gpu: nvgpu: allow bind to be interrupted

This change solves two problems:

(*) the possibility of a crash due to interrupting the gpu
initialization following a bind
(*) a IOVA memory leak that could prevent the GPU from binding after
about 200 bind/unbind cycles

A detailed list of fixes:

- chek that arbiter is initialized before freeing it.
- do not re-enable interrupts when MSI is enabled on unbind.
- free the semaphore sea on unbind.
- ensure we dont double load the vbios.
- check return value of nvgpu_mutex_init for semaphores.
- add corresponding nvgpu_mutex_destroy calls.

bug 1816516

Change-Id: Ia8af73019e0e1183998855d55bb3eea09672a8b7
Signed-off-by: David Nieto <dmartineznie@nvidia.com>
Reviewed-on: http://git-master/r/1465302
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: Thomas Fleury <tfleury@nvidia.com>
Reviewed-by: David Jarrett <djarrett@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1563019
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>

6 files changed, 29 insertions, 11 deletions

diff --git a/drivers/gpu/nvgpu/clk/clk_arb.c b/drivers/gpu/nvgpu/clk/clk_arb.c
index 4f09da74..0df9545a 100644
--- a/drivers/gpu/nvgpu/clk/clk_arb.c
+++ b/drivers/gpu/nvgpu/clk/clk_arb.c
@@ -490,9 +490,9 @@ void nvgpu_clk_arb_cleanup_arbiter(struct gk20a *g)
                         nvgpu_kfree(g, arb->vf_table_pool[index].mclk_points);
                 }
                 nvgpu_mutex_destroy(&g->clk_arb->pstate_lock);
+                nvgpu_kfree(g, g->clk_arb);
+                g->clk_arb = NULL;
         }
-        nvgpu_kfree(g, g->clk_arb);
-        g->clk_arb = NULL;
 }
 
 static int nvgpu_clk_arb_install_fd(struct gk20a *g,
diff --git a/drivers/gpu/nvgpu/common/semaphore.c b/drivers/gpu/nvgpu/common/semaphore.c
index 5496f5ec..0d512e8f 100644
--- a/drivers/gpu/nvgpu/common/semaphore.c
+++ b/drivers/gpu/nvgpu/common/semaphore.c
@@ -66,11 +66,12 @@ out:
         return ret;
 }
 
-void gk20a_semaphore_sea_destroy(struct gk20a *g)
+void nvgpu_semaphore_sea_destroy(struct gk20a *g)
 {
         if (!g->sema_sea)
                 return;
 
+        nvgpu_dma_free(g, &g->sema_sea->sea_mem);
         nvgpu_mutex_destroy(&g->sema_sea->sea_lock);
         nvgpu_kfree(g, g->sema_sea);
         g->sema_sea = NULL;
diff --git a/drivers/gpu/nvgpu/gk20a/gk20a.h b/drivers/gpu/nvgpu/gk20a/gk20a.h
index 13c62691..0cd77d1e 100644
--- a/drivers/gpu/nvgpu/gk20a/gk20a.h
+++ b/drivers/gpu/nvgpu/gk20a/gk20a.h
@@ -1233,6 +1233,7 @@ struct gk20a {
 #ifdef CONFIG_DEBUG_FS
         struct debugfs_blob_wrapper bios_blob;
 #endif
+        bool bios_is_init;
 
         struct nvgpu_clk_arb *clk_arb;
 
diff --git a/drivers/gpu/nvgpu/gk20a/mm_gk20a.c b/drivers/gpu/nvgpu/gk20a/mm_gk20a.c
index 0e0326dd..dd8b900d 100644
--- a/drivers/gpu/nvgpu/gk20a/mm_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/mm_gk20a.c
@@ -393,7 +393,7 @@ static void gk20a_remove_mm_support(struct mm_gk20a *mm)
         nvgpu_vm_put(mm->pmu.vm);
         nvgpu_vm_put(mm->cde.vm);
 
-        gk20a_semaphore_sea_destroy(g);
+        nvgpu_semaphore_sea_destroy(g);
         gk20a_vidmem_destroy(g);
         nvgpu_pd_cache_fini(g);
 }
diff --git a/drivers/gpu/nvgpu/gp106/bios_gp106.c b/drivers/gpu/nvgpu/gp106/bios_gp106.c
index f772e267..c728433f 100644
--- a/drivers/gpu/nvgpu/gp106/bios_gp106.c
+++ b/drivers/gpu/nvgpu/gp106/bios_gp106.c
@@ -177,6 +177,9 @@ int gp106_bios_init(struct gk20a *g)
 
         gk20a_dbg_fn("");
 
+        if (g->bios_is_init)
+                return 0;
+
         gk20a_dbg_info("reading bios from EEPROM");
         g->bios.size = BIOS_SIZE;
         g->bios.data = nvgpu_vmalloc(g, BIOS_SIZE);
@@ -195,12 +198,13 @@ int gp106_bios_init(struct gk20a *g)
 
         err = nvgpu_bios_parse_rom(g);
         if (err)
-                return err;
+                goto free_firmware;
 
         if (g->gpu_characteristics.vbios_version < g->vbios_min_version) {
                 nvgpu_err(g, "unsupported VBIOS version %08x",
                                 g->gpu_characteristics.vbios_version);
-                return -EINVAL;
+                err = -EINVAL;
+                goto free_firmware;
         }
 
         /* WAR for HW2.5 RevA (INA3221 is missing) */
@@ -216,25 +220,37 @@ int gp106_bios_init(struct gk20a *g)
 
         d = debugfs_create_blob("bios", S_IRUGO, l->debugfs,
                         &g->bios_blob);
-        if (!d)
+        if (!d) {
+                err = -EINVAL;
                 nvgpu_err(g, "No debugfs?");
+                goto free_firmware;
+        }
 #endif
-
         gk20a_dbg_fn("done");
 
         err = gp106_bios_devinit(g);
         if (err) {
                 nvgpu_err(g, "devinit failed");
-                return err;
+                goto free_debugfs;
         }
 
         if (nvgpu_is_enabled(g, NVGPU_PMU_RUN_PREOS)) {
                 err = gp106_bios_preos(g);
                 if (err) {
                         nvgpu_err(g, "pre-os failed");
-                        return err;
+                        goto free_debugfs;
                 }
         }
+        g->bios_is_init = true;
 
         return 0;
+free_debugfs:
+#ifdef CONFIG_DEBUG_FS
+        debugfs_remove(d);
+#endif
+free_firmware:
+        if (g->bios.data)
+                nvgpu_vfree(g, g->bios.data);
+        return err;
 }
+
diff --git a/drivers/gpu/nvgpu/include/nvgpu/semaphore.h b/drivers/gpu/nvgpu/include/nvgpu/semaphore.h
index 5c0019ae..8915b722 100644
--- a/drivers/gpu/nvgpu/include/nvgpu/semaphore.h
+++ b/drivers/gpu/nvgpu/include/nvgpu/semaphore.h
@@ -160,7 +160,7 @@ struct nvgpu_semaphore_sea {
  * Semaphore sea functions.
  */
 struct nvgpu_semaphore_sea *nvgpu_semaphore_sea_create(struct gk20a *gk20a);
-void gk20a_semaphore_sea_destroy(struct gk20a *g);
+void nvgpu_semaphore_sea_destroy(struct gk20a *g);
 int nvgpu_semaphore_sea_map(struct nvgpu_semaphore_pool *sea,
                             struct vm_gk20a *vm);
 void nvgpu_semaphore_sea_unmap(struct nvgpu_semaphore_pool *sea,