Re: [PATCH 1/9] Squashfs: move zlib decompression wrapper code into a separate file

[Artem Bityutskiy <dedekind1@gmail.com>]

On Tue, 2009-12-08 at 22:20 +0000, Phillip Lougher wrote:
> Andrew Morton wrote:
> > On Mon, 07 Dec 2009 02:25:08 +0000
> > Phillip Lougher <phillip@lougher.demon.co.uk> wrote:
> > 
> >> +
> >> +int zlib_uncompress(struct squashfs_sb_info *msblk, void **buffer,
> >> +	struct buffer_head **bh, int b, int offset, int length, int srclength,
> >> +	int pages)
> > 
> > This isn't a very good function name.  zlib_uncompress() now becomes a
> > kernel-wide identifier, but it's part of squashfs, not part of zlib.
> > 
> > Maybe that gets fixed in a later patch.  If so, thwap me.
> > 
> 
> Yes, they get fixed up in [PATCH 3/9] Squashfs: add a decompressor framework.
> That patch makes the functions static, and instead exports them via a
> suitably named decompressor ops structure.
> 
> +const struct squashfs_decompressor squashfs_zlib_comp_ops = {
> +	.init = zlib_init,
> +	.free = zlib_free,
> +	.decompress = zlib_uncompress,
> +	.id = ZLIB_COMPRESSION,
> +	.name = "zlib",
> +	.supported = 1
> +};
> 
> I split the patches up to make them easier to review.  The first two patches
> move the zlib code out to a separate file (ready for adding the framework).
> The third patch adds the framework.  At the time of the second patch, however,
> to not break compilation, the functions have to be global.  In hindsight
> I should have made named the functions squashfs_xxx, and removed the squashfs_
> when they were made static in the third patch.

Did you consider using cryptoapi? UBIFS uses zlib/lzo in cryptoapi - it
is a very clean way.

-- 
Best Regards,
Artem Bityutskiy (Артём Битюцкий)