On Fri, Sep 04, 2015 at 01:20:38PM -0400, Tom Rini wrote: > On Mon, Aug 31, 2015 at 04:46:04PM +0200, Maxime Ripard wrote: > > > To check the alignment of the image blocks to the storage blocks, the > > current code uses a convoluted syntax, while a simple mod also does the > > work. > > > > Signed-off-by: Maxime Ripard <[email protected]> > > --- > > common/aboot.c | 3 +-- > > 1 file changed, 1 insertion(+), 2 deletions(-) > > > > diff --git a/common/aboot.c b/common/aboot.c > > index 65e633acfcb9..c8556d9b23f4 100644 > > --- a/common/aboot.c > > +++ b/common/aboot.c > > @@ -269,8 +269,7 @@ void write_sparse_image(block_dev_desc_t *dev_desc, > > } > > > > /* verify sparse_header->blk_sz is an exact multiple of info->blksz */ > > - if (sparse_header->blk_sz != > > - (sparse_header->blk_sz & ~(info->blksz - 1))) { > > + if (sparse_header->blk_sz % info->blksz) { > > So, sometimes we have convoluted syntax like this to avoid what ends up > as floating point math on 32bit platforms.
Now that you speak of this, we did have some compilers on some platforms that ended up generating floating point related failures on our branch, but I didn't have the time to look into it. However, I don't really know how we can end up with floating point math with a mod, is this some optimisations done by gcc? > Maybe this needs to be one of the helpers in include/linux/math64.h? Probably then, yes. Thanks! Maxime -- Maxime Ripard, Free Electrons Embedded Linux, Kernel and Android engineering http://free-electrons.com
signature.asc
Description: Digital signature
_______________________________________________ U-Boot mailing list [email protected] http://lists.denx.de/mailman/listinfo/u-boot
