What type of file system is the image? What is the block / cluster
size? Is it for all files or just some of them? Can you run 'istat'
on one of the files that has a different output and send me the output?
brian
On Nov 11, 2004, at 5:38 AM, neutrino neutrino wrote:
> I have been doing tests that was meant to show that the tool "sorter"
> used on a common image will produce the same output independant of
> Operation System. However, when I ran "sorter" on a FreeBSD-5.2.1
> systems it produce files that were different to other systems (solaris
> sparc/ redhat i386).
>
> Comparing output files from the redhat system to what was produced on
> the FreeBSD system in a hex editor, I found that the FreeBSD system
> always differs at the offset 0x3400. I ran "ktrace" (Kernel Trace) on
> the "sorter" command that I was using in FreeBSD. What I found was
> when "icat" was reading the target file, it got to a point where it
> switched from doing a direct read, to using lseek before a read
> statement.
>
> Further investigation using a hex editor I found that the problem
> seems to be a miscalcuation on the lseek. It seems to skip 1024 bytes
> at offset 0x3400.
>
> I am not a programmer, and I have dont have the time to investigate to
> much further, but I thought I would report it. Having a quick look at
> the source, I was guessing the problem may be in the OFF_T macro. But
> I could easily be wrong.
|
