Merge branch 'etnaviv/fixes' of https://git.pengutronix.de/git/lst/linux into drm-fixes

Lucas wrote:
"a couple of small fixes:
- 2 patches from Fabio to fix module reloading
- one patch to fix a userspace visible regression, where the job
timeout is a bit too eager and kills legitimate jobs"

Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1530868450.15725.8.camel@pengutronix.de

3 files changed, 47 insertions, 4 deletions

diff --git a/drivers/gpu/drm/etnaviv/etnaviv_drv.c b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
index e5013a999147..540b59fb4103 100644
--- a/drivers/gpu/drm/etnaviv/etnaviv_drv.c
+++ b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
@@ -631,8 +631,11 @@ static struct platform_driver etnaviv_platform_driver = {
         },
 };
 
+static struct platform_device *etnaviv_drm;
+
 static int __init etnaviv_init(void)
 {
+        struct platform_device *pdev;
         int ret;
         struct device_node *np;
 
@@ -644,7 +647,7 @@ static int __init etnaviv_init(void)
 
         ret = platform_driver_register(&etnaviv_platform_driver);
         if (ret != 0)
-                platform_driver_unregister(&etnaviv_gpu_driver);
+                goto unregister_gpu_driver;
 
         /*
          * If the DT contains at least one available GPU device, instantiate
@@ -653,20 +656,33 @@ static int __init etnaviv_init(void)
         for_each_compatible_node(np, NULL, "vivante,gc") {
                 if (!of_device_is_available(np))
                         continue;
-
-                platform_device_register_simple("etnaviv", -1, NULL, 0);
+                pdev = platform_device_register_simple("etnaviv", -1,
+                                                       NULL, 0);
+                if (IS_ERR(pdev)) {
+                        ret = PTR_ERR(pdev);
+                        of_node_put(np);
+                        goto unregister_platform_driver;
+                }
+                etnaviv_drm = pdev;
                 of_node_put(np);
                 break;
         }
 
+        return 0;
+
+unregister_platform_driver:
+        platform_driver_unregister(&etnaviv_platform_driver);
+unregister_gpu_driver:
+        platform_driver_unregister(&etnaviv_gpu_driver);
         return ret;
 }
 module_init(etnaviv_init);
 
 static void __exit etnaviv_exit(void)
 {
-        platform_driver_unregister(&etnaviv_gpu_driver);
+        platform_device_unregister(etnaviv_drm);
         platform_driver_unregister(&etnaviv_platform_driver);
+        platform_driver_unregister(&etnaviv_gpu_driver);
 }
 module_exit(etnaviv_exit);
 
diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gpu.h b/drivers/gpu/drm/etnaviv/etnaviv_gpu.h
index dd430f0f8ff5..90f17ff7888e 100644
--- a/drivers/gpu/drm/etnaviv/etnaviv_gpu.h
+++ b/drivers/gpu/drm/etnaviv/etnaviv_gpu.h
@@ -131,6 +131,9 @@ struct etnaviv_gpu {
         struct work_struct sync_point_work;
         int sync_point_event;
 
+        /* hang detection */
+        u32 hangcheck_dma_addr;
+
         void __iomem *mmio;
         int irq;
 
diff --git a/drivers/gpu/drm/etnaviv/etnaviv_sched.c b/drivers/gpu/drm/etnaviv/etnaviv_sched.c
index a74eb57af15b..50d6b88cb7aa 100644
--- a/drivers/gpu/drm/etnaviv/etnaviv_sched.c
+++ b/drivers/gpu/drm/etnaviv/etnaviv_sched.c
@@ -10,6 +10,7 @@
 #include "etnaviv_gem.h"
 #include "etnaviv_gpu.h"
 #include "etnaviv_sched.h"
+#include "state.xml.h"
 
 static int etnaviv_job_hang_limit = 0;
 module_param_named(job_hang_limit, etnaviv_job_hang_limit, int , 0444);
@@ -85,6 +86,29 @@ static void etnaviv_sched_timedout_job(struct drm_sched_job *sched_job)
 {
         struct etnaviv_gem_submit *submit = to_etnaviv_submit(sched_job);
         struct etnaviv_gpu *gpu = submit->gpu;
+        u32 dma_addr;
+        int change;
+
+        /*
+         * If the GPU managed to complete this jobs fence, the timout is
+         * spurious. Bail out.
+         */
+        if (fence_completed(gpu, submit->out_fence->seqno))
+                return;
+
+        /*
+         * If the GPU is still making forward progress on the front-end (which
+         * should never loop) we shift out the timeout to give it a chance to
+         * finish the job.
+         */
+        dma_addr = gpu_read(gpu, VIVS_FE_DMA_ADDRESS);
+        change = dma_addr - gpu->hangcheck_dma_addr;
+        if (change < 0 || change > 16) {
+                gpu->hangcheck_dma_addr = dma_addr;
+                schedule_delayed_work(&sched_job->work_tdr,
+                                      sched_job->sched->timeout);
+                return;
+        }
 
         /* block scheduler */
         kthread_park(gpu->sched.thread);