diff options
Diffstat (limited to 'drivers/gpu/host1x')
| -rw-r--r-- | drivers/gpu/host1x/Kconfig | 1 | ||||
| -rw-r--r-- | drivers/gpu/host1x/bus.c | 80 | ||||
| -rw-r--r-- | drivers/gpu/host1x/channel.c | 8 | ||||
| -rw-r--r-- | drivers/gpu/host1x/debug.c | 15 | ||||
| -rw-r--r-- | drivers/gpu/host1x/dev.c | 185 | ||||
| -rw-r--r-- | drivers/gpu/host1x/dev.h | 5 | ||||
| -rw-r--r-- | drivers/gpu/host1x/hw/channel_hw.c | 44 | ||||
| -rw-r--r-- | drivers/gpu/host1x/intr.c | 3 | ||||
| -rw-r--r-- | drivers/gpu/host1x/job.c | 160 | ||||
| -rw-r--r-- | drivers/gpu/host1x/job.h | 6 | ||||
| -rw-r--r-- | drivers/gpu/host1x/syncpt.c | 5 |
11 files changed, 347 insertions, 165 deletions
diff --git a/drivers/gpu/host1x/Kconfig b/drivers/gpu/host1x/Kconfig index 6dab94adf25e..6815b4db17c1 100644 --- a/drivers/gpu/host1x/Kconfig +++ b/drivers/gpu/host1x/Kconfig @@ -2,6 +2,7 @@ config TEGRA_HOST1X tristate "NVIDIA Tegra host1x driver" depends on ARCH_TEGRA || (ARM && COMPILE_TEST) + select DMA_SHARED_BUFFER select IOMMU_IOVA help Driver for the NVIDIA Tegra host1x hardware. diff --git a/drivers/gpu/host1x/bus.c b/drivers/gpu/host1x/bus.c index 218e3718fd68..bdee16a0bb8e 100644 --- a/drivers/gpu/host1x/bus.c +++ b/drivers/gpu/host1x/bus.c @@ -5,6 +5,7 @@ */ #include <linux/debugfs.h> +#include <linux/dma-mapping.h> #include <linux/host1x.h> #include <linux/of.h> #include <linux/seq_file.h> @@ -742,6 +743,7 @@ EXPORT_SYMBOL(host1x_driver_unregister); */ void __host1x_client_init(struct host1x_client *client, struct lock_class_key *key) { + host1x_bo_cache_init(&client->cache); INIT_LIST_HEAD(&client->list); __mutex_init(&client->lock, "host1x client lock", key); client->usecount = 0; @@ -761,7 +763,6 @@ EXPORT_SYMBOL(host1x_client_exit); /** * __host1x_client_register() - register a host1x client * @client: host1x client - * @key: lock class key for the client-specific mutex * * Registers a host1x client with each host1x controller instance. Note that * each client will only match their parent host1x controller and will only be @@ -830,6 +831,8 @@ int host1x_client_unregister(struct host1x_client *client) mutex_unlock(&clients_lock); + host1x_bo_cache_destroy(&client->cache); + return 0; } EXPORT_SYMBOL(host1x_client_unregister); @@ -904,3 +907,78 @@ unlock: return err; } EXPORT_SYMBOL(host1x_client_resume); + +struct host1x_bo_mapping *host1x_bo_pin(struct device *dev, struct host1x_bo *bo, + enum dma_data_direction dir, + struct host1x_bo_cache *cache) +{ + struct host1x_bo_mapping *mapping; + + if (cache) { + mutex_lock(&cache->lock); + + list_for_each_entry(mapping, &cache->mappings, entry) { + if (mapping->bo == bo && mapping->direction == dir) { + kref_get(&mapping->ref); + goto unlock; + } + } + } + + mapping = bo->ops->pin(dev, bo, dir); + if (IS_ERR(mapping)) + goto unlock; + + spin_lock(&mapping->bo->lock); + list_add_tail(&mapping->list, &bo->mappings); + spin_unlock(&mapping->bo->lock); + + if (cache) { + INIT_LIST_HEAD(&mapping->entry); + mapping->cache = cache; + + list_add_tail(&mapping->entry, &cache->mappings); + + /* bump reference count to track the copy in the cache */ + kref_get(&mapping->ref); + } + +unlock: + if (cache) + mutex_unlock(&cache->lock); + + return mapping; +} +EXPORT_SYMBOL(host1x_bo_pin); + +static void __host1x_bo_unpin(struct kref *ref) +{ + struct host1x_bo_mapping *mapping = to_host1x_bo_mapping(ref); + + /* + * When the last reference of the mapping goes away, make sure to remove the mapping from + * the cache. + */ + if (mapping->cache) + list_del(&mapping->entry); + + spin_lock(&mapping->bo->lock); + list_del(&mapping->list); + spin_unlock(&mapping->bo->lock); + + mapping->bo->ops->unpin(mapping); +} + +void host1x_bo_unpin(struct host1x_bo_mapping *mapping) +{ + struct host1x_bo_cache *cache = mapping->cache; + + if (cache) + mutex_lock(&cache->lock); + + kref_put(&mapping->ref, __host1x_bo_unpin); + + if (cache) + mutex_unlock(&cache->lock); +} +EXPORT_SYMBOL(host1x_bo_unpin); diff --git a/drivers/gpu/host1x/channel.c b/drivers/gpu/host1x/channel.c index 4cd212bb570d..2a9a3a8d5931 100644 --- a/drivers/gpu/host1x/channel.c +++ b/drivers/gpu/host1x/channel.c @@ -75,6 +75,14 @@ struct host1x_channel *host1x_channel_get_index(struct host1x *host, return ch; } +void host1x_channel_stop(struct host1x_channel *channel) +{ + struct host1x *host = dev_get_drvdata(channel->dev->parent); + + host1x_hw_cdma_stop(host, &channel->cdma); +} +EXPORT_SYMBOL(host1x_channel_stop); + static void release_channel(struct kref *kref) { struct host1x_channel *channel = diff --git a/drivers/gpu/host1x/debug.c b/drivers/gpu/host1x/debug.c index 8a14880c61bb..18d9c8d206e3 100644 --- a/drivers/gpu/host1x/debug.c +++ b/drivers/gpu/host1x/debug.c @@ -7,6 +7,7 @@ */ #include <linux/debugfs.h> +#include <linux/pm_runtime.h> #include <linux/seq_file.h> #include <linux/uaccess.h> @@ -52,6 +53,11 @@ static int show_channel(struct host1x_channel *ch, void *data, bool show_fifo) { struct host1x *m = dev_get_drvdata(ch->dev->parent); struct output *o = data; + int err; + + err = pm_runtime_resume_and_get(m->dev); + if (err < 0) + return err; mutex_lock(&ch->cdma.lock); mutex_lock(&debug_lock); @@ -64,6 +70,8 @@ static int show_channel(struct host1x_channel *ch, void *data, bool show_fifo) mutex_unlock(&debug_lock); mutex_unlock(&ch->cdma.lock); + pm_runtime_put(m->dev); + return 0; } @@ -71,9 +79,14 @@ static void show_syncpts(struct host1x *m, struct output *o) { struct list_head *pos; unsigned int i; + int err; host1x_debug_output(o, "---- syncpts ----\n"); + err = pm_runtime_resume_and_get(m->dev); + if (err < 0) + return; + for (i = 0; i < host1x_syncpt_nb_pts(m); i++) { u32 max = host1x_syncpt_read_max(m->syncpt + i); u32 min = host1x_syncpt_load(m->syncpt + i); @@ -101,6 +114,8 @@ static void show_syncpts(struct host1x *m, struct output *o) base_val); } + pm_runtime_put(m->dev); + host1x_debug_output(o, "\n"); } diff --git a/drivers/gpu/host1x/dev.c b/drivers/gpu/host1x/dev.c index fbb6447b8659..6994f8c0e02e 100644 --- a/drivers/gpu/host1x/dev.c +++ b/drivers/gpu/host1x/dev.c @@ -6,18 +6,26 @@ */ #include <linux/clk.h> +#include <linux/delay.h> #include <linux/dma-mapping.h> #include <linux/io.h> #include <linux/list.h> #include <linux/module.h> #include <linux/of_device.h> #include <linux/of.h> +#include <linux/pm_runtime.h> #include <linux/slab.h> +#include <soc/tegra/common.h> + #define CREATE_TRACE_POINTS #include <trace/events/host1x.h> #undef CREATE_TRACE_POINTS +#if IS_ENABLED(CONFIG_ARM_DMA_USE_IOMMU) +#include <asm/dma-iommu.h> +#endif + #include "bus.h" #include "channel.h" #include "debug.h" @@ -132,6 +140,12 @@ static const struct host1x_sid_entry tegra186_sid_table[] = { .offset = 0x30, .limit = 0x34 }, + { + /* NVDEC */ + .base = 0x1b00, + .offset = 0x30, + .limit = 0x34 + }, }; static const struct host1x_info host1x06_info = { @@ -156,6 +170,18 @@ static const struct host1x_sid_entry tegra194_sid_table[] = { .offset = 0x30, .limit = 0x34 }, + { + /* NVDEC */ + .base = 0x1b00, + .offset = 0x30, + .limit = 0x34 + }, + { + /* NVDEC1 */ + .base = 0x1bc0, + .offset = 0x30, + .limit = 0x34 + }, }; static const struct host1x_info host1x07_info = { @@ -190,6 +216,9 @@ static void host1x_setup_sid_table(struct host1x *host) const struct host1x_info *info = host->info; unsigned int i; + if (!info->has_hypervisor) + return; + for (i = 0; i < info->num_sid_entries; i++) { const struct host1x_sid_entry *entry = &info->sid_table[i]; @@ -238,6 +267,17 @@ static struct iommu_domain *host1x_iommu_attach(struct host1x *host) struct iommu_domain *domain = iommu_get_domain_for_dev(host->dev); int err; +#if IS_ENABLED(CONFIG_ARM_DMA_USE_IOMMU) + if (host->dev->archdata.mapping) { + struct dma_iommu_mapping *mapping = + to_dma_iommu_mapping(host->dev); + arm_iommu_detach_device(host->dev); + arm_iommu_release_mapping(mapping); + + domain = iommu_get_domain_for_dev(host->dev); + } +#endif + /* * We may not always want to enable IOMMU support (for example if the * host1x firewall is already enabled and we don't support addressing @@ -347,6 +387,27 @@ static void host1x_iommu_exit(struct host1x *host) } } +static int host1x_get_resets(struct host1x *host) +{ + int err; + + host->resets[0].id = "mc"; + host->resets[1].id = "host1x"; + host->nresets = ARRAY_SIZE(host->resets); + + err = devm_reset_control_bulk_get_optional_exclusive_released( + host->dev, host->nresets, host->resets); + if (err) { + dev_err(host->dev, "failed to get reset: %d\n", err); + return err; + } + + if (WARN_ON(!host->resets[1].rstc)) + return -ENOENT; + + return 0; +} + static int host1x_probe(struct platform_device *pdev) { struct host1x *host; @@ -386,6 +447,7 @@ static int host1x_probe(struct platform_device *pdev) if (syncpt_irq < 0) return syncpt_irq; + host1x_bo_cache_init(&host->cache); mutex_init(&host->devices_lock); INIT_LIST_HEAD(&host->devices); INIT_LIST_HEAD(&host->list); @@ -423,12 +485,9 @@ static int host1x_probe(struct platform_device *pdev) return err; } - host->rst = devm_reset_control_get(&pdev->dev, "host1x"); - if (IS_ERR(host->rst)) { - err = PTR_ERR(host->rst); - dev_err(&pdev->dev, "failed to get reset: %d\n", err); + err = host1x_get_resets(host); + if (err) return err; - } err = host1x_iommu_init(host); if (err < 0) { @@ -443,22 +502,10 @@ static int host1x_probe(struct platform_device *pdev) goto iommu_exit; } - err = clk_prepare_enable(host->clk); - if (err < 0) { - dev_err(&pdev->dev, "failed to enable clock\n"); - goto free_channels; - } - - err = reset_control_deassert(host->rst); - if (err < 0) { - dev_err(&pdev->dev, "failed to deassert reset: %d\n", err); - goto unprepare_disable; - } - err = host1x_syncpt_init(host); if (err) { dev_err(&pdev->dev, "failed to initialize syncpts\n"); - goto reset_assert; + goto free_channels; } err = host1x_intr_init(host, syncpt_irq); @@ -467,10 +514,18 @@ static int host1x_probe(struct platform_device *pdev) goto deinit_syncpt; } - host1x_debug_init(host); + pm_runtime_enable(&pdev->dev); + + err = devm_tegra_core_dev_init_opp_table_common(&pdev->dev); + if (err) + goto pm_disable; - if (host->info->has_hypervisor) - host1x_setup_sid_table(host); + /* the driver's code isn't ready yet for the dynamic RPM */ + err = pm_runtime_resume_and_get(&pdev->dev); + if (err) + goto pm_disable; + + host1x_debug_init(host); err = host1x_register(host); if (err < 0) @@ -486,13 +541,14 @@ unregister: host1x_unregister(host); deinit_debugfs: host1x_debug_deinit(host); + + pm_runtime_put_sync_suspend(&pdev->dev); +pm_disable: + pm_runtime_disable(&pdev->dev); + host1x_intr_deinit(host); deinit_syncpt: host1x_syncpt_deinit(host); -reset_assert: - reset_control_assert(host->rst); -unprepare_disable: - clk_disable_unprepare(host->clk); free_channels: host1x_channel_list_free(&host->channel_list); iommu_exit: @@ -507,19 +563,94 @@ static int host1x_remove(struct platform_device *pdev) host1x_unregister(host); host1x_debug_deinit(host); + + pm_runtime_force_suspend(&pdev->dev); + host1x_intr_deinit(host); host1x_syncpt_deinit(host); - reset_control_assert(host->rst); - clk_disable_unprepare(host->clk); host1x_iommu_exit(host); + host1x_bo_cache_destroy(&host->cache); return 0; } +static int __maybe_unused host1x_runtime_suspend(struct device *dev) +{ + struct host1x *host = dev_get_drvdata(dev); + int err; + + host1x_intr_stop(host); + host1x_syncpt_save(host); + + err = reset_control_bulk_assert(host->nresets, host->resets); + if (err) { + dev_err(dev, "failed to assert reset: %d\n", err); + goto resume_host1x; + } + + usleep_range(1000, 2000); + + clk_disable_unprepare(host->clk); + reset_control_bulk_release(host->nresets, host->resets); + + return 0; + +resume_host1x: + host1x_setup_sid_table(host); + host1x_syncpt_restore(host); + host1x_intr_start(host); + + return err; +} + +static int __maybe_unused host1x_runtime_resume(struct device *dev) +{ + struct host1x *host = dev_get_drvdata(dev); + int err; + + err = reset_control_bulk_acquire(host->nresets, host->resets); + if (err) { + dev_err(dev, "failed to acquire reset: %d\n", err); + return err; + } + + err = clk_prepare_enable(host->clk); + if (err) { + dev_err(dev, "failed to enable clock: %d\n", err); + goto release_reset; + } + + err = reset_control_bulk_deassert(host->nresets, host->resets); + if (err < 0) { + dev_err(dev, "failed to deassert reset: %d\n", err); + goto disable_clk; + } + + host1x_setup_sid_table(host); + host1x_syncpt_restore(host); + host1x_intr_start(host); + + return 0; + +disable_clk: + clk_disable_unprepare(host->clk); +release_reset: + reset_control_bulk_release(host->nresets, host->resets); + + return err; +} + +static const struct dev_pm_ops host1x_pm_ops = { + SET_RUNTIME_PM_OPS(host1x_runtime_suspend, host1x_runtime_resume, + NULL) + /* TODO: add system suspend-resume once driver will be ready for that */ +}; + static struct platform_driver tegra_host1x_driver = { .driver = { .name = "tegra-host1x", .of_match_table = host1x_of_match, + .pm = &host1x_pm_ops, }, .probe = host1x_probe, .remove = host1x_remove, diff --git a/drivers/gpu/host1x/dev.h b/drivers/gpu/host1x/dev.h index fa6d4bc46e98..ca4b082f0cd4 100644 --- a/drivers/gpu/host1x/dev.h +++ b/drivers/gpu/host1x/dev.h @@ -118,7 +118,8 @@ struct host1x { struct host1x_syncpt_base *bases; struct device *dev; struct clk *clk; - struct reset_control *rst; + struct reset_control_bulk_data resets[2]; + unsigned int nresets; struct iommu_group *group; struct iommu_domain *domain; @@ -149,6 +150,8 @@ struct host1x { struct list_head list; struct device_dma_parameters dma_parms; + + struct host1x_bo_cache cache; }; void host1x_hypervisor_writel(struct host1x *host1x, u32 r, u32 v); diff --git a/drivers/gpu/host1x/hw/channel_hw.c b/drivers/gpu/host1x/hw/channel_hw.c index 1999780a7203..6b40e9af1e88 100644 --- a/drivers/gpu/host1x/hw/channel_hw.c +++ b/drivers/gpu/host1x/hw/channel_hw.c @@ -159,6 +159,27 @@ static void host1x_channel_set_streamid(struct host1x_channel *channel) #endif } +static void host1x_enable_gather_filter(struct host1x_channel *ch) +{ +#if HOST1X_HW >= 6 + struct host1x *host = dev_get_drvdata(ch->dev->parent); + u32 val; + + if (!host->hv_regs) + return; + + val = host1x_hypervisor_readl( + host, HOST1X_HV_CH_KERNEL_FILTER_GBUFFER(ch->id / 32)); + val |= BIT(ch->id % 32); + host1x_hypervisor_writel( + host, val, HOST1X_HV_CH_KERNEL_FILTER_GBUFFER(ch->id / 32)); +#elif HOST1X_HW >= 4 + host1x_ch_writel(ch, + HOST1X_CHANNEL_CHANNELCTRL_KERNEL_FILTER_GBUFFER(1), + HOST1X_CHANNEL_CHANNELCTRL); +#endif +} + static int channel_submit(struct host1x_job *job) { struct host1x_channel *ch = job->channel; @@ -190,6 +211,7 @@ static int channel_submit(struct host1x_job *job) } host1x_channel_set_streamid(ch); + host1x_enable_gather_filter(ch); /* begin a CDMA submit */ err = host1x_cdma_begin(&ch->cdma, job); @@ -249,27 +271,6 @@ error: return err; } -static void enable_gather_filter(struct host1x *host, - struct host1x_channel *ch) -{ -#if HOST1X_HW >= 6 - u32 val; - - if (!host->hv_regs) - return; - - val = host1x_hypervisor_readl( - host, HOST1X_HV_CH_KERNEL_FILTER_GBUFFER(ch->id / 32)); - val |= BIT(ch->id % 32); - host1x_hypervisor_writel( - host, val, HOST1X_HV_CH_KERNEL_FILTER_GBUFFER(ch->id / 32)); -#elif HOST1X_HW >= 4 - host1x_ch_writel(ch, - HOST1X_CHANNEL_CHANNELCTRL_KERNEL_FILTER_GBUFFER(1), - HOST1X_CHANNEL_CHANNELCTRL); -#endif -} - static int host1x_channel_init(struct host1x_channel *ch, struct host1x *dev, unsigned int index) { @@ -278,7 +279,6 @@ static int host1x_channel_init(struct host1x_channel *ch, struct host1x *dev, #else ch->regs = dev->regs + index * 0x100; #endif - enable_gather_filter(dev, ch); return 0; } diff --git a/drivers/gpu/host1x/intr.c b/drivers/gpu/host1x/intr.c index 45b6be927ec4..965ba21818b1 100644 --- a/drivers/gpu/host1x/intr.c +++ b/drivers/gpu/host1x/intr.c @@ -297,14 +297,11 @@ int host1x_intr_init(struct host1x *host, unsigned int irq_sync) "host1x_sp_%02u", id); } - host1x_intr_start(host); - return 0; } void host1x_intr_deinit(struct host1x *host) { - host1x_intr_stop(host); } void host1x_intr_start(struct host1x *host) diff --git a/drivers/gpu/host1x/job.c b/drivers/gpu/host1x/job.c index 0eef6df7c89e..5e8c183167b7 100644 --- a/drivers/gpu/host1x/job.c +++ b/drivers/gpu/host1x/job.c @@ -134,20 +134,20 @@ 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; - struct iommu_domain *domain; - struct sg_table *sgt; unsigned int i; int err; - domain = iommu_get_domain_for_dev(dev); job->num_unpins = 0; for (i = 0; i < job->num_relocs; i++) { struct host1x_reloc *reloc = &job->relocs[i]; - dma_addr_t phys_addr, *phys; + 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) { @@ -155,64 +155,44 @@ static unsigned int pin_job(struct host1x *host, struct host1x_job *job) goto unpin; } - /* - * If the client device is not attached to an IOMMU, the - * physical address of the buffer object can be used. - * - * Similarly, when an IOMMU domain is shared between all - * host1x clients, the IOVA is already available, so no - * need to map the buffer object again. - * - * XXX Note that this isn't always safe to do because it - * relies on an assumption that no cache maintenance is - * needed on the buffer objects. - */ - if (!domain || client->group) - phys = &phys_addr; - else - phys = NULL; - - sgt = host1x_bo_pin(dev, reloc->target.bo, phys); - if (IS_ERR(sgt)) { - err = PTR_ERR(sgt); - goto unpin; - } + bo = reloc->target.bo; - if (sgt) { - unsigned long mask = HOST1X_RELOC_READ | - HOST1X_RELOC_WRITE; - enum dma_data_direction dir; - - switch (reloc->flags & mask) { - case HOST1X_RELOC_READ: - dir = DMA_TO_DEVICE; - break; + switch (reloc->flags & mask) { + case HOST1X_RELOC_READ: + direction = DMA_TO_DEVICE; + break; - case HOST1X_RELOC_WRITE: - dir = DMA_FROM_DEVICE; - break; + case HOST1X_RELOC_WRITE: + direction = DMA_FROM_DEVICE; + break; - case HOST1X_RELOC_READ | HOST1X_RELOC_WRITE: - dir = DMA_BIDIRECTIONAL; - break; + case HOST1X_RELOC_READ | HOST1X_RELOC_WRITE: + direction = DMA_BIDIRECTIONAL; + break; - default: - err = -EINVAL; - goto unpin; - } + default: + err = -EINVAL; + goto unpin; + } - err = dma_map_sgtable(dev, sgt, dir, 0); - if (err) - goto unpin; + map = host1x_bo_pin(dev, bo, direction, &client->cache); + if (IS_ERR(map)) { + err = PTR_ERR(map); + goto unpin; + } - job->unpins[job->num_unpins].dev = dev; - job->unpins[job->num_unpins].dir = dir; - phys_addr = sg_dma_address(sgt->sgl); + /* + * 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] = phys_addr; - job->unpins[job->num_unpins].bo = reloc->target.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++; } @@ -224,12 +204,11 @@ static unsigned int pin_job(struct host1x *host, struct host1x_job *job) return 0; for (i = 0; i < job->num_cmds; i++) { + struct host1x_bo_mapping *map; size_t gather_size = 0; struct scatterlist *sg; - dma_addr_t phys_addr; unsigned long shift; struct iova *alloc; - dma_addr_t *phys; unsigned int j; if (job->cmds[i].is_wait) @@ -243,25 +222,16 @@ static unsigned int pin_job(struct host1x *host, struct host1x_job *job) goto unpin; } - /** - * If the host1x is not attached to an IOMMU, there is no need - * to map the buffer object for the host1x, since the physical - * address can simply be used. - */ - if (!iommu_get_domain_for_dev(host->dev)) - phys = &phys_addr; - else - phys = NULL; - - sgt = host1x_bo_pin(host->dev, g->bo, phys); - if (IS_ERR(sgt)) { - err = PTR_ERR(sgt); - goto put; + map = host1x_bo_pin(host->dev, g->bo, DMA_TO_DEVICE, &host->cache); + if (IS_ERR(map)) { + err = PTR_ERR(map); + goto unpin; } if (host->domain) { - for_each_sgtable_sg(sgt, sg, j) + 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); @@ -272,33 +242,23 @@ static unsigned int pin_job(struct host1x *host, struct host1x_job *job) goto put; } - err = iommu_map_sgtable(host->domain, - iova_dma_addr(&host->iova, alloc), - sgt, 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 put; } - job->unpins[job->num_unpins].size = gather_size; - phys_addr = iova_dma_addr(&host->iova, alloc); - } else if (sgt) { - err = dma_map_sgtable(host->dev, sgt, DMA_TO_DEVICE, 0); - if (err) - goto put; - - job->unpins[job->num_unpins].dir = DMA_TO_DEVICE; - job->unpins[job->num_unpins].dev = host->dev; - phys_addr = sg_dma_address(sgt->sgl); + map->phys = iova_dma_addr(&host->iova, alloc); + map->size = gather_size; } - job->addr_phys[job->num_unpins] = phys_addr; - job->gather_addr_phys[i] = phys_addr; - - 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; @@ -690,22 +650,16 @@ 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]; - struct device *dev = unpin->dev ?: host->dev; - struct sg_table *sgt = unpin->sgt; - - if (!job->enable_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 (unpin->dev && sgt) - dma_unmap_sgtable(unpin->dev, sgt, unpin->dir, 0); + 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(dev, unpin->bo, sgt); - host1x_bo_put(unpin->bo); + host1x_bo_unpin(map); + host1x_bo_put(bo); } job->num_unpins = 0; diff --git a/drivers/gpu/host1x/job.h b/drivers/gpu/host1x/job.h index b4428c5495c9..dad5a1946693 100644 --- a/drivers/gpu/host1x/job.h +++ b/drivers/gpu/host1x/job.h @@ -35,11 +35,7 @@ struct host1x_job_cmd { }; struct host1x_job_unpin_data { - struct host1x_bo *bo; - struct sg_table *sgt; - struct device *dev; - size_t size; - enum dma_data_direction dir; + struct host1x_bo_mapping *map; }; /* diff --git a/drivers/gpu/host1x/syncpt.c b/drivers/gpu/host1x/syncpt.c index d198a10848c6..e08e331e46ae 100644 --- a/drivers/gpu/host1x/syncpt.c +++ b/drivers/gpu/host1x/syncpt.c @@ -143,6 +143,8 @@ void host1x_syncpt_restore(struct host1x *host) for (i = 0; i < host1x_syncpt_nb_bases(host); i++) host1x_hw_syncpt_restore_wait_base(host, sp_base + i); + host1x_hw_syncpt_enable_protection(host); + wmb(); } @@ -366,9 +368,6 @@ int host1x_syncpt_init(struct host1x *host) host->syncpt = syncpt; host->bases = bases; - host1x_syncpt_restore(host); - host1x_hw_syncpt_enable_protection(host); - /* Allocate sync point to use for clearing waits for expired fences */ host->nop_sp = host1x_syncpt_alloc(host, 0, "reserved-nop"); if (!host->nop_sp) |