AppleVIAATA over 4gb RAM bug solved (at least i believe so)! JMicron and AppleNForceATA at development

[El Massman]

i'm getting the same problem

 

can't determine immediate dependencies for com.jmicron.JMicronATA
can't determine dependencies for com.jmicron.JMicronATA
can't alloc class="JMicronATAPIroot"

 

let me know if there's anything i can do to help testing wise. or any information you need.

 

thanks,

 

Marc

[slashack]

Ok i just got a working JmicronATA.kext form JMICRON ftp. It still has the RAM problem but at least it loads. So you can try modify this one hope this will help.

 

I've uploaded a new version, using Info.plist from your kext. Did someone tried with the unmodified JMIcron? does it loads correctly? I think the problem might be in the original source code i'm working.

[adusak]

Just tried the update and it gives me kernel panic while booting had to unplug my pata devices just to get back to the system

[slashack]

Just tried the update and it gives me kernel panic while booting had to unplug my pata devices just to get back to the system

 

Can u give me some info on the panic? (boot with debug=0x100)

Have u tried the original kext i uploaded? That's the source code without any modification by me. If this works, then my modification should work too, but if it doesn't work, then the source code is not usefull.

[adusak]

Yes i tried your unmodified kext but after booting it didnt find my pata devices so it probably didnt load.

I will upload some pictures of the kernel panic in a few moments.

 

EDIT: Ok i attached images of the panic

[slashack]

Yes i tried your unmodified kext but after booting it didnt find my pata devices so it probably didnt load.

I will upload some pictures of the kernel panic in a few moments.

 

EDIT: Ok i attached images of the panic

 

Ok, the driver is working now. That's good news. This issue is known to me. In AppleVIAATA, dma needed to be 2 byte aligned, but in JMicron, it needs to be 4 byte aligned. I'll fix that asap.

[slashack]

Yes i tried your unmodified kext but after booting it didnt find my pata devices so it probably didnt load.

I will upload some pictures of the kernel panic in a few moments.

 

EDIT: Ok i attached images of the panic

 

I've updated JMicron kext (v0.4) Try them and tell me what happened.

 

Sorry for all this, but i don't have such hardware to test them.

[rfuilrez]

I've updated JMicron kext (v0.4) Try them and tell me what happened.

 

Sorry for all this, but i don't have such hardware to test them.

 

No dice on that one either. Attached is a photo of the Kernel Panic. Occurred during boot..

[slashack]

No dice on that one either. Attached is a photo of the Kernel Panic. Occurred during boot..

 

Theres a new version (0.5). I forgot to update some pointer references in the transfer.

Be patient, the 4gb issue is in the method i'm actually doing changes (createChannelCommands), the one that panics. I believe this kext might work.

 

Thanks to the testers!

[mitch_de]

Theres a new version (0.5). I forgot to update some pointer references in the transfer.

Be patient, the 4gb issue is in the method i'm actually doing changes (createChannelCommands), the one that panics. I believe this kext might work.

 

Thanks to the testers!

Thanks to you also !

I buyed yesterday an 10 Euro IDE>SATA module to avoid that > 3 GB RAM problem with my JM until we get it fixed. WORKED. I now have disabled all IDE on my GA board.

My solution may be helpful if you only have one IDE left (i have SATA DVD and one SATA HD, only one IDE HD). If you have more than one IDE i would wait for that great work or sell the IDEws to ebay and buy SATA therefor.

[rfuilrez]

Theres a new version (0.5). I forgot to update some pointer references in the transfer.

Be patient, the 4gb issue is in the method i'm actually doing changes (createChannelCommands), the one that panics. I believe this kext might work.

 

Thanks to the testers!

 

Well, I'm not going to jump to any conclusions and say that it's 100% fixed for sure. However, this is me playing a video from my IDE Drive with the majority of 4GB allocated.

 

 

And here's the only stuff I saw in the boot log.

 

 

It seems to me, you may have hit the nail on the head with this fix.

[adusak]

Ok i think this could be the thing. But i still experienced panic i dont believe it is related but i attache pictures anyway.

[mitch_de]

Ok i think this could be the thing. But i still experienced panic i dont believe it is related but i attache pictures anyway.

if you set maxmem=2000 (as kernel boot flag)do you also get panics ? I f yes something other may be wrong.

 

I would also check RAM with REMBER / memtest - problematic RAM can also make much troubles.

 

I got (now i use an IDE>SATA contr) kernel panic really early at boottime with JM and 4 GB. I never saw the desktop , crash was way before.

[slashack]

Ok i think this could be the thing. But i still experienced panic i dont believe it is related but i attache pictures anyway.

 

I don't think those panics are related, but as rfuilrez said, don't jump to conclusions.

 

The panic ocurred somewhere in hfs filesystem, the thread was backupd. Maybe your hdd has something wrong, and not the ide controller. The other issue might be memory. Try testint them (verify hdd in disk utility) and if panics is recurrent, try booting with maxmem=2048. If u keep having panics, then the issue is not this kext.

[bikinifarm]

I tried the latest version (1.06) from the JMicron web site as indicated by adusak, and with my PATA DVD drive (PIONEER DVD-RW DVR-115D), the inserted DVD looks like an external drive. I believe the its plist causes this.

 

Version 1.0 JMicron driver worked properly in this regard, except causing kernel panics with the memory issue.

 

Then I tried the latest slashack version (0.5) with 64 bit addressing, and it displays DVD the same way. [did not test for the memory issue]

 

Unfortunately, this is not just a display issue, it confuses the finder as to whether it can burn blank disks. Ejection is also goofed up.

 

Slashack, if your recompile is based on the latest version 1.06 from JMicron's web site, you would have inadvertently carried these issues over. Based on my limited understanding, you may need to revisit the plist based on the original 1.0 version, instead of the 1.06 version, or at least use the version 1.0 as a base.

 

Thanks for all of your help. Really appreciated.

[slashack]

I tried the latest version (1.06) from the JMicron web site as indicated by adusak, and with my PATA DVD drive (PIONEER DVD-RW DVR-115D), the inserted DVD looks like an external drive. I believe the its plist causes this.

 

Version 1.0 JMicron driver worked properly in this regard, except causing kernel panics with the memory issue.

 

Then I tried the latest slashack version (0.5) with 64 bit addressing, and it displays DVD the same way. [did not test for the memory issue]

 

The memory issue is the one i solved. I'll try and solve the other issue. If u could test if kext work it would be great. (run rember to test all memory and try and open files from IDE disks)

Unfortunately, this is not just a display issue, it confuses the finder as to whether it can burn blank disks. Ejection is also goofed up.

 

Slashack, if your recompile is based on the latest version 1.06 from JMicron's web site, you would have inadvertently carried these issues over. Based on my limited understanding, you may need to revisit the plist based on the original 1.0 version, instead of the 1.06 version, or at least use the version 1.0 as a base.

Can you attach a plist from v1.0?

Thanks for all of your help. Really appreciated.

[bikinifarm]

The memory issue is the one i solved. I'll try and solve the other issue. If u could test if kext work it would be great. (run rember to test all memory and try and open files from IDE disks)

 

Can you attach a plist from v1.0?

 

I am attaching the plist as well as a few screen shots as to how the DVD looks on the desktop. Many thanks slashack!

 

The original JMicron v1.0:

 

 

The JMicron 1.06 from their web site:

 

 

The 64 bit corrected version by slashack:

 

 

and the plist from the JMicron version 1.0:

 

Info.plist.zip

[rfuilrez]

One thing I did notice with THIS driver is the allocated memory

didn't appear to come back to the free pool or it took its time.

 

That's actually how OS X manages memory. Even without this kext it will do it the same way. It keeps the memory allocated, but flags it as inactive. So if you use the program again, it already has the data loaded. And then if you need memory for another program, it will free up what's needed, and allocate it for that app instead.

[Volcacius]

Tested modified AppleNForceATA.kext and it works perfectly!! it's incredible, no more KP in situations that used to give me panics.

[Zulu.Walker]

Altered JMicron (0.5) works well with JMB363 controller set as IDE under BIOS - 2 SATA and 2 IDE detected, IDE has one DVD writer that has burned a few discs with no problems so far - all while running Highload memory stresstest on 4, 6, and 8GB memory configurations. Speeds are the same as the 32-bit version, but without the bugs. I'm guessing this kext (and all that are based around it) will still work under Snow Leopard - making a lot of JMicronATA/AppleVIAATA.kext dependent mobos future-proof.

 

You're a godsend slashack. People have been complaining about this issue and none have made any moves to fix it - you've done it all in one fell swoop. Your AppleVIAATA is also working on another system on my mini-render "farm" for a 2-port Silicon Image 3112A that's been flashed under windows with the BASE SATALink firmware straight from SiI.

 

Thank you. Really. It's about time someone had the brains to address this issue.

[mitch_de]

Altered JMicron (0.5) works well with JMB363 controller set as IDE under BIOS - 2 SATA and 2 IDE detected, IDE has one DVD writer that has burned a few discs with no problems so far - all while running Highload memory stresstest on 4, 6, and 8GB memory configurations.

Thank you. Really. It's about time someone had the brains to address this issue.

Can you tell me which DL-LINK this Altered JMicron (0.5) has ?

[slashack]

I am attaching the plist as well as a few screen shots as to how the DVD looks on the desktop. Many thanks slashack!

 

The original JMicron v1.0:

 

 

The JMicron 1.06 from their web site:

 

 

The 64 bit corrected version by slashack:

 

 

and the plist from the JMicron version 1.0:

 

Info.plist.zip

There are basically 2 differences between plists:

1)

slashack:

Under IOKitPersonalities -> JMicronATA Driver:

			   
		<key>IOMediaIcon</key>
		<dict>
			<key>CFBundleIdentifier</key>
			<string>com.apple.iokit.IOStorageFamily</string>
			<key>IOBundleResourceFile</key>
			<string>External.icns</string>
		</dict>

I believe this is the icon issue. In 1.0 plist this isn't there.

 

2)

Under IOKitPersonalities -> JMicronATA PCI:

slashack:

			   
		<key>IOPCIPrimaryMatch</key>
		<string>0x2361197b 0x2363197b 0x2365197b 0x2366197b 0x2368197b</string>

1.0:

		  
		<key>IOPCIClassMatch</key>
		<string>0x01010000&0xffff0000</string>
		<key>IOPCIPrimaryMatch</key>
		<string>0x2368197b 0x2360197b&0xfff0ffff</string>

 

To solve the icon issue, you might remove the code from Info.plist, and i think that will solve it. (remove everything i quoted in difference n1).

I don't think that changing the code in difference n2 from original 1.0 plist might solve the other issues you are having trouble with. This keys indicates the OS when to load this driver.

 

 

Can you tell me which DL-LINK this Altered JMicron (0.5) has ?

 

In first post (i've updated the links) or in Darwin ATA googlecode project

[bikinifarm]

There are basically 2 differences between plists:

1)

slashack:

Under IOKitPersonalities -> JMicronATA Driver:

			   
		<key>IOMediaIcon</key>
		<dict>
			<key>CFBundleIdentifier</key>
			<string>com.apple.iokit.IOStorageFamily</string>
			<key>IOBundleResourceFile</key>
			<string>External.icns</string>
		</dict>

I believe this is the icon issue. In 1.0 plist this isn't there.

 

2)

Under IOKitPersonalities -> JMicronATA PCI:

slashack:

			   
		<key>IOPCIPrimaryMatch</key>
		<string>0x2361197b 0x2363197b 0x2365197b 0x2366197b 0x2368197b</string>

1.0:

		  
		<key>IOPCIClassMatch</key>
		<string>0x01010000&0xffff0000</string>
		<key>IOPCIPrimaryMatch</key>
		<string>0x2368197b 0x2360197b&0xfff0ffff</string>

 

To solve the icon issue, you might remove the code from Info.plist, and i think that will solve it. (remove everything i quoted in difference n1).

I don't think that changing the code in difference n2 from original 1.0 plist might solve the other issues you are having trouble with. This keys indicates the OS when to load this driver.

In first post (i've updated the links) or in Darwin ATA googlecode project

 

 

Slashack;

 

I followed your advise, and removed the entire IOMediaIcon key (per note #1) from the plist. I made no other changes. The DVD and CD disks are now recognized and displayed as such. I also tested burning a CD, which completed with success.

 

I noticed that blank DVDs are taking a long time to show up, (1-2 minutes), but it may very well be normal. This was one of the odd behaviors that I had noticed before. I also did testing with rember, and DVD read test you had recommended in your previous post, and I am happy to report that there are NO kernel panics. So, it looks like everything is just fine. You may want to update the plist so that the driver is usable with DVD burners.

 

MANY MANY Thanks for your help in putting this JMicron issue to bed at last. Well done...

[mitch_de]

Thanks posting the google JM 0.5 link again!

I will try it with my 4 GB GA-EP35-DS3 (using IDE>SATA converter to avoid JM > 3 GB crashes )

[rfuilrez]

Thanks posting the google JM 0.5 link again!

I will try it with my 4 GB GA-EP35-DS3 (using IDE>SATA converter to avoid JM > 3 GB crashes )

 

Then really, what's the point? The whole purpose of the modified kext, is to allow IDE controllers to work with > 3 GB of RAM. I can testify that it will work on your board. I'm using the GA-EP35C-DS3R board with 4GB of RAM flawlessly now for over a week. Current uptime is over 3 days, and I've allocated the majority of 4GB on multiple occasions while using an IDE hard drive.