[PATCH] read_file: Make it work on tftp servers which do not pass size

[Alexander Aring]

Hi Sascha,

On Thu, Jun 20, 2013 at 08:42:17AM +0200, Sascha Hauer wrote:
> On Wed, Jun 19, 2013 at 11:57:01PM +0200, Alexander Aring wrote:
> > Hi,
> > 
> > On Wed, Jun 19, 2013 at 11:32:39PM +0200, Alexander Aring wrote:
> > > Hi Sascha,
> > > 
> > > only a very small hint.
> > > 
> > > On Wed, Jun 19, 2013 at 10:58:48PM +0200, Sascha Hauer wrote:
> > > > Some tftp servers (for example netkit-tftp) do not pass the filesize.
> > > > Add a workaround for read_file which reads the file into a temporary
> > > > file which then is copied to a buffer.
> > > > 
> > > > Signed-off-by: Sascha Hauer <s.hauer at pengutronix.de>
> > > > ---
> > > >  fs/fs.c      | 18 ++++++++++++++++++
> > > >  fs/tftp.c    |  5 ++++-
> > > >  include/fs.h |  2 ++
> > > >  3 files changed, 24 insertions(+), 1 deletion(-)
> > > > 
> > > > diff --git a/fs/fs.c b/fs/fs.c
> > > > index dc3a6e3..7046f2c 100644
> > > > --- a/fs/fs.c
> > > > +++ b/fs/fs.c
> > > > @@ -38,10 +38,21 @@ void *read_file(const char *filename, size_t *size)
> > > >  	int fd;
> > > 
> > > ...
> > > 
> > > >  
> > > >  	tftp_do_close(priv);
> > > >  
> > > > diff --git a/include/fs.h b/include/fs.h
> > > > index 8ff7300..fa6a8da 100644
> > > > --- a/include/fs.h
> > > > +++ b/include/fs.h
> > > > @@ -147,6 +147,8 @@ int protect(int fd, size_t count, unsigned long offset, int prot);
> > > >  int protect_file(const char *file, int prot);
> > > >  void *memmap(int fd, int flags);
> > > >  
> > > > +#define FILESIZE_MAX	((size_t)-1)
> > > 
> > > The type of st_size in struct stat is loff_t. I check this and
> > > ((size_t)-1) is different than ((loff_t)-1), so I think we need to cast
> > > to loff_t. Maybe it's better to use a define from limits.h, but we don't
> > > have some header file like this.
> > > 
> > 
> > ah, I see it now. The fs layer use size_t instead of loff_t. Sorry for
> > the noise.
> 
> No no, you were right. This indeed has to be a loff_t. Where did you
> find that the fs layer uses size_t? It should do so only for the length
> arguments of read/write and friends.
> 

In function read_file :). The parameter size should be changed to
loff_t instead of size_t, because we make "*size = s.st_size;".



The point is, that a file can be greater than 4GB, but we _can't_
read/write a "block" from a file that is greather than 4GB.

But in this function we do "read(fd, buf, s.st_size) < s.st_size", this
need to be in small (max) 4GB pieces(if necessary). In other words, we
can't use s.st_size here if the file is greater than ((size_t)-1).

I don't think if we need something like this, because we never handle
files which are greater than 4GB.

That's my point of view. This example is only for a 32Bit architecture.

Regards
Alex