems that a simple restart of alsa service and 
then reinserting of the hdsp pcmcia did the trick. Yet it is odd that this 
does happen nonetheless...

This brings me to another interesting issue and that is that my system fails 
to unload the snd-hdsp once it is pulled out. As a matter of fact no logs 
point to the fact that the card was even pulled out, yet /proc/asound 
reflects that it is missing. Perhaps I should simply write some ugly daemon 
that checks whether the soundcard still exists in /proc/asound and if it 
doesn't to unload the module so that upon the next load it again 
automatically uploads firmware... Any ideas how I could do this with little 
or no impact on system performance?

Apologies to all for my incessant barrage of stuff. Just trying my best to 
provide devs with as much info as possible in order to get this pickle 
solved :-)

Best wishes,

Ico


-------------------------------------------------------
The SF.Net email is sponsored by: Beat the post-holiday blues
Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt