OS2AHCI not working on TP T510

[Silvan Scherrer]

When booting with Danis and in Bios compatibility mode it works.

When i switch to AHCI in Bios AHCI gives me:
Found AHCI device Intel PCH AHCI (0:31:2 8086:3b2f) class:0x010601

And after that i get a:
OS/2 is unable to operate your hard disk or diskette drive.
The System is stopped

The OS2AHCI.ADD is version 1.31 from 8th August 2013

[abwillis]

Did any version 1.26 or greater work? I am wondering if it is a regression or something that was actually never fixed for that system. Check out #11

[Silvan Scherrer]

I tried 1.30 and 1.28. But both showed the same error.

[Silvan Scherrer]

find the testlog.cmd output attached

[David Azarewicz]

Since the AHCI driver is not loaded, the log you provided does not contain any useful data. The PCI output also does not show any compatible AHCI controllers. Please only attach logs that are taken when the system is in the configuration with the reported problem. Logs taken in other configurations are not useful.

Please note that there is no official supported version 1.31. Only test builds, and there have been several.

You can try the current test build if you like: ​http://88watts.net/download/ahci-test-1.31-20130822.zip

If you actually have a compatible AHCI controller and this build does not work, I will need a startup debug log. Please see: ​http://svn.ecomstation.nl/ahci/wiki/Troubleshooting

[Silvan Scherrer]

i looked at the Troubleshooting section already several times. But i could not figure out how to do a "Startup Debug Log", when the system does not boot.

I will try the version you mentioned above soon and let you know.

[David Azarewicz]

Replying to diver:

i looked at the Troubleshooting section already several times. But i could not figure out how to do a "Startup Debug Log", when the system does not boot.

I will try the version you mentioned above soon and let you know.

If it won't boot, the serial debug log, or a trapdump is the only way.

[Silvan Scherrer]

it boots from DVD, so i try to get a log out of this boot and copy it to a USB stick. i guess that should work.
as the system of course has no serial port and a trapdump with 4GB ram doesn't work afaik.

[David Azarewicz]

That is unlikely to work. What I need is the contents of OS2AHCI$, so if you copy OS2AHCI$ to a file and it says no such file, then it didn't work.

A trap dump at the point in time when you have your failure will be very small, only a few hundred K.

Doesn't the T510 have a serial port when connected to a docking station?

[abwillis]

The T410 and up do not have a serial in the dock (at least not the docks I have seen, though it may be possible to find such).
Silvan may have better success than I have but I have never been successful in getting any output to the dumpfs with a 4G memory chip installed.

[Silvan Scherrer]

hmm my USB scenario seemd to work. At least i got a file.

[Silvan Scherrer]

for the completeness, the attached driver behaves exactly the same.

[David Azarewicz]

Yes, your boot off the DVD and copying to a USB stick did work. However, it was the old driver from 08-Aug-2013 that was used. Nevertheless, some useful information was obtained.

From the data in the log, the problem is that the drive in port 0 is failing every transaction. The communication with the port is working OK. The communication with the drive is working OK. It is the drive itself that is failing the commands. By contrast, the drive in port 1 is working OK.

Please try another test. Using the newest driver from 22-Aug-2013, add the /!r switch and see if that makes any difference. You must use the 22-Aug-2013 driver for this test. This switch will not have the desired effect with the older drivers.

[Silvan Scherrer]

hmm now I need to figure out how to boot with this driver, as the DVD has the old of course.

[David Azarewicz]

I assumed you could test booting off the hard disk.

If you need me to build you a small test ISO, I can do that.

[abwillis]

Do you have a USB floppy drive? If so there are instructions on the CD on how to setup a floppy to copy it over during the boot. The other option, boot with Dani's driver and then make the change to the config.sys and make sure the latest driver is on the harddrive and then boot and make the change in the bios to ahci and first see if it just boots.

[Silvan Scherrer]

Booting from harddrive direct is not working. Even with the driver from 22. August and the /!r parameter. So i can only boot from DVD/CD. I guess a small ISO with USB drivers would be good. So i can test with that.

[David Azarewicz]

Ok, thanks. That means the result of the test is that it didn't work. I'll have to think about this a bit. If I come up with another test, I'll provide a test ISO. Thanks.

[David Azarewicz]

Ok, try this build. ​http://88watts.net/download/ahci-test-1.31-20130904.zip

A simple works/doesn't work is OK, so you can just try it on your hard disk. I assume you have more than one boot partition so you can boot the partition that works, copy the driver into place and reboot to test.

Also, in case you didn't already know, you don't have to turn AHCI mode off and on in the BIOS. You can just leave it set to AHCI mode. DaniS506 can run the hardware just fine that way.

If the system boots with this driver, please attach a new log file. (use the /d:2 switch).

[Silvan Scherrer]

Sorry to say, but same result. Means not working.

[David Azarewicz]

Very strange. I thought that might work.

Here is a test ISO: ​http://88watts.net/download/TestEcsIso.7z The file is 49MB and only 80MB unzipped so it can be written to a CD.

Be sure to specify the /d:2 switch on the setup screen for the AHCI driver and capture a log for me.

Also, do you happen to know if that disk works in AHCI mode with Linux? The only way to to know for sure is to look in the Linux boot log and see how the adapter was configured.

[Silvan Scherrer]

did you change more drivers on that ISO? As soon as i stick in my USB device i get a trap 000d in Module USBMSD

[Silvan Scherrer]

new log

[Silvan Scherrer]

As the USB way did not work i tried the way with an external SATA device. This device got a drive letter and so i could copy the os2ahci$ to that. I hope the log still worked.
And i have no clue if the AHCI works on Linux. I know it worked on Windows.

[David Azarewicz]

Unfortunately, whether or not it works on Windows is pretty useless information.

I see the new drive attached on port 5, and of course, it worked fine. It doesn't have the problem that the drive in port 0 does.

The problem is that the drive in port 0 disconnects and reconnects itself every time a transaction is requested. This sets a bit in the error register. Apparently this is really a non-fatal message so I am just masking it off so it gets ignored. I suspect that there really is something wrong with that drive since properly working drives don't do that. However if it works, I don't see anything seriously wrong with ignoring this particular error since this only occurs at init time. After init time, errors are handled differently by the interrupt handler.

There were 2 spots where this status was checked and I missed the second one in the last build. This new build should work given the previous test results.

​http://88watts.net/download/ahci-test-1.31-20130906.zip

Please let me know if it works.

[Silvan Scherrer]

Just tried that driver, but still no joy.

[Silvan Scherrer]

ok the above was with /!r in place. When w/o that switch it behaves different. Means after loading os2ahci i see a lot disk activity for several minutes. And then the known message appears.

[abwillis]

Do you have another disk to try in the machine?

[David Azarewicz]

That just proves that that drive is producing too many errors. At this point we know that it is the drive that is giving us problems. We know that the adapter is working OK (at least for the other ports) and we know that the driver is working OK. The problem is dealing with the errors that drive is producing.

The next test I would do are:

1. Try a different drive in that port. It could be the disk or the port.

2. Try that disk in a different port. It could be the disk or the port.

It is probably non-productive to continue debugging this, but if you want to I suggest you setup that system to boot using a different disk. That way you can try different drivers easily and get logs easily.

What you really need is a drive that works properly.

You could also choose to only operate that disk in EIDE mode, although if it really is having problems, it may eventually fail in that mode, too.

[Silvan Scherrer]

i tried a different disk and saw the exact same error. so i guess the disk is not the problem.
how to try at another port is not that easy (i guess) with a ThinkPad?.

[David Azarewicz]

Please try ​http://88watts.net/download/ahci-test-1.32-20131030.zip on your system and let me know if your disk on port 0 works better. Thanks.

[Silvan Scherrer]

hmm i don't know what you did, but now it found my disk and it could boot in ahci mode.

now i wonder if i should work with that mode. do you think it's safe? or should i wait till you release the new version?

[David Azarewicz]

Excellent. I discovered that some hardware reports bad (incorrect) status bits. I implemented a fix so it will work even if the status bits are wrong. It is safe to use that driver until it expires in 30 days. I expect I will release an official version before then.

[David Azarewicz]

Fixed in version 1.32.