Hi,

> If possible I would like to understand a little more about some of the
> technical challenges associated with such a port.
> Any thoughts, references, directions etc. you might have regarding PVFS2
> on win32 would be greatly appreciated.

I am not going to ask why you would want to do this :)

There are several components that need porting. The most significant ones
that I can think of are
a) userspace libraries (especially the BMI and Trove Components that are
heavily OS dependant)
b) client-core daemon (since it relies on device nodes etc. but this
should be the easiest thing to fix if (a) is accomplished)
c) the kernel module component (that allows PVFS2 to be mounted as a
regular file system). This is probably going to be the hardest part since
this has to interface with the Windows kernel.

(a) and (b) should build with the Cygwin environments, although I have
never done that myself, but accomplishing (c) would require significant
time and effort I think.
Apart from these, you will have to deal with presence/absence of header
files and myriad build related issues.

If you wish to undertake this effort, perhaps you could also take a look at
the paper below that describes some of the challenges that the Coda file
system designers faced in porting Coda to Windows.
http://www.cs.cmu.edu/~coda/docdir/freenix99.pdf
("Porting the Coda File System to Windows")
thanks,
Murali

_______________________________________________
Pvfs2-developers mailing list
[email protected]
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers

Reply via email to