| Message ID | 1445685744-21616-1-git-send-email-lokeshvutla@ti.com |
|---|---|
| State | Accepted |
| Delegated to: | Tom Rini |
| Headers | show |
On 24 October 2015 at 05:22, Lokesh Vutla <lokeshvutla@ti.com> wrote: > In case if one of the bank that is passed is of size zero, then u-boot > will be updating memory node with a bank of size zero. There is no need > to update memory node if size is zero, so check for bank size before > updating. > > Reviewed-by: Tom Rini <trini@konsulko.com> > Signed-off-by: Lokesh Vutla <lokeshvutla@ti.com> > --- > Changes since v1: > - Fixed a typo in commit message. > > common/fdt_support.c | 3 +++ > 1 file changed, 3 insertions(+) Acked-by: Simon Glass <sjg@chromium.org>
On Sat, Oct 24, 2015 at 04:52:24PM +0530, Lokesh Vutla wrote: > In case if one of the bank that is passed is of size zero, then u-boot > will be updating memory node with a bank of size zero. There is no need > to update memory node if size is zero, so check for bank size before > updating. > > Reviewed-by: Tom Rini <trini@konsulko.com> > Signed-off-by: Lokesh Vutla <lokeshvutla@ti.com> > Acked-by: Simon Glass <sjg@chromium.org> So, I thought about this more, and looked at device trees, more, and concluded, we're stuck with the interface we have. Today there's still many boards (such as the whole Beaglebone family!) that are setting incorrect sizes in the DTS but being correctly fixed up. We'll need to continue with the other tricks that can be done to pass in more memory in the memory node.
diff --git a/common/fdt_support.c b/common/fdt_support.c index f86365e..0019eef 100644 --- a/common/fdt_support.c +++ b/common/fdt_support.c @@ -401,6 +401,9 @@ static int fdt_pack_reg(const void *fdt, void *buf, u64 *address, u64 *size, char *p = buf; for (i = 0; i < n; i++) { + if (size[i] == 0) + continue; + if (address_cells == 2) *(fdt64_t *)p = cpu_to_fdt64(address[i]); else