Alexander Viro wrote:
> a) if you make struct inode a part of ext2_inode - WTF bother with pointer?

ok, nobody likes the pointer but me, gone :)

> b) ->destroy_inode() / ->clear_inode().  Merge them - that way it's one
> method.

I believe this belongs in a separate patch after my 1-7, because this
involves breaking some other filesystems.  easier migration -step- if we
add, then shoot me if clear_inode is not destroyed by the end of 2.5.x.


> c) get_empty_inode() must die.  Make it new_inode() and be done with that.
> And have socket.c explicitly set ->i_dev to NODEV afterwards.

done


> d) ext2/balloc.c cleanup probably should be merged before.

hum...  if you are sending one patch for this to linus, I would say
after, but if you are breaking the cleanup into all the steps in
balloc.c,v then I agree with you.

> We will need to set very strict rules on passing around/storing pointers to
> ext2_inode and its ilk, though.  There will be bugs when somebody just decides
> that keeping such pointers might be a good idea and forgets to be nice with
> ->i_count.  Or decrement it manually instead of calling iput(), etc.
> 
> It _MUST_ be explicitly documented - preferably beaten into skulls.

documented in patch6.

Updated patch7 attached (the "meat" of the changes).  For updated
patches 1-7, grab them from kernel.org at

ftp://ftp.kernel.org/pub/linux/kernel/people/jgarzik/patchkits/2.5/2.5.2-pre9-jg1.tar.bz2

-- 
Jeff Garzik      | Alternate titles for LOTR:
Building 1024    | Fast Times at Uruk-Hai
MandrakeSoft     | The Took, the Elf, His Daughter and Her Lover
                 | Samwise Gamgee: International Hobbit of Mystery