2005-11-27 21:23:42

by Andries E. Brouwer

[permalink] [raw]
Subject: s_maxbytes on isofs for 2.4

I got a problem report on the handling of large (2.4GB) files
with isofs, where 2.6 was fine and 2.4 failed. Replied

>I suspect that the difference between 2.4 and 2.6 is the assignment
> s->s_maxbytes = 0xffffffff;
>in isofs/inode.c. Could you try to add that after
> s->s_magic = ISOFS_SUPER_MAGIC;
>in the 2.4 source?

and got the confirmation that that solves the problems.
Maybe one should consider adding this in 2.4.
No, I have not audited the source. If in fact there is
a reason why files this size are not handled correctly,
there should probably be an assignment with the largest
value that is handled correctly, together with a comment.

Andries


2005-11-28 19:28:58

by Marcelo Tosatti

[permalink] [raw]
Subject: Re: s_maxbytes on isofs for 2.4

Hi Andries,

On Sun, Nov 27, 2005 at 10:23:40PM +0100, [email protected] wrote:
> I got a problem report on the handling of large (2.4GB) files
> with isofs, where 2.6 was fine and 2.4 failed. Replied
>
> >I suspect that the difference between 2.4 and 2.6 is the assignment
> > s->s_maxbytes = 0xffffffff;
> >in isofs/inode.c. Could you try to add that after
> > s->s_magic = ISOFS_SUPER_MAGIC;
> >in the 2.4 source?
>
> and got the confirmation that that solves the problems.
> Maybe one should consider adding this in 2.4.
> No, I have not audited the source. If in fact there is
> a reason why files this size are not handled correctly,
> there should probably be an assignment with the largest
> value that is handled correctly, together with a comment.

My knowledge is quite limited, but I can't spot any issues with
files upto 4GB. Who was the reporter?

Yes, suppose we could include it if there is interest to confirm safety.

2005-11-28 20:33:23

by Andries E. Brouwer

[permalink] [raw]
Subject: Re: s_maxbytes on isofs for 2.4

On Mon, Nov 28, 2005 at 11:46:43AM -0200, Marcelo Tosatti wrote:

> > I got a problem report on the handling of large (2.4GB) files
> > with isofs, where 2.6 was fine and 2.4 failed. Replied
> >
> > >I suspect that the difference between 2.4 and 2.6 is the assignment
> > > s->s_maxbytes = 0xffffffff;
> > >in isofs/inode.c. Could you try to add that after
> > > s->s_magic = ISOFS_SUPER_MAGIC;
> > >in the 2.4 source?
> >
> > and got the confirmation that that solves the problems.
> > Maybe one should consider adding this in 2.4.
>
> I can't spot any issues with files upto 4GB.
> Who was the reporter?

Giulio Orsero <giulioo \x40 pobox \x2e com>