drm/i915/gt: Ignore forcewake acquisition for posting_reads

We don't care about the result of the read, so it may be garbage, we
only care that the mmio is flushed. As such, we can forgo using an
individual forcewake and lock around any posting-read for an engine.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190703155225.9501-4-chris@chris-wilson.co.uk

drivers/gpu/drm/i915/gt/intel_engine.h

@@ -51,7 +51,7 @@ struct drm_printer;
 #define ENGINE_READ16(...)	__ENGINE_READ_OP(read16, __VA_ARGS__)
 #define ENGINE_READ(...)	__ENGINE_READ_OP(read, __VA_ARGS__)
 #define ENGINE_READ_FW(...)	__ENGINE_READ_OP(read_fw, __VA_ARGS__)
-#define ENGINE_POSTING_READ(...) __ENGINE_READ_OP(posting_read, __VA_ARGS__)
+#define ENGINE_POSTING_READ(...) __ENGINE_READ_OP(posting_read_fw, __VA_ARGS__)
 #define ENGINE_POSTING_READ16(...) __ENGINE_READ_OP(posting_read16, __VA_ARGS__)
 
 #define ENGINE_READ64(engine__, lower_reg__, upper_reg__) \