中国DOS联盟

1. For two partitions with overlapping areas, will data loss necessarily occur? When the data of the previous partition has not yet involved the overlapping area, the data will still be fine. But the problems that haven't occurred currently don't mean there are no errors, nor do they mean they won't occur in the future. When the data of the previous partition is written to the overlapping area, basically the entire subsequent partition is completely lost and cannot be recovered (because the overwritten part happens to be the most important part).
2. The basic unit for the system to access a logical disk is the cluster. Currently, each cluster of a hard disk partition is usually 4 - 32K. Then, how can a partition that is sector-based ensure that this partition just has an integer number of clusters (of course, the possibility of just an integer number of clusters also exists)? If it's not an integer number of clusters, then the range of the last cluster will cover the starting part of the next partition, that is, an implicit partition overlap occurs!!! If the subsequent area happens to be the extended partition table, and when that cluster is written with more data, a disaster occurs, and all logical partitions on the hard disk except the primary partition are lost (this kind of example is not rare at all).
3. When the partition is in the space behind 8G, the software will fill in FE, 00, 01 (not very sure if it's this value) in the start CHS and FE, FF, FF in the end CHS. If there are multiple partition entries after 8G, then there will be two or more entries with exactly the same CHS (from the early perspective, it's very similar to a logical lock, and it's very likely to get locked up with DOS 6.x or earlier versions). Then, when PQ encounters this situation, it will act smartly and think the partition entry is wrong, and start to modify it randomly. Then, the modified partition is almost all reduced to only 8G in size (this kind of example is more common).
***As long as you pay attention to the posts in various forums, you will find that almost every month the above two failures occur, and their common point is using PQ to partition.***
***By the way: The DOS 7.1 that can avoid the logical lock also has a bug. That is, if a FAT/FAT32 primary partition is in the space behind 8G (this situation often occurs in one-click restoration) and is used to boot the system, the boot will fail. Using the DOS of 98 has no this problem. This problem is not encountered by many people for a long time, but it does exist and is very serious.***

Last edited by netwinxp on 2008-12-22 at 11:48 ]

Haven't used PartitionMagic (PQ) for many years. Now I can't tell which one it is. I used to like using them just to adjust partition sizes. Often, the partition table would get messed up for no reason, then it would report errors by itself, and then I had to use DiskGenius's partition repair function.

Basically, I've been using DiskGenius and GPT fdisk (gdisk) all the time. Now I don't have the hobby of adjusting partitions like before.

I seem to have encountered the situation of cross partitions too. When I installed Linux before, I changed a Fireball LCT20 20G to 40G, which made me happy. But then I found that data written in one partition would overwrite another partition. Then with DiskGenius, the partition table was wrong. Oh no

Your explanation is basically clear to me. Now my question is: Does this problem exist in all PQ partitioning software, or is it only an issue with early versions of PQ software? The partitioning software I'm using right now is actually just the PQ series, but the actual name is Volume Magic, which is equivalent to the server version of PQ.

If this problem exists in all versions of PQ software, then I would be really scared, because basically all the machines I've installed in the past few years have been partitioned with it.

Now the question is, how can I judge whether my current partition table has a problem. Can diskgen be used to judge?

What I care about most now is what partition is good? Please advise

Originally posted by Climbing at 2008-12-20 09:51:
I basically understand your explanation. Now my question is: Does this problem exist in all PQ partitioning software, or is it only the early versions of PQ software that have this problem? The partitioning software I am using actually only...

I have only analyzed the most popular PQ8 and an earlier version online.
The way to identify whether other versions have this risk is to check if the end of all partitions is the last sector of the track (usually 63). ***Now most use LBA addresses, so conversion may be needed** If yes, then even if it is partitioned with PQ, there should be no problem. If not, there may be a risk.
Personally, since it is an algorithm issue, it is very likely that a series of products have this risk. :o

According to netwinxp's statement,
Take a look with the ptedit that comes with pq. Pay attention to this 254 and 63.
Also, for partitions after 8GB, the start cylinder and end cylinder are both 1023.

If the CHS parameters of the partition are not like this, there may be a problem.
Generally, it can be used normally under Windows, but in grub, there may be warnings or the partition cannot be accessed, which is the so-called "bad partition".

As can be seen from the above, the "well - partitioned" partition size should be a multiple of 7.844MB, which is (254 + 1) * 63 * 512 = 8225280B = 7.844MB.

And Windows Disk Management allows creating partitions that are not multiples of the above value, and it often results in "bad partitions" or gaps between two partitions (in order to make the partition start from cylinder 0, head 1, sector 1).

Well, I found two machines that I previously partitioned with my Volume Magic. I ran ptedit32 to check, and indeed there are no problems as described by netwinxp. Everything is quite standard and compliant. A big stone in my heart finally landed, and I can rest assured to continue using it for partitioning in the future.

en, it may also be a problem introduced by the author of the Chinese localization and CRACK. The two I tried are both relatively popular traditional Chinese versions.

I basically don't use localization software, and there's no need to localize such software.

I also checked, everything is normal.
Using the English version 8.05 server edition.

If it's the problem of the Chinese version, then it can't be blamed on pq, heh heh.

I usually first choose Microsoft's Disk Manager for partitioning, and secondly use PQ8.05 Server Edition

I have encountered the problem mentioned by 'netwinxp'. When using other partition software to partition, PQ mistakenly reported an error (actually there was no error, and it had been used normally all the time), and it cleverly fixed it on its own, resulting in all disk partitions being messed up and data loss. That loss was terrible!!!

Climbing said that you have used PQ without problems for so many years and there were no errors. I think you were talking about a new machine, then there would naturally be no errors. It partitioned the area by itself, so it naturally would not report an error and would also be used normally. The problem is that PQ sometimes thinks that partitioning by other partition software is wrong and repairs it on its own, with serious consequences.

The friend upstairs said that PQ partitioning would make grub think it was a non-standard partition. I also encountered it, but I didn't pay attention to what partition software was used to partition the area. However, grub would not cleverly repair the "wrong" partition on its own, so there would be no loss.

Originally posted by xsrdg at 2008-12-24 05:27 PM:
Some friends upstairs said that PQ partitioning will make GRUB think it's a non-standard partition. I also encountered this, but I didn't pay attention to which partitioning software was used to create the partitions. However, GRUB won't smartly repair "incorrect" partitions, so there won't be any loss.

I'm not saying that partitions created by PQ will make GRUB think they are non-standard partitions.

Instead, for hard drives that PQ thinks have partition errors, GRUB often thinks they are non-standard partitions.

As far as the partitioning software I have used so far is concerned, PQ is the strictest in checking the partition table (it must be as shown in my above ptedit screenshot), and as long as there is the slightest deviation from the partition parameters, it will go to "fix" it, but the repair function is very bad, and it makes mistakes as soon as it fixes.
GRUB is also relatively strict in checking the partition table, so sometimes there will be problems of "bad partitions" that cannot be accessed. But if the hard drive that PQ thinks has a normal partition table, GRUB will definitely be able to access it normally.

On the contrary, the disk manager in Windows has the loosest requirements for partition table parameters, almost not caring about the CHS values of the partition table, and only looking at the LBA values. And I once tried that even if two partitions were overlapped, or four partition entries were written in the extended partition table (generally, there should be no more than two in the extended partition table), the disk manager could recognize them (usually, both DOS couldn't boot when these two errors occurred).

I agree with the view of brother fujianabc above.

I use PQ for partitioning. Of course, it's not all for new machines. When I completely repartition, for many times when I install the system for others, there are already a lot of data (partitioned) on the other party's hard drive, and I need to use the lossless partitioning function of PQ to reorganize the unreasonable partition scheme. In addition to using PQ for partitioning, I often use Acronis. Sometimes, when PQ starts, it prompts a partition table error (Error 106, etc.) and then cannot see the partitions of the original hard drive, but Acronis can normally recognize the original partitions. At this time, I use Acronis to partition the hard drive, but after processing with Acronis, PQ can recognize normally again. Of course, I hardly ever use Windows Disk Management for partitioning. Even if I do, at most I delete an original partition and then rebuild it.