On 2012-11-23 08:14, Steve Sakoman wrote:
> On Thu, Nov 22, 2012 at 12:47 AM, Tomi Valkeinen <tomi.valkeinen@xxxxxx> wrote:
>
>>> I'll add the printk's to omapfb_setup_plane() that were requested by
>>> Tomi and report back.
>
> Today was a holiday, so family obligations didn't allow much time to
> look at this.
>
> I did however do a quick build with some printk's in
> omapfb_setup_plane() so I could narrow down where to start looking.
>
> I discovered that the error (omapdss OVERLAY error: check_overlay:
> paddr cannot be 0) is happening as a result of the
> ovl->set_overlay_info() call in the else clause of the code below:
>
> if (pi->enabled) {
> r = omapfb_setup_overlay(fbi, ovl, pi->pos_x, pi->pos_y,
> pi->out_width, pi->out_height);
> if (r) {
> printk("omapfb_setup_plane: pi->enabled &
> omapfb_setup_overlay()\n");
> goto undo;
> }
> } else {
> struct omap_overlay_info info;
>
> ovl->get_overlay_info(ovl, &info);
>
> info.pos_x = pi->pos_x;
> info.pos_y = pi->pos_y;
> info.out_width = pi->out_width;
> info.out_height = pi->out_height;
>
> r = ovl->set_overlay_info(ovl, &info);
> if (r) {
> printk("omapfb_setup_plane: !pi->enabled &
> ovl->set_overlay_info failed\n");
> goto undo;
> }
> }
Ok, so hmm... So the overlay is disabled, and the paddr is zero. Can you
put a printk at dss/apply.c:dss_olv_set_info, and print ovl->id and
info->paddr? And perhaps also to omapfb-main.c:omapfb_setup_overlay, at
the point where it sets info.paddr, as I think that's the only place
where omapfb sets paddr.
I don't know what's going on there, why the paddr is suddenly zero.
Looking at the log from Enric, SETUP_PLANE works fine just fine, and
then, for no reason, next SETUP_PLANE fails, and the log doesn't show
anything much happening between.
However, I think omapdss may have been more permissive in the past,
allowing paddr 0 if the overlay is disabled. I could add that back, but
I'd rather first understand what's happening here.
> I'll look at this more over as I get time over the coming days.
>
>> How can I reproduce this? Is there a downloadable rootfs somewhere that
>> I could try?
>
> The only downloadable rootfs I have is 3.5 based. I'll try to get a
> 3.6 image posted in the next couple of days.
>
>> Could you also copy full kernel boot log to pastebin or such?
>
> OK, will do that with my next build. I'll assume you want dss
> debugging enabled for that.
That's probably not needed. I mostly wanted to know if there's some dss
error/warning seen on the boot. But probably not if the problem happens
in the code above.
Tomi
Attachment:
signature.asc
Description: OpenPGP digital signature
