>> I have some other logic mostly complete that keeps offset parameter in
>> the vm_area struct at 32 bits, and hopefully a greater chunck of the
>> page cache.
MA> You mean 'vm_offset' field ?
MA> There are 37 files with (sub)string 'vm_offset' in them.
MA> Changeing its type from current loff_t to pgoff_t would
MA> help finding its instances, I guess. (And thus ease locating
MA> places using it for page (non-)aligned things.)
I changed the name from vm_offset to vm_index. Works equally well.
As far as non page alinged things, the only instance I have heard of
to date was with a.out.
There are a lot of implications to changing vm_offset in filemap.c
that I'm just finishing working out.
That and putting tests to see if the file size is bigger than we can
handle in the page cache at approprate places.
I believe I started at the really efficient end and you started at the
functional end.
Makes life fun . . .
MA> That part about "a greater chunk" I don't understand, though.
Poorly worded, (Thinking to fast probably!)
My meaning was changing the meaning of vm_offset you remove potential
64bit calculations from a lot of places.
Eric
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/