| Message ID | 1251908825-3477-1-git-send-email-zhenyuw@linux.intel.com (mailing list archive) |
|---|---|
| State | Accepted |
| Headers | show |
On Thu, 2009-09-03 at 00:27 +0800, Zhenyu Wang wrote: > Don't do dynamic clock frequence change if not in kernel > modesetting. But this one keeps render clock control which > should be fine. > > Btw, I don't have machine with dual frequence feature to > really test this now, but this fixed kernel oops when kernel > modesetting is disabled. > > Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com> > --- > drivers/gpu/drm/i915/intel_display.c | 6 ++++++ > 1 files changed, 6 insertions(+), 0 deletions(-) > > diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c > index 6a58e1d..81ba1a2 100644 > --- a/drivers/gpu/drm/i915/intel_display.c > +++ b/drivers/gpu/drm/i915/intel_display.c > @@ -3418,6 +3418,9 @@ static void intel_idle_update(struct work_struct *work) > intel_decrease_displayclock(dev); > } > > + if (!drm_core_check_feature(dev, DRIVER_MODESET)) > + return; > + Is this required? This function is never called if the KMS is disabled. > list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { > /* Skip inactive CRTCs */ > if (!crtc->fb) > @@ -3451,6 +3454,9 @@ void intel_mark_busy(struct drm_device *dev, struct drm_gem_object *obj) > dev_priv->busy = true; > intel_increase_renderclock(dev, true); > > + if (!drm_core_check_feature(dev, DRIVER_MODESET)) > + return; > + Should this be moved before increasing the render clock? In the UMS mode we had better not touch the render clock. > list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { > if (!crtc->fb) > continue;
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c index 6a58e1d..81ba1a2 100644 --- a/drivers/gpu/drm/i915/intel_display.c +++ b/drivers/gpu/drm/i915/intel_display.c @@ -3418,6 +3418,9 @@ static void intel_idle_update(struct work_struct *work) intel_decrease_displayclock(dev); } + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return; + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { /* Skip inactive CRTCs */ if (!crtc->fb) @@ -3451,6 +3454,9 @@ void intel_mark_busy(struct drm_device *dev, struct drm_gem_object *obj) dev_priv->busy = true; intel_increase_renderclock(dev, true); + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return; + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { if (!crtc->fb) continue;
Don't do dynamic clock frequence change if not in kernel modesetting. But this one keeps render clock control which should be fine. Btw, I don't have machine with dual frequence feature to really test this now, but this fixed kernel oops when kernel modesetting is disabled. Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com> --- drivers/gpu/drm/i915/intel_display.c | 6 ++++++ 1 files changed, 6 insertions(+), 0 deletions(-)