gpu: host1x: Remove wait check support

The job submission userspace ABI doesn't support this and there are no
plans to implement it, so all of this code is dead and can be removed.
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Tested-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>

drivers/gpu/drm/tegra/drm.c

@@ -321,46 +321,14 @@ static int host1x_reloc_copy_from_user(struct host1x_reloc *dest,
 	return 0;
 }
 
-static int host1x_waitchk_copy_from_user(struct host1x_waitchk *dest,
-					 struct drm_tegra_waitchk __user *src,
-					 struct drm_file *file)
-{
-	u32 cmdbuf;
-	int err;
-
-	err = get_user(cmdbuf, &src->handle);
-	if (err < 0)
-		return err;
-
-	err = get_user(dest->offset, &src->offset);
-	if (err < 0)
-		return err;
-
-	err = get_user(dest->syncpt_id, &src->syncpt);
-	if (err < 0)
-		return err;
-
-	err = get_user(dest->thresh, &src->thresh);
-	if (err < 0)
-		return err;
-
-	dest->bo = host1x_bo_lookup(file, cmdbuf);
-	if (!dest->bo)
-		return -ENOENT;
-
-	return 0;
-}
-
 int tegra_drm_submit(struct tegra_drm_context *context,
 		     struct drm_tegra_submit *args, struct drm_device *drm,
 		     struct drm_file *file)
 {
 	unsigned int num_cmdbufs = args->num_cmdbufs;
 	unsigned int num_relocs = args->num_relocs;
-	unsigned int num_waitchks = args->num_waitchks;
 	struct drm_tegra_cmdbuf __user *user_cmdbufs;
 	struct drm_tegra_reloc __user *user_relocs;
-	struct drm_tegra_waitchk __user *user_waitchks;
 	struct drm_tegra_syncpt __user *user_syncpt;
 	struct drm_tegra_syncpt syncpt;
 	struct host1x *host1x = dev_get_drvdata(drm->dev->parent);
@@ -372,7 +340,6 @@ int tegra_drm_submit(struct tegra_drm_context *context,
 
 	user_cmdbufs = u64_to_user_ptr(args->cmdbufs);
 	user_relocs = u64_to_user_ptr(args->relocs);
-	user_waitchks = u64_to_user_ptr(args->waitchks);
 	user_syncpt = u64_to_user_ptr(args->syncpts);
 
 	/* We don't yet support other than one syncpt_incr struct per submit */
@@ -384,12 +351,11 @@ int tegra_drm_submit(struct tegra_drm_context *context,
 		return -EINVAL;
 
 	job = host1x_job_alloc(context->channel, args->num_cmdbufs,
-			       args->num_relocs, args->num_waitchks);
+			       args->num_relocs);
 	if (!job)
 		return -ENOMEM;
 
 	job->num_relocs = args->num_relocs;
-	job->num_waitchk = args->num_waitchks;
 	job->client = (u32)args->context;
 	job->class = context->client->base.class;
 	job->serialize = true;
@@ -398,7 +364,7 @@ int tegra_drm_submit(struct tegra_drm_context *context,
 	 * Track referenced BOs so that they can be unreferenced after the
 	 * submission is complete.
 	 */
-	num_refs = num_cmdbufs + num_relocs * 2 + num_waitchks;
+	num_refs = num_cmdbufs + num_relocs * 2;
 
 	refs = kmalloc_array(num_refs, sizeof(*refs), GFP_KERNEL);
 	if (!refs) {
@@ -489,30 +455,6 @@ int tegra_drm_submit(struct tegra_drm_context *context,
 		}
 	}
 
-	/* copy and resolve waitchks from submit */
-	while (num_waitchks--) {
-		struct host1x_waitchk *wait = &job->waitchk[num_waitchks];
-		struct tegra_bo *obj;
-
-		err = host1x_waitchk_copy_from_user(
-			wait, &user_waitchks[num_waitchks], file);
-		if (err < 0)
-			goto fail;
-
-		obj = host1x_to_tegra_bo(wait->bo);
-		refs[num_refs++] = &obj->gem;
-
-		/*
-		 * The unaligned offset will cause an unaligned write during
-		 * of the waitchks patching, corrupting the commands stream.
-		 */
-		if (wait->offset & 3 ||
-		    wait->offset >= obj->gem.size) {
-			err = -EINVAL;
-			goto fail;
-		}
-	}
-
 	if (copy_from_user(&syncpt, user_syncpt, sizeof(syncpt))) {
 		err = -EFAULT;
 		goto fail;

drivers/gpu/host1x/dev.h

@@ -78,7 +78,6 @@ struct host1x_syncpt_ops {
 	void (*load_wait_base)(struct host1x_syncpt *syncpt);
 	u32 (*load)(struct host1x_syncpt *syncpt);
 	int (*cpu_incr)(struct host1x_syncpt *syncpt);
-	int (*patch_wait)(struct host1x_syncpt *syncpt, void *patch_addr);
 	void (*assign_to_channel)(struct host1x_syncpt *syncpt,
 	                          struct host1x_channel *channel);
 	void (*enable_protection)(struct host1x *host);
@@ -183,13 +182,6 @@ static inline int host1x_hw_syncpt_cpu_incr(struct host1x *host,
 	return host->syncpt_op->cpu_incr(sp);
 }
 
-static inline int host1x_hw_syncpt_patch_wait(struct host1x *host,
-					      struct host1x_syncpt *sp,
-					      void *patch_addr)
-{
-	return host->syncpt_op->patch_wait(sp, patch_addr);
-}
-
 static inline void host1x_hw_syncpt_assign_to_channel(
 	struct host1x *host, struct host1x_syncpt *sp,
 	struct host1x_channel *ch)

drivers/gpu/host1x/hw/channel_hw.c

@@ -104,8 +104,7 @@ static int channel_submit(struct host1x_job *job)
 	sp = host->syncpt + job->syncpt_id;
 	trace_host1x_channel_submit(dev_name(ch->dev),
 				    job->num_gathers, job->num_relocs,
-				    job->num_waitchk, job->syncpt_id,
-				    job->syncpt_incrs);
+				    job->syncpt_id, job->syncpt_incrs);
 
 	/* before error checks, return current max */
 	prev_max = job->syncpt_end = host1x_syncpt_read_max(sp);

drivers/gpu/host1x/hw/syncpt_hw.c

@@ -96,16 +96,6 @@ static int syncpt_cpu_incr(struct host1x_syncpt *sp)
 	return 0;
 }
 
-/* remove a wait pointed to by patch_addr */
-static int syncpt_patch_wait(struct host1x_syncpt *sp, void *patch_addr)
-{
-	u32 override = host1x_class_host_wait_syncpt(HOST1X_SYNCPT_RESERVED, 0);
-
-	*((u32 *)patch_addr) = override;
-
-	return 0;
-}
-
 /**
  * syncpt_assign_to_channel() - Assign syncpoint to channel
  * @sp: syncpoint
@@ -156,7 +146,6 @@ static const struct host1x_syncpt_ops host1x_syncpt_ops = {
 	.load_wait_base = syncpt_read_wait_base,
 	.load = syncpt_load,
 	.cpu_incr = syncpt_cpu_incr,
-	.patch_wait = syncpt_patch_wait,
 	.assign_to_channel = syncpt_assign_to_channel,
 	.enable_protection = syncpt_enable_protection,
 };

drivers/gpu/host1x/job.c

@@ -34,8 +34,7 @@
 #define HOST1X_WAIT_SYNCPT_OFFSET 0x8
 
 struct host1x_job *host1x_job_alloc(struct host1x_channel *ch,
-				    u32 num_cmdbufs, u32 num_relocs,
-				    u32 num_waitchks)
+				    u32 num_cmdbufs, u32 num_relocs)
 {
 	struct host1x_job *job = NULL;
 	unsigned int num_unpins = num_cmdbufs + num_relocs;
@@ -46,7 +45,6 @@ struct host1x_job *host1x_job_alloc(struct host1x_channel *ch,
 	total = sizeof(struct host1x_job) +
 		(u64)num_relocs * sizeof(struct host1x_reloc) +
 		(u64)num_unpins * sizeof(struct host1x_job_unpin_data) +
-		(u64)num_waitchks * sizeof(struct host1x_waitchk) +
 		(u64)num_cmdbufs * sizeof(struct host1x_job_gather) +
 		(u64)num_unpins * sizeof(dma_addr_t) +
 		(u64)num_unpins * sizeof(u32 *);
@@ -66,8 +64,6 @@ 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->waitchk = num_waitchks ? mem : NULL;
-	mem += num_waitchks * sizeof(struct host1x_waitchk);
 	job->gathers = num_cmdbufs ? mem : NULL;
 	mem += num_cmdbufs * sizeof(struct host1x_job_gather);
 	job->addr_phys = num_unpins ? mem : NULL;
@@ -111,73 +107,6 @@ void host1x_job_add_gather(struct host1x_job *job, struct host1x_bo *bo,
 }
 EXPORT_SYMBOL(host1x_job_add_gather);
 
-/*
- * NULL an already satisfied WAIT_SYNCPT host method, by patching its
- * args in the command stream. The method data is changed to reference
- * a reserved (never given out or incr) HOST1X_SYNCPT_RESERVED syncpt
- * with a matching threshold value of 0, so is guaranteed to be popped
- * by the host HW.
- */
-static void host1x_syncpt_patch_offset(struct host1x_syncpt *sp,
-				       struct host1x_bo *h, u32 offset)
-{
-	void *patch_addr = NULL;
-
-	/* patch the wait */
-	patch_addr = host1x_bo_kmap(h, offset >> PAGE_SHIFT);
-	if (patch_addr) {
-		host1x_syncpt_patch_wait(sp,
-					 patch_addr + (offset & ~PAGE_MASK));
-		host1x_bo_kunmap(h, offset >> PAGE_SHIFT, patch_addr);
-	} else
-		pr_err("Could not map cmdbuf for wait check\n");
-}
-
-/*
- * Check driver supplied waitchk structs for syncpt thresholds
- * that have already been satisfied and NULL the comparison (to
- * avoid a wrap condition in the HW).
- */
-static int do_waitchks(struct host1x_job *job, struct host1x *host,
-		       struct host1x_job_gather *g)
-{
-	struct host1x_bo *patch = g->bo;
-	int i;
-
-	/* compare syncpt vs wait threshold */
-	for (i = 0; i < job->num_waitchk; i++) {
-		struct host1x_waitchk *wait = &job->waitchk[i];
-		struct host1x_syncpt *sp =
-			host1x_syncpt_get(host, wait->syncpt_id);
-
-		/* validate syncpt id */
-		if (wait->syncpt_id > host1x_syncpt_nb_pts(host))
-			continue;
-
-		/* skip all other gathers */
-		if (patch != wait->bo)
-			continue;
-
-		trace_host1x_syncpt_wait_check(wait->bo, wait->offset,
-					       wait->syncpt_id, wait->thresh,
-					       host1x_syncpt_read_min(sp));
-
-		if (host1x_syncpt_is_expired(sp, wait->thresh)) {
-			dev_dbg(host->dev,
-				"drop WAIT id %u (%s) thresh 0x%x, min 0x%x\n",
-				wait->syncpt_id, sp->name, wait->thresh,
-				host1x_syncpt_read_min(sp));
-
-			host1x_syncpt_patch_offset(sp, patch,
-						   g->offset + wait->offset);
-		}
-
-		wait->bo = NULL;
-	}
-
-	return 0;
-}
-
 static unsigned int pin_job(struct host1x *host, struct host1x_job *job)
 {
 	unsigned int i;
@@ -331,17 +260,6 @@ static bool check_reloc(struct host1x_reloc *reloc, struct host1x_bo *cmdbuf,
 	return true;
 }
 
-static bool check_wait(struct host1x_waitchk *wait, struct host1x_bo *cmdbuf,
-		       unsigned int offset)
-{
-	offset *= sizeof(u32);
-
-	if (wait->bo != cmdbuf || wait->offset != offset)
-		return false;
-
-	return true;
-}
-
 struct host1x_firewall {
 	struct host1x_job *job;
 	struct device *dev;
@@ -349,9 +267,6 @@ struct host1x_firewall {
 	unsigned int num_relocs;
 	struct host1x_reloc *reloc;
 
-	unsigned int num_waitchks;
-	struct host1x_waitchk *waitchk;
-
 	struct host1x_bo *cmdbuf;
 	unsigned int offset;
 
@@ -378,20 +293,6 @@ static int check_register(struct host1x_firewall *fw, unsigned long offset)
 		fw->reloc++;
 	}
 
-	if (offset == HOST1X_WAIT_SYNCPT_OFFSET) {
-		if (fw->class != HOST1X_CLASS_HOST1X)
-			return -EINVAL;
-
-		if (!fw->num_waitchks)
-			return -EINVAL;
-
-		if (!check_wait(fw->waitchk, fw->cmdbuf, fw->offset))
-			return -EINVAL;
-
-		fw->num_waitchks--;
-		fw->waitchk++;
-	}
-
 	return 0;
 }
 
@@ -556,8 +457,6 @@ static inline int copy_gathers(struct host1x_job *job, struct device *dev)
 	fw.dev = dev;
 	fw.reloc = job->relocarray;
 	fw.num_relocs = job->num_relocs;
-	fw.waitchk = job->waitchk;
-	fw.num_waitchks = job->num_waitchk;
 	fw.class = job->class;
 
 	for (i = 0; i < job->num_gathers; i++) {
@@ -604,8 +503,8 @@ static inline int copy_gathers(struct host1x_job *job, struct device *dev)
 		offset += g->words * sizeof(u32);
 	}
 
-	/* No relocs and waitchks should remain at this point */
-	if (fw.num_relocs || fw.num_waitchks)
+	/* No relocs should remain at this point */
+	if (fw.num_relocs)
 		return -EINVAL;
 
 	return 0;
@@ -616,19 +515,6 @@ int host1x_job_pin(struct host1x_job *job, struct device *dev)
 	int err;
 	unsigned int i, j;
 	struct host1x *host = dev_get_drvdata(dev->parent);
-	DECLARE_BITMAP(waitchk_mask, host1x_syncpt_nb_pts(host));
-
-	bitmap_zero(waitchk_mask, host1x_syncpt_nb_pts(host));
-	for (i = 0; i < job->num_waitchk; i++) {
-		u32 syncpt_id = job->waitchk[i].syncpt_id;
-
-		if (syncpt_id < host1x_syncpt_nb_pts(host))
-			set_bit(syncpt_id, waitchk_mask);
-	}
-
-	/* get current syncpt values for waitchk */
-	for_each_set_bit(i, waitchk_mask, host1x_syncpt_nb_pts(host))
-		host1x_syncpt_load(host->syncpt + i);
 
 	/* pin memory */
 	err = pin_job(host, job);
@@ -663,10 +549,6 @@ int host1x_job_pin(struct host1x_job *job, struct device *dev)
 		err = do_relocs(job, g);
 		if (err)
 			break;
-
-		err = do_waitchks(job, host, g);
-		if (err)
-			break;
 	}
 
 out:

drivers/gpu/host1x/syncpt.c

@@ -373,12 +373,6 @@ bool host1x_syncpt_is_expired(struct host1x_syncpt *sp, u32 thresh)
 		return (s32)(current_val - thresh) >= 0;
 }
 
-/* remove a wait pointed to by patch_addr */
-int host1x_syncpt_patch_wait(struct host1x_syncpt *sp, void *patch_addr)
-{
-	return host1x_hw_syncpt_patch_wait(sp->host, sp, patch_addr);
-}
-
 int host1x_syncpt_init(struct host1x *host)
 {
 	struct host1x_syncpt_base *bases;

drivers/gpu/host1x/syncpt.h

@@ -124,7 +124,4 @@ static inline int host1x_syncpt_is_valid(struct host1x_syncpt *sp)
 	return sp->id < host1x_syncpt_nb_pts(sp->host);
 }
 
-/* Patch a wait by replacing it with a wait for syncpt 0 value 0 */
-int host1x_syncpt_patch_wait(struct host1x_syncpt *sp, void *patch_addr);
-
 #endif

include/linux/host1x.h

@@ -192,13 +192,6 @@ struct host1x_reloc {
 	unsigned long shift;
 };
 
-struct host1x_waitchk {
-	struct host1x_bo *bo;
-	u32 offset;
-	u32 syncpt_id;
-	u32 thresh;
-};
-
 struct host1x_job {
 	/* When refcount goes to zero, job can be freed */
 	struct kref ref;
@@ -215,11 +208,6 @@ struct host1x_job {
 	struct host1x_job_gather *gathers;
 	unsigned int num_gathers;
 
-	/* Wait checks to be processed at submit time */
-	struct host1x_waitchk *waitchk;
-	unsigned int num_waitchk;
-	u32 waitchk_mask;
-
 	/* Array of handles to be pinned & unpinned */
 	struct host1x_reloc *relocarray;
 	unsigned int num_relocs;
@@ -261,8 +249,7 @@ struct host1x_job {
 };
 
 struct host1x_job *host1x_job_alloc(struct host1x_channel *ch,
-				    u32 num_cmdbufs, u32 num_relocs,
-				    u32 num_waitchks);
+				    u32 num_cmdbufs, u32 num_relocs);
 void host1x_job_add_gather(struct host1x_job *job, struct host1x_bo *mem_id,
 			   u32 words, u32 offset);
 struct host1x_job *host1x_job_get(struct host1x_job *job);

include/trace/events/host1x.h

@@ -115,16 +115,15 @@ TRACE_EVENT(host1x_cdma_push_gather,
 );
 
 TRACE_EVENT(host1x_channel_submit,
-	TP_PROTO(const char *name, u32 cmdbufs, u32 relocs, u32 waitchks,
-			u32 syncpt_id, u32 syncpt_incrs),
+	TP_PROTO(const char *name, u32 cmdbufs, u32 relocs, u32 syncpt_id,
+		 u32 syncpt_incrs),
 
-	TP_ARGS(name, cmdbufs, relocs, waitchks, syncpt_id, syncpt_incrs),
+	TP_ARGS(name, cmdbufs, relocs, syncpt_id, syncpt_incrs),
 
 	TP_STRUCT__entry(
 		__field(const char *, name)
 		__field(u32, cmdbufs)
 		__field(u32, relocs)
-		__field(u32, waitchks)
 		__field(u32, syncpt_id)
 		__field(u32, syncpt_incrs)
 	),
@@ -133,15 +132,14 @@ TRACE_EVENT(host1x_channel_submit,
 		__entry->name = name;
 		__entry->cmdbufs = cmdbufs;
 		__entry->relocs = relocs;
-		__entry->waitchks = waitchks;
 		__entry->syncpt_id = syncpt_id;
 		__entry->syncpt_incrs = syncpt_incrs;
 	),
 
-	TP_printk("name=%s, cmdbufs=%u, relocs=%u, waitchks=%d,"
-		"syncpt_id=%u, syncpt_incrs=%u",
-	  __entry->name, __entry->cmdbufs, __entry->relocs, __entry->waitchks,
-	  __entry->syncpt_id, __entry->syncpt_incrs)
+	TP_printk("name=%s, cmdbufs=%u, relocs=%u, syncpt_id=%u, "
+		  "syncpt_incrs=%u",
+		  __entry->name, __entry->cmdbufs, __entry->relocs,
+		  __entry->syncpt_id, __entry->syncpt_incrs)
 );
 
 TRACE_EVENT(host1x_channel_submitted,