On 02/05/2018 10:39 PM, Sagar Arun Kamble wrote:
+/**
+ * intel_guc_wopcm_init_hw() - Setup GuC WOPCM registers.
+ * @guc: intel guc.
+ *
+ * Setup the GuC WOPCM size and offset registers with the stored
values. It will
+ * also check the registers locking status to determine whether
these registers
+ * are unlocked and can be updated.
+ */
+void intel_guc_wopcm_init_hw(struct intel_guc *guc)
+{
+ bool locked = guc_wopcm_locked(guc);
+
+ GEM_BUG_ON(!(guc->wopcm.flags & INTEL_GUC_WOPCM_VALID));
+
+ /*
+ * Bug if driver hasn't updated the HW Registers and GuC WOPCM
has been
+ * locked. Return directly if WOPCM was locked and we have updated
+ * the registers.
+ */
+ if (locked) {
+ /*
+ * Mark as updated if registers contained correct values.
+ * This will happen while reloading the driver module without
+ * rebooting the system.
+ */
+ if (guc_wopcm_regs_valid(guc))
+ goto out;
+
+ GEM_BUG_ON(!(guc->wopcm.flags & INTEL_GUC_WOPCM_HW_UPDATED));
We should not go ahead with uc_init_hw in this case so returning error
from here or
checking wopcm.flags to abort uc_init_hw is needed with debug message.
In a second thought, I think we need do more work here to check whether the
locked values are sufficient for current fw sizes. if yes then return
success else
fail the wopcm init.
I agree we only check the status and return the control to uc_init_hw.
Thanks & Regards,
-Jackie
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
