Re: File systems on 4k sector devices?

[Taylor R Campbell <campbell%mumble.net@localhost>]

   Date: Thu, 7 Jun 2012 21:23:00 -0400
   From: Thor Lancelot Simon <tls%panix.com@localhost>

   On Thu, Jun 07, 2012 at 09:14:45PM +0000, Taylor R Campbell wrote:
   > 
   > What atomicity issues would there be in splitting larger blocks into
   > individually encrypted chunks of 512 bytes?

   Hm.  I guess there is a very inefficient RMW cycle if you actually just
   write 512 bytes, but, aside from that, it is OK.

That would happen only if the user of the cgd (e.g., a file system)
actually did I/O in 512-byte blocks, wouldn't it?  Surely it shouldn't
make a difference whether cgd is interposed between the physical disk
and the file system, no?