Ext HDD's greater than 137GB under Win ME

[jaclaz]

@ jaclaz:

As I posted before -

the 4th column in 1st line is shown in cylinders not bytes so no "0-127"

I used your numbers with an "m" BUT starting with the 2nd partition there's no way to set "sectors" - its either "c" (cylinders) or "m" (megabytes). Therefore it ONLY shows 0 - 2031 so I cannot put "128". I used 2032 to "3000m" & so on using cylinder numbers.

The problem lies in this.

Obviously the data is shown in cylinders (noone ever talked of bytes).

In a "normal" VM with attached a "normal" hard disk image (8 Gb) in size, partitionlogic behaves correctly, proposing "127 cylinders before" on second iteration.

Some simple CHS math:

(127)x(254+1)x63=2,040,255 Sectors x 512 bytes = 1,044,610,560 bytes i.e. roughly 1 Gb or 1,004 Mbytes, i.e. the 1000m

BUT from your previous screenshot, partition logic detects your hard disk (INcorrectly) as having a geometry of nx16x63 INSTEAD of the "right one" of mx255x63

So, numbers change:

(2031)x(15+1)x63=2,047,248 Sectors x 512 bytes = 1,048,190,976 bytes i.e. roughly 1 Gb or 1,048 Mbytes, i.e. the 1000m

RPM detects the drive as having mx255x63 geometry thus all the values set by partition logic are "crazy" from a CHS point of view.

To have a "robust" (meaning both CHS and LBA correct) partitioning, using this approach we have some limitations.

Since 16x255=255x16 we need to find Cylinder numbers on a 16 head device that "fall" on the same boundary on a 255 head one.

Since 255 and 16 have no "common divider" we have a "minimal step" of 255x16=4080.

In other words:

4080x16x63= 4,112,640 x 512 = 2,105,671,680 i.e. roughly 2 Gb

and:

256x255x63= 4,112,640 x 512 = 2,105,671,680 i.e. roughly 2 Gb

This is the "minimum" partition size you can have and all partition sizes must be a multiple of this.

Following the idea of 4 LBA FAT 32 partitions, let's see which can be the "right" number(s) for each in partition logic.

Let's see:

RPM "allows" for:

121,600x255x63= 1,953,504,000 sectors

the screenshot from partition logic is not too readable and some numbers are "strange", but it looks like reasonable:

1,938,000x16x63= 1,953,504,000 sectors

But we need 63 sectors for MBR+hidden sectors (Track 0) and 63 sectors for the Extended partition.

So you could use:

63<-MBR+hidden sectors

485520<-first primary partition

63<-Extended

485520<-first logical volume inside extended

481440<-first logical volume inside extended

481440<-first logical volume inside extended

3954<-unused

You should get the primary partition and one of the logical volumes:

485,520x16x63=489,404,160 x 512 = 250,574,929,920 bytes

or

30,464x255x63=489,404,160 x 512 = 250,574,929,920 bytes

and the other two logical volumes:

481,440x16x63=485,291,520 x 512 = 248,469,258,240 bytes

or

30,208x255x63=485,291,520 x 512 = 248,469,258,240 bytes

So, now you have to use the "c" instead of the "m" and:

1. Primary FAT32 LBA 485520c
   
2. Logical FAT32 LBA 485520c
   
3. Logical FAT32 LBA 481440c
   
4. Logical FAT32 LBA 481440c
   

from #2 always using the "sectors before supplied by partition logic.

Once done that, in RPM you should have all partitions going to the right boundaries, i.e. for each partition starting head only 0 or 1 and Ending Head only 254.

I hope that the above is clear.

jaclaz

Edited September 22, 2010 by jaclaz

[piikea]

@piikea: would you please give me a screeshot of the above using RPM 2.43, with partition 5 selected, instead of 0 (the MBR). Be sure to use 2.43, since the above is from 2.40, despite your saying 2.43.

I just wish to check out what RPM is not liking in 5 (since it's shown in red).

Oops - there's 2.40 AND 2.43 in the same zip file. It initially shows MBR/EMBR "warning" -

and then w/ partition 5 selected -

side note:

RPM 2.40 showed a couple partitions on my internal Drive #1 in RED too but it has been installed & working fine for years now so it must be pretty "picky / finicky".

I will hold off for the moment trying to re-partition in case these new results change anything. Since this drive "may" someday migrate into a newer PC as the main HDD (#1) - I was going to have a "small" partition 1 for a possible OS to reside on by itself & then evenly sized partitions of roughly 250 to 300GB each.

Since the degree of difficulty has turned out to be so high, at this point I just want it to reliably store data, etc. w/ minimum partition sizes of 250GB (except for MBR or primary as necessary).

As to

I hope that the above is clear.

jaclaz

uh....vaguely. RPM will show it as: 0 - 2031, 2032 - 5000, 5001 - 10000 or whatever numbers you use. Or, 0 - 485520c, 485521 - ?, etc

Edited September 22, 2010 by piikea

[dencorso]

@piikea: Thanks! All versions of RPM are indeed pretty "picky / finicky". Bur whereas 2.40 always shows what it dislikes in red, 2.43 goes a step forward and refuses to "import" what it really hates, and instead gives the warning you saw. And, yes, things may work OK despite of this, only be "non standard" or may not work. It depends on how the actual OS in use is picky enough or not. In any case, I think you should try exactly the partitioning scheme proposed by jaclaz:

So, now you have to use the "c" instead of the "m" and:

1. Primary FAT32 LBA 485520c
   
2. Logical FAT32 LBA 485520c
   
3. Logical FAT32 LBA 481440c
   
4. Logical FAT32 LBA 481440c
   

from #2 always using the "sectors before supplied by partition logic.

Let's see how it turns out. Bear in mind that until you actually start adding data to the HDD, we can repartition it until we're satisfied, but once there's data in it, to change the partitioning scheme will be a much more unsavoury thing to do, since it'll require to save data elsewhere, change the scheme, and then put the data back. So *now* is the moment to experiment. Changing subjects somewhat, notice that, despite the fact that RPM's default view is CHS, it'll show you LBA sectors if you press "F4", and pressing "F4" again brings you back to CHS.

As for what RPM will show, as long as it doesn't give you an "overlap warning" and all partitions created show in white, all is well.

===========================================================================

@piikea: please do ignore what follows, since it's irrelevant for solving your present problem.

@jaclaz:

The bootsector is the 1st sector (= sector 0) but the backup is the 7th sector (= sector 6)...

Moreover, the boot code is split and overflows to the 3rd sector (= sector 2) and its backup is duly at the 9th sector (= sector 8), whith the intervening sectors being the FSInfo (at sector 1) and its backup (at sector 7). All other sectors in the first 32 are usually zeroed out already, so there is no need, nor, however, any harm in zeroing them out.

Since we're at it, I've done some spelunking and found one thing I didn't know, nor have I seen it described elsewhere:

Both my XP SP3 boot partitions, and the new machine I'm working into (An Asus Eee PC 701 boting XP SP3 from a 32 GB class 10 SDHC card) have XP SP3 installed in FAT-32 LBA primary active partitions. And looking into either shows the following surprise:

LBA0 is the boot sector that loads NTLDR and LBA6 is its backup;

LBA1 is the FSInfo sector and LBA7 is its backup;

LBA2 is blank, and so is LBA8!

LBA12 has the second part of the boot code, and has *NO* backup!!!

All other sectors in the first 32 are blank, as expected.

By zeroing out LBA12 the system stops booting.

Running Bootsect /nt52 C: restores to LBA12 exactly the same code as it had previously, before being zeroed out. And this was done using Bootsect v. 6.1.7100.0 (from Win 7 WAIK), so it must be pretty standard code.

Were you aware of this? Well, if I know you, of course you do!

But it was a surprise to me, particularly the fact that LBA12 has no backup...

PS: A quick search led me to this transverse reference on a Tutorial about ReactOS:

Note:

If the active partition uses a FAT32 filesystem the bootcode does not fit into a single sector but occupies two sectors.

Microsoft uses Sector 0 (bootsector) and Sector 12.

FreeLoader uses sector 0 and Sector 14.

So, there should not be any conflicts with existing implementations.

[piikea]

So *now* is the moment to experiment.

Yes I know, thats why I want to make sure its "well prepared" & as standard as it can be. I will attempt re-partitioning w/ those numbers.

BUT, my confusion is: 1st I can put 0c-485520c then the 2nd has to be 485520c-? (since 1st one takes up 485519c). If I go 485520c + 485520c = 971040c it's "invalid ending sector #" ?

Edited September 23, 2010 by piikea

[dencorso]

If the counting starts from zero, it'd be:

0c-485519c

485520c-971039c

etc.

But since an extended partition will be created, it has to be taken into account, too.

Then again, it seems, from your screenshot, that Partiton Logic takes care of it.

[piikea]

If the counting starts from zero, it'd be:

0c-485519c

485520c-971039c

etc.

Unfortunately, I have no idea whats going on.

PL will list under cylinders:

0 - 485519

485520 - 1456558

1456559 - ?

so at this point IF I add 485519 to 1456559 it tells me "invalid ending cylinder"

IF I add 485520 same thing.

IF I add 481440 same thing.

IF I use the numbers on the PL pop up STILL "invalid ending cylinder".

So, I give up.

[jaclaz]

And looking into either shows the following surprise:

LBA0 is the boot sector that loads NTLDR and LBA6 is its backup;

LBA1 is the FSInfo sector and LBA7 is its backup;

LBA2 is blank, and so is LBA8!

LBA12 has the second part of the boot code, and has *NO* backup!!!

All other sectors in the first 32 are blank, as expected.

By zeroing out LBA12 the system stops booting.

You should post this under "News":

http://www.boot-land.net/forums/index.php?showtopic=7739&st=92

http://www.boot-land.net/forums/index.php?showtopic=7739&st=100

http://mirror.href.com/thestarman/asm/mbr/index.html#ntfat32

http://mirror.href.com/thestarman/asm/mbr/index.html#mswin41

So, I give up.

Well, NO.

Giving up is NOT an option.

I know I am grumpy, but if I ask you to use 485520c, why don't you use it?

DO just the first partition and use for it 485520c (and not ANY other number).

Then, post TWO screenshots of the drive in RPM, one like the ones you already posted (CHS) and one showing the LBA (press F4).

Then, try creating the other three partitions ALL as primary (it is likely that the "shift" is caused by the Extended)

Then, post TWO screenshots of the drive in RPM, one like the ones you already posted (CHS) and one showing the LBA (press F4).

jaclaz

Edited September 23, 2010 by jaclaz

[dencorso]

@jaclaz: Well, I'm happy to see I'm just about half again a year late in finding out about LBA12 (and LBA14). Your reference to the RMPrepUSB is quite accurate and dates things from early 2009. Of course, I should be reading Boot-Land more closely than I've been doing... That thread eluded me completely then, and yesterday, when I seached, google pointed me to it, but Boot-Land wasn't loading at that precise moment. Now, the reference to the Starman isn't fair at all: that's exactly the first place I looked, before posting, and he never mentions either LBA12 or LBA14 at all (and the Starman *is* the Boot Bible!).

Or if he does mention them he does it in such a cryptical way that I've not found it, even now, on close rereading. All in all, it still begs the question: why use LBA12 or LBA14 without a backup, when LBA2 and LBA8 sit there unused? Just to allow double-booting with DOS by substituting just the 1st sector of the boot loader? Ain't that kind of perverse?

Thanks a lot, you do rock!

[jaclaz]

Now, the reference to the Starman isn't fair at all: that's exactly the first place I looked, before posting, and he never mentions either LBA12 or LBA14 at all (and the Starman *is* the Boot Bible!).

Yes , it is.

The info may not be seemingly there, but it is .

http://mirror.href.com/thestarman/asm/mbr/MSWIN41.htm

The Boot Record itself:

Microsoft's MSWIN4.1 Boot Record is actually 3 sectors long, and is found at Logical Sectors 0 through 2 for any volume

http://mirror.href.com/thestarman/asm/mbr/ntFAT32BR.htm

7CB5 E94803 JMP 8000 ; Jump to new code loaded from

; 13th sector of Boot Record.

And :

http://mirror.href.com/thestarman/asm/mbr/FAT32brcomp.htm

Ain't that kind of perverse?

Hey, that's MS , you remember.

Compare with the neat cmdcons trick :

http://www.boot-land.net/forums/index.php?showtopic=2362&st=6

but in this case they probably did it to keep compatibility with a number of utilities that only "understood" first sector.

The ReactOS guys were probably "forced" to use yet another unused sector for the same compatibility reasons.

jaclaz

Edited September 23, 2010 by jaclaz

[piikea]

I know I am grumpy, but if I ask you to use 485520c, why don't you use it?

Well dencorso mentioned 0 - 485519 in post #80. Besides it shows up as that when you do use 485520. Perhaps I should be putting a "1" in the "starting cylinder" box? Also didn't add "c" to the numbers - maybe that made some difference.

DO just the first partition and use for it 485520c (and not ANY other number).

I put 485520 & I'm pretty sure it displayed as 0 - 485519.

Then, post TWO screenshots of the drive in RPM, one like the ones you already posted (CHS) & one showing the LBA (press F4).

Then, try creating the other three partitions ALL as primary (it is likely that the "shift" is caused by the Extended)

Used 485520 for the 1st 3 partitions then had to use 1938020 as ending cylinder on the last (4th) partition becuase all other numbers gave me "invalid ending cylinder".

[piikea]

Here are the 2nd two screens since trying to post them in previous post gave me:

Error This upload failed

Because this whole endeavor is trying to drive me insane.

Then, post TWO screenshots of the drive in RPM, one like the ones you already posted (CHS) and one showing the LBA (press F4).

Actually it wont let me upload them here either. Same .jpg's, created w. same program 2 minutes after the other ones, same size. Idk???????????/

[dencorso]

There's a limit to how many attachments you have in the forum. Enter your profile, then settings, then manage your attachments and see it. Then delete some older attachments to make room for the new ones. BTW, for maximum information create the screenshots with the partition of interest highlighted, not the MBR, please.

I'm almost proposing that we do the converse: partition with RPM 2.43, and Partition Logic shall not complain.

And... please, arm yourself with a lot of patience... we're almost there, by now, all main hurdles are already overcome.

[piikea]

There's a limit to how many attachments you have in the forum. Enter your profile, then settings, then manage your attachments and see it. Then delete some older attachments to make room for the new ones. BTW, for maximum information create the screenshots with the partition of interest highlighted, not the MBR, please.

I'm almost proposing that we do the converse: partition with RPM 2.43, and Partition Logic shall not complain.

And... please, arm yourself with a lot of patience... we're almost there, by now, all main hurdles are already overcome.

I thought there may be a limit but didnt think I could've possibly hit it already. Anyway, here are the 2nd two screens (in this case I'm not sure which partition is the main one of interest) -

I'm assuming all the red is still NOT good!

BTW - I appreciate all the efforts, this stuff just makes me a mental case after awhile.

[dencorso]

OK. Let's do something with RPM 2.43:

1) Take a deep breathe, relax, and

2) Highlight partition 2, go to the ending head select it and type-in 255, in place of the current 254, then accept, save select format latter, and see whether it becomes white.

3) Whatever the result, post again the pair of screenshots, with partiton 2 highlited.

[piikea]

OK. Let's do something with RPM 2.43:

1) Take a deep breathe, relax, and

2) Highlight partition 2, go to the ending head select it and type-in 255, in place of the current 254, then accept, save select format latter, and see whether it becomes white.

3) Whatever the result, post again the pair of screenshots, with partiton 2 highlited.

I need ULTRA patience because this drive is cursed (or I am). I changed it to 255, it didnt ask "accept" but I hit F2 to save which I'm pretty sure it did & the line never turned white - before OR after save. It half froze computer so had to close out of it to do anything. BUT, when I reboot & run RPM it takes up the whole screen & therefore I can't get a screen. There is no "minimize" that I can locate. So I delete RPM & re-download it - same thing. So screenshots of RPM are out.

It does still give the "MBR/ENBR....." error message.

PLUS -

It was on DRive 1 & trying to figure out how to minimize it, hit something & it said:

"MBR written reboot without saving"

It wouldn't let me undo, or save or not save so all I could do was exit & re-start. Luckily it did but had a black screen & words "rebooting HD1"....... then on to my regular desktop & boot up. So, I almost screwed up the actual system.

I've also started getting Windows bluescreen errors when throught MY Computer I click to "see" a different drive (or partition). Then have to reboot & usually it works ok....(until it does again eventually). Idk where this came from.

My worry is this ill-fated ext drive causing more problems than its worth.

Edited September 24, 2010 by piikea