drm/amdkfd: Adding new IOCTL for scratch memory v2

v2:
* Renamed ALLOC_MEMORY_OF_SCRATCH to SET_SCRATCH_BACKING_VA
* Removed size parameter from the ioctl, it was unused
* Removed hole in ioctl number space
* No more call to write_config_static_mem
* Return correct error code from ioctl

Signed-off-by: Moses Reuben <moses.reuben@amd.com>
Signed-off-by: Ben Goz <ben.goz@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>

drivers/gpu/drm/amd/amdkfd/kfd_chardev.c

@@ -848,6 +848,40 @@ static int kfd_ioctl_wait_events(struct file *filp, struct kfd_process *p,
 
 	return err;
 }
+static int kfd_ioctl_set_scratch_backing_va(struct file *filep,
+					struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_set_scratch_backing_va_args *args = data;
+	struct kfd_process_device *pdd;
+	struct kfd_dev *dev;
+	long err;
+
+	dev = kfd_device_by_id(args->gpu_id);
+	if (!dev)
+		return -EINVAL;
+
+	mutex_lock(&p->mutex);
+
+	pdd = kfd_bind_process_to_device(dev, p);
+	if (IS_ERR(pdd)) {
+		err = PTR_ERR(pdd);
+		goto bind_process_to_device_fail;
+	}
+
+	pdd->qpd.sh_hidden_private_base = args->va_addr;
+
+	mutex_unlock(&p->mutex);
+
+	if (sched_policy == KFD_SCHED_POLICY_NO_HWS && pdd->qpd.vmid != 0)
+		dev->kfd2kgd->set_scratch_backing_va(
+			dev->kgd, args->va_addr, pdd->qpd.vmid);
+
+	return 0;
+
+bind_process_to_device_fail:
+	mutex_unlock(&p->mutex);
+	return err;
+}
 
 #define AMDKFD_IOCTL_DEF(ioctl, _func, _flags) \
 	[_IOC_NR(ioctl)] = {.cmd = ioctl, .func = _func, .flags = _flags, \
@@ -902,6 +936,9 @@ static const struct amdkfd_ioctl_desc amdkfd_ioctls[] = {
 
 	AMDKFD_IOCTL_DEF(AMDKFD_IOC_DBG_WAVE_CONTROL,
 			kfd_ioctl_dbg_wave_control, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_SET_SCRATCH_BACKING_VA,
+			kfd_ioctl_set_scratch_backing_va, 0),
 };
 
 #define AMDKFD_CORE_IOCTL_COUNT	ARRAY_SIZE(amdkfd_ioctls)

drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c

@@ -270,6 +270,9 @@ static int create_compute_queue_nocpsch(struct device_queue_manager *dqm,
 	pr_debug("Loading mqd to hqd on pipe %d, queue %d\n",
 			q->pipe, q->queue);
 
+	dqm->dev->kfd2kgd->set_scratch_backing_va(
+			dqm->dev->kgd, qpd->sh_hidden_private_base, qpd->vmid);
+
 	retval = mqd->load_mqd(mqd, q->mqd, q->pipe, q->queue, &q->properties,
 			       q->process->mm);
 	if (retval)

drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c

@@ -24,6 +24,7 @@
 #include "kfd_device_queue_manager.h"
 #include "cik_regs.h"
 #include "oss/oss_2_4_sh_mask.h"
+#include "gca/gfx_7_2_sh_mask.h"
 
 static bool set_cache_memory_policy_cik(struct device_queue_manager *dqm,
 				   struct qcm_process_device *qpd,
@@ -123,6 +124,7 @@ static int register_process_cik(struct device_queue_manager *dqm,
 	} else {
 		temp = get_sh_mem_bases_nybble_64(pdd);
 		qpd->sh_mem_bases = compute_sh_mem_bases_64bit(temp);
+		qpd->sh_mem_config |= 1  << SH_MEM_CONFIG__PRIVATE_ATC__SHIFT;
 	}
 
 	pr_debug("is32bit process: %d sh_mem_bases nybble: 0x%X and register 0x%X\n",

drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c

@@ -135,6 +135,8 @@ static int register_process_vi(struct device_queue_manager *dqm,
 		qpd->sh_mem_bases = compute_sh_mem_bases_64bit(temp);
 		qpd->sh_mem_config |= SH_MEM_ADDRESS_MODE_HSA64 <<
 			SH_MEM_CONFIG__ADDRESS_MODE__SHIFT;
+		qpd->sh_mem_config |= 1  <<
+			SH_MEM_CONFIG__PRIVATE_ATC__SHIFT;
 	}
 
 	pr_debug("is32bit process: %d sh_mem_bases nybble: 0x%X and register 0x%X\n",

drivers/gpu/drm/amd/amdkfd/kfd_priv.h

@@ -432,6 +432,7 @@ struct qcm_process_device {
 	uint32_t gds_size;
 	uint32_t num_gws;
 	uint32_t num_oac;
+	uint32_t sh_hidden_private_base;
 };
 
 /* Data that is per-process-per device. */