CF, EWF, file read errors after commit

  • Thread starter Thread starter Guest
  • Start date Start date
G

Guest

Hi everybody,

I (our customers) have problems with file read errors on CFs after some
write operations. The CFs are protected with EWF. The errors only occur after
EWF commit. The most fatal and frequently error is that XPe does not boot and
stops with the message that C:\Windows\system32\config\system (registry
content) can't be read.
This is the time I have to tell you something about the system configuration.
XPe boots from the EWF protected CF which is primary master on IDE. The CF
is NTFS formatted and not configured as fixed disk (Kingston Elite Pro 2GB).
Anyway the system boots from the CF and everything works fine except the
described problem. The second master on IDE is a hard disk where users can
put their data. It's not very often in use and therefore most of the time
suspended and quite.
Well, I did some research on the problem and have the suspicion that the
registry error occurs primilary after plugging a new USB stick and EWF
committing. There are also cases of file read errors in "C:\Documents and
Settings" after the creation of a link on the users desktop.
I already read some articles which describes similar problems in the
combination with immediate shutdown after the commit.
Are there any ideas out there what could be the cause of these problems?

Thank you
Best regards
Dirk
 
Dirk,

How do you shutdown your image after EWF commit after the changes you mentioned below?
Since you card is removable it is even more important to wait until all the buffers are cached.

Also,
Have you tried testing the system on a fixed CF? That should help since the default disk write cache policy for fixed disks is
different than for removable media.
Have you tried a different (more reliable) CF brand? E.g., SanDisk or Lexar?
 
KM,

your question how I shutdown my image after EWF commit seems to be in fact
the key to the solution.
After a little bit more research of the problem, I find out that the error
does not occure when I restart the PC unlike to a shutdown with power off at
the end of the shutdown.
My theory is now, that the power off may occure to early for the cache write
back. The problem should not exist, when I have the system on a fixed CF
because of the different cache policy for fixed disks - right? This is how I
understand it now.
I will also try a different CF brand, because Kingston dont't want to
provide me with a tool to mark their CFs as fixed disks.

Thanks
Dirk
 
Dirk,

Well, having the card marked as fixed may help a bit. However, it is not the best solution.

You should always wait for some timeout after you do commit on EWF. Basically I'd suggest you to use "-live" switch of EWF to be
sure the commit happens the time you initiated it, not at shutdown. (or use corresponding EWF API)
 
Back
Top