arga

Samsung 1TB HD103UJ: HDDScan problem (476 green blocks)

Recommended Posts

But the beta 6c says nothing about this problem. It says it is for 45nm CPU.

I will see if I try it later. I am quite tired of playing with things... :)

I have another Gigabyte mobo with G31 chipset: the GA-G31M-S2L.

I will try this one with the Samsung.

My BIOS in this one is F5. And it seems they addressed the Samsung problem in F3:

"Fix: Some of 1000 GB (1TB) HDDs will be detected size error. "

-----

Well, now I am running again the CMRR Secure Eraser. And there is something weird:

this time, the program has not warned me about the special areas HPA and/or DCO.

I don't know if they have dissapeared.

Are they important? Can I restore them?

Share this post


Link to post
Share on other sites
Well, now I am running again the CMRR Secure Eraser. And there is something weird:

this time, the program has not warned me about the special areas HPA and/or DCO.

I don't know if they have dissapeared.

Are they important? Can I restore them?

Alas, I found the cause. When I run the CMRR eraser with the Samsung drive as only drive (as Secondary master) , HPA and DCO areas appear in the disc, and are detected by CMRR.

Also, the drive will shrink to 32MB, even if the CMRR Erasure is not finally run.

So both things are related: When the drive is the only drive, size goes down to 32MB, and HPA/DCO areas appear.

But if I run CMRR with 2 disks, one Primary Master, and the Samsung as Secondary Master, then these areas doesn't exist, and the 32MB problem doesn't appear.

More: I have run tjhe CMRR erase a second time, but this time there was no improvement regarding slow blocks. I am now above 100 'greens'.

Share this post


Link to post
Share on other sites

@weaker:

I have just tried the Samsung in my other Gigabyte mobo (G31 chipset).

The 32MB problem doesn't happen here (I have tried the drive alone, of course).

BUT size decreases by 1 MB.

It seems that the mobo is reserving 1 MB for itself. Maybe this is the HPA/DCO area.

Share this post


Link to post
Share on other sites

Did you run a scan on this drive with mhdd?

Whenever there is a problem with a drive on windows, common consensus is that you must boot into pure DOS and run a disk utility there. This way utility may bypass OS, BIOS, controller, etc and work directly with the drive.

Well, if all is fine, i dont bother. But if there is a problem with the drive, that is what i do.

Boot into mhdd. Type "port" <enter>. Press "1" <enter>. Type "ID" <enter>. If your drive details come up, good. If not, repeat with 2-3-4-5, etc instead of 1.

When you found your drive, type "scan" <enter>. Then press "CTL + ENTER". It should scan your drive. Make sure EraseWaits, Remap, etc... are all off (defaults are non-destructive).

Also make sure you have a single drive in the system to eliminate problems with BIOS, drives, etc.

And after this, i would call it a day for you.

Share this post


Link to post
Share on other sites
Did you run a scan on this drive with mhdd?

Whenever there is a problem with a drive on windows, common consensus is that you must boot into pure DOS and run a disk utility there. This way utility may bypass OS, BIOS, controller, etc and work directly with the drive.

Well, if all is fine, i dont bother. But if there is a problem with the drive, that is what i do.

Boot into mhdd. Type "port" <enter>. Press "1" <enter>. Type "ID" <enter>. If your drive details come up, good. If not, repeat with 2-3-4-5, etc instead of 1.

When you found your drive, type "scan" <enter>. Then press "CTL + ENTER". It should scan your drive. Make sure EraseWaits, Remap, etc... are all off (defaults are non-destructive).

Also make sure you have a single drive in the system to eliminate problems with BIOS, drives, etc.

And after this, i would call it a day for you.

Hello 666:

Yes, I tried MHDD scanning, and the results are approximately the same as those of HDDScan in Windows. I am running HDDScan on a pretty clean installation of Windows, and nothing else is running, no internet connection, nothing. So the results are reliable. But it is good to know MHDD, just in case I don't have a clean installation of Windows available, then I will use MHDD in DOS.

I erased the disk with MHDD too, but no improvement in slow sectors beyond what I achieved with the first erasure using CMRR. By the way, as I have said before, CMRR didn't cause the shrinkage to 32MB. It was the Gigabyte mobo. Gigabyte mobos normally reserve 1 MB of disk space on the primary disk. But this mobo reserved much more :)

My other Gigabyte mobo (G31 chipset) works correctly. It only reserves 1 MB.

Furthermore, I have upgraded the BIOS of the buggy mobo,to version F6c, and now works correctly too, reserving just 1 MB on the Samsung.

So this has been a good day, I guess. Yes, now is time to rest :)

@ weaker:

I have flashed my Gigabyte mobo (the one who shrinked the drive to 32MB) to the latest version, the F6c, and voilá, the problem has been solved, and now the mobo only reserves 1 MB on the disk when it is the primary disk. It is funny because this improvement is not documented in the F6c legend.

The buggy Gigabyte mobos, instead of reserving 1MB, they reserve almost the full Terabyte, and you are left with only 32MB. I think this is what happens.

But now both GB mobos have the problem solved. :)

Edited by arga

Share this post


Link to post
Share on other sites
I am running HDDScan on a pretty clean installation of Windows, and nothing else is running, no internet connection, nothing. So the results are reliable.

MHDD bypasses DOS and BIOS and talks directly from controller to the drive. You eliminate OS/driver issues in this scenario. In windows, there are too many factors to consider to single out a problem. For example, my jmicron controller driver is buggy. They were listing PATA drives as removable storage in windows till just a few weeks ago! It bails out every once in a while and windows test results get screwed. In this scenario, i drop to DOS and test drive fitness from there. Almost all the time problems are not arising from the drive. In fact, I never had any drive fail after running stress tests for a day on it. If it did not fail, or if it was not a defective drive, i do not get problems with it later on [except when i really fiddle with a drive quite a lot].

HDDScan... has some bugs... For example, when chosen READ test, it is supposed to transfer the data through the controller. But it does not. I think it just does verify when READ is chosen.

SOLUTION: Run Victoria for Windows 4.3 for READ tests.

It also does not create proper MHT reports for archival purposes. I was shocked to find the graph for one of my drives was replaced with the current scan. Apparently, it does not include the image in the MHT file but makes a reference to its location in hddscan folder. This is not good. By the time i found this out, i lost many reports. Because after you run the test, the report works fine even if you move it elsewhere. So you think it is done with, and move it and run another drive. Only to find out in the future that none of your graphs are present!

SOLUTION: Open the hddscan report file in IE. Save it as html. Open the html, save as MHT.

Edited by 6_6_6

Share this post


Link to post
Share on other sites
I am running HDDScan on a pretty clean installation of Windows, and nothing else is running, no internet connection, nothing. So the results are reliable.

MHDD bypasses DOS and BIOS and talks directly from controller to the drive. You eliminate OS/driver issues in this scenario. In windows, there are too many factors to consider to single out a problem. For example, my jmicron controller driver is buggy. They were listing PATA drives as removable storage in windows till just a few weeks ago! It bails out every once in a while and windows test results get screwed. In this scenario, i drop to DOS and test drive fitness from there. Almost all the time problems are not arising from the drive. In fact, I never had any drive fail after running stress tests for a day on it. If it did not fail, or if it was not a defective drive, i do not get problems with it later on [except when i really fiddle with a drive quite a lot].

HDDScan... has some bugs... For example, when chosen READ test, it is supposed to transfer the data through the controller. But it does not. I think it just does verify when READ is chosen.

SOLUTION: Run Victoria for Windows 4.3 for READ tests.

It also does not create proper MHT reports for archival purposes. I was shocked to find the graph for one of my drives was replaced with the current scan. Apparently, it does not include the image in the MHT file but makes a reference to its location in hddscan folder. This is not good. By the time i found this out, i lost many reports. Because after you run the test, the report works fine even if you move it elsewhere. So you think it is done with, and move it and run another drive. Only to find out in the future that none of your graphs are present!

SOLUTION: Open the hddscan report file in IE. Save it as html. Open the html, save as MHT.

Hello 666:

thanks for your many advices.

Okay, I will use MHDD in DOS to scan the drives, instead of HDDScan in Windows. In the hddguru forum, I have received the same advice from several posters. Now I have just realized, that this would be particularly useful when you need to scan the boot, primary drive.

I only use "verify" in HDDScan. I didn't know the difference between read and verify. I decided to use "verify" long ago, I don't remember why I guess I tried both and I chose verify as giving more meaningful results. Now, reading your post, I understand that if you choose "verify", then the data is read by the disk, and verified, and you don't get the data, but only the result of the verification made by the disk. And if you choose "read" then the data should be read by the controller, and the program.

Ok, I will try Victoria and the READ test.

Yes, the MHT files lose the images. I noticed it very quickly. I tried to open the mht files in another computer and instead of images, there were little crosses. And, as you say, if I open the mht file in the same PC, then the images are shown, because it finds the image in the HDDScan folder.

What I was doing to keep the result, was simply copying image captures into a text file (.rtf). and I copy the log too (I highlight the slow sector list, copy and paste).

But now I see your technique, and it is better: open the MHT file in IE, and resave it.

I have tried to save it directly in MHT format, and the images are kept. You don't need the extra step of saving in HTML first, unless you really want to keep both formats.

cheers

Edited by arga

Share this post


Link to post
Share on other sites

just another question: If I create bad blocks with MHDD, will I be able to regenerate them (eg, with HDDregenerator), or at least map them out?

I say this, because if I create a bad block to force the RMA, but Samsung doesn't accept the RMA, then I would get the disk back with the bad block.

Edited by arga

Share this post


Link to post
Share on other sites

arga, of course bad blocks will be mapped out by the drive. they will show up under SMART "reallocated sector... etc". HDDRegenerator is snake oil i believe. If there was something like this, manufacturers would have done it themselves and saved lots of money.

Why shouldn't samsung accept a drive with bad blocks? They will. Just trash it beyond repair.

I bought a silent fridge few years back. You cannot test these fridges :) When i was purchasing it, i specifically insisted on silent, quiet and and all, and the showroom confirmed fridge is silent, that is it is a hitachi with a japanese compressor blah blah. They installed the fridge, i wanted to return it within the first hour! I thought, let me wait for a day and let the compressor settle the gas and complete its cycles... I waited, same thing, no improvement. Went to the shop right away. It was still the same day. They said they cannot take a fridge back because of noise. We don't have 30 days unconditional returns or any consumer rights here. You have to take matters into your own hands.

There was no way i was going to tolerate that compressor noise. So i wrapped the fridge with sound insulation material. You know, they require the fridges to have 10 cms clearance from all sides for air circulation.... Well, my fridge had no clearance whatsoever from anywhere. All areas that air/sound can escape were covered with insulation material. It was almost air-tight. I thought, alright in a week or so, compressor will fail and they will exchange / refund my fridge. Almost 3 years and that compressor still did not bail out with zero clearance and zero air-circulation. Damn, Japanse make good compressors! I must put my ear to the wall of the fridge to hear if it is working :)

Some might call it unethical... Well, i dont claim something which is not mine. I just paid for it, and it is my every right to get a satisfactory product. It is not a random buy -- i was very very specific as to what i needed to the sales people.

just another question: If I create bad blocks with MHDD, will I be able to regenerate them (eg, with HDDregenerator), or at least map them out?

I say this, because if I create a bad block to force the RMA, but Samsung doesn't accept the RMA, then I would get the disk back with the bad block.

Share this post


Link to post
Share on other sites

@arga: Sorry for not posting earlier. Yes, the Gigabyte BIOS Backup is exactly the cause. The HPA changes (1 MB shrink or more exactly 2113 sectors to take the BIOS copy) and this was done wrongly for 1 TB drives. Sometimes Gigabyte is quite reluctant to write lengthy changelogs which means that things change that are not written explicitely in the changelog.

It seems to happen to the first drive the BIOS sees. That's why it didn't happen when you had the 1 TB drive as secondary.

Thanks for doing the tests!

Share this post


Link to post
Share on other sites

Hello 666:

I once tried HDDRegenerator in one of my drives.

It was a Samsung P80, 160GB. I bought it in 2003.

To keep the story short, it had a bad block.

I used HDDRegenerator, and the bad block dissapeared.

I though this bad block was fixed, but now, looking at the Smart Report carefully, I see it was really mapped out,

because

Reallocating events (RAW) =1

Uncorrectable sector count (RAW) =1.

This are the Raw values. The "processed value" is =100=perfect.

The semaphore is yellow, but back then, I thought it was an error, because the "Processed Values"=100=Perfect.

But now I realize that there was 1 sector reallocated.

So maybe HDDRegenerator is not as good as I thought... :)

I have scanned this drive now with MHDD, and it shows 30 "greens". Considering that its only 160GB,

its pretty high.

The rest of my drives are ok. No problems.

Ok, I think I will try to create a Bad Block with MHHD. Can I create a single bad block in the location I choose?

Another doubt: will I get a new one, or a refurbished one?

I am sorry to hear your experience with that fridge. You wanted a specific product, and the salesman fooled you.

Then you had to spend time modding it, and you had to take risks.

Just imagine if it fails just after the warranty expiration.

I am also very demanding about noise.

I have read testimonies of people buying TFT monitors with noisy power supplies. This is also a delicate situation because

the manufacturer could deny a replacement.

In Spain, there is a mandatory 7-day no-question warranty. Some shops extend it to 10, 15, or even 30 days. Internet shops

normally give the minimum (7 days).

I was slow, and didn't test the Samsung F1 quickly enough. Next time I will test it in the first day, and it will go back to the shop immediately if there are too many greens.

Share this post


Link to post
Share on other sites
@arga: Sorry for not posting earlier. Yes, the Gigabyte BIOS Backup is exactly the cause. The HPA changes (1 MB shrink or more exactly 2113 sectors to take the BIOS copy) and this was done wrongly for 1 TB drives. Sometimes Gigabyte is quite reluctant to write lengthy changelogs which means that things change that are not written explicitely in the changelog.

It seems to happen to the first drive the BIOS sees. That's why it didn't happen when you had the 1 TB drive as secondary.

Thanks for doing the tests!

ok, I understand.

thanks to you too!

Share this post


Link to post
Share on other sites

weaker, I don't think it is BIOS backup. I think it is Easy Recovery... Easy Something... That gigabyte uses to clone drives or make images.

@arga: Sorry for not posting earlier. Yes, the Gigabyte BIOS Backup is exactly the cause. The HPA changes (1 MB shrink or more exactly 2113 sectors to take the BIOS copy) and this was done wrongly for 1 TB drives. Sometimes Gigabyte is quite reluctant to write lengthy changelogs which means that things change that are not written explicitely in the changelog.

It seems to happen to the first drive the BIOS sees. That's why it didn't happen when you had the 1 TB drive as secondary.

Thanks for doing the tests!

Share this post


Link to post
Share on other sites
Reallocating events (RAW) =1

Uncorrectable sector count (RAW) =1.

This are the Raw values. The "processed value" is =100=perfect.

The semaphore is yellow, but back then, I thought it was an error, because the "Processed Values"=100=Perfect.

But now I realize that there was 1 sector reallocated.

That is why i suggested HD Sentinel. It does read these values properly and warns appropriately. There are probably freeware tools for that too.

I have scanned this drive now with MHDD, and it shows 30 "greens". Considering that its only 160GB,

its pretty high.

It is an old and used drive. I think that is normal.

Ok, I think I will try to create a Bad Block with MHHD. Can I create a single bad block in the location I choose?

No idea. But you must be able to. For all i know, the standard drive tool that comes with stock linux installs, hdparm, can create bad sectors at the place of your choosing.

Another doubt: will I get a new one, or a refurbished one?

Probably refurbished. I remember getting a new drive from the shop though. But i gave them the drive and told them i am not willing to wait 3 weeks for a replacement. So they just gave me a new one and RMA'ed the one i gave them.

Just imagine if it fails just after the warranty expiration.

I have a friend. Half of her stuff fails within the first week of warranty expiration :)

I was slow, and didn't test the Samsung F1 quickly enough. Next time I will test it in the first day, and it will go back to the shop immediately if there are too many greens.

Anytime a drive is purchased, it must be thoroughly tested prior to committing data on it. Usually one day of stress testing would break a defective drive. I think, in all these years, i had 1 or 2 drives that failed initial break-in. And i had none (touch wood) fail afterwards. I made one or two drives bad by fiddling way too much with them, but they still hold data (on one of them, i must swap the PCB with an identical one, controller board is busted. 500 gb total, 2 physical drives, 1 PCB :)) ). I had way too many drives. So I have hard time understanding how people experience high failure rate with their drives.

Share this post


Link to post
Share on other sites
>>That is why i suggested HD Sentinel. It does read these values properly and warns appropriately. There are probably freeware tools for that too.<<

Yep, most freeware SMART toold show the raw value. But at that time, I didn't pay attention to the RAW values, as I thought they were "scrambled", hehe.

>>It is an old and used drive. I think that is normal.<<

Well, not really. This 160GB drive already had a high green count from the beginning. And the bad sector came from the beginning.

I just didn't have the knowledgement to react appropriately when I bought it.

>>No idea. But you must be able to. For all i know, the standard drive tool that comes with stock linux installs, hdparm, can create bad sectors at the place of your choosing.<<

thats a luxury OS :)

someday I will build a Linux PC.

>>Probably refurbished.<<

maybe the refurbished is worse :)

>>Usually one day of stress testing <<

I use Bartstuff for 1 or 2 hours.But I am not expert in burning in.

>>I had way too many drives. So I have hard time understanding how people experience high failure rate with their drives. <<

I also have very low failure rates. Just one drive came DOA.

Sometimes, there's a drive that looks like it is dying, but then I find that it is just the molex that is loose.

Edited by arga

Share this post


Link to post
Share on other sites
weaker, I don't think it is BIOS backup. I think it is Easy Recovery... Easy Something... That gigabyte uses to clone drives or make images.
@arga: Sorry for not posting earlier. Yes, the Gigabyte BIOS Backup is exactly the cause. The HPA changes (1 MB shrink or more exactly 2113 sectors to take the BIOS copy) and this was done wrongly for 1 TB drives. Sometimes Gigabyte is quite reluctant to write lengthy changelogs which means that things change that are not written explicitely in the changelog.

It seems to happen to the first drive the BIOS sees. That's why it didn't happen when you had the 1 TB drive as secondary.

Thanks for doing the tests!

Another possible correction: I think the Gigabyte mobo *creates* the HPA, which otherwise doesn't exist.

Share this post


Link to post
Share on other sites

AFAICT the Gigabyte BIOS creates a HPA and it is part of the VirtualDualBIOS function. To my knowledge it has nothing to do with XPressRecovery (which you probably meant).

Share this post


Link to post
Share on other sites

guys, after my disk was shrinked to 32MB, I went to Samsung ESTOOL, and out of desperation, I erased the MBR. Is this important? Is there any distinct info in the MBR that can't be restored?

The drive is working ok now, doesn't look like there is any problem with that.

Edited by arga

Share this post


Link to post
Share on other sites
AFAICT the Gigabyte BIOS creates a HPA and it is part of the VirtualDualBIOS function. To my knowledge it has nothing to do with XPressRecovery (which you probably meant).

http://hdat2.com/hdat2_faq.html

Q13: Host Protected Area (HPA) from BIOS.

A13: Some motherboards has incorporated the HPA feature in the BIOS. This feature allows information contained in the first partition of the Hard Drive to be copied to a hidden HPA partition on the same drive where it is immune from attack.

If the HPA is removed from the HD the area of use will be available at the end of the drive as 'Unallocated'.

E.g. Gigabyte's Xpress Recovery.

Share this post


Link to post
Share on other sites
>>No idea. But you must be able to. For all i know, the standard drive tool that comes with stock linux installs, hdparm, can create bad sectors at the place of your choosing.<<

thats a luxury OS :)

There is a windows version of hdparm [umm, i just looked at it and i guess it does not have makebad option under windows]. Hdparm has been around for i think 15-20 years and it is open source... and it gets installed in all linux versions by default. I would not trust my drives with Russian tools (mhdd, hddscan, victoria, etc) but i would be certain that the code of hdparm has been carefully scrutinized by a bazillion of people over time, bugs have been ironed out and it is stable. Mind you, don't get me wrong, it is not that anything wrong with these tools, they all work great, but their source codes have not been peer reviewed. They are more capable than hdparm, but i would choose hdparm if i can do the same thing there.

No idea about MBR. You must be safe i believe.

Share this post


Link to post
Share on other sites

To give you an example, one of the most popular FTP servers for linux... back in days, gave FULL ROOT access to my server (and all the servers it is installed on), when installed to support ANONYMOUS FTP. The author was not available, but someone else on the net fixed the source code and released a patch in 10 mins. You are not bound for a single person to fix the problems with open source software whereas you must wait for the author to take corrective action (if he has time, resources, etc) with proprietary software.

Share this post


Link to post
Share on other sites

ok, as I don't have linux, I will do fine with the russians. I trust them :)

I am using a software called MyTheatre, for satellite TV. This is a russian program too, and it is great. ProgDVB is also russian. Don't underestimate russian programmers, baby. :P

I don't know anything about FTP servers, but I understand what you say about the benefits of open-source code. :)

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now