| Author |
Message |
_Dmitry_
Guest
|
 Posted:
Fri Oct 22, 2004 3:19 pm Post subject:
Windows FAT32: cluster allocation question |
|
|
Hi,
I'd appreciate if a guru could answer following questions:
1. When creating a file on an empty Windows FAT32 volume, can I assume that
the file will be placed from the first available data cluster (that is, right
after the Root Directory)?
2. When populating a file on an empty FAT32 volume, can I assume that
cluster allocation will be linear (that is, written data will be stored in a
consecutive chain of clusters)?
------------------------------------------------------
Background: Our in-house hardware logs data to Compact Flash cards, and our
software extracts logged data afterwards. Before hardware can log data, we
need to initialise our cards by creating 1 file on the card that should take
the whole volume of the card.
OS: Windows XP, Windows 2000
Media: 512Mb, 1Gb, 4Gb Compact Flash cards formatted with FAT32
File creation technique: we use standard C library ( fopen, fwrite )
Encountered problems:
1) On Windows 2000 PCs we discovered that our file is not placed from the
beginning of the data area on the card but at an offset of 64K clusters from
the end of the Root Directory.
2) In some situations cluster allocation was not linear.
Thank you very much in advance.
Best regards,
Dmitry |
|
| Back to top |
|
 |
Dave August
Guest
|
| Posted:
Tue Oct 26, 2004 3:23 am Post subject:
Re: Windows FAT32: cluster allocation question |
|
|
Dmitry,
In the old days what you said was true... but there are LOTS of caviets
here.. "empty" HAD to be freshly formatted, just deleting files is NOT good
enough... Again in the 'old days' Windows always allocated new files from
the unused area and not from where any files had been but were erased (most
likely to help keep things as consectuve as possible and to help file
recovery programs)...
You mat find that your best bet is to create "disk images" of what your
embedded hardware wants to see and use some other method to initilize your
cards... you can eaysily write a program using C that opens the PHYSICAL
device and does sector I/O to it... in effect 'cloning' an image that your
hardware will understand... Go check out the documentation ofn
CreateFile()... FYI I've done this for several clients who had similar
problems.... You can eaysilly write a special fromat program that does this
also....(creates the images it self instead of reading a file)
Dave
"_Dmitry_" <_Dmitry_@discussions.microsoft.com> wrote in message
news:6AC14657-E87E-4C15-AB96-F7C671CAEB6F@microsoft.com...
| Quote: | Hi,
I'd appreciate if a guru could answer following questions:
1. When creating a file on an empty Windows FAT32 volume, can I assume
that
the file will be placed from the first available data cluster (that is,
right
after the Root Directory)?
2. When populating a file on an empty FAT32 volume, can I assume that
cluster allocation will be linear (that is, written data will be stored in
a
consecutive chain of clusters)?
------------------------------------------------------
Background: Our in-house hardware logs data to Compact Flash cards, and
our
software extracts logged data afterwards. Before hardware can log data, we
need to initialise our cards by creating 1 file on the card that should
take
the whole volume of the card.
OS: Windows XP, Windows 2000
Media: 512Mb, 1Gb, 4Gb Compact Flash cards formatted with FAT32
File creation technique: we use standard C library ( fopen, fwrite )
Encountered problems:
1) On Windows 2000 PCs we discovered that our file is not placed from the
beginning of the data area on the card but at an offset of 64K clusters
from
the end of the Root Directory.
2) In some situations cluster allocation was not linear.
Thank you very much in advance.
Best regards,
Dmitry
|
|
|
| Back to top |
|
|
Alex Nichol
Guest
|
| Posted:
Tue Oct 26, 2004 1:10 pm Post subject:
Re: Windows FAT32: cluster allocation question |
|
|
_Dmitry_ wrote:
| Quote: | Encountered problems:
1) On Windows 2000 PCs we discovered that our file is not placed from the
beginning of the data area on the card but at an offset of 64K clusters from
the end of the Root Directory.
|
Without being certain, that I think is quite likely policy. Possibly to
ensure there is space available to expand the root directory if needed.
Also space is needed in the root of the disk to hold the core data when
doing a dump; that again may need to be at this point so it can be
accessed without needing to explore.
| Quote: | 2) In some situations cluster allocation was not linear.
|
On a fresh formatted disk I would not expect that, provided a single
file was made in a manner that did not involve putting other files on
the disk as well (indexes, restore point directories in XP and the
like)
--
Alex Nichol MS MVP (Windows Technologies)
Bournemouth, U.K. Alexn@mvps.D8E8L.org (remove the D8 bit) |
|
| Back to top |
|
|
_Dmitry_
Guest
|
| Posted:
Fri Nov 12, 2004 8:39 pm Post subject:
Re: Windows FAT32: cluster allocation question |
|
|
Dave,
Thank you very much for your answers.
Direct IO is the way we are going to go, and it's great to hear you've done
it successfully on several occasions. It means it's not going to be too many
pitfalls :)
Again, thanks a million!
Dmitry
"Dave August" wrote:
| Quote: | Dmitry,
In the old days what you said was true... but there are LOTS of caviets
here.. "empty" HAD to be freshly formatted, just deleting files is NOT good
enough... Again in the 'old days' Windows always allocated new files from
the unused area and not from where any files had been but were erased (most
likely to help keep things as consectuve as possible and to help file
recovery programs)...
You mat find that your best bet is to create "disk images" of what your
embedded hardware wants to see and use some other method to initilize your
cards... you can eaysily write a program using C that opens the PHYSICAL
device and does sector I/O to it... in effect 'cloning' an image that your
hardware will understand... Go check out the documentation ofn
CreateFile()... FYI I've done this for several clients who had similar
problems.... You can eaysilly write a special fromat program that does this
also....(creates the images it self instead of reading a file)
Dave
"_Dmitry_" <_Dmitry_@discussions.microsoft.com> wrote in message
news:6AC14657-E87E-4C15-AB96-F7C671CAEB6F@microsoft.com...
Hi,
I'd appreciate if a guru could answer following questions:
1. When creating a file on an empty Windows FAT32 volume, can I assume
that
the file will be placed from the first available data cluster (that is,
right
after the Root Directory)?
2. When populating a file on an empty FAT32 volume, can I assume that
cluster allocation will be linear (that is, written data will be stored in
a
consecutive chain of clusters)?
------------------------------------------------------
Background: Our in-house hardware logs data to Compact Flash cards, and
our
software extracts logged data afterwards. Before hardware can log data, we
need to initialise our cards by creating 1 file on the card that should
take
the whole volume of the card.
OS: Windows XP, Windows 2000
Media: 512Mb, 1Gb, 4Gb Compact Flash cards formatted with FAT32
File creation technique: we use standard C library ( fopen, fwrite )
Encountered problems:
1) On Windows 2000 PCs we discovered that our file is not placed from the
beginning of the data area on the card but at an offset of 64K clusters
from
the end of the Root Directory.
2) In some situations cluster allocation was not linear.
Thank you very much in advance.
Best regards,
Dmitry
|
|
|
| Back to top |
|
|
|
|
|
|
FAQ
Memberlist
Register
Profile
Log in to check your private messages
Log in