1 files changed, 176 insertions, 89 deletions

diff --git a/drivers/gpu/host1x/job.c b/drivers/gpu/host1x/job.c
index 527a1cddb14f..3ed49e1fd933 100644
--- a/drivers/gpu/host1x/job.c
+++ b/drivers/gpu/host1x/job.c
@@ -1,24 +1,14 @@
+// SPDX-License-Identifier: GPL-2.0-only
 /*
  * Tegra host1x Job
  *
  * Copyright (c) 2010-2015, NVIDIA Corporation.
- *
- * This program is free software; you can redistribute it and/or modify it
- * under the terms and conditions of the GNU General Public License,
- * version 2, as published by the Free Software Foundation.
- *
- * This program is distributed in the hope it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
- * more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program.  If not, see <http://www.gnu.org/licenses/>.
  */
 
 #include <linux/dma-mapping.h>
 #include <linux/err.h>
 #include <linux/host1x.h>
+#include <linux/iommu.h>
 #include <linux/kref.h>
 #include <linux/module.h>
 #include <linux/scatterlist.h>
@@ -34,18 +24,25 @@
 #define HOST1X_WAIT_SYNCPT_OFFSET 0x8
 
 struct host1x_job *host1x_job_alloc(struct host1x_channel *ch,
-				    u32 num_cmdbufs, u32 num_relocs)
+				    u32 num_cmdbufs, u32 num_relocs,
+				    bool skip_firewall)
 {
 	struct host1x_job *job = NULL;
-	unsigned int num_unpins = num_cmdbufs + num_relocs;
+	unsigned int num_unpins = num_relocs;
+	bool enable_firewall;
 	u64 total;
 	void *mem;
 
+	enable_firewall = IS_ENABLED(CONFIG_TEGRA_HOST1X_FIREWALL) && !skip_firewall;
+
+	if (!enable_firewall)
+		num_unpins += num_cmdbufs;
+
 	/* Check that we're not going to overflow */
 	total = sizeof(struct host1x_job) +
 		(u64)num_relocs * sizeof(struct host1x_reloc) +
 		(u64)num_unpins * sizeof(struct host1x_job_unpin_data) +
-		(u64)num_cmdbufs * sizeof(struct host1x_job_gather) +
+		(u64)num_cmdbufs * sizeof(struct host1x_job_cmd) +
 		(u64)num_unpins * sizeof(dma_addr_t) +
 		(u64)num_unpins * sizeof(u32 *);
 	if (total > ULONG_MAX)
@@ -55,6 +52,8 @@ struct host1x_job *host1x_job_alloc(struct host1x_channel *ch,
 	if (!job)
 		return NULL;
 
+	job->enable_firewall = enable_firewall;
+
 	kref_init(&job->ref);
 	job->channel = ch;
 
@@ -64,8 +63,8 @@ struct host1x_job *host1x_job_alloc(struct host1x_channel *ch,
 	mem += num_relocs * sizeof(struct host1x_reloc);
 	job->unpins = num_unpins ? mem : NULL;
 	mem += num_unpins * sizeof(struct host1x_job_unpin_data);
-	job->gathers = num_cmdbufs ? mem : NULL;
-	mem += num_cmdbufs * sizeof(struct host1x_job_gather);
+	job->cmds = num_cmdbufs ? mem : NULL;
+	mem += num_cmdbufs * sizeof(struct host1x_job_cmd);
 	job->addr_phys = num_unpins ? mem : NULL;
 
 	job->reloc_addr_phys = job->addr_phys;
@@ -86,6 +85,22 @@ static void job_free(struct kref *ref)
 {
 	struct host1x_job *job = container_of(ref, struct host1x_job, ref);
 
+	if (job->release)
+		job->release(job);
+
+	if (job->fence) {
+		/*
+		 * remove_callback is atomic w.r.t. fence signaling, so
+		 * after the call returns, we know that the callback is not
+		 * in execution, and the fence can be safely freed.
+		 */
+		dma_fence_remove_callback(job->fence, &job->fence_cb);
+		dma_fence_put(job->fence);
+	}
+
+	if (job->syncpt)
+		host1x_syncpt_put(job->syncpt);
+
 	kfree(job);
 }
 
@@ -98,18 +113,37 @@ EXPORT_SYMBOL(host1x_job_put);
 void host1x_job_add_gather(struct host1x_job *job, struct host1x_bo *bo,
 			   unsigned int words, unsigned int offset)
 {
-	struct host1x_job_gather *gather = &job->gathers[job->num_gathers];
+	struct host1x_job_gather *gather = &job->cmds[job->num_cmds].gather;
 
 	gather->words = words;
 	gather->bo = bo;
 	gather->offset = offset;
 
-	job->num_gathers++;
+	job->num_cmds++;
 }
 EXPORT_SYMBOL(host1x_job_add_gather);
 
+void host1x_job_add_wait(struct host1x_job *job, u32 id, u32 thresh,
+			 bool relative, u32 next_class)
+{
+	struct host1x_job_cmd *cmd = &job->cmds[job->num_cmds];
+
+	cmd->is_wait = true;
+	cmd->wait.id = id;
+	cmd->wait.threshold = thresh;
+	cmd->wait.next_class = next_class;
+	cmd->wait.relative = relative;
+
+	job->num_cmds++;
+}
+EXPORT_SYMBOL(host1x_job_add_wait);
+
 static unsigned int pin_job(struct host1x *host, struct host1x_job *job)
 {
+	unsigned long mask = HOST1X_RELOC_READ | HOST1X_RELOC_WRITE;
+	struct host1x_client *client = job->client;
+	struct device *dev = client->dev;
+	struct host1x_job_gather *g;
 	unsigned int i;
 	int err;
 
@@ -117,8 +151,9 @@ static unsigned int pin_job(struct host1x *host, struct host1x_job *job)
 
 	for (i = 0; i < job->num_relocs; i++) {
 		struct host1x_reloc *reloc = &job->relocs[i];
-		struct sg_table *sgt;
-		dma_addr_t phys_addr;
+		enum dma_data_direction direction;
+		struct host1x_bo_mapping *map;
+		struct host1x_bo *bo;
 
 		reloc->target.bo = host1x_bo_get(reloc->target.bo);
 		if (!reloc->target.bo) {
@@ -126,35 +161,83 @@ static unsigned int pin_job(struct host1x *host, struct host1x_job *job)
 			goto unpin;
 		}
 
-		phys_addr = host1x_bo_pin(reloc->target.bo, &sgt);
+		bo = reloc->target.bo;
 
-		job->addr_phys[job->num_unpins] = phys_addr;
-		job->unpins[job->num_unpins].bo = reloc->target.bo;
-		job->unpins[job->num_unpins].sgt = sgt;
+		switch (reloc->flags & mask) {
+		case HOST1X_RELOC_READ:
+			direction = DMA_TO_DEVICE;
+			break;
+
+		case HOST1X_RELOC_WRITE:
+			direction = DMA_FROM_DEVICE;
+			break;
+
+		case HOST1X_RELOC_READ | HOST1X_RELOC_WRITE:
+			direction = DMA_BIDIRECTIONAL;
+			break;
+
+		default:
+			err = -EINVAL;
+			goto unpin;
+		}
+
+		map = host1x_bo_pin(dev, bo, direction, NULL);
+		if (IS_ERR(map)) {
+			err = PTR_ERR(map);
+			goto unpin;
+		}
+
+		/*
+		 * host1x clients are generally not able to do scatter-gather themselves, so fail
+		 * if the buffer is discontiguous and we fail to map its SG table to a single
+		 * contiguous chunk of I/O virtual memory.
+		 */
+		if (map->chunks > 1) {
+			err = -EINVAL;
+			goto unpin;
+		}
+
+		job->addr_phys[job->num_unpins] = map->phys;
+		job->unpins[job->num_unpins].map = map;
 		job->num_unpins++;
 	}
 
-	for (i = 0; i < job->num_gathers; i++) {
-		struct host1x_job_gather *g = &job->gathers[i];
+	/*
+	 * We will copy gathers BO content later, so there is no need to
+	 * hold and pin them.
+	 */
+	if (job->enable_firewall)
+		return 0;
+
+	for (i = 0; i < job->num_cmds; i++) {
+		struct host1x_bo_mapping *map;
 		size_t gather_size = 0;
 		struct scatterlist *sg;
-		struct sg_table *sgt;
-		dma_addr_t phys_addr;
 		unsigned long shift;
 		struct iova *alloc;
 		unsigned int j;
 
+		if (job->cmds[i].is_wait)
+			continue;
+
+		g = &job->cmds[i].gather;
+
 		g->bo = host1x_bo_get(g->bo);
 		if (!g->bo) {
 			err = -EINVAL;
 			goto unpin;
 		}
 
-		phys_addr = host1x_bo_pin(g->bo, &sgt);
+		map = host1x_bo_pin(host->dev, g->bo, DMA_TO_DEVICE, NULL);
+		if (IS_ERR(map)) {
+			err = PTR_ERR(map);
+			goto unpin;
+		}
 
-		if (!IS_ENABLED(CONFIG_TEGRA_HOST1X_FIREWALL) && host->domain) {
-			for_each_sg(sgt->sgl, sg, sgt->nents, j)
+		if (host->domain) {
+			for_each_sgtable_sg(map->sgt, sg, j)
 				gather_size += sg->length;
+
 			gather_size = iova_align(&host->iova, gather_size);
 
 			shift = iova_shift(&host->iova);
@@ -162,34 +245,32 @@ static unsigned int pin_job(struct host1x *host, struct host1x_job *job)
 					   host->iova_end >> shift, true);
 			if (!alloc) {
 				err = -ENOMEM;
-				goto unpin;
+				goto put;
 			}
 
-			err = iommu_map_sg(host->domain,
-					iova_dma_addr(&host->iova, alloc),
-					sgt->sgl, sgt->nents, IOMMU_READ);
+			err = iommu_map_sgtable(host->domain, iova_dma_addr(&host->iova, alloc),
+						map->sgt, IOMMU_READ);
 			if (err == 0) {
 				__free_iova(&host->iova, alloc);
 				err = -EINVAL;
-				goto unpin;
+				goto put;
 			}
 
-			job->addr_phys[job->num_unpins] =
-				iova_dma_addr(&host->iova, alloc);
-			job->unpins[job->num_unpins].size = gather_size;
-		} else {
-			job->addr_phys[job->num_unpins] = phys_addr;
+			map->phys = iova_dma_addr(&host->iova, alloc);
+			map->size = gather_size;
 		}
 
-		job->gather_addr_phys[i] = job->addr_phys[job->num_unpins];
-
-		job->unpins[job->num_unpins].bo = g->bo;
-		job->unpins[job->num_unpins].sgt = sgt;
+		job->addr_phys[job->num_unpins] = map->phys;
+		job->unpins[job->num_unpins].map = map;
 		job->num_unpins++;
+
+		job->gather_addr_phys[i] = map->phys;
 	}
 
 	return 0;
 
+put:
+	host1x_bo_put(g->bo);
 unpin:
 	host1x_job_unpin(job);
 	return err;
@@ -197,8 +278,7 @@ unpin:
 
 static int do_relocs(struct host1x_job *job, struct host1x_job_gather *g)
 {
-	u32 last_page = ~0;
-	void *cmdbuf_page_addr = NULL;
+	void *cmdbuf_addr = NULL;
 	struct host1x_bo *cmdbuf = g->bo;
 	unsigned int i;
 
@@ -213,35 +293,29 @@ static int do_relocs(struct host1x_job *job, struct host1x_job_gather *g)
 		if (cmdbuf != reloc->cmdbuf.bo)
 			continue;
 
-		if (IS_ENABLED(CONFIG_TEGRA_HOST1X_FIREWALL)) {
+		if (job->enable_firewall) {
 			target = (u32 *)job->gather_copy_mapped +
 					reloc->cmdbuf.offset / sizeof(u32) +
 						g->offset / sizeof(u32);
 			goto patch_reloc;
 		}
 
-		if (last_page != reloc->cmdbuf.offset >> PAGE_SHIFT) {
-			if (cmdbuf_page_addr)
-				host1x_bo_kunmap(cmdbuf, last_page,
-						 cmdbuf_page_addr);
+		if (!cmdbuf_addr) {
+			cmdbuf_addr = host1x_bo_mmap(cmdbuf);
 
-			cmdbuf_page_addr = host1x_bo_kmap(cmdbuf,
-					reloc->cmdbuf.offset >> PAGE_SHIFT);
-			last_page = reloc->cmdbuf.offset >> PAGE_SHIFT;
-
-			if (unlikely(!cmdbuf_page_addr)) {
+			if (unlikely(!cmdbuf_addr)) {
 				pr_err("Could not map cmdbuf for relocation\n");
 				return -ENOMEM;
 			}
 		}
 
-		target = cmdbuf_page_addr + (reloc->cmdbuf.offset & ~PAGE_MASK);
+		target = cmdbuf_addr + reloc->cmdbuf.offset;
 patch_reloc:
 		*target = reloc_addr;
 	}
 
-	if (cmdbuf_page_addr)
-		host1x_bo_kunmap(cmdbuf, last_page, cmdbuf_page_addr);
+	if (cmdbuf_addr)
+		host1x_bo_munmap(cmdbuf, cmdbuf_addr);
 
 	return 0;
 }
@@ -447,7 +521,8 @@ out:
 	return err;
 }
 
-static inline int copy_gathers(struct host1x_job *job, struct device *dev)
+static inline int copy_gathers(struct device *host, struct host1x_job *job,
+			       struct device *dev)
 {
 	struct host1x_firewall fw;
 	size_t size = 0;
@@ -460,8 +535,13 @@ static inline int copy_gathers(struct host1x_job *job, struct device *dev)
 	fw.num_relocs = job->num_relocs;
 	fw.class = job->class;
 
-	for (i = 0; i < job->num_gathers; i++) {
-		struct host1x_job_gather *g = &job->gathers[i];
+	for (i = 0; i < job->num_cmds; i++) {
+		struct host1x_job_gather *g;
+
+		if (job->cmds[i].is_wait)
+			continue;
+
+		g = &job->cmds[i].gather;
 
 		size += g->words * sizeof(u32);
 	}
@@ -470,12 +550,12 @@ static inline int copy_gathers(struct host1x_job *job, struct device *dev)
 	 * Try a non-blocking allocation from a higher priority pools first,
 	 * as awaiting for the allocation here is a major performance hit.
 	 */
-	job->gather_copy_mapped = dma_alloc_wc(dev, size, &job->gather_copy,
+	job->gather_copy_mapped = dma_alloc_wc(host, size, &job->gather_copy,
 					       GFP_NOWAIT);
 
 	/* the higher priority allocation failed, try the generic-blocking */
 	if (!job->gather_copy_mapped)
-		job->gather_copy_mapped = dma_alloc_wc(dev, size,
+		job->gather_copy_mapped = dma_alloc_wc(host, size,
 						       &job->gather_copy,
 						       GFP_KERNEL);
 	if (!job->gather_copy_mapped)
@@ -483,10 +563,14 @@ static inline int copy_gathers(struct host1x_job *job, struct device *dev)
 
 	job->gather_copy_size = size;
 
-	for (i = 0; i < job->num_gathers; i++) {
-		struct host1x_job_gather *g = &job->gathers[i];
+	for (i = 0; i < job->num_cmds; i++) {
+		struct host1x_job_gather *g;
 		void *gather;
 
+		if (job->cmds[i].is_wait)
+			continue;
+		g = &job->cmds[i].gather;
+
 		/* Copy the gather */
 		gather = host1x_bo_mmap(g->bo);
 		memcpy(job->gather_copy_mapped + offset, gather + g->offset,
@@ -522,28 +606,33 @@ int host1x_job_pin(struct host1x_job *job, struct device *dev)
 	if (err)
 		goto out;
 
-	if (IS_ENABLED(CONFIG_TEGRA_HOST1X_FIREWALL)) {
-		err = copy_gathers(job, dev);
+	if (job->enable_firewall) {
+		err = copy_gathers(host->dev, job, dev);
 		if (err)
 			goto out;
 	}
 
 	/* patch gathers */
-	for (i = 0; i < job->num_gathers; i++) {
-		struct host1x_job_gather *g = &job->gathers[i];
+	for (i = 0; i < job->num_cmds; i++) {
+		struct host1x_job_gather *g;
+
+		if (job->cmds[i].is_wait)
+			continue;
+		g = &job->cmds[i].gather;
 
 		/* process each gather mem only once */
 		if (g->handled)
 			continue;
 
 		/* copy_gathers() sets gathers base if firewall is enabled */
-		if (!IS_ENABLED(CONFIG_TEGRA_HOST1X_FIREWALL))
+		if (!job->enable_firewall)
 			g->base = job->gather_addr_phys[i];
 
-		for (j = i + 1; j < job->num_gathers; j++) {
-			if (job->gathers[j].bo == g->bo) {
-				job->gathers[j].handled = true;
-				job->gathers[j].base = g->base;
+		for (j = i + 1; j < job->num_cmds; j++) {
+			if (!job->cmds[j].is_wait &&
+			    job->cmds[j].gather.bo == g->bo) {
+				job->cmds[j].gather.handled = true;
+				job->cmds[j].gather.base = g->base;
 			}
 		}
 
@@ -567,24 +656,22 @@ void host1x_job_unpin(struct host1x_job *job)
 	unsigned int i;
 
 	for (i = 0; i < job->num_unpins; i++) {
-		struct host1x_job_unpin_data *unpin = &job->unpins[i];
-
-		if (!IS_ENABLED(CONFIG_TEGRA_HOST1X_FIREWALL) &&
-		    unpin->size && host->domain) {
-			iommu_unmap(host->domain, job->addr_phys[i],
-				    unpin->size);
-			free_iova(&host->iova,
-				iova_pfn(&host->iova, job->addr_phys[i]));
+		struct host1x_bo_mapping *map = job->unpins[i].map;
+		struct host1x_bo *bo = map->bo;
+
+		if (!job->enable_firewall && map->size && host->domain) {
+			iommu_unmap(host->domain, job->addr_phys[i], map->size);
+			free_iova(&host->iova, iova_pfn(&host->iova, job->addr_phys[i]));
 		}
 
-		host1x_bo_unpin(unpin->bo, unpin->sgt);
-		host1x_bo_put(unpin->bo);
+		host1x_bo_unpin(map);
+		host1x_bo_put(bo);
 	}
 
 	job->num_unpins = 0;
 
 	if (job->gather_copy_size)
-		dma_free_wc(job->channel->dev, job->gather_copy_size,
+		dma_free_wc(host->dev, job->gather_copy_size,
 			    job->gather_copy_mapped, job->gather_copy);
 }
 EXPORT_SYMBOL(host1x_job_unpin);
@@ -594,7 +681,7 @@ EXPORT_SYMBOL(host1x_job_unpin);
  */
 void host1x_job_dump(struct device *dev, struct host1x_job *job)
 {
-	dev_dbg(dev, "    SYNCPT_ID   %d\n", job->syncpt_id);
+	dev_dbg(dev, "    SYNCPT_ID   %d\n", job->syncpt->id);
 	dev_dbg(dev, "    SYNCPT_VAL  %d\n", job->syncpt_end);
 	dev_dbg(dev, "    FIRST_GET   0x%x\n", job->first_get);
 	dev_dbg(dev, "    TIMEOUT     %d\n", job->timeout);