On Wed, Nov 19, 2014 at 01:04:49PM -0500, Sean Paul wrote: > In finish pageflip, the driver was not selecting the root > window when dispatching events. This exposed a race where > a plane update would change the window selection and cause > tegra_dc_finish_page_flip to check the wrong base address. > > This patch also protects access to the window selection register > as well as the registers affected by it. > > Signed-off-by: Sean Paul <seanpaul@xxxxxxxxxxxx> > --- > drivers/gpu/drm/tegra/dc.c | 20 +++++++++++++++++--- > 1 file changed, 17 insertions(+), 3 deletions(-) I had hoped to be done with the conversion to atomic modesetting by now at which point this patch would become significantly different. But I've applied it to the fixes branch now and rebased the atomic modesetting patches on top since it'll take a while longer for those to be ready. Note also that we should be able to get rid of this spinlock by using the per-window registers directly. As I understand it the window select register is the legacy way of programming windows. There is also a flattened address space available for them. According to the TRM these are located at: 0x0a00: window A 0x0c00: window B 0x0e00: window C Supporting that would require a larger rewrite, but I think it'd be a good idea to do that at some point to avoid the spinlock. Thierry
Attachment:
pgp42kG4GHDZu.pgp
Description: PGP signature
