Remove vpr refetch from kernel
Do VPR refetch in below sequence :
- first force idle the GPU
- then do SMC call to update VPR size
- unidle the GPU
- while GPU is resuming from rail gated, it will fetch latest
VPR settings on its own
- Hence there is no need to explicitly call vpr_refetch call
in kernel
Bug
1487804
Change-Id: I310acfa6d812e71fe3038b601763b43e8cf44d23
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: http://git-master/r/421702
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
GVS: Gerrit_Virtual_Submit
static void tegra_update_resize_cfg(phys_addr_t base , size_t size)
{
#ifdef CONFIG_TRUSTED_LITTLE_KERNEL
- /* Config VPR_BOM/_SIZE in MC */
- te_set_vpr_params((void *)(uintptr_t)base, size);
- /* trigger GPU to refetch VPR base and size. */
- nvhost_vpr_info_fetch();
+ int err = 0;
+
+ err = gk20a_do_idle();
+ if (!err) {
+ /* Config VPR_BOM/_SIZE in MC */
+ te_set_vpr_params((void *)(uintptr_t)base, size);
+ gk20a_do_unidle();
+ }
#endif
}