On Tue, Oct 09, 2012 at 02:08:13PM -0500, Scott Wood wrote:
On 10/09/2012 01:51:14 PM, Tom Rini wrote:
On Tue, Oct 09, 2012 at 01:09:22PM -0500, Scott Wood wrote:
On 10/06/2012 11:59:17 AM, Albert ARIBAUD wrote:
Hi Anatolij,
On Sat, 6 Oct 2012 11:31:03 +0200, Anatolij Gustschin
wrote:
Fix: yaffs_guts.c: In function 'yaffs_check_chunk_erased': yaffs_guts.c:324:6: warning: variable 'result' set but not used [-Wunused-but-set-variable]
[snip]
Tested-by: Albert ARIBAUD albert.u.boot@aribaud.net
This effectively makes VCMA9 and smdk2410 (the two boards which had the yaffs warnings) build clean.
It takes care of the warnings, but I still see this:
Configuring for VCMA9 board... make: *** [u-boot] Error 139
The linaro and ELDK 4.2, 5.2 toolchains build this fine. I would suggest grabbing one of those for ARM.
I'll give one of them a try, but is it expected that the compiler will be able to do 64/32 division without __udivdi3? Shouldn't yaffs2 be using lldiv() or do_div()?
Based on a very short skim, yes, yaffs2 looks to be buggy here and for example "end_block = mtd->size / mtd->erasesize - 1;" from yaffs_uboot_glue.c should be using mtd->erase_shift and shifting.