• Welcome to OS2World OLD-STATIC-BACKUP Forum.
 

News:

This is an old OS2World backup forum for reference only. IT IS READ ONLY!!!

If you need help with OS/2 - eComStation visit http://www.os2world.com/forum

Main Menu

(FIXED?) (New FF304 Tip?) Losing Multimedia Sound Access... sort of...

Started by RobertM, 2008.11.29, 00:01:17

Previous topic - Next topic

RobertM

Hello all,

I am having some interesting issues with Firefox 3.0x (all versions from 3.01 to 3.04) and UniAudio...

After opening Firefox (besides it eventually crashing with an out of resources error regardless of the virtualadresslimit settings), certain programs lose access to the sound card.



BackTalk:
It is 16 hundred hours 26 minutes and 14 seconds.
208.54.90.67    . . , . A I Built New Computer Page Loaded
86.42.246.237   . . , . Star Trek Phase 2 Wallpaper Page Loaded
(I load Firefox 3.04 at this point)
86.42.246.237   . . , . Star Trek Phase 2 Posters Page Loaded
MCI error: Error encountered in SPI.INI file.  Switching to quiet mode.
MCI error: Invalid device ID given.  Switching to quiet mode.
It is 16 hundred hours 30 minutes.
MCI error: Error encountered in SPI.INI file.  Switching to quiet mode.
MCI error: Invalid device ID given.  Switching to quiet mode.






Z!:
error: MCI Error 5010: Cannot load MMPM2 driver.



Oddly, the WPS still plays sounds fine (even when Firefox is running). Shutting down Firefox does not restore the sound card availability.

I am using UniAud v114rc3.



I have tried updating the drivers but:
Updating the drivers to the latest versions produce no sound (though installing the v1.20 version does seem to initialize the card - you can hear the "pop" during boot-up of it being turned on - but that's it).

I have tried changing the volumes using UniMixPM and by using UniMix via loading a dump of the working volumes (from the old drivers), but neither produces sound.

I am presuming that I need to install the new drivers in a different fashion - or need a newer uniaud.dll? Perhaps an incremental upgrade?

Here's the only odd thing I noticed about my current setup. These are the current files installed from the older version of UniAud:
uniaud.dll
uniaud16.sys
uniaud32.sym
uniaud32.sys
uniaud32kee.sys


The bold ones are the ones not in the updated package on Paul's site - nor apparently in some of the later builds that come with installers.


The drivers being loaded in the config.sys file are:
DEVICE=C:\MMOS2\UNIAUD32.SYS /V
DEVICE=C:\MMOS2\UNIAUD16.SYS /V /C


So, I dont think there is an issue with the kee version not being in the updated drivers...

Thanks for any help,
Robert


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


The Blue Warper

Hello, Robert!

Have you tried disabling the high resolution timer in Firefox?  Quoting from the readme (this has been there for years):

- set NSPR_OS2_NO_HIRES_TIMER=1
This causes Firefox not to use OS/2's high resolution timer. Set this if other applications using the high resolution timer (multimedia apps) act strangely.

MrJinx

Unimixpm is no longer usable in the later builds from Paul. I have to use the unimix command line and manually invoke the mixer changes. once I got everything working again, I saved the unimix dump and reload it from startup CMD .
If you had unimixpm working before a uniaud upgrade, delete the unimixpm.ini and when you launch unimixpm the next time, It is reset, and master volume is the only option. (That doesn't work either.) The MMOS2 volume object will work once you get all the mixer levels set right in unimix.

Pete

#3
Hi Robert

Wonder if Firefox is "trampling" memory space ie grabbing memory resources without worrying about anything else.

Do you have the following line in config.sys? - it may help if your answer is "No".

EARLYMEMINIT=TRUE


Might help if you specified the hardware involved as different builds of uniaud work better with some hardware and maybe someone would be able to point you to the "best" for the hardware involved eg some of the SBLive (Audigy2 based + possibly others) cards work with 114RC1 or 114RC6 but nothing in between.

If you are using HDA hardware then you should start with 114RC6 as the "base" and then experiment with the HDA builds from Paul; these either include hda in the filename or have build versions 1.9.xx in the filename. Paul's builds are here http://download.smedley.info/

If not using hda then check the PCI Soundcard Matrix to see if the hardware is listed and which driver package was used http://www.os2usr.org/os2sound.html - otherwise start with 114RC6 and work backwards until you get a working driver set. Then let the uniaud guys know what works at http://svn.netlabs.org/uniaud/

The posts by Mr Jinx and The Blue Warper cover the other problems I was going to discuss but lead into:-

Have you checked the current switch settings in the mixer? Sometimes changing the uniaud driver seems to cause switches to turn on/off.

Use the following command to generate a text file containing mixer switch Ids and settings

[L:\MMOS2]unimix -list > mixer.txt


Hope the above is helpful/useful

Pete

RobertM

Hi Pete,

Quote from: Pete on 2008.11.29, 02:53:09
Hi Robert

Wonder if Firefox is "trampling" memory space ie grabbing memory resources without worrying about anything else.

Do you have the following line in config.sys? - it may help if your answer is "No".

EARLYMEMINIT=TRUE

That was my first thought based off some posts I read elsewhere. So...

I have WSeB CP2 PF with the 1.104a kernel (non-SMP), on an ancient AMD Athlon 1200MHz (HP board of some sort with onboard audio), 512MB RAM.

The system is running HPFS386 and JFS. Using EARLYMEMINIT makes the HPFS386 driver load fail due to lack of memory for heap space (HPFS386 currently configured at 8MB cache and automatic heap allocation).

PCI.EXE (from the tool on Hobbes) claims it is:
Device 1371h ES1371, ES1373 AudioPCI

Subsystem ID 13711274h Creative Sound Blaster AudioPCI64V, AudioPCI 128 (Generic ID)
Subsystem Vendor 1274h Creative (Was: Ensoniq)
Address 0 is an I/O Port : 1800h..183Fh
System IRQ 5, INT# A



Quote from: Pete on 2008.11.29, 02:53:09
Might help if you specified the hardware involved as different builds of uniaud work better with some hardware and maybe someone would be able to point you to the "best" for the hardware involved eg some of the SBLive (Audigy2 based + possibly others) cards work with 114RC1 or 114RC6 but nothing in between.

If you are using HDA hardware then you should start with 114RC6 as the "base" and then experiment with the HDA builds from Paul; these either include hda in the filename or have build versions 1.9.xx in the filename. Paul's builds are here http://download.smedley.info/

If not using hda then check the PCI Soundcard Matrix to see if the hardware is listed and which driver package was used http://www.os2usr.org/os2sound.html - otherwise start with 114RC6 and work backwards until you get a working driver set.

It suggests the es1371.zip (for the es1373/SB128) file from Hobbes or a driver on SWC. So, I may try the es1371 driver and see what happens.

Quote from: Pete on 2008.11.29, 02:53:09
Then let the uniaud guys know what works at http://svn.netlabs.org/uniaud/

The posts by Mr Jinx and The Blue Warper cover the other problems I was going to discuss but lead into:-

Have you checked the current switch settings in the mixer? Sometimes changing the uniaud driver seems to cause switches to turn on/off.

Use the following command to generate a text file containing mixer switch Ids and settings

[L:\MMOS2]unimix -list > mixer.txt


Hope the above is helpful/useful

Pete

Thanks Pete, gonna play some more and see what happens...

Robert


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


RobertM

Quote from: The Blue Warper on 2008.11.29, 00:35:22
Hello, Robert!

Have you tried disabling the high resolution timer in Firefox?  Quoting from the readme (this has been there for years):

- set NSPR_OS2_NO_HIRES_TIMER=1
This causes Firefox not to use OS/2's high resolution timer. Set this if other applications using the high resolution timer (multimedia apps) act strangely.


Hi TBW,

Sadly, that is already set.  :(


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


RobertM

Quote from: MrJinx on 2008.11.29, 02:06:21
Unimixpm is no longer usable in the later builds from Paul. I have to use the unimix command line and manually invoke the mixer changes. once I got everything working again, I saved the unimix dump and reload it from startup CMD .
If you had unimixpm working before a uniaud upgrade, delete the unimixpm.ini and when you launch unimixpm the next time, It is reset, and master volume is the only option. (That doesn't work either.) The MMOS2 volume object will work once you get all the mixer levels set right in unimix.

Hi MrJinx,

Thanks much. From what you are saying (and my experiences in trying this) then I did find a working version of UniAudio and simply needed to set the mixer levels using UniMix - and then use the Warp Volume control. Ah well.

I am going to try the Warp card specific driver from Hobbes, and failing that, will go back to a newer build of Uniaud and see what happens.

Oh, as a side note, I have 1.1.4 RC4 running currently... (I said rc3 ealier... but MMOS2's installer thinks it's RC4).

Thanks,
Rob


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


RobertM

#7
UGH!!!!!

OK.... seems the v114rc6 drivers are the last that work (same results with FF3... any of the 114rc6 and lower drivers properly work with FF2 and below).

The newer than 114rc6 drivers load, but that is about it. Manually setting the volumes (using unimix and verifying) have no effect on sound playback (the unimix app does show the volumes properly being set though).

The ES1371 driver finds no card installed. I am guessing it may be due to the IRQ being shared?

Installed an SBLive EMU10K1 card, sblive081b2 driver does not find card (tried the KEE version instead of the W4 version). Running PCI.EXE shows the card using IRQ10 (A) (finally not a shared IRQ) - but shows it disabled (the IRQ?) - "Interrupt (disabled) 0A".

Enabling full HW detection, as always (on this system), locks up the system - which also means I cannot apparently use the IRQ/Card config tools from Veit.

Going into the crappy BIOS on this system does not provide a "NON PNP OS" setting. It instead offers:
- Other
- Win95
- Win98/2000 (which is what I was previously using)
- WinXP

Tried Win95 (since it was barely sorta kinda PnP), same results according to PCI.EXE

May try "Other" and see what happens. The WinXP setting locks the machine nicely too...

May also try two other ES1373 cards with the ES1371 driver (it's really a 1371/1373 driver)

As a side note, the sound card is NOT on the mobo (sorry for my earlier mistake). I kinda have a feeling that I may have limited sound options on this thing simply because the damn board will not properly assign IRQs and activate the IRQ line - and locks when "Full Hardware Detection" is enabled.

Sounds like it is finally time to replace this ancient thing - I've been avoiding it simply because I have so much stuff installed on it and it will be a "wonderful" hassle to recreate my work environment on another machine.

I'm still baffled by what the issue with FF3 on this system is. All file system caches are set to near nothing (8192) and threads are at 800, VirtualAddressLimit has been REM'd out, set at 1024, set at 1320, set at 1536; and regardless, will always cause an "out of resources" hard error (system modal dialog) after FF3 being open for a while (OS/2 World, Google Mail, Star Trek New Voyages Forum)... oddly, sometimes while running (also noticed less and less things work the longer FF3 is running - like printing, running other apps such as PMView or Embellish, etc), or oddly, FF3 will seem to work fine, but on exit will crash with the "Out of resources" error (or outright hang the GUI requiring a C-A-D reboot).

Well, when my system is finished it's horrendously long chkdsk of Drive E (HPFS386 partition with a few million files...) I will try my other SBLive! card and failing that, try one of the other two ES1373 based cards (one is from Ensoniq, the other from Creative (presumably after Creative bought Ensoniq).


Woulda been nice getting the SBLive! to work, as multiple WAV streams would be really nice. Kinda wishing as ancient as this machine is, that it had an ISA slot... I've got 3 MWave 2780 Multimedia Modem Plus cards lying around, and wouldnt mind dropping one of those in (if I can find the newer drivers).

Will post my results on the ES137x series and UniAud to the support page in a bit... though I kinda have a feeling it is something to do with my hardware and the newer driver (shared IRQ?, not properly enabling the IRQ?). Sadly, the ES137x card insists on grabbing IRQ5 (pre-set at factory or by the Windows machine it was in? Dunno)... and this BIOS refuses to change any of that stuff. Hopefully selecting "Other" may resolve that (or lock the system and require another half hour for chkdsking). LoL!

Robert



|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


RobertM

#8
Ummm... Neat... I think.

So, after a hang on boot with full hardware detection enabled and the spiirq driver installed, I disabled full hardware detection (left the driver in), and rebooted (and chkdsk'd), and the driver (SBLive!) loaded and recognized the card at IRQ10. Then the machine locked up (while still in the boot process).

One step further at least...

So, now I am going to try to take out the SPI stuff (the IRQ setting stuff) as I suspect that is where it is hanging and see what the results are now that the IRQ is set... hopefully, one final chkdsk and reboot and it will be up and running.

Not a fun way of getting a sound card to work...  ???


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


RobertM

Without the SPIIRQ driver, the card is no longer recognized... sadly, it seems the IRQ needs to be set and enabled each boot. Will move that line up in the config.sys file and see what happens...



Yeah, long winded thread, but at least anyone else running into these issues can see what I tried, what happened, what didnt work, and hopefully know which pitfalls/traps to avoid...



Rob


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


RobertM

(in using the SBLive! and the sbliveos2-081 beta driver)

ok... so, selecting "Other" makes the motherboard configure the IRQ and other resources. The stupid board nicely insists on sharing an IRQ no matter how many are free (I have 4 free). Changing slots makes it share a different IRQ. I went as far as disabling COM ports and USB to free up IRQs, and the board dutifully moved the onboard NIC to IRQ3 and the sound card to the same IRQ. Changing slots makes it share the IRQ with the video card.

No longer need the spcirq driver - which is good, because it causes this system to hang during boot.

Got some neat tones out of the sound card - until I went back into the BIOS and enabled "Busmaster" on the slot the sound card is in (SoundBlaster Live! EMU10K1 based). Enabling busmastering for that slot has given me normal sound.

Tested multiple audio streams (two instances of Z! and once instance of Backtalk)... works fine.

So, here I am figuring this should end the issues with Firefox3 on my system... with 32 "handles" to the mixer/sound device, it should leave 31 available, right? Or maybe 26 available (assuming it tries using a separate handle for each tab I have open)...

Wrong - once again nothing but the WPS can touch the sound card... and I suspect that is only because it opened it's "connection" to it before I started Firefox. BackTalk opens and closes it's connection, and Z! was closed when I opened Firefox 3.04...

So I am still at a loss as to what is causing this.

I am going to try a "bare minimum" install of Firefox (ie: no plugins and addons - not that I have many installed) and see what happens.

Failing that, I think I am rebooting (to regain sound card use) and switching back to FF 2.0.0.18 for now.

Thanks everyone for your help...
Robert


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


RobertM

#11
Thanks to everyone for their help...

And a special thanks to Henk Kelder for his WPTools and CheckINI. Removed a lot of crap and broken stuff from my INI files, and unexpectedly, FF304 now suddenly does not lock out the sound card. AND... restarting my WPS actually restarts it (instead of killing the running one and not restarting it forcing me to reboot).



As a side note:
xFix destroyed the INI file - and all I did was remove handles to non-existing objects (such as H:\Something and U:\SomethingElse) which neither the drive or the object existed. When I restarted, most of my desktop objects were missing (all of which had Drive C or Drive D references), including key OS/2 objects.

xFix has been great for me in the past... but I would not recommend it on an already damaged INI file. I would suggest using it only to clean up extra entries in an INI file you know to be otherwise good.

...

Going back to an archived copy and running CheckINI on it, I found all sorts of references that were almost correct (or in computer terms, totally different) such as "Object ID B82A points to reference 2B82A" and so on... the number 2 (or occassionally 20) seems to have gotten appended to the beginning of a lot of them. A few thousand errors, and it handled them expertly.

One more note though. If your INI file is severely damaged (or slightly damaged but with this particular serious issue), just keep an eye on what you are allowing it to remove (use the interactive mode) as it may suggest removing the reference to your Desktop folder. You probably dont want to do that (or if you do, for instance, if it is a corrupted reference, then you want to have a tool handy that will recreate that reference for the <WP_Desktop> object - otherwise you wont have a desktop, and thus no working WPS).



So, in the end, the Ensoniq/ES137x sound card was not the issue (though newer drivers dont seem to work with it - but that could be because they require busmastering, which was not enabled in the BIOS)... but the gains for the effort were worth it.. the SBLive! supports 32 audio streams at once, and my INI files are a lot cleaner and more stable.


Will see soon if it helps with the resources issue as well... as sometimes when FF would crash (v2.x, v3.x) the INI files would need to be restored from an archive, I'm wondering if the resource issue is related to the INI file errors and some fatalistic loop that some part of FF gets stuck in when accessing it for something (perhaps certain parts that read the INI files for association info and icon info for objects?).

Thanks to all!!!

Robert


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


RobertM

No "Out of resources" errors yet...
Firefox 3.04 is no longer hanging on close
And oddly, it seems to be using less memory as well.

Could all of that have been due to a severely damaged INI file? I am surfing the exact same sites as always, the exact same tabs open - with two exceptions (in addition to the rest, I loaded two more sites that I only occassionally visit).

Weird... if this is the case, then a little tip about such issues maybe being added to the ones already mentioned such as virtualaddresslimit and the timer about checking the OS/2 INI files for corruption may be a wise idea.

Other issues that have gone away that always happened after FF304 was loaded (after fixing the OS/2 INI files) include parts of WPS Wizard no longer working (for instance, the various things on the left side bar ("Open Drives Folder", etc) would all change to "Error:", "Error:", etc after running FF304, and the toolbar would be blank or occassionally contain one or two empty buttons (no icon image on them, no functionality). This previously would require restarting to resolve. Now, it doesnt happen at all (at least, so far).


I am guessing that an intact set of OS/2 INI files may be a requirement for FF304 to work properly... anyone?

Robert


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


Pete

Hi Robert

All sorts of weird things start happening on my systems when the INI files get "munged" - helpful indications of a need to run checkini and cleanini.

My experience with INI file problems seems to indicate problems arise from 2 main causes:

1] System Shutdown - either failed or the system was not Shutdown properly ie someone simply turned off the power - can result in the INI files getting trashed and an error message about the problem on the next boot; an example is the "cannot find the Desktop" message.

2] Size! - the bigger the files get the more chance of corruption it seems. Shame there is not an inbuilt automatic "tidy up" mechanism that could get rid of unnecessary entries at Shutdown or next Boot.


If you do not think the problem was caused by 1] then it almost certainly was a case of 2]

The only "preventative action" for 2] is to schedule checkin and cleanini runs on a regular basis - weekly suits me "it's Friday, it's 4pm, it's checkin and cleanini time"   :-)

I would have expected various system indications that the INI files were in need of a run with the checkini and cleanini tools - not just FF being a bit stroppy. I guess the fact that Z! (which I have never installed/used so know nothing of) could not connect to the multimedia bits was probably *the* clue...  :-)

Never used xFix - is that part of xWorkplace? - so cannot comment on it except to say that I will certainly avoid it in the future.

Glad you have resolved the problem(s) anyway  :-)

Pete