Re: EWF - force write

Tech-Archive recommends: Fix windows errors by optimizing your registry

From: KM (konstmor_at_no_spam_yahoo.com)
Date: 11/19/04 

Date: Thu, 18 Nov 2004 21:34:56 -0800

Slobodan,

> Hi Dale,
>
> We are talking here about RAM EWF not DISK overlay, right?
>
> In SP2 You can do live commitanddisable operation. (This will leave EWF disabled from that point on)
>
> Live commit option is not possible to implement on running OS partition. Well technically it is easily implemented but consider
> consequences.
> You have open file system and registry used by OS and other programs. Now you choose to commit this state to disk and some data
> get
> written to disk, some not depending on what your system is actually doing when you commit.

I don't think this would be hard to avoid such problems since read/write disk operations are redirected to one code - EWF.

Theoretically (we have no sources for EWF so we have to guess) you just have a cache (an array) with read and write operations on it
happening at the same time.

In this case, read operations - committing cache data to the disk, write operations - OS continues to redirecting disk block calls
to EWF overlay (EWF is still on).
Assuming you sychronized the operations, you do not disable the filter (caching) until all blocks are written to the disk. You are
just marking (as "committed") block by block writing it to the disk. If system happens to use a block again, you un-marked it as
"non-commited". A few iterations through the entire cache and you are finally getting a zero "non-commited" blocks.

You can even help the filter to commit data with less iterations if you highered its committing thread priority.

Does this make sense?

Konstantin

> Next time when you try to boot you will have the same situation as if power went off on regular OS in best case. In worst case
> data
> on your disk will look like cheese, since files are not committed to disk but rather data clumps that could for that matter start
> from FS and then extend to data.
>
> Live RAM EWF commit would be possible only on non OS partitions if you can unmount safely file system.
> So EWF could be made to detect that FS is closed and then if commit is issued data could be written to disk.
>
> In some cases this could be useful, if you need this then please send feedback to MS.
>
> http://msdn.microsoft.com/embedded/community/community/feedback/feedxp/default.aspx
>
> Regards,
> Slobodan
>
> "Dale Hughes" <dale.hughes@gmail.com> wrote in message news:1100830279.806810.142550@c13g2000cwb.googlegroups.com...
>>
>> Nestor wrote:
>> > Hi,
>> >
>> > I work with san disk compact flash and I would like to use the EWF
>> > component. But I have a small problem, I have understood that writes
>> on
>> > overlay disk occurs when system shut down, and my system is often
>> shut down
>> > directly (without window shutdown call).
>> >
>> > Is it possible to force writes on overlay before shuting down ?
>> >
>> > thanks
>> > Mike
>>
>> Welcome to the world of half *** code solutions, LOL. From the
>> embedded world I am used to solutions that have at least been thought
>> out to some degree. This is just another example of Microsoft failing
>> miserably at trying to enter the embedded world. I had great hopes for
>> XPE, however my patience with working with Microsoft OS's for real
>> world solutions is running out.
>> We are currently being "stung" with the exact same problem. I see no
>> reason why microsoft developers could not have designed a commit
>> immediate mode for ewfmgr. Many people could handle a small pause while
>> changes are committed, if you are willing to wait for the eternity it
>> requires for XP to load and unload, waiting for a commit pause to save
>> your changes is minimal.
>>
>> If you have any solutions to this disaster of an oversight, please let
>> me know.
>>
>> Thanks,
>>
>
>

Quantcast