Patch to mitigate HDMI Suspend/Resume crash/noise during Audio Playback.

We were sending HAD DRM_DPMS_ON event when resuming from sleep state
in gfx_late_resume. Which was causing the HAD HOT_PLUG sequence to
run again causing:
 1. Audio Noise
 2. Random crash.

With this patch, we handle the situation by checking if audio stream
is busy in resume sequence. If yes, we don't send DRM_DPMS_ON in
gfx_late_resume to both Local MIPI and HDMI as we never send
DRM_DPMS_OFF in case HDMI audio is busy becasue of a known JIRA
issue - ANDROID-1553. Otherwise we turn them ON.
After the Jira issue is resolved, we can turn the local MIPI
sequence independent of HDMI Audio state.

Refer JIRA: ANDROID-1553 for more details.

ISSUE: HSD-206943

Signed-off-by: Mohan Kandra <mohan.b.kandra@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

diff --git a/drivers/staging/mrst/drv/psb_powermgmt.c b/drivers/staging/mrst/drv/psb_powermgmt.c
index efd6ae3..48e538c 100644 (file)
--- a/drivers/staging/mrst/drv/psb_powermgmt.c
+++ b/drivers/staging/mrst/drv/psb_powermgmt.c
@@ -390,7 +390,18 @@ static void gfx_late_resume(struct early_suspend *es)
                struct drm_encoder_helper_funcs *ehf = encoder->helper_private;
 
                if (drm_helper_encoder_in_use(encoder) && ehf && ehf->mode_set
-                   && ehf->dpms) {
+                   && ehf->dpms
+#ifndef JIRA_ANDROID-1553
+       /*
+         Local MIPI fails to turn back on from a DPMS off/on cycle if HDMI
+         audio returns busy to disallow system suspend.
+         Once ANDROID-1553 is fixed, the expectation is to turn off MIPI but
+         keep display island on if there is active audio playback over HDMI
+         Refer Jira bug# Android-1553 for more details.
+       */
+                       && !(dev_priv->hdmi_audio_busy)
+#endif
+               ) {
                        struct drm_crtc *crtc = encoder->crtc;
 
                        if (crtc)