@@ -2710,6 +2710,21 @@ static const struct cparams {
{ 0, 800, 231, 23784 },
};
+/**
+ * Lock protecting IPS related data structures
+ * - i915_mch_dev
+ * - dev_priv->max_delay
+ * - dev_priv->min_delay
+ * - dev_priv->fmax
+ * - dev_priv->gpu_busy
+ * - dev_priv->gfx_power
+ */
+static DEFINE_SPINLOCK(mchdev_lock);
+
+/* Global for IPS driver to get at the current i915 device. Protected by
+ * mchdev_lock. */
+static struct drm_i915_private *i915_mch_dev;
+
unsigned long i915_chipset_val(struct drm_i915_private *dev_priv)
{
u64 total_count, diff, ret;
@@ -2717,6 +2732,8 @@ unsigned long i915_chipset_val(struct drm_i915_private *dev_priv)
unsigned long now = jiffies_to_msecs(jiffies), diff1;
int i;
+ assert_spin_locked(&mchdev_lock);
+
diff1 = now - dev_priv->last_time1;
/* Prevent division-by-zero if we are asking too fast.
@@ -2918,15 +2935,14 @@ static u16 pvid_to_extvid(struct drm_i915_private *dev_priv, u8 pxvid)
return v_table[pxvid].vd;
}
-void i915_update_gfx_val(struct drm_i915_private *dev_priv)
+static void __i915_update_gfx_val(struct drm_i915_private *dev_priv)
{
struct timespec now, diff1;
u64 diff;
unsigned long diffms;
u32 count;
- if (dev_priv->info->gen != 5)
- return;
+ assert_spin_locked(&mchdev_lock);
getrawmonotonic(&now);
diff1 = timespec_sub(now, dev_priv->last_time2);
@@ -2954,11 +2970,25 @@ void i915_update_gfx_val(struct drm_i915_private *dev_priv)
dev_priv->gfx_power = diff;
}
+void i915_update_gfx_val(struct drm_i915_private *dev_priv)
+{
+ if (dev_priv->info->gen != 5)
+ return;
+
+ spin_lock(&mchdev_lock);
+
+ __i915_update_gfx_val(dev_priv);
+
+ spin_unlock(&mchdev_lock);
+}
+
unsigned long i915_gfx_val(struct drm_i915_private *dev_priv)
{
unsigned long t, corr, state1, corr2, state2;
u32 pxvid, ext_v;
+ assert_spin_locked(&mchdev_lock);
+
pxvid = I915_READ(PXVFREQ_BASE + (dev_priv->cur_delay * 4));
pxvid = (pxvid >> 24) & 0x7f;
ext_v = pvid_to_extvid(dev_priv, pxvid);
@@ -2984,23 +3014,11 @@ unsigned long i915_gfx_val(struct drm_i915_private *dev_priv)
state2 = (corr2 * state1) / 10000;
state2 /= 100; /* convert to mW */
- i915_update_gfx_val(dev_priv);
+ __i915_update_gfx_val(dev_priv);
return dev_priv->gfx_power + state2;
}
-/* Global for IPS driver to get at the current i915 device */
-static struct drm_i915_private *i915_mch_dev;
-/*
- * Lock protecting IPS related data structures
- * - i915_mch_dev
- * - dev_priv->max_delay
- * - dev_priv->min_delay
- * - dev_priv->fmax
- * - dev_priv->gpu_busy
- */
-static DEFINE_SPINLOCK(mchdev_lock);
-
/**
* i915_read_mch_val - return value for IPS use
*
@@ -3163,6 +3181,8 @@ ips_ping_for_i915_load(void)
void intel_gpu_ips_init(struct drm_i915_private *dev_priv)
{
+ /* We only register the i915 ips part with intel-ips once everything is
+ * set up, to avoid intel-ips sneaking in and reading bogus values. */
spin_lock(&mchdev_lock);
i915_mch_dev = dev_priv;
dev_priv->mchdev_lock = &mchdev_lock;
The update_gfx_val function called from mark_busy wasn't taking the mchdev_lock, as it should have. Also sprinkle a few spinlock asserts over the code to document things better. Things are still rather confusing, especially since a few variables in dev_priv are used by both the gen6+ rps code and the ilk ips code. But protected by totally different locks. Follow-on patches will clean that up. v2: Don't add a deadlock ... hence split up update_gfx_val into a wrapper that grabs the lock and an internal __ variant for callsites within intel_pm.c that already have taken the lock. v3: Mark the internal helper as static, noticed by Ben Widawsky. v4: Damien Lespiau had questions about the safety of the ips setup sequence, explain in a comment why it works. Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> --- drivers/gpu/drm/i915/intel_pm.c | 52 +++++++++++++++++++++++++++------------ 1 file changed, 36 insertions(+), 16 deletions(-)