Re: [PATCH] allow to create a minix3fs with a blocksize != 1K

[Davidlohr Bueso <dave@gnu.org>]

On Sun, 2012-06-10 at 22:21 +0200, Vladimir 'φ-coder/phcoder' Serbinenko
wrote:
> On 10.06.2012 21:58, Davidlohr Bueso wrote:
> 
> > On Wed, 2012-05-09 at 23:56 +0200, Vladimir 'φ-coder/phcoder' Serbinenko
> > wrote:
> >> Have used it to test minixfs support with various block sizes.
> >>
> > 
> > Hi Vladimir,
> > 
> > A few comments:
> > 
> > +       if (MINIX_BLOCK_SIZE % 512 || MINIX_BLOCK_SIZE < 1024
> > +           || MINIX_BLOCK_SIZE >= 65536)
> > +               usage();
> > 
> > This check should be part of the switch statement, perhaps in its own
> > function.
> > +       if (*tmp && ...)
> > 
> > 
> 
> Ok
> 
> > +       if (fs_version != 3 && MINIX_BLOCK_SIZE != MINIX12_BLOCK_SIZE)
> > +               usage();
> > 
> > We should set -1/-2 options to be mutually exclusive with -B.
> > 
> 
> Ok
> 
> > +#define MAX_MINIX_BLOCK_SIZE 65536
> > Isn't the max minix block size 4096?
> > 
> 
> No, Maximum for the on-disk format is 65535. Minix itself supports
> everything from 1025 to 65536-512 as long as it's divisible by 512.
> Linux driver is limited to 1024, 2048 and 4096

hmmm, do you have any documentation I can look at? Both the native mkfs
minix tool and the mfs server, also set the max block size to 4K.
Unfortunately I don't have a copy of ast's minix book.

> 
> > +#define MINIX12_BLOCK_SIZE   1024
> > This is the minimal block size, including v3. MINIX12_ doesn't describe
> > it well.
> > 
> 
> I use this constant specifically in places which only relate to v1 and v2
> 
> > 
> > Also, please send the patches in the body of the message - just as with
> > the lkml.
> > 
> 
> Ok
> 
> > Thanks,
> > Davidlohr
> > 
> > 
> 
> 
>