7 files changed, 107 insertions, 45 deletions

diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.c b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
index a60584763b61..a0b9d8a95b16 100644
--- a/drivers/gpu/drm/msm/adreno/adreno_gpu.c
+++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
@@ -124,6 +124,8 @@ void adreno_recover(struct msm_gpu *gpu)
 
         /* reset completed fence seqno, just discard anything pending: */
         adreno_gpu->memptrs->fence = gpu->submitted_fence;
+        adreno_gpu->memptrs->rptr  = 0;
+        adreno_gpu->memptrs->wptr  = 0;
 
         gpu->funcs->pm_resume(gpu);
         ret = gpu->funcs->hw_init(gpu);
@@ -229,7 +231,7 @@ void adreno_idle(struct msm_gpu *gpu)
                         return;
         } while(time_before(jiffies, t));
 
-        DRM_ERROR("timeout waiting for %s to drain ringbuffer!\n", gpu->name);
+        DRM_ERROR("%s: timeout waiting to drain ringbuffer!\n", gpu->name);
 
         /* TODO maybe we need to reset GPU here to recover from hang? */
 }
@@ -256,11 +258,17 @@ void adreno_wait_ring(struct msm_gpu *gpu, uint32_t ndwords)
 {
         struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu);
         uint32_t freedwords;
+        unsigned long t = jiffies + ADRENO_IDLE_TIMEOUT;
         do {
                 uint32_t size = gpu->rb->size / 4;
                 uint32_t wptr = get_wptr(gpu->rb);
                 uint32_t rptr = adreno_gpu->memptrs->rptr;
                 freedwords = (rptr + (size - 1) - wptr) % size;
+
+                if (time_after(jiffies, t)) {
+                        DRM_ERROR("%s: timeout waiting for ringbuffer space\n", gpu->name);
+                        break;
+                }
         } while(freedwords < ndwords);
 }
 
diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c
index 864c9773636b..008d772384c7 100644
--- a/drivers/gpu/drm/msm/msm_drv.c
+++ b/drivers/gpu/drm/msm/msm_drv.c
@@ -499,25 +499,41 @@ int msm_wait_fence_interruptable(struct drm_device *dev, uint32_t fence,
                 struct timespec *timeout)
 {
         struct msm_drm_private *priv = dev->dev_private;
-        unsigned long timeout_jiffies = timespec_to_jiffies(timeout);
-        unsigned long start_jiffies = jiffies;
-        unsigned long remaining_jiffies;
         int ret;
 
-        if (time_after(start_jiffies, timeout_jiffies))
-                remaining_jiffies = 0;
-        else
-                remaining_jiffies = timeout_jiffies - start_jiffies;
-
-        ret = wait_event_interruptible_timeout(priv->fence_event,
-                        priv->completed_fence >= fence,
-                        remaining_jiffies);
-        if (ret == 0) {
-                DBG("timeout waiting for fence: %u (completed: %u)",
-                                fence, priv->completed_fence);
-                ret = -ETIMEDOUT;
-        } else if (ret != -ERESTARTSYS) {
-                ret = 0;
+        if (!priv->gpu)
+                return 0;
+
+        if (fence > priv->gpu->submitted_fence) {
+                DRM_ERROR("waiting on invalid fence: %u (of %u)\n",
+                                fence, priv->gpu->submitted_fence);
+                return -EINVAL;
+        }
+
+        if (!timeout) {
+                /* no-wait: */
+                ret = fence_completed(dev, fence) ? 0 : -EBUSY;
+        } else {
+                unsigned long timeout_jiffies = timespec_to_jiffies(timeout);
+                unsigned long start_jiffies = jiffies;
+                unsigned long remaining_jiffies;
+
+                if (time_after(start_jiffies, timeout_jiffies))
+                        remaining_jiffies = 0;
+                else
+                        remaining_jiffies = timeout_jiffies - start_jiffies;
+
+                ret = wait_event_interruptible_timeout(priv->fence_event,
+                                fence_completed(dev, fence),
+                                remaining_jiffies);
+
+                if (ret == 0) {
+                        DBG("timeout waiting for fence: %u (completed: %u)",
+                                        fence, priv->completed_fence);
+                        ret = -ETIMEDOUT;
+                } else if (ret != -ERESTARTSYS) {
+                        ret = 0;
+                }
         }
 
         return ret;
diff --git a/drivers/gpu/drm/msm/msm_drv.h b/drivers/gpu/drm/msm/msm_drv.h
index 80d75094bf0a..df8f1d084bc1 100644
--- a/drivers/gpu/drm/msm/msm_drv.h
+++ b/drivers/gpu/drm/msm/msm_drv.h
@@ -153,7 +153,7 @@ void *msm_gem_vaddr(struct drm_gem_object *obj);
 int msm_gem_queue_inactive_work(struct drm_gem_object *obj,
                 struct work_struct *work);
 void msm_gem_move_to_active(struct drm_gem_object *obj,
-                struct msm_gpu *gpu, uint32_t fence);
+                struct msm_gpu *gpu, bool write, uint32_t fence);
 void msm_gem_move_to_inactive(struct drm_gem_object *obj);
 int msm_gem_cpu_prep(struct drm_gem_object *obj, uint32_t op,
                 struct timespec *timeout);
@@ -191,6 +191,12 @@ u32 msm_readl(const void __iomem *addr);
 #define DBG(fmt, ...) DRM_DEBUG(fmt"\n", ##__VA_ARGS__)
 #define VERB(fmt, ...) if (0) DRM_DEBUG(fmt"\n", ##__VA_ARGS__)
 
+static inline bool fence_completed(struct drm_device *dev, uint32_t fence)
+{
+        struct msm_drm_private *priv = dev->dev_private;
+        return priv->completed_fence >= fence;
+}
+
 static inline int align_pitch(int width, int bpp)
 {
         int bytespp = (bpp + 7) / 8;
diff --git a/drivers/gpu/drm/msm/msm_gem.c b/drivers/gpu/drm/msm/msm_gem.c
index 6b5a6c8c7658..29eacfa29cfb 100644
--- a/drivers/gpu/drm/msm/msm_gem.c
+++ b/drivers/gpu/drm/msm/msm_gem.c
@@ -40,9 +40,9 @@ static struct page **get_pages(struct drm_gem_object *obj)
                 }
 
                 msm_obj->sgt = drm_prime_pages_to_sg(p, npages);
-                if (!msm_obj->sgt) {
+                if (IS_ERR(msm_obj->sgt)) {
                         dev_err(dev->dev, "failed to allocate sgt\n");
-                        return ERR_PTR(-ENOMEM);
+                        return ERR_CAST(msm_obj->sgt);
                 }
 
                 msm_obj->pages = p;
@@ -159,7 +159,6 @@ out_unlock:
 out:
         switch (ret) {
         case -EAGAIN:
-                set_need_resched();
         case 0:
         case -ERESTARTSYS:
         case -EINTR:
@@ -393,11 +392,14 @@ int msm_gem_queue_inactive_work(struct drm_gem_object *obj,
 }
 
 void msm_gem_move_to_active(struct drm_gem_object *obj,
-                struct msm_gpu *gpu, uint32_t fence)
+                struct msm_gpu *gpu, bool write, uint32_t fence)
 {
         struct msm_gem_object *msm_obj = to_msm_bo(obj);
         msm_obj->gpu = gpu;
-        msm_obj->fence = fence;
+        if (write)
+                msm_obj->write_fence = fence;
+        else
+                msm_obj->read_fence = fence;
         list_del_init(&msm_obj->mm_list);
         list_add_tail(&msm_obj->mm_list, &gpu->active_list);
 }
@@ -411,7 +413,8 @@ void msm_gem_move_to_inactive(struct drm_gem_object *obj)
         WARN_ON(!mutex_is_locked(&dev->struct_mutex));
 
         msm_obj->gpu = NULL;
-        msm_obj->fence = 0;
+        msm_obj->read_fence = 0;
+        msm_obj->write_fence = 0;
         list_del_init(&msm_obj->mm_list);
         list_add_tail(&msm_obj->mm_list, &priv->inactive_list);
 
@@ -433,8 +436,18 @@ int msm_gem_cpu_prep(struct drm_gem_object *obj, uint32_t op,
         struct msm_gem_object *msm_obj = to_msm_bo(obj);
         int ret = 0;
 
-        if (is_active(msm_obj) && !(op & MSM_PREP_NOSYNC))
-                ret = msm_wait_fence_interruptable(dev, msm_obj->fence, timeout);
+        if (is_active(msm_obj)) {
+                uint32_t fence = 0;
+
+                if (op & MSM_PREP_READ)
+                        fence = msm_obj->write_fence;
+                if (op & MSM_PREP_WRITE)
+                        fence = max(fence, msm_obj->read_fence);
+                if (op & MSM_PREP_NOSYNC)
+                        timeout = NULL;
+
+                ret = msm_wait_fence_interruptable(dev, fence, timeout);
+        }
 
         /* TODO cache maintenance */
 
@@ -455,9 +468,10 @@ void msm_gem_describe(struct drm_gem_object *obj, struct seq_file *m)
         uint64_t off = drm_vma_node_start(&obj->vma_node);
 
         WARN_ON(!mutex_is_locked(&dev->struct_mutex));
-        seq_printf(m, "%08x: %c(%d) %2d (%2d) %08llx %p %d\n",
+        seq_printf(m, "%08x: %c(r=%u,w=%u) %2d (%2d) %08llx %p %d\n",
                         msm_obj->flags, is_active(msm_obj) ? 'A' : 'I',
-                        msm_obj->fence, obj->name, obj->refcount.refcount.counter,
+                        msm_obj->read_fence, msm_obj->write_fence,
+                        obj->name, obj->refcount.refcount.counter,
                         off, msm_obj->vaddr, obj->size);
 }
 
diff --git a/drivers/gpu/drm/msm/msm_gem.h b/drivers/gpu/drm/msm/msm_gem.h
index d746f13d283c..0676f32e2c6a 100644
--- a/drivers/gpu/drm/msm/msm_gem.h
+++ b/drivers/gpu/drm/msm/msm_gem.h
@@ -36,7 +36,7 @@ struct msm_gem_object {
          */
         struct list_head mm_list;
         struct msm_gpu *gpu;     /* non-null if active */
-        uint32_t fence;
+        uint32_t read_fence, write_fence;
 
         /* Transiently in the process of submit ioctl, objects associated
          * with the submit are on submit->bo_list.. this only lasts for
diff --git a/drivers/gpu/drm/msm/msm_gem_submit.c b/drivers/gpu/drm/msm/msm_gem_submit.c
index 3e1ef3a00f60..5281d4bc37f7 100644
--- a/drivers/gpu/drm/msm/msm_gem_submit.c
+++ b/drivers/gpu/drm/msm/msm_gem_submit.c
@@ -78,7 +78,7 @@ static int submit_lookup_objects(struct msm_gem_submit *submit,
                 }
 
                 if (submit_bo.flags & BO_INVALID_FLAGS) {
-                        DBG("invalid flags: %x", submit_bo.flags);
+                        DRM_ERROR("invalid flags: %x\n", submit_bo.flags);
                         ret = -EINVAL;
                         goto out_unlock;
                 }
@@ -92,7 +92,7 @@ static int submit_lookup_objects(struct msm_gem_submit *submit,
                  */
                 obj = idr_find(&file->object_idr, submit_bo.handle);
                 if (!obj) {
-                        DBG("invalid handle %u at index %u", submit_bo.handle, i);
+                        DRM_ERROR("invalid handle %u at index %u\n", submit_bo.handle, i);
                         ret = -EINVAL;
                         goto out_unlock;
                 }
@@ -100,7 +100,7 @@ static int submit_lookup_objects(struct msm_gem_submit *submit,
                 msm_obj = to_msm_bo(obj);
 
                 if (!list_empty(&msm_obj->submit_entry)) {
-                        DBG("handle %u at index %u already on submit list",
+                        DRM_ERROR("handle %u at index %u already on submit list\n",
                                         submit_bo.handle, i);
                         ret = -EINVAL;
                         goto out_unlock;
@@ -216,8 +216,9 @@ static int submit_bo(struct msm_gem_submit *submit, uint32_t idx,
                 struct msm_gem_object **obj, uint32_t *iova, bool *valid)
 {
         if (idx >= submit->nr_bos) {
-                DBG("invalid buffer index: %u (out of %u)", idx, submit->nr_bos);
-                return EINVAL;
+                DRM_ERROR("invalid buffer index: %u (out of %u)\n",
+                                idx, submit->nr_bos);
+                return -EINVAL;
         }
 
         if (obj)
@@ -239,7 +240,7 @@ static int submit_reloc(struct msm_gem_submit *submit, struct msm_gem_object *ob
         int ret;
 
         if (offset % 4) {
-                DBG("non-aligned cmdstream buffer: %u", offset);
+                DRM_ERROR("non-aligned cmdstream buffer: %u\n", offset);
                 return -EINVAL;
         }
 
@@ -266,7 +267,7 @@ static int submit_reloc(struct msm_gem_submit *submit, struct msm_gem_object *ob
                         return -EFAULT;
 
                 if (submit_reloc.submit_offset % 4) {
-                        DBG("non-aligned reloc offset: %u",
+                        DRM_ERROR("non-aligned reloc offset: %u\n",
                                         submit_reloc.submit_offset);
                         return -EINVAL;
                 }
@@ -276,7 +277,7 @@ static int submit_reloc(struct msm_gem_submit *submit, struct msm_gem_object *ob
 
                 if ((off >= (obj->base.size / 4)) ||
                                 (off < last_offset)) {
-                        DBG("invalid offset %u at reloc %u", off, i);
+                        DRM_ERROR("invalid offset %u at reloc %u\n", off, i);
                         return -EINVAL;
                 }
 
@@ -374,14 +375,15 @@ int msm_ioctl_gem_submit(struct drm_device *dev, void *data,
                         goto out;
 
                 if (submit_cmd.size % 4) {
-                        DBG("non-aligned cmdstream buffer size: %u",
+                        DRM_ERROR("non-aligned cmdstream buffer size: %u\n",
                                         submit_cmd.size);
                         ret = -EINVAL;
                         goto out;
                 }
 
-                if (submit_cmd.size >= msm_obj->base.size) {
-                        DBG("invalid cmdstream size: %u", submit_cmd.size);
+                if ((submit_cmd.size + submit_cmd.submit_offset) >=
+                                msm_obj->base.size) {
+                        DRM_ERROR("invalid cmdstream size: %u\n", submit_cmd.size);
                         ret = -EINVAL;
                         goto out;
                 }
diff --git a/drivers/gpu/drm/msm/msm_gpu.c b/drivers/gpu/drm/msm/msm_gpu.c
index e1e1ec9321ff..3bab937965d1 100644
--- a/drivers/gpu/drm/msm/msm_gpu.c
+++ b/drivers/gpu/drm/msm/msm_gpu.c
@@ -29,13 +29,14 @@
 static void bs_init(struct msm_gpu *gpu, struct platform_device *pdev)
 {
         struct drm_device *dev = gpu->dev;
-        struct kgsl_device_platform_data *pdata = pdev->dev.platform_data;
+        struct kgsl_device_platform_data *pdata;
 
         if (!pdev) {
                 dev_err(dev->dev, "could not find dtv pdata\n");
                 return;
         }
 
+        pdata = pdev->dev.platform_data;
         if (pdata->bus_scale_table) {
                 gpu->bsc = msm_bus_scale_register_client(pdata->bus_scale_table);
                 DBG("bus scale client: %08x", gpu->bsc);
@@ -230,6 +231,8 @@ static void hangcheck_timer_reset(struct msm_gpu *gpu)
 static void hangcheck_handler(unsigned long data)
 {
         struct msm_gpu *gpu = (struct msm_gpu *)data;
+        struct drm_device *dev = gpu->dev;
+        struct msm_drm_private *priv = dev->dev_private;
         uint32_t fence = gpu->funcs->last_fence(gpu);
 
         if (fence != gpu->hangcheck_fence) {
@@ -237,14 +240,22 @@ static void hangcheck_handler(unsigned long data)
                 gpu->hangcheck_fence = fence;
         } else if (fence < gpu->submitted_fence) {
                 /* no progress and not done.. hung! */
-                struct msm_drm_private *priv = gpu->dev->dev_private;
                 gpu->hangcheck_fence = fence;
+                dev_err(dev->dev, "%s: hangcheck detected gpu lockup!\n",
+                                gpu->name);
+                dev_err(dev->dev, "%s:     completed fence: %u\n",
+                                gpu->name, fence);
+                dev_err(dev->dev, "%s:     submitted fence: %u\n",
+                                gpu->name, gpu->submitted_fence);
                 queue_work(priv->wq, &gpu->recover_work);
         }
 
         /* if still more pending work, reset the hangcheck timer: */
         if (gpu->submitted_fence > gpu->hangcheck_fence)
                 hangcheck_timer_reset(gpu);
+
+        /* workaround for missing irq: */
+        queue_work(priv->wq, &gpu->retire_work);
 }
 
 /*
@@ -265,7 +276,8 @@ static void retire_worker(struct work_struct *work)
                 obj = list_first_entry(&gpu->active_list,
                                 struct msm_gem_object, mm_list);
 
-                if (obj->fence <= fence) {
+                if ((obj->read_fence <= fence) &&
+                                (obj->write_fence <= fence)) {
                         /* move to inactive: */
                         msm_gem_move_to_inactive(&obj->base);
                         msm_gem_put_iova(&obj->base, gpu->id);
@@ -321,7 +333,11 @@ int msm_gpu_submit(struct msm_gpu *gpu, struct msm_gem_submit *submit,
                                         submit->gpu->id, &iova);
                 }
 
-                msm_gem_move_to_active(&msm_obj->base, gpu, submit->fence);
+                if (submit->bos[i].flags & MSM_SUBMIT_BO_READ)
+                        msm_gem_move_to_active(&msm_obj->base, gpu, false, submit->fence);
+
+                if (submit->bos[i].flags & MSM_SUBMIT_BO_WRITE)
+                        msm_gem_move_to_active(&msm_obj->base, gpu, true, submit->fence);
         }
         hangcheck_timer_reset(gpu);
         mutex_unlock(&dev->struct_mutex);