gpu: nvgpu: Add nvgpu_mem_is_valid() call

Add a function to check if an nvgpu_mem is allocated (valid) or not.
Also fix possibly leaked state in nvgpu_mems when they fail to
allocate.

Also ensure that in the case of a failure to allocate no state is
accidentally leaked to the caller. This should hopefully make it
less likely that a caller thinks a buffer that failed to allocate
is actually allocated.

Change-Id: I43224ece7da84e63b2f43f36f04941126fabf3c7
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1559419
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>

1 files changed, 2 insertions, 0 deletions

diff --git a/drivers/gpu/nvgpu/common/linux/dma.c b/drivers/gpu/nvgpu/common/linux/dma.c
index 13c1c347..43009fca 100644
--- a/drivers/gpu/nvgpu/common/linux/dma.c
+++ b/drivers/gpu/nvgpu/common/linux/dma.c
@@ -167,6 +167,7 @@ fail_free:
         dma_free_coherent(d, size, mem->cpu_va, iova);
         mem->cpu_va = NULL;
         mem->priv.sgt = NULL;
+        mem->size = 0;
         return err;
 }
 
@@ -253,6 +254,7 @@ fail_kfree:
         nvgpu_kfree(g, mem->priv.sgt);
 fail_physfree:
         nvgpu_free(&g->mm.vidmem.allocator, addr);
+        mem->size = 0;
         return err;
 #else
         return -ENOSYS;