[deliverable/linux.git] / drivers / gpu / drm / amd / amdkfd / kfd_device.c

/*
 * Copyright 2014 Advanced Micro Devices, Inc.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
 * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
 * OTHER DEALINGS IN THE SOFTWARE.
 */

#include <linux/amd-iommu.h>
#include <linux/bsearch.h>
#include <linux/pci.h>
#include <linux/slab.h>
#include "kfd_priv.h"

#define MQD_SIZE_ALIGNED 768

static const struct kfd_device_info kaveri_device_info = {
	.max_pasid_bits = 16,
	.mqd_size_aligned = MQD_SIZE_ALIGNED
};

struct kfd_deviceid {
	unsigned short did;
	const struct kfd_device_info *device_info;
};

/* Please keep this sorted by increasing device id. */
static const struct kfd_deviceid supported_devices[] = {
	{ 0x1304, &kaveri_device_info },	/* Kaveri */
	{ 0x1305, &kaveri_device_info },	/* Kaveri */
	{ 0x1306, &kaveri_device_info },	/* Kaveri */
	{ 0x1307, &kaveri_device_info },	/* Kaveri */
	{ 0x1309, &kaveri_device_info },	/* Kaveri */
	{ 0x130A, &kaveri_device_info },	/* Kaveri */
	{ 0x130B, &kaveri_device_info },	/* Kaveri */
	{ 0x130C, &kaveri_device_info },	/* Kaveri */
	{ 0x130D, &kaveri_device_info },	/* Kaveri */
	{ 0x130E, &kaveri_device_info },	/* Kaveri */
	{ 0x130F, &kaveri_device_info },	/* Kaveri */
	{ 0x1310, &kaveri_device_info },	/* Kaveri */
	{ 0x1311, &kaveri_device_info },	/* Kaveri */
	{ 0x1312, &kaveri_device_info },	/* Kaveri */
	{ 0x1313, &kaveri_device_info },	/* Kaveri */
	{ 0x1315, &kaveri_device_info },	/* Kaveri */
	{ 0x1316, &kaveri_device_info },	/* Kaveri */
	{ 0x1317, &kaveri_device_info },	/* Kaveri */
	{ 0x1318, &kaveri_device_info },	/* Kaveri */
	{ 0x131B, &kaveri_device_info },	/* Kaveri */
	{ 0x131C, &kaveri_device_info },	/* Kaveri */
	{ 0x131D, &kaveri_device_info },	/* Kaveri */
};

static const struct kfd_device_info *lookup_device_info(unsigned short did)
{
	size_t i;

	for (i = 0; i < ARRAY_SIZE(supported_devices); i++) {
		if (supported_devices[i].did == did) {
			BUG_ON(supported_devices[i].device_info == NULL);
			return supported_devices[i].device_info;
		}
	}

	return NULL;
}

struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd, struct pci_dev *pdev)
{
	struct kfd_dev *kfd;

	const struct kfd_device_info *device_info =
					lookup_device_info(pdev->device);

	if (!device_info)
		return NULL;

	kfd = kzalloc(sizeof(*kfd), GFP_KERNEL);
	if (!kfd)
		return NULL;

	kfd->kgd = kgd;
	kfd->device_info = device_info;
	kfd->pdev = pdev;
	kfd->init_complete = false;

	return kfd;
}

static bool device_iommu_pasid_init(struct kfd_dev *kfd)
{
	const u32 required_iommu_flags = AMD_IOMMU_DEVICE_FLAG_ATS_SUP |
					AMD_IOMMU_DEVICE_FLAG_PRI_SUP |
					AMD_IOMMU_DEVICE_FLAG_PASID_SUP;

	struct amd_iommu_device_info iommu_info;
	unsigned int pasid_limit;
	int err;

	err = amd_iommu_device_info(kfd->pdev, &iommu_info);
	if (err < 0) {
		dev_err(kfd_device,
			"error getting iommu info. is the iommu enabled?\n");
		return false;
	}

	if ((iommu_info.flags & required_iommu_flags) != required_iommu_flags) {
		dev_err(kfd_device, "error required iommu flags ats(%i), pri(%i), pasid(%i)\n",
		       (iommu_info.flags & AMD_IOMMU_DEVICE_FLAG_ATS_SUP) != 0,
		       (iommu_info.flags & AMD_IOMMU_DEVICE_FLAG_PRI_SUP) != 0,
		       (iommu_info.flags & AMD_IOMMU_DEVICE_FLAG_PASID_SUP) != 0);
		return false;
	}

	pasid_limit = min_t(unsigned int,
			(unsigned int)1 << kfd->device_info->max_pasid_bits,
			iommu_info.max_pasids);
	/*
	 * last pasid is used for kernel queues doorbells
	 * in the future the last pasid might be used for a kernel thread.
	 */
	pasid_limit = min_t(unsigned int,
				pasid_limit,
				kfd->doorbell_process_limit - 1);

	err = amd_iommu_init_device(kfd->pdev, pasid_limit);
	if (err < 0) {
		dev_err(kfd_device, "error initializing iommu device\n");
		return false;
	}

	if (!kfd_set_pasid_limit(pasid_limit)) {
		dev_err(kfd_device, "error setting pasid limit\n");
		amd_iommu_free_device(kfd->pdev);
		return false;
	}

	return true;
}

static void iommu_pasid_shutdown_callback(struct pci_dev *pdev, int pasid)
{
	struct kfd_dev *dev = kfd_device_by_pci_dev(pdev);

	if (dev)
		kfd_unbind_process_from_device(dev, pasid);
}

bool kgd2kfd_device_init(struct kfd_dev *kfd,
			 const struct kgd2kfd_shared_resources *gpu_resources)
{
	unsigned int size;

	kfd->shared_resources = *gpu_resources;

	/* calculate max size of mqds needed for queues */
	size = max_num_of_processes *
		max_num_of_queues_per_process *
		kfd->device_info->mqd_size_aligned;

	/* add another 512KB for all other allocations on gart */
	size += 512 * 1024;

	if (kfd2kgd->init_sa_manager(kfd->kgd, size)) {
		dev_err(kfd_device,
			"Error initializing sa manager for device (%x:%x)\n",
			kfd->pdev->vendor, kfd->pdev->device);
		goto out;
	}

	kfd_doorbell_init(kfd);

	if (kfd_topology_add_device(kfd) != 0) {
		dev_err(kfd_device,
			"Error adding device (%x:%x) to topology\n",
			kfd->pdev->vendor, kfd->pdev->device);
		goto kfd_topology_add_device_error;
	}

	if (!device_iommu_pasid_init(kfd)) {
		dev_err(kfd_device,
			"Error initializing iommuv2 for device (%x:%x)\n",
			kfd->pdev->vendor, kfd->pdev->device);
		goto device_iommu_pasid_error;
	}
	amd_iommu_set_invalidate_ctx_cb(kfd->pdev,
						iommu_pasid_shutdown_callback);

	kfd->init_complete = true;
	dev_info(kfd_device, "added device (%x:%x)\n", kfd->pdev->vendor,
		 kfd->pdev->device);

	goto out;

device_iommu_pasid_error:
	kfd_topology_remove_device(kfd);
kfd_topology_add_device_error:
	kfd2kgd->fini_sa_manager(kfd->kgd);
	dev_err(kfd_device,
		"device (%x:%x) NOT added due to errors\n",
		kfd->pdev->vendor, kfd->pdev->device);
out:
	return kfd->init_complete;
}

void kgd2kfd_device_exit(struct kfd_dev *kfd)
{
	if (kfd->init_complete) {
		amd_iommu_free_device(kfd->pdev);
		kfd_topology_remove_device(kfd);
	}

	kfree(kfd);
}

void kgd2kfd_suspend(struct kfd_dev *kfd)
{
	BUG_ON(kfd == NULL);

	if (kfd->init_complete)
		amd_iommu_free_device(kfd->pdev);
}

int kgd2kfd_resume(struct kfd_dev *kfd)
{
	unsigned int pasid_limit;
	int err;

	BUG_ON(kfd == NULL);

	pasid_limit = kfd_get_pasid_limit();

	if (kfd->init_complete) {
		err = amd_iommu_init_device(kfd->pdev, pasid_limit);
		if (err < 0)
			return -ENXIO;
		amd_iommu_set_invalidate_ctx_cb(kfd->pdev,
						iommu_pasid_shutdown_callback);
	}

	return 0;
}

void kgd2kfd_interrupt(struct kfd_dev *dev, const void *ih_ring_entry)
{
}
Commit	Line	Data
4a488a7a OG	1	/*
	2	* Copyright 2014 Advanced Micro Devices, Inc.
	3	*
	4	* Permission is hereby granted, free of charge, to any person obtaining a
	5	* copy of this software and associated documentation files (the "Software"),
	6	* to deal in the Software without restriction, including without limitation
	7	* the rights to use, copy, modify, merge, publish, distribute, sublicense,
	8	* and/or sell copies of the Software, and to permit persons to whom the
	9	* Software is furnished to do so, subject to the following conditions:
	10	*
	11	* The above copyright notice and this permission notice shall be included in
	12	* all copies or substantial portions of the Software.
	13	*
	14	* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
	15	* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
	16	* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
	17	* THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
	18	* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
	19	* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
	20	* OTHER DEALINGS IN THE SOFTWARE.
	21	*/
	22
	23	#include <linux/amd-iommu.h>
	24	#include <linux/bsearch.h>
	25	#include <linux/pci.h>
	26	#include <linux/slab.h>
	27	#include "kfd_priv.h"
	28
19f6d2a6 OG	29	#define MQD_SIZE_ALIGNED 768
19f6d2a6 OG	30
4a488a7a OG	31	static const struct kfd_device_info kaveri_device_info = {
4a488a7a OG	32	.max_pasid_bits = 16,
19f6d2a6	33	.mqd_size_aligned = MQD_SIZE_ALIGNED
4a488a7a OG	34	};
	35
	36	struct kfd_deviceid {
	37	unsigned short did;
	38	const struct kfd_device_info *device_info;
	39	};
	40
	41	/* Please keep this sorted by increasing device id. */
	42	static const struct kfd_deviceid supported_devices[] = {
	43	{ 0x1304, &kaveri_device_info }, /* Kaveri */
	44	{ 0x1305, &kaveri_device_info }, /* Kaveri */
	45	{ 0x1306, &kaveri_device_info }, /* Kaveri */
	46	{ 0x1307, &kaveri_device_info }, /* Kaveri */
	47	{ 0x1309, &kaveri_device_info }, /* Kaveri */
	48	{ 0x130A, &kaveri_device_info }, /* Kaveri */
	49	{ 0x130B, &kaveri_device_info }, /* Kaveri */
	50	{ 0x130C, &kaveri_device_info }, /* Kaveri */
	51	{ 0x130D, &kaveri_device_info }, /* Kaveri */
	52	{ 0x130E, &kaveri_device_info }, /* Kaveri */
	53	{ 0x130F, &kaveri_device_info }, /* Kaveri */
	54	{ 0x1310, &kaveri_device_info }, /* Kaveri */
	55	{ 0x1311, &kaveri_device_info }, /* Kaveri */
	56	{ 0x1312, &kaveri_device_info }, /* Kaveri */
	57	{ 0x1313, &kaveri_device_info }, /* Kaveri */
	58	{ 0x1315, &kaveri_device_info }, /* Kaveri */
	59	{ 0x1316, &kaveri_device_info }, /* Kaveri */
	60	{ 0x1317, &kaveri_device_info }, /* Kaveri */
	61	{ 0x1318, &kaveri_device_info }, /* Kaveri */
	62	{ 0x131B, &kaveri_device_info }, /* Kaveri */
	63	{ 0x131C, &kaveri_device_info }, /* Kaveri */
	64	{ 0x131D, &kaveri_device_info }, /* Kaveri */
	65	};
	66
	67	static const struct kfd_device_info *lookup_device_info(unsigned short did)
	68	{
	69	size_t i;
	70
	71	for (i = 0; i < ARRAY_SIZE(supported_devices); i++) {
	72	if (supported_devices[i].did == did) {
	73	BUG_ON(supported_devices[i].device_info == NULL);
	74	return supported_devices[i].device_info;
	75	}
	76	}
	77
	78	return NULL;
	79	}
	80
	81	struct kfd_dev kgd2kfd_probe(struct kgd_dev kgd, struct pci_dev *pdev)
	82	{
	83	struct kfd_dev *kfd;
	84
	85	const struct kfd_device_info *device_info =
	86	lookup_device_info(pdev->device);
	87
	88	if (!device_info)
	89	return NULL;
	90
	91	kfd = kzalloc(sizeof(*kfd), GFP_KERNEL);
	92	if (!kfd)
	93	return NULL;
	94
	95	kfd->kgd = kgd;
	96	kfd->device_info = device_info;
	97	kfd->pdev = pdev;
19f6d2a6	98	kfd->init_complete = false;
4a488a7a OG	99
	100	return kfd;
	101	}
	102
b17f068a OG	103	static bool device_iommu_pasid_init(struct kfd_dev *kfd)
	104	{
	105	const u32 required_iommu_flags = AMD_IOMMU_DEVICE_FLAG_ATS_SUP \|
	106	AMD_IOMMU_DEVICE_FLAG_PRI_SUP \|
	107	AMD_IOMMU_DEVICE_FLAG_PASID_SUP;
	108
	109	struct amd_iommu_device_info iommu_info;
	110	unsigned int pasid_limit;
	111	int err;
	112
	113	err = amd_iommu_device_info(kfd->pdev, &iommu_info);
	114	if (err < 0) {
	115	dev_err(kfd_device,
	116	"error getting iommu info. is the iommu enabled?\n");
	117	return false;
	118	}
	119
	120	if ((iommu_info.flags & required_iommu_flags) != required_iommu_flags) {
	121	dev_err(kfd_device, "error required iommu flags ats(%i), pri(%i), pasid(%i)\n",
	122	(iommu_info.flags & AMD_IOMMU_DEVICE_FLAG_ATS_SUP) != 0,
	123	(iommu_info.flags & AMD_IOMMU_DEVICE_FLAG_PRI_SUP) != 0,
	124	(iommu_info.flags & AMD_IOMMU_DEVICE_FLAG_PASID_SUP) != 0);
	125	return false;
	126	}
	127
	128	pasid_limit = min_t(unsigned int,
	129	(unsigned int)1 << kfd->device_info->max_pasid_bits,
	130	iommu_info.max_pasids);
	131	/*
	132	* last pasid is used for kernel queues doorbells
	133	* in the future the last pasid might be used for a kernel thread.
	134	*/
	135	pasid_limit = min_t(unsigned int,
	136	pasid_limit,
	137	kfd->doorbell_process_limit - 1);
	138
	139	err = amd_iommu_init_device(kfd->pdev, pasid_limit);
	140	if (err < 0) {
	141	dev_err(kfd_device, "error initializing iommu device\n");
	142	return false;
	143	}
	144
	145	if (!kfd_set_pasid_limit(pasid_limit)) {
	146	dev_err(kfd_device, "error setting pasid limit\n");
	147	amd_iommu_free_device(kfd->pdev);
	148	return false;
	149	}
	150
	151	return true;
	152	}
	153
	154	static void iommu_pasid_shutdown_callback(struct pci_dev *pdev, int pasid)
	155	{
	156	struct kfd_dev *dev = kfd_device_by_pci_dev(pdev);
	157
	158	if (dev)
	159	kfd_unbind_process_from_device(dev, pasid);
	160	}
	161
4a488a7a OG	162	bool kgd2kfd_device_init(struct kfd_dev *kfd,
	163	const struct kgd2kfd_shared_resources *gpu_resources)
	164	{
19f6d2a6 OG	165	unsigned int size;
19f6d2a6 OG	166
4a488a7a OG	167	kfd->shared_resources = *gpu_resources;
4a488a7a OG	168
19f6d2a6 OG	169	/* calculate max size of mqds needed for queues */
	170	size = max_num_of_processes *
	171	max_num_of_queues_per_process *
	172	kfd->device_info->mqd_size_aligned;
	173
	174	/* add another 512KB for all other allocations on gart */
	175	size += 512 * 1024;
	176
	177	if (kfd2kgd->init_sa_manager(kfd->kgd, size)) {
	178	dev_err(kfd_device,
	179	"Error initializing sa manager for device (%x:%x)\n",
	180	kfd->pdev->vendor, kfd->pdev->device);
	181	goto out;
	182	}
	183
	184	kfd_doorbell_init(kfd);
	185
	186	if (kfd_topology_add_device(kfd) != 0) {
	187	dev_err(kfd_device,
	188	"Error adding device (%x:%x) to topology\n",
	189	kfd->pdev->vendor, kfd->pdev->device);
	190	goto kfd_topology_add_device_error;
	191	}
	192
b17f068a OG	193	if (!device_iommu_pasid_init(kfd)) {
	194	dev_err(kfd_device,
	195	"Error initializing iommuv2 for device (%x:%x)\n",
	196	kfd->pdev->vendor, kfd->pdev->device);
	197	goto device_iommu_pasid_error;
	198	}
	199	amd_iommu_set_invalidate_ctx_cb(kfd->pdev,
	200	iommu_pasid_shutdown_callback);
5b5c4e40	201
4a488a7a OG	202	kfd->init_complete = true;
	203	dev_info(kfd_device, "added device (%x:%x)\n", kfd->pdev->vendor,
	204	kfd->pdev->device);
	205
19f6d2a6 OG	206	goto out;
19f6d2a6 OG	207
b17f068a OG	208	device_iommu_pasid_error:
b17f068a OG	209	kfd_topology_remove_device(kfd);
19f6d2a6 OG	210	kfd_topology_add_device_error:
	211	kfd2kgd->fini_sa_manager(kfd->kgd);
	212	dev_err(kfd_device,
	213	"device (%x:%x) NOT added due to errors\n",
	214	kfd->pdev->vendor, kfd->pdev->device);
	215	out:
	216	return kfd->init_complete;
4a488a7a OG	217	}
	218
	219	void kgd2kfd_device_exit(struct kfd_dev *kfd)
	220	{
b17f068a OG	221	if (kfd->init_complete) {
	222	amd_iommu_free_device(kfd->pdev);
	223	kfd_topology_remove_device(kfd);
	224	}
5b5c4e40	225
4a488a7a OG	226	kfree(kfd);
	227	}
	228
	229	void kgd2kfd_suspend(struct kfd_dev *kfd)
	230	{
	231	BUG_ON(kfd == NULL);
b17f068a OG	232
	233	if (kfd->init_complete)
	234	amd_iommu_free_device(kfd->pdev);
4a488a7a OG	235	}
	236
	237	int kgd2kfd_resume(struct kfd_dev *kfd)
	238	{
b17f068a OG	239	unsigned int pasid_limit;
	240	int err;
	241
4a488a7a OG	242	BUG_ON(kfd == NULL);
4a488a7a OG	243
b17f068a OG	244	pasid_limit = kfd_get_pasid_limit();
	245
	246	if (kfd->init_complete) {
	247	err = amd_iommu_init_device(kfd->pdev, pasid_limit);
	248	if (err < 0)
	249	return -ENXIO;
	250	amd_iommu_set_invalidate_ctx_cb(kfd->pdev,
	251	iommu_pasid_shutdown_callback);
	252	}
	253
4a488a7a OG	254	return 0;
	255	}
	256
	257	void kgd2kfd_interrupt(struct kfd_dev dev, const void ih_ring_entry)
	258	{
	259	}