SeaTools

[HughR]

I just bought an ST31500341AS.  This is the 1.5T drive that has had a lot of problems reports, apparently due to firmware bugs.  Mine came with CC1J firmware which seems to be the latest.

As a confidence building measure, I decided to run SeaTools for DOS to check the drive before putting it into service.  I happened to have a copy that I downloaded a couple of weeks ago to diagnose a problem with another drive.  I am using SeaTools for DOS since I'm testing this as the only drive in the system and I don't want to install MS Windows on it (my system of choice is Linux).

The long test starts out fine.  But when I come back hours later, I'm at a DOS prompt. The DOS screen shows boilerplate messages about looking at the *.LOG files.

This seems wrong to me: previous experience with the SeaTools for DOS long test (on my other drive) didn't end this way.  In those tests, the log appeared in real-time in a "window".  When the test ended, the log said so, and I was allowed to continue using the SeaTools GUI.

So I think that SeaTools is crashing back to DOS.  But it isn't crashing instantly: I've checked on the progress perhaps an hour in and all looked well.

When I look at the .LOG file, it mentions the test startup but not completion.  The timestamp on the file is identical to the start time.

This seems repeatable: I've tried it twice with the same result.

When I use the short test, all is as I expect: upon completion, I can continue using the SeaTools GUI.

Computer: an Acer  Aspire E380.  About a year and a half old.  AMD dual-core processor, socket AM2, nVidia chipset.  nVidia 6200-based add-in video card.  No other hard drive.  SeaTools for DOS is booted from a CD.

SeaTools for DOS version 2.13 bPGE.

How can I determine the nature of the termination of SeaTools?  Why did it exit?  Did it crash?  If so, why?

What other tesing methods should I try?

[HughR]

I removed my video card (XFX GeForce 6200 card) and ran the long test again.  This time it finished as I had expected.  No errors were found and the run terminated with the SeaTools GUI ready to run more tests.

What made me try without the video card?  In previous runs, with the video card, after SeaTools terminated and the DOS prompt appeared, I used DOS and found that scrolling was extremely slow.  This made me wonder whether something in the video handling was broken.

Any idea what might be going wrong and how I could fix it?

Of course this result may not actually be related to the video card.

[HughR]

I bought another of these drives, flashed the latest firmware supplied to me by Seagate support, and ran a long test on the same machine again, with the video card removed.

Again, SeaTools for DOS exited without diagnostic.  So the video card doesn't appear to be key.

I removed some RAM, taking the machine back to the state in which Acer shipped it.  This time SeaTools ran to completion.  I have no confidence that the removing of components had any relevance.

Note: when SeaTools exits, the system is left in FreeDOS.  So the machine has not crashed.  SeaTools probably did crash but I don't know how to tell a crash from a clean exit.

During these latest tests, I left the old SATA  (not a Seagate) drive attached.

==> how can I determine why SeaTools is exiting?

==> are there known SeaTools bugs with these symptoms?

==> could this be a hardware problem?  What kind?

[HughR]

I did a second long test without rebooting or restarting SeaTools today.  Although the first test had succeeded, the second terminated when SeaTools exited.  The progress bar had reached 49% the last time I looked (I didn't look often).

[HughR]

I talked to Seagate support using chat.  No answers, but suggestions:

- try another machine

- check for memory problems

- try SeaTools for Windows (but I use Linux; the Linux version only supports SCSI)

- try SeaTools for DOS version 1.10 (CLI version)

Message Edited by HughR on 01-12-2009 11:19 PM

[HughR]

I tried SeaTools for DOS version 1.10.  It isn't CLI.  It is graphical with but with graphics drawn by characters.  It looks crude but it seems fully functional and reasonable to use.

It crashed too.  But since the screen was drawn as characters, it didn't get cleared when it went back to DOS.  Here's what I could glean from the screen:

- 68% complete when it crashed.

- LBA 2012325888 was the last reported.  But I don't think it reports every LBA it tries because that would slow it down.

The program crashed with the following information:

Int 0x1d at eip=15c7; flags=3002
eax=00000001 ebx=0006c360 ecx=00069b60 edx=0011d990 esi=aea87330 edi=0011d9c0
ebp=0011d8f8 esp=000024f4 cs=2b ds=b7 ex=b7 fs=a7 gs=bf ss=33 error=0000

Thank you for using SeaTools for DOS.  A list of test log files is
...

I don't know what Interrupt 0x1d is.  It isn't generated by a standard 8086 exception: the number is too high.

Ralf Brown's Interrupt List says it is a BIOS call that it names "SYSTEM DATA - VIDEO PARAMETER TABLES".  I don't know if that is what we have here.

Is there any way that I could gather more diagnostic information?

In the mean time, I'm running a memory test on the machine.

Message Edited by HughR on 01-12-2009 11:19 PM

[HughR]

This thread might be related: http://forums.seagate.com/stx/board/message?board.id=SeaTool&message.id=367&jump=true#M367

This blog entry seems to describe similar symptoms: http://whyiseverytitletaken.blogspot.com/2008/10/general-computer-woes.html

[HughR]

Over 13 hours of memtest86+ 2.01 found no errors.  Note: I'm still running with my add-on RAM removed, as described in an earlier post.

[HughR]

More testing:

I ran the SeaTools for DOS v1.10 long test again.  It  worked!  Without restarting SeaTools, I ran the test again and it worked. Without restarting, I ran the long test a third time and again it worked.

I power cycled the machine and ran the long test again.  If failed.  Here is the interesting output:

58% complete
LBA 1718552576
Int 0x1d at eip=15c7; flags=3002
eax=0000407c ebx=0006c360 ecx=0024af78 edx=0024eff4 esi=aea87330 edi=0011d9c0
ebp=0011d8f8 esp=000024f4 cs=2b ds=b7 es=b7 fs=a7 gs=bf ss=33 error=0000

 I power cycled yet again and ran the long test.  It failed much sooner.  Here is the similar output:

1% complete
LBA 53811200
Int 0x1d at eip=15c7; flags=3002
eax=0022383c ebx=0006c360 ecx=00069b60 edx=00223838 esi=aea87330 edi=0011d9c0
ebp=0011d898 esp=000024f4 cs=2b ds=b7 es=b7 fs=a7 gs=bf ss=33 error=0000

 All three crashes have produced similar diagnostics.  I hope that this is enough information for Seagate to debug this.
 

[HughR]

Still more testing:

I had a theory that once a long test worked, a subsequent test would too, as long as SeaTools was not exited (clearly the machine would not be rebooted either).

I ran a long test 4 times in one invocation of SeaTools for DOS.  The first three worked.  The fourth failed in the usual way (I didn't bother transcribing the diagnostic, but it started with "Int 0x1d" ).

Is this enough information for Seagate to fix SeaTools?  If not tell me what is need.

Message Edited by HughR on 01-23-2009 11:42 PM