EDIT: DaveAE has what seems to be a working solution in an older eth3com.device driver, jump to pg 9 or so. Problem root cause still not identified (although DaveAE may know it )


Ok, as the Audio is still crashing OS4 thread is now HUGE, I'd like to break this up if at all possible. New information has come out, and it appears that we are able to reproduce this with some regularity at this time.

Even if you have already commented in the other thread, please post yes or no, along with your system specs (A1-SE, XE, uA1-C) and soundcard used, and run the following test and post results:

1. Open AmigaAmp or other music player.
2. Load a playlist, and set to repeat/loop so it doesn't run out of music.
3. using IBrowse, shell ftp, or other, download 2-3 ISO images AT THE SAME TIME. Try Fedora Core images if you'd like, 2-3 at the same time, most easily done via IBrowse.

The expected results here are your system will have a hard lockup in < 2 hours. In many cases, it's 30 minutes or less. This may affect more than one Creative soundcard, but affects at least the SB Live and possibly other Creative chipsets chipset apparently.

The current theory that testing seems to bear out is an interaction between the Creative chipset(s?) and the Via 686B, something that has been documented back in ~2002 or so in detail (see links down the page.)

There were some comments in the old 686B/Via threads (see links below) stating a 'fixed' version of the SB Live was supposed to be produced, but I don't know if this is/was true, or how to distinguish between the differ DB board/chipset revs, if they exist.

NOTE: If you are seeing audio distortion only on a uA1-C, please see this thread instead.

NOTE: For some people occasional surfing while playing music is enough, for others it is not. PLEASE use the above test downloading ISOs (or multiple other large files that will take > an hour to download) for this test.[/b]

NOTE2: PLEASE make sure your system is set up properly with respect to CPU settings and VCore. Overheated CPUs can certainly cause system lockups, although we believe we have reproduced this with verified settings on several/most occasions, and on systems adequetely cooled. See IntuitionBase for CPU settings and VCore settings if needed.

Finally, if at all possible, please run memtester on your system if you haven't already, to verify good RAM. Memtester can be had from OS4Depot, here
Please note if you've memsted your RAM if possible, thanks.

Summary from other thread regarding reproduction (from Fuzion, edited slightly by me):
Quote:

(Audio only) Playing audio (Using many different players), while offline or an idle network (ie. No surfing, RDesktopping etc...) allows the playing of audio til the cows come home. - (Network only) Networking without audio & the user can continually surf (For example) - (Audio + Network) Playing audio (Again, any player) & networking is causing an audio lock up where the whole system locks up & the audio gets 'jammed' in a very short loop, causing a buzzing noise. Only a reboot (Soft or Hard) can escape the crash - (Network only) Playing 'movie' files with audio muted has not caused the crash even when networking is also being executed. - (VCore & 3rd Party tests) Varying VCores have been tried out by people, completely clean installs (For example, this crash has occured since the ORIGINAL OS4 Pre CD installation), with 'user-startup' files removed & 'WBStartup' directories removed - (Heat tests) New fans have been installed, cases left open & even tests with minimal cooling have been carried out to see if heat causes any change in the stability - Some people appear to have reduced the frequency of crashes (In some rare cases, completely) but others continue to get this crash but no single conclusion has been brought to light. - (Audio + Network) We're had tests with single LARGE files being downloaded & also small files (eg. Websites with a few images) to see if 'chunks' of downloads cause more or less chance of a crash, still while audio is being played. We get the crash in both cases. - (Audio + Network) We've had tests with single streaming audio (TuneNet, Amplifier, AmigaAmp) with no other networking that has caused the crash/ - (PCI tests) Some have tried moving their sound card to other slots - (A1 model) Without checking the entire thread, I believe the crash has occured on A1-XE's, A1-SE's & uA1 motherboards (anyone confirmed on uA1-C same behavior, or only SB cards?) - Running from RAM:, changing mpega.library, using 68k software (removing ppc.library temporarily) has no change on behavior. - The crash causes a complete lock up so a Reaper report cannot be submitted, nor has serial output shown anything relevant. - Several members (Wardy, Steff) have replaced their SB boards with Terratec/Envy24HT cards and no longer have the same problem, on the same machine. Envy24HT Supported cards: Quote: OS4 native AHI driver for Terratec Aureon Sky, Phase28 and M-Audio Revolution 7.1. The Aureon Space and Revolution 5.1 should also work but are untested. It SEEMS that ANY audio being played (whether it be MP3, AIFF's, audio from a movie (ie. MPlayer) or audio from games causes this lock up while the system is actively (Not just online but idle) networking in one way or another.

Background info that may be relevant to this issue:
Via link to SBLive/686b interaction problem
SB/686b issues
Via Official and commonly used fixes
TechReport comments on Via 686B/SB issues

Some snippets of info from the background links:
Quote:

A German hardware site discovered a glitch bug in the 686B southbridge of the popular KT133A chipset for AMD CPUs. Large DMA data transfers between hard drives on separate IDE channels caused data corruption. The presence of a Sound Blaster Live! worsens the problem. See this news article at The Register. Read the English translation of the tests performed. VIA has acknowledged the problem and will issue a BIOS patch soon. Meanwhile, there is a fix available here. Alternatively, you can use these BIOS settings that have been reported to work: PCI Delay Transaction: 0 or off PCI Master Read Caching: 0 or off PCI Latency: any value from 0 to 32

Quote:

reative may have designed the Sound Blaster Live! and its drivers to utilize the PCI bus inefficiently, or pushed it close to the PCI specs' limits that makes it sensitive to differing PCI bus & timings. Here's a quote from Pinnacle Systems, a company specializing in video capture solutions which have problems with Sound Blaster cards (taken from page 3 of the test report): "According to Pinnacles observations, Liveware seems to permanently transfer null-bytes over the PCI bus to the SB Live, even at times when it does not have to play notes. In the worst case, this destroys up to 30 % of a system's PCI bandwidth (that's why we noticed it on video-capturing ...). Deactivating SB Live's MIDI functions stops this behavior." PCI timing, which seems to be somewhat critical anyway, is extremely burdened in such a case, thus making the error rate explode. And what's got Creative Support to say? Right: nothing! Sounds like bad driver programming, or inefficiencies in the EMU10K1 design, requiring it to be fed useless null-bytes to keep it operating. Either way, the SBLive! is a PCI bus hog. On the other hand, VIA chipsets have been known to be more flaky in their PCI support (among other things). Either way, a PCI bus and PCI peripheral have to work hand in hand for things to work. When one side is less tolerant of the other, something gives and you'll get problems like data corruption and the dreaded static, clicking & popping.

Quote:

When using the original 4in1’s v4.29 or 4.32, only the SB Live had problems. After I transferred the files over the network, I checked the files for data corruption; mainly checking the Windows 2000 SP2 full install since it is sensitive to corruption. Out of the 6 cards tested, the SB live was the only card to cause data corruption and cause mp3’s to skip. There are many possible answers to why the data corruption occurs. The most common answer to the issue points to Creative to be at fault. When Creative designed the SB Live, they didn’t design it to support PCI Bus mastering, therefore when in use, the SB Live hogs the PCI bus. More can be read about this issue in WiNC’s 686B FAQ. ... ... In the end who is to blame for the cause of the issue? Data corruption can be blamed on both Creative and motherboard manufacturers since it only occurs on certain 686B motherboards. Audio skipping can be blamed all on Creative because it not only appears on VIA platforms, but Intel platforms as well.

Quote:

Official statement from VIA: The data corruption error, which some web sites and people have reported with the 686b Southbridge, is caused by incorrect bios registry setting to the Northbridge. These bios settings were made by motherboard manufacturers, in an attempt to fix a conflict with the Sound Blaster Live Value cards. Information has been provided to all motherboard manufacturers on how to correctly resolve both the data corruption error and the Sound Blaster Live conflict. The patch released by VIA in the 4in1 4.31 drivers’ replicates the correct bios settings. We provided this patch to make sure as many people got a fix to their Sound Blaster Live problems as soon as possible. Most motherboard manufacturers have now corrected their bios and the patch is not necessarily needed although it will not harm any VIA based system if installed.

Relevant quotes from the original thread:
DrBombCrater:
Quote:

I think the SBLive issue that happened on VIA-based PCs is a bit of a red herring here. The corruption and lockup problems only ever occured with machines that had both the 686 and a VIA north bridge. Systems with the 686 and a non-VIA NB (such as the AMD 750) were okay. That's not to say it's an impossibility. The Articia S may be reacting as badly to the 686B's problems as the VIA north bridges. ... All those 'fixes' listed concern register locations on the KT133(A) North Bridge, so they have no relevance at all to the A1. The specific functions controlled by those registers may be present on the Articia S (but by no means all, the Articia isn't as tweakable as most x86 type designs) but the hardware behind them is different so even figuring out how to do the same things on the Articia probably wouldn't help. I still believe that the SBLive is not really a factor and that either the 686B's effect on the PCI bus has far more widespread effects than just the UDMA/ethernet lockup, or the Articia's PCI controller is not performing quite as it should.

DaveAE:
Quote:

Well, I've known for a long time that the SB Live! is a bitch, certainly together with the VIA686, but I can't tell if it's hogging the PCI bus on the A1. Anyway, since the microA1 has problems too with its CMI chip, I think there is more to it than this.

DWolfman:
Quote:

BTW, I did get to try this out on my XE system. I could not get it to hang even once. I tried with both AmigaAMP (using the WarpUP decoder) and Amplifier (using the OS4 mpega.library), while transferring the YDL 3 ISO images across my network to my server. No problems whatsoever. I have my drives on a Sii0680 card, and the sound card is a Soundblaster 16 (works with the SB128 driver), so I know it is not one of the "Live" cards or even using the EMU10K chipset. ... But mine is a SB16, not a 128. It just happens to have a chip that the SB128 driver supports. Maybe there is something in the card itself, not necessarily the chipset per se, that is the problem? Or there could be multiple different causes (as has been mentioned before) that all result in the same symptom, the lockup.

Wardyone:
Quote:

I would reccommend another sound card, just not a SB Audigy or SB live.. I've done every test imaginable and not solved the audio lockup. As soon as I replaced my SB live 5.1 I've never had a lock up since. As I understand it, there is known issues with the SB live in cobination with the Via 686b. ... ... What we do is fill a usb storage device full of mp3s and play them with AmigaAmp. 2 out of the 3 A1's here froze, but it does take long time to happen and sometimes it does not happen. It was just something we stumbled across when we were looking for this Audio freeze. To add to the stress we sometimes copy large files to a usb storage device. Your A1 might lockup, but to be honest I doubt that it will.

Jack:
Quote:

I tried two sblive chips here: CT4832 SB0220 Both freeze on demand: 3 simultaneous downloads of 100Mb file to ram: along with tunenet0.56 or amigaamp2.35 never survived more than one cycle of download (wget, 1.5MB/s each, from local http server). Without streaming audio 6-7 download cycles went without a blink.

Last edited by wegster on 21-Jul-2005 at 02:09 AM.
Last edited by wegster on 16-Jul-2005 at 05:34 PM.
Last edited by wegster on 10-Jul-2005 at 11:23 PM.
Last edited by wegster on 10-Jul-2005 at 10:57 PM.
Last edited by wegster on 10-Jul-2005 at 10:51 PM.
Last edited by wegster on 10-Jul-2005 at 10:38 PM.

_________________
Are we not done with the same silly arguments and flames yet??!

@wegster

No problems here, using an XE. SB Live! card installed BUT I also have a sii0680 installed so not the onboard VIA for ide. 4 large downloads, amigaamp playing a big playlist and continuing browsing. Everything happy.

Last edited by mtreasure on 10-Jul-2005 at 10:55 PM.

_________________


AmigaOne XE-G4, 512MB RAM (AmigaOS 4.1)
Radeon 9250, Terratec Aureon 5.1
CDRW, DVDRW, SII0680

@wegster
System:
A1XE-G4-933 with hardware fix
upgraded CPU heatsink/fan
512MB RAM, memtester checked
using a1ide.device for all IDE
SB Live/emu10k driver
OS4 pre-release update 3

Results:
reproducible lockup generally in < 30 minutes
Have reproduced > 10 times consistently.

No changes with or without 68k vs PPC, using ANY music software (AmigaAmp, Amplitutude, Songplayer), no changes running from RAM, no changes regardless of means used to download ISOs. No changes with using PIO mode instead of DMA.

Last edited by wegster on 10-Jul-2005 at 11:24 PM.
Last edited by wegster on 10-Jul-2005 at 10:56 PM.

_________________
Are we not done with the same silly arguments and flames yet??!

@mtreasure
Length of time of test? Is the onboard Via used for ANYTHING? (CDRW?)

Is the a1ide.device disabled via uBoot?

Was the test done as outlined above, multiple downloads (say 1GB or more, enough to keep going for 1-2 hours straight) ast same time?

_________________
Are we not done with the same silly arguments and flames yet??!

@wegster

Test ran until the downloads finished. 3 ISOs as in your link plus a download of the OS4 update 1 ISO and I continued to browse etc.

No, the onboard VIA is running nothing here. The a1ide is disabled in uboot, yes.

Test done as per your instructions. CPU is clocked at 866MHz though.

Last edited by mtreasure on 10-Jul-2005 at 11:01 PM.

_________________


AmigaOne XE-G4, 512MB RAM (AmigaOS 4.1)
Radeon 9250, Terratec Aureon 5.1
CDRW, DVDRW, SII0680

@mtreasure

Great, thanks. Any chance to get you to enable the via and re-test?

(Sorry, just trying to get as much info as possible here..)

_________________
Are we not done with the same silly arguments and flames yet??!

@wegster

Reproduced here, downloaded three Debian ISOs and the system crashed (hard lock-up, repeated looped audio) after around 10MB of all files had downloaded.

AmigaOne G4-XE @ 800MHz (unfixed)
SB Live! (Model CT4760 - I would guess different revisions would have different model numbers as I can't see any revision number on it, at least not without removing it and examining thoroughly)
a1ide.device disabled in kicklayout but not in U-Boot (due to not re-enabling it properly after some other tests)

Chris

Last edited by Chris_Y on 10-Jul-2005 at 11:18 PM.
Last edited by Chris_Y on 10-Jul-2005 at 11:17 PM.

_________________
"Miracles we do at once, the impossible takes a little longer" - AJS on Hyperion
Avatar is Tabitha by Eric W Schwartz

@wegster

Quote:

The expected results here are your system will have a hard lockup in < 2 hours. In many cases, it's 30 minutes or less. This may affect more than one Creative soundcard, but affects at least the SB Live and possibly other Creative chipsets chipset apparently.

Has anyone tested this with Linux as well?

_________________
Site admins are people too..pooff!

@wegster

Has anyone confirmed a different sound card than a creative that doesn't have these problems and is reliable?

@RC

Quote:

RC wrote: @wegster Has anyone confirmed a different sound card than a creative that doesn't have these problems and is reliable?

Yes, from the first post in this thread:
Quote:

- Several members (Wardy, Steff) have replaced their SB boards with Terratec/Envy24HT cards and no longer have the same problem, on the same machine.

PS- for those of you with uBoot disabling onboard IDE, what command are you using in uBoot?
setenv a1ide_maxbus 0
or something else?

_________________
Are we not done with the same silly arguments and flames yet??!

@wegster

OK! I got talked into do the same tests on my machine:

A1 XE G4@800MHz, original fan and settings.
SII0680
SB Live 5.1

I disabled the internal ide port both in UBOOT and kicklayout. And the winner is, eh... the results:

I still get the crashes.

Regards,
Jocke 'Zerohero' Birging

_________________
Common sense - So rare it's almost like a super power

@wegster

I should have read the above more thoroughly. I think I will try the Revolution 7.1 and see what happens.

Last edited by RC on 11-Jul-2005 at 02:01 AM.

@wegster

OK, one more time for the record!

A1G4 XE 7455 (933mhz)chip with a MicroCool Northpole heatsink and a 8200 rpm fan moving 10.8 CFM (theoretically?) in a Coolermaster Centurion 5 case with one 120x 120x 25mm Rear Fan(Exhaust) and one 80 x 80 x 20 mm Front Fan (Intake) and a Q-Tec 400W Dual Fan Gold PSU providing power.

I have overclocked my system to 1.066ghz and have set vcore to 1.84 (had it at 1.79 without any problems before except for the audio/network crash)!

Have Radeon 9100 (same chip as 8500) a Sii 0686 controller in the third from bottom pci slot and in the very bottom slot a Juli@ soundcard (this is replacing the SBLive 5.1 Digital Live card I had earlier). The SB has resided in both the bottom slot and the next fro the bottom slot under different tests.

Harddrive is connected to the Sil and a Dvd rom and CDrw is connected to A1ide.device.

Networking using the onboard ethernet.

Have tested downloading 1 single large ISO (4.2gb) separately and playing AmigaAmp and never had any lockups using the SBlive card.

Downloading several ISO's (up to 7 at a time of 680 mb each) separately gives no problems but when repeated with AmigaAmp and playlist it will lockup in less than 30min. Have done this at least 10x with lockups and 2-3 times without audio and no lockups.

Have done the same thing with m-player and AMPlifier and had similar lockups.

Have tried with and without mpega.library etc. same problems.

Have almost always had DNetC running in the background.

Since I replaced the SB 5.1 Live! with the juli@ soundcard and Envy24HT driver similar tests above provoke no crashes. Have done 3 tests with 10 ISO's of 680mb and AmigaAmp playing from a playlist with no problems at all. All files downloaded and music still playing on, DNetC chugging away as usual.

Right now I'm playing from a playlist put on a USB stick (as some people have suggested) in the front USB slot on my case. Sorry but I have not run this test with the SBLive before the switch.

If I have forgotten anything please ask.



@wegster

Quote:

- Several members (Wardy, Steff) have replaced their SB boards with Terratec/Envy24HT cards and no longer have the same problem, on the same machine.

My soundcard is NOT a Terratec card but a very sexy thing from ESI! It does, however, use the same Envy24HT chipset!

Hey my post has been chopped off at the bottom?

Last edited by Steff on 11-Jul-2005 at 01:47 AM.
Last edited by Steff on 11-Jul-2005 at 01:42 AM.
Last edited by Steff on 11-Jul-2005 at 01:40 AM.

_________________
Fixed A1G4XE 7455 RX933PC with fried CPU
Sapphire Radeon 9100 128mb
ESI Juli@ 24bit 192kHz Envy24HT
Sil 680 Ultra Ata 133 E-ide
SeaGate Barracuda 120gb 8mb cache

@Steff

Hey what's up my post above has been chopped off at the bottom. Can see it in edit mode but not after I post?

Oooooops! Have been using illegal characters I guess. All fixed now!

Last edited by Steff on 11-Jul-2005 at 01:48 AM.
Last edited by Steff on 11-Jul-2005 at 01:46 AM.

_________________
Fixed A1G4XE 7455 RX933PC with fried CPU
Sapphire Radeon 9100 128mb
ESI Juli@ 24bit 192kHz Envy24HT
Sil 680 Ultra Ata 133 E-ide
SeaGate Barracuda 120gb 8mb cache

@RC

Quote:

RC wrote: @wegster Is only the Terratec card the good one or is it the Envy24HT chipset that makes the difference. My real question, I guess, is how about the M-Audio Revolution 7,1, which seems to be more available and seems to have pretty good reviews?

Quote:

Envy24HT Supported cards: Quote: OS4 native AHI driver for Terratec Aureon Sky, Phase28 and M-Audio Revolution 7.1. The Aureon Space and Revolution 5.1 should also work but are untested.

Steff has the Juli. It seems _some_ SB cards are also not affected (SB16?), while SB128, Live, Audigy I believe are. Somewhat up in the air on exact models of SB cards, but it would appear SB Live is certainly included in the batch, if we're on the right track.

_________________
Are we not done with the same silly arguments and flames yet??!

@wegster
Quote:

reproducible lockup generally in < 30 minutes Have reproduced > 10 times consistently.

Since you are so good at reproducing this problem (I've never seen it) I will make a suggestion.

I suggest hooking up a null modem cable and capturing the output (if any).

Then try with a debug kernel load and setting the level to about 3 and see if you can see anything. Vary the debug level to taste.

In 99% of cases I've experienced any "lockup" I've been able to capture output from a debug kernel load which pinpointed the problem rather quickly.

_________________
ExecSG Team Lead

@ssolie

Probably time to read the "other" thread (towards the end) where wegster has already done that, a bit.



Quote:

Since you are so good at reproducing this problem (I've never seen it) I will make a suggestion.

Do you have a SBlive in your setup? What does it look like otherwise?

Last edited by Steff on 11-Jul-2005 at 02:21 AM.

_________________
Fixed A1G4XE 7455 RX933PC with fried CPU
Sapphire Radeon 9100 128mb
ESI Juli@ 24bit 192kHz Envy24HT
Sil 680 Ultra Ata 133 E-ide
SeaGate Barracuda 120gb 8mb cache

Does this lockup appear with every Sound Blaster or just the Live?
Is the emu10k driver used for all Sound Blasters?

If changing sound card (& driver) remove the lockups it must surely be
the emu10k driver that has a bug? Or something it does differently
causes a bug to appear elsewhere?
So what does it do differently compared to the Envy24HT driver Dave?

I don't believe this can be caused by heat either. Yes, in some cases
users may have wrong setup causing similar problems. But they would
surely get the same lockups if they used VIA IDE+emu10k sound card.

Seems to me that everyone that says they've got no problems either
not use internal LAN port, VIA IDE or different sound card (driver).

_________________
/Harry

[SOLD] µA1-C - 750GX 800MHz - 512MB - Antec Aria case

Avatar by HNL_DK!

@Samwel
Please read the topmost thread- it answers your questions, at least what we THIN is the problem as of this writing

_________________
Are we not done with the same silly arguments and flames yet??!

@ssolie
Will do so again, trying without AHI output this time, as AHI spams with writes, and seems to offer no help in tracking this down.

_________________
Are we not done with the same silly arguments and flames yet??!