[bug report] drm/amdgpu: switch ih handling to two levels (v3)
Dan Carpenter
dan.carpenter at oracle.com
Tue Apr 4 13:04:13 UTC 2017
Hello Alex Deucher,
The patch d766e6a39338: "drm/amdgpu: switch ih handling to two levels
(v3)" from Mar 29, 2016, leads to the following static checker
warning:
drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c:319 amdgpu_irq_add_id()
warn: double check that we're allocating correct size: 8 vs 32
drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
305 int amdgpu_irq_add_id(struct amdgpu_device *adev,
306 unsigned client_id, unsigned src_id,
307 struct amdgpu_irq_src *source)
308 {
309 if (client_id >= AMDGPU_IH_CLIENTID_MAX)
310 return -EINVAL;
311
312 if (src_id >= AMDGPU_MAX_IRQ_SRC_ID)
313 return -EINVAL;
314
315 if (!source->funcs)
316 return -EINVAL;
317
318 if (!adev->irq.client[client_id].sources) {
319 adev->irq.client[client_id].sources = kcalloc(AMDGPU_MAX_IRQ_SRC_ID,
320 sizeof(struct amdgpu_irq_src),
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
We want to allocate an array of pointers, so probably this could be
changed to sizeof(void *)?
321 GFP_KERNEL);
322 if (!adev->irq.client[client_id].sources)
323 return -ENOMEM;
324 }
regards,
dan carpenter
More information about the amd-gfx
mailing list