diff options
Diffstat (limited to 'drivers/gpu/host1x')
-rw-r--r-- | drivers/gpu/host1x/cdma.c | 42 | ||||
-rw-r--r-- | drivers/gpu/host1x/channel.c | 5 | ||||
-rw-r--r-- | drivers/gpu/host1x/debug.c | 38 | ||||
-rw-r--r-- | drivers/gpu/host1x/dev.c | 16 | ||||
-rw-r--r-- | drivers/gpu/host1x/dev.h | 38 | ||||
-rw-r--r-- | drivers/gpu/host1x/hw/cdma_hw.c | 23 | ||||
-rw-r--r-- | drivers/gpu/host1x/hw/channel_hw.c | 5 | ||||
-rw-r--r-- | drivers/gpu/host1x/hw/debug_hw.c | 36 | ||||
-rw-r--r-- | drivers/gpu/host1x/hw/intr_hw.c | 30 | ||||
-rw-r--r-- | drivers/gpu/host1x/hw/syncpt_hw.c | 10 | ||||
-rw-r--r-- | drivers/gpu/host1x/intr.c | 16 | ||||
-rw-r--r-- | drivers/gpu/host1x/intr.h | 4 | ||||
-rw-r--r-- | drivers/gpu/host1x/job.c | 8 | ||||
-rw-r--r-- | drivers/gpu/host1x/syncpt.c | 58 | ||||
-rw-r--r-- | drivers/gpu/host1x/syncpt.h | 8 |
15 files changed, 205 insertions, 132 deletions
diff --git a/drivers/gpu/host1x/cdma.c b/drivers/gpu/host1x/cdma.c index a18db4d5347c..c5d82a8a2ec9 100644 --- a/drivers/gpu/host1x/cdma.c +++ b/drivers/gpu/host1x/cdma.c @@ -96,12 +96,12 @@ fail: */ static void host1x_pushbuffer_push(struct push_buffer *pb, u32 op1, u32 op2) { - u32 pos = pb->pos; - u32 *p = (u32 *)((void *)pb->mapped + pos); - WARN_ON(pos == pb->fence); + u32 *p = (u32 *)((void *)pb->mapped + pb->pos); + + WARN_ON(pb->pos == pb->fence); *(p++) = op1; *(p++) = op2; - pb->pos = (pos + 8) & (pb->size_bytes - 1); + pb->pos = (pb->pos + 8) & (pb->size_bytes - 1); } /* @@ -134,14 +134,19 @@ unsigned int host1x_cdma_wait_locked(struct host1x_cdma *cdma, enum cdma_event event) { for (;;) { + struct push_buffer *pb = &cdma->push_buffer; unsigned int space; - if (event == CDMA_EVENT_SYNC_QUEUE_EMPTY) + switch (event) { + case CDMA_EVENT_SYNC_QUEUE_EMPTY: space = list_empty(&cdma->sync_queue) ? 1 : 0; - else if (event == CDMA_EVENT_PUSH_BUFFER_SPACE) { - struct push_buffer *pb = &cdma->push_buffer; + break; + + case CDMA_EVENT_PUSH_BUFFER_SPACE: space = host1x_pushbuffer_space(pb); - } else { + break; + + default: WARN_ON(1); return -EINVAL; } @@ -159,12 +164,14 @@ unsigned int host1x_cdma_wait_locked(struct host1x_cdma *cdma, mutex_lock(&cdma->lock); continue; } + cdma->event = event; mutex_unlock(&cdma->lock); down(&cdma->sem); mutex_lock(&cdma->lock); } + return 0; } @@ -234,6 +241,7 @@ static void update_cdma_locked(struct host1x_cdma *cdma) /* Start timer on next pending syncpt */ if (job->timeout) cdma_start_timer_locked(cdma, job); + break; } @@ -247,7 +255,9 @@ static void update_cdma_locked(struct host1x_cdma *cdma) /* Pop push buffer slots */ if (job->num_slots) { struct push_buffer *pb = &cdma->push_buffer; + host1x_pushbuffer_pop(pb, job->num_slots); + if (cdma->event == CDMA_EVENT_PUSH_BUFFER_SPACE) signal = true; } @@ -269,11 +279,9 @@ static void update_cdma_locked(struct host1x_cdma *cdma) void host1x_cdma_update_sync_queue(struct host1x_cdma *cdma, struct device *dev) { - u32 restart_addr; - u32 syncpt_incrs; - struct host1x_job *job = NULL; - u32 syncpt_val; struct host1x *host1x = cdma_to_host1x(cdma); + u32 restart_addr, syncpt_incrs, syncpt_val; + struct host1x_job *job = NULL; syncpt_val = host1x_syncpt_load(cdma->timeout.syncpt); @@ -342,9 +350,11 @@ void host1x_cdma_update_sync_queue(struct host1x_cdma *cdma, syncpt_val += syncpt_incrs; } - /* The following sumbits from the same client may be dependent on the + /* + * The following sumbits from the same client may be dependent on the * failed submit and therefore they may fail. Force a small timeout - * to make the queue cleanup faster */ + * to make the queue cleanup faster. + */ list_for_each_entry_from(job, &cdma->sync_queue, list) if (job->client == cdma->timeout.client) @@ -375,6 +385,7 @@ int host1x_cdma_init(struct host1x_cdma *cdma) err = host1x_pushbuffer_init(&cdma->push_buffer); if (err) return err; + return 0; } @@ -410,6 +421,7 @@ int host1x_cdma_begin(struct host1x_cdma *cdma, struct host1x_job *job) /* init state on first submit with timeout value */ if (!cdma->timeout.initialized) { int err; + err = host1x_hw_cdma_timeout_init(host1x, cdma, job->syncpt_id); if (err) { @@ -418,6 +430,7 @@ int host1x_cdma_begin(struct host1x_cdma *cdma, struct host1x_job *job) } } } + if (!cdma->running) host1x_hw_cdma_start(host1x, cdma); @@ -448,6 +461,7 @@ void host1x_cdma_push(struct host1x_cdma *cdma, u32 op1, u32 op2) slots_free = host1x_cdma_wait_locked(cdma, CDMA_EVENT_PUSH_BUFFER_SPACE); } + cdma->slots_free = slots_free - 1; cdma->slots_used++; host1x_pushbuffer_push(pb, op1, op2); diff --git a/drivers/gpu/host1x/channel.c b/drivers/gpu/host1x/channel.c index b4ae3affb987..8f437d924c10 100644 --- a/drivers/gpu/host1x/channel.c +++ b/drivers/gpu/host1x/channel.c @@ -83,9 +83,10 @@ EXPORT_SYMBOL(host1x_channel_put); struct host1x_channel *host1x_channel_request(struct device *dev) { struct host1x *host = dev_get_drvdata(dev->parent); - int max_channels = host->info->nb_channels; + unsigned int max_channels = host->info->nb_channels; struct host1x_channel *channel = NULL; - int index, err; + unsigned long index; + int err; mutex_lock(&host->chlist_mutex); diff --git a/drivers/gpu/host1x/debug.c b/drivers/gpu/host1x/debug.c index ee3d12b51c50..d9330fcc62ad 100644 --- a/drivers/gpu/host1x/debug.c +++ b/drivers/gpu/host1x/debug.c @@ -39,6 +39,7 @@ void host1x_debug_output(struct output *o, const char *fmt, ...) va_start(args, fmt); len = vsnprintf(o->buf, sizeof(o->buf), fmt, args); va_end(args); + o->fn(o->ctx, o->buf, len); } @@ -48,13 +49,17 @@ static int show_channels(struct host1x_channel *ch, void *data, bool show_fifo) struct output *o = data; mutex_lock(&ch->reflock); + if (ch->refcount) { mutex_lock(&ch->cdma.lock); + if (show_fifo) host1x_hw_show_channel_fifo(m, ch, o); + host1x_hw_show_channel_cdma(m, ch, o); mutex_unlock(&ch->cdma.lock); } + mutex_unlock(&ch->reflock); return 0; @@ -62,22 +67,27 @@ static int show_channels(struct host1x_channel *ch, void *data, bool show_fifo) static void show_syncpts(struct host1x *m, struct output *o) { - int i; + unsigned int i; + host1x_debug_output(o, "---- syncpts ----\n"); + 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); + if (!min && !max) continue; - host1x_debug_output(o, "id %d (%s) min %d max %d\n", + + host1x_debug_output(o, "id %u (%s) min %d max %d\n", i, m->syncpt[i].name, min, max); } for (i = 0; i < host1x_syncpt_nb_bases(m); i++) { u32 base_val; + base_val = host1x_syncpt_load_wait_base(m->syncpt + i); if (base_val) - host1x_debug_output(o, "waitbase id %d val %d\n", i, + host1x_debug_output(o, "waitbase id %u val %d\n", i, base_val); } @@ -114,7 +124,9 @@ static int host1x_debug_show_all(struct seq_file *s, void *unused) .fn = write_to_seqfile, .ctx = s }; + show_all(s->private, &o); + return 0; } @@ -124,7 +136,9 @@ static int host1x_debug_show(struct seq_file *s, void *unused) .fn = write_to_seqfile, .ctx = s }; + show_all_no_fifo(s->private, &o); + return 0; } @@ -134,10 +148,10 @@ static int host1x_debug_open_all(struct inode *inode, struct file *file) } static const struct file_operations host1x_debug_all_fops = { - .open = host1x_debug_open_all, - .read = seq_read, - .llseek = seq_lseek, - .release = single_release, + .open = host1x_debug_open_all, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, }; static int host1x_debug_open(struct inode *inode, struct file *file) @@ -146,10 +160,10 @@ static int host1x_debug_open(struct inode *inode, struct file *file) } static const struct file_operations host1x_debug_fops = { - .open = host1x_debug_open, - .read = seq_read, - .llseek = seq_lseek, - .release = single_release, + .open = host1x_debug_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, }; static void host1x_debugfs_init(struct host1x *host1x) @@ -201,6 +215,7 @@ void host1x_debug_dump(struct host1x *host1x) struct output o = { .fn = write_to_printk }; + show_all(host1x, &o); } @@ -209,5 +224,6 @@ void host1x_debug_dump_syncpts(struct host1x *host1x) struct output o = { .fn = write_to_printk }; + show_syncpts(host1x, &o); } diff --git a/drivers/gpu/host1x/dev.c b/drivers/gpu/host1x/dev.c index ff348690df94..a62317af76ad 100644 --- a/drivers/gpu/host1x/dev.c +++ b/drivers/gpu/host1x/dev.c @@ -63,13 +63,13 @@ u32 host1x_ch_readl(struct host1x_channel *ch, u32 r) } static const struct host1x_info host1x01_info = { - .nb_channels = 8, - .nb_pts = 32, - .nb_mlocks = 16, - .nb_bases = 8, - .init = host1x01_init, - .sync_offset = 0x3000, - .dma_mask = DMA_BIT_MASK(32), + .nb_channels = 8, + .nb_pts = 32, + .nb_mlocks = 16, + .nb_bases = 8, + .init = host1x01_init, + .sync_offset = 0x3000, + .dma_mask = DMA_BIT_MASK(32), }; static const struct host1x_info host1x02_info = { @@ -102,7 +102,7 @@ static const struct host1x_info host1x05_info = { .dma_mask = DMA_BIT_MASK(34), }; -static struct of_device_id host1x_of_match[] = { +static const struct of_device_id host1x_of_match[] = { { .compatible = "nvidia,tegra210-host1x", .data = &host1x05_info, }, { .compatible = "nvidia,tegra124-host1x", .data = &host1x04_info, }, { .compatible = "nvidia,tegra114-host1x", .data = &host1x02_info, }, diff --git a/drivers/gpu/host1x/dev.h b/drivers/gpu/host1x/dev.h index dace124994bb..5220510f39da 100644 --- a/drivers/gpu/host1x/dev.h +++ b/drivers/gpu/host1x/dev.h @@ -45,7 +45,7 @@ struct host1x_cdma_ops { void (*start)(struct host1x_cdma *cdma); void (*stop)(struct host1x_cdma *cdma); void (*flush)(struct host1x_cdma *cdma); - int (*timeout_init)(struct host1x_cdma *cdma, u32 syncpt_id); + int (*timeout_init)(struct host1x_cdma *cdma, unsigned int syncpt); void (*timeout_destroy)(struct host1x_cdma *cdma); void (*freeze)(struct host1x_cdma *cdma); void (*resume)(struct host1x_cdma *cdma, u32 getptr); @@ -82,21 +82,21 @@ struct host1x_intr_ops { int (*init_host_sync)(struct host1x *host, u32 cpm, void (*syncpt_thresh_work)(struct work_struct *work)); void (*set_syncpt_threshold)( - struct host1x *host, u32 id, u32 thresh); - void (*enable_syncpt_intr)(struct host1x *host, u32 id); - void (*disable_syncpt_intr)(struct host1x *host, u32 id); + struct host1x *host, unsigned int id, u32 thresh); + void (*enable_syncpt_intr)(struct host1x *host, unsigned int id); + void (*disable_syncpt_intr)(struct host1x *host, unsigned int id); void (*disable_all_syncpt_intrs)(struct host1x *host); int (*free_syncpt_irq)(struct host1x *host); }; struct host1x_info { - int nb_channels; /* host1x: num channels supported */ - int nb_pts; /* host1x: num syncpoints supported */ - int nb_bases; /* host1x: num syncpoints supported */ - int nb_mlocks; /* host1x: number of mlocks */ - int (*init)(struct host1x *); /* initialize per SoC ops */ - int sync_offset; - u64 dma_mask; /* mask of addressable memory */ + unsigned int nb_channels; /* host1x: number of channels supported */ + unsigned int nb_pts; /* host1x: number of syncpoints supported */ + unsigned int nb_bases; /* host1x: number of syncpoint bases supported */ + unsigned int nb_mlocks; /* host1x: number of mlocks supported */ + int (*init)(struct host1x *host1x); /* initialize per SoC ops */ + unsigned int sync_offset; /* offset of syncpoint registers */ + u64 dma_mask; /* mask of addressable memory */ }; struct host1x { @@ -109,7 +109,6 @@ struct host1x { struct clk *clk; struct mutex intr_mutex; - struct workqueue_struct *intr_wq; int intr_syncpt_irq; const struct host1x_syncpt_ops *syncpt_op; @@ -183,19 +182,20 @@ static inline int host1x_hw_intr_init_host_sync(struct host1x *host, u32 cpm, } static inline void host1x_hw_intr_set_syncpt_threshold(struct host1x *host, - u32 id, u32 thresh) + unsigned int id, + u32 thresh) { host->intr_op->set_syncpt_threshold(host, id, thresh); } static inline void host1x_hw_intr_enable_syncpt_intr(struct host1x *host, - u32 id) + unsigned int id) { host->intr_op->enable_syncpt_intr(host, id); } static inline void host1x_hw_intr_disable_syncpt_intr(struct host1x *host, - u32 id) + unsigned int id) { host->intr_op->disable_syncpt_intr(host, id); } @@ -212,9 +212,9 @@ static inline int host1x_hw_intr_free_syncpt_irq(struct host1x *host) static inline int host1x_hw_channel_init(struct host1x *host, struct host1x_channel *channel, - int chid) + unsigned int id) { - return host->channel_op->init(channel, host, chid); + return host->channel_op->init(channel, host, id); } static inline int host1x_hw_channel_submit(struct host1x *host, @@ -243,9 +243,9 @@ static inline void host1x_hw_cdma_flush(struct host1x *host, static inline int host1x_hw_cdma_timeout_init(struct host1x *host, struct host1x_cdma *cdma, - u32 syncpt_id) + unsigned int syncpt) { - return host->cdma_op->timeout_init(cdma, syncpt_id); + return host->cdma_op->timeout_init(cdma, syncpt); } static inline void host1x_hw_cdma_timeout_destroy(struct host1x *host, diff --git a/drivers/gpu/host1x/hw/cdma_hw.c b/drivers/gpu/host1x/hw/cdma_hw.c index 305ea8f3382d..659c1bbfeeba 100644 --- a/drivers/gpu/host1x/hw/cdma_hw.c +++ b/drivers/gpu/host1x/hw/cdma_hw.c @@ -41,7 +41,7 @@ static void cdma_timeout_cpu_incr(struct host1x_cdma *cdma, u32 getptr, { struct host1x *host1x = cdma_to_host1x(cdma); struct push_buffer *pb = &cdma->push_buffer; - u32 i; + unsigned int i; for (i = 0; i < syncpt_incrs; i++) host1x_syncpt_incr(cdma->timeout.syncpt); @@ -58,6 +58,7 @@ static void cdma_timeout_cpu_incr(struct host1x_cdma *cdma, u32 getptr, &pb->phys, getptr); getptr = (getptr + 8) & (pb->size_bytes - 1); } + wmb(); } @@ -162,12 +163,14 @@ static void cdma_stop(struct host1x_cdma *cdma) struct host1x_channel *ch = cdma_to_channel(cdma); mutex_lock(&cdma->lock); + if (cdma->running) { host1x_cdma_wait_locked(cdma, CDMA_EVENT_SYNC_QUEUE_EMPTY); host1x_ch_writel(ch, HOST1X_CHANNEL_DMACTRL_DMASTOP, HOST1X_CHANNEL_DMACTRL); cdma->running = false; } + mutex_unlock(&cdma->lock); } @@ -213,11 +216,11 @@ static void cdma_resume(struct host1x_cdma *cdma, u32 getptr) u32 cmdproc_stop; dev_dbg(host1x->dev, - "resuming channel (id %d, DMAGET restart = 0x%x)\n", + "resuming channel (id %u, DMAGET restart = 0x%x)\n", ch->id, getptr); cmdproc_stop = host1x_sync_readl(host1x, HOST1X_SYNC_CMDPROC_STOP); - cmdproc_stop &= ~(BIT(ch->id)); + cmdproc_stop &= ~BIT(ch->id); host1x_sync_writel(host1x, cmdproc_stop, HOST1X_SYNC_CMDPROC_STOP); cdma->torndown = false; @@ -231,14 +234,11 @@ static void cdma_resume(struct host1x_cdma *cdma, u32 getptr) */ static void cdma_timeout_handler(struct work_struct *work) { + u32 prev_cmdproc, cmdproc_stop, syncpt_val; struct host1x_cdma *cdma; struct host1x *host1x; struct host1x_channel *ch; - u32 syncpt_val; - - u32 prev_cmdproc, cmdproc_stop; - cdma = container_of(to_delayed_work(work), struct host1x_cdma, timeout.wq); host1x = cdma_to_host1x(cdma); @@ -277,9 +277,9 @@ static void cdma_timeout_handler(struct work_struct *work) return; } - dev_warn(host1x->dev, "%s: timeout: %d (%s), HW thresh %d, done %d\n", - __func__, cdma->timeout.syncpt->id, cdma->timeout.syncpt->name, - syncpt_val, cdma->timeout.syncpt_val); + dev_warn(host1x->dev, "%s: timeout: %u (%s), HW thresh %d, done %d\n", + __func__, cdma->timeout.syncpt->id, cdma->timeout.syncpt->name, + syncpt_val, cdma->timeout.syncpt_val); /* stop HW, resetting channel/module */ host1x_hw_cdma_freeze(host1x, cdma); @@ -291,7 +291,7 @@ static void cdma_timeout_handler(struct work_struct *work) /* * Init timeout resources */ -static int cdma_timeout_init(struct host1x_cdma *cdma, u32 syncpt_id) +static int cdma_timeout_init(struct host1x_cdma *cdma, unsigned int syncpt) { INIT_DELAYED_WORK(&cdma->timeout.wq, cdma_timeout_handler); cdma->timeout.initialized = true; @@ -306,6 +306,7 @@ static void cdma_timeout_destroy(struct host1x_cdma *cdma) { if (cdma->timeout.initialized) cancel_delayed_work(&cdma->timeout.wq); + cdma->timeout.initialized = false; } diff --git a/drivers/gpu/host1x/hw/channel_hw.c b/drivers/gpu/host1x/hw/channel_hw.c index 946c332c3906..5e8df78b7acd 100644 --- a/drivers/gpu/host1x/hw/channel_hw.c +++ b/drivers/gpu/host1x/hw/channel_hw.c @@ -46,6 +46,7 @@ static void trace_write_gather(struct host1x_cdma *cdma, struct host1x_bo *bo, */ for (i = 0; i < words; i += TRACE_MAX_LENGTH) { u32 num_words = min(words - i, TRACE_MAX_LENGTH); + offset += i * sizeof(u32); trace_host1x_cdma_push_gather(dev_name(dev), bo, @@ -66,6 +67,7 @@ static void submit_gathers(struct host1x_job *job) struct host1x_job_gather *g = &job->gathers[i]; u32 op1 = host1x_opcode_gather(g->words); u32 op2 = g->base + g->offset; + trace_write_gather(cdma, g->bo, g->offset, op1 & 0xffff); host1x_cdma_push(cdma, op1, op2); } @@ -75,7 +77,8 @@ static inline void synchronize_syncpt_base(struct host1x_job *job) { struct host1x *host = dev_get_drvdata(job->channel->dev->parent); struct host1x_syncpt *sp = host->syncpt + job->syncpt_id; - u32 id, value; + unsigned int id; + u32 value; value = host1x_syncpt_read_max(sp); id = sp->base->id; diff --git a/drivers/gpu/host1x/hw/debug_hw.c b/drivers/gpu/host1x/hw/debug_hw.c index cc3f1825c735..7a4a3286e4a7 100644 --- a/drivers/gpu/host1x/hw/debug_hw.c +++ b/drivers/gpu/host1x/hw/debug_hw.c @@ -40,8 +40,7 @@ enum { static unsigned int show_channel_command(struct output *o, u32 val) { - unsigned mask; - unsigned subop; + unsigned int mask, subop; switch (val >> 28) { case HOST1X_OPCODE_SETCLASS: @@ -51,12 +50,11 @@ static unsigned int show_channel_command(struct output *o, u32 val) val >> 6 & 0x3ff, val >> 16 & 0xfff, mask); return hweight8(mask); - } else { - host1x_debug_output(o, "SETCL(class=%03x)\n", - val >> 6 & 0x3ff); - return 0; } + host1x_debug_output(o, "SETCL(class=%03x)\n", val >> 6 & 0x3ff); + return 0; + case HOST1X_OPCODE_INCR: host1x_debug_output(o, "INCR(offset=%03x, [", val >> 16 & 0xfff); @@ -143,7 +141,8 @@ static void show_channel_gathers(struct output *o, struct host1x_cdma *cdma) struct host1x_job *job; list_for_each_entry(job, &cdma->sync_queue, list) { - int i; + unsigned int i; + host1x_debug_output(o, "\n%p: JOB, syncpt_id=%d, syncpt_val=%d, first_get=%08x, timeout=%d num_slots=%d, num_handles=%d\n", job, job->syncpt_id, job->syncpt_end, job->first_get, job->timeout, @@ -190,7 +189,7 @@ static void host1x_debug_show_channel_cdma(struct host1x *host, cbread = host1x_sync_readl(host, HOST1X_SYNC_CBREAD(ch->id)); cbstat = host1x_sync_readl(host, HOST1X_SYNC_CBSTAT(ch->id)); - host1x_debug_output(o, "%d-%s: ", ch->id, dev_name(ch->dev)); + host1x_debug_output(o, "%u-%s: ", ch->id, dev_name(ch->dev)); if (HOST1X_CHANNEL_DMACTRL_DMASTOP_V(dmactrl) || !ch->cdma.push_buffer.mapped) { @@ -200,14 +199,13 @@ static void host1x_debug_show_channel_cdma(struct host1x *host, if (HOST1X_SYNC_CBSTAT_CBCLASS_V(cbstat) == HOST1X_CLASS_HOST1X && HOST1X_SYNC_CBSTAT_CBOFFSET_V(cbstat) == - HOST1X_UCLASS_WAIT_SYNCPT) + HOST1X_UCLASS_WAIT_SYNCPT) host1x_debug_output(o, "waiting on syncpt %d val %d\n", cbread >> 24, cbread & 0xffffff); else if (HOST1X_SYNC_CBSTAT_CBCLASS_V(cbstat) == - HOST1X_CLASS_HOST1X && - HOST1X_SYNC_CBSTAT_CBOFFSET_V(cbstat) == - HOST1X_UCLASS_WAIT_SYNCPT_BASE) { - + HOST1X_CLASS_HOST1X && + HOST1X_SYNC_CBSTAT_CBOFFSET_V(cbstat) == + HOST1X_UCLASS_WAIT_SYNCPT_BASE) { base = (cbread >> 16) & 0xff; baseval = host1x_sync_readl(host, HOST1X_SYNC_SYNCPT_BASE(base)); @@ -236,7 +234,7 @@ static void host1x_debug_show_channel_fifo(struct host1x *host, u32 val, rd_ptr, wr_ptr, start, end; unsigned int data_count = 0; - host1x_debug_output(o, "%d: fifo:\n", ch->id); + host1x_debug_output(o, "%u: fifo:\n", ch->id); val = host1x_ch_readl(ch, HOST1X_CHANNEL_FIFOSTAT); host1x_debug_output(o, "FIFOSTAT %08x\n", val); @@ -290,20 +288,22 @@ static void host1x_debug_show_channel_fifo(struct host1x *host, static void host1x_debug_show_mlocks(struct host1x *host, struct output *o) { - int i; + unsigned int i; host1x_debug_output(o, "---- mlocks ----\n"); + for (i = 0; i < host1x_syncpt_nb_mlocks(host); i++) { u32 owner = host1x_sync_readl(host, HOST1X_SYNC_MLOCK_OWNER(i)); if (HOST1X_SYNC_MLOCK_OWNER_CH_OWNS_V(owner)) - host1x_debug_output(o, "%d: locked by channel %d\n", + host1x_debug_output(o, "%u: locked by channel %u\n", i, HOST1X_SYNC_MLOCK_OWNER_CHID_V(owner)); else if (HOST1X_SYNC_MLOCK_OWNER_CPU_OWNS_V(owner)) - host1x_debug_output(o, "%d: locked by cpu\n", i); + host1x_debug_output(o, "%u: locked by cpu\n", i); else - host1x_debug_output(o, "%d: unlocked\n", i); + host1x_debug_output(o, "%u: unlocked\n", i); } + host1x_debug_output(o, "\n"); } diff --git a/drivers/gpu/host1x/hw/intr_hw.c b/drivers/gpu/host1x/hw/intr_hw.c index e1e31e9e67cd..dacb8009a605 100644 --- a/drivers/gpu/host1x/hw/intr_hw.c +++ b/drivers/gpu/host1x/hw/intr_hw.c @@ -38,14 +38,14 @@ static void host1x_intr_syncpt_handle(struct host1x_syncpt *syncpt) host1x_sync_writel(host, BIT_MASK(id), HOST1X_SYNC_SYNCPT_THRESH_CPU0_INT_STATUS(BIT_WORD(id))); - queue_work(host->intr_wq, &syncpt->intr.work); + schedule_work(&syncpt->intr.work); } static irqreturn_t syncpt_thresh_isr(int irq, void *dev_id) { struct host1x *host = dev_id; unsigned long reg; - int i, id; + unsigned int i, id; for (i = 0; i < DIV_ROUND_UP(host->info->nb_pts, 32); i++) { reg = host1x_sync_readl(host, @@ -62,7 +62,7 @@ static irqreturn_t syncpt_thresh_isr(int irq, void *dev_id) static void _host1x_intr_disable_all_syncpt_intrs(struct host1x *host) { - u32 i; + unsigned int i; for (i = 0; i < DIV_ROUND_UP(host->info->nb_pts, 32); ++i) { host1x_sync_writel(host, 0xffffffffu, @@ -72,10 +72,12 @@ static void _host1x_intr_disable_all_syncpt_intrs(struct host1x *host) } } -static int _host1x_intr_init_host_sync(struct host1x *host, u32 cpm, - void (*syncpt_thresh_work)(struct work_struct *)) +static int +_host1x_intr_init_host_sync(struct host1x *host, u32 cpm, + void (*syncpt_thresh_work)(struct work_struct *)) { - int i, err; + unsigned int i; + int err; host1x_hw_intr_disable_all_syncpt_intrs(host); @@ -106,18 +108,21 @@ static int _host1x_intr_init_host_sync(struct host1x *host, u32 cpm, } static void _host1x_intr_set_syncpt_threshold(struct host1x *host, - u32 id, u32 thresh) + unsigned int id, + u32 thresh) { host1x_sync_writel(host, thresh, HOST1X_SYNC_SYNCPT_INT_THRESH(id)); } -static void _host1x_intr_enable_syncpt_intr(struct host1x *host, u32 id) +static void _host1x_intr_enable_syncpt_intr(struct host1x *host, + unsigned int id) { host1x_sync_writel(host, BIT_MASK(id), HOST1X_SYNC_SYNCPT_THRESH_INT_ENABLE_CPU0(BIT_WORD(id))); } -static void _host1x_intr_disable_syncpt_intr(struct host1x *host, u32 id) +static void _host1x_intr_disable_syncpt_intr(struct host1x *host, + unsigned int id) { host1x_sync_writel(host, BIT_MASK(id), HOST1X_SYNC_SYNCPT_THRESH_INT_DISABLE(BIT_WORD(id))); @@ -127,8 +132,13 @@ static void _host1x_intr_disable_syncpt_intr(struct host1x *host, u32 id) static int _host1x_free_syncpt_irq(struct host1x *host) { + unsigned int i; + devm_free_irq(host->dev, host->intr_syncpt_irq, host); - flush_workqueue(host->intr_wq); + + for (i = 0; i < host->info->nb_pts; i++) + cancel_work_sync(&host->syncpt[i].intr.work); + return 0; } diff --git a/drivers/gpu/host1x/hw/syncpt_hw.c b/drivers/gpu/host1x/hw/syncpt_hw.c index 56e85395ac24..c93f74fcce72 100644 --- a/drivers/gpu/host1x/hw/syncpt_hw.c +++ b/drivers/gpu/host1x/hw/syncpt_hw.c @@ -26,8 +26,9 @@ */ static void syncpt_restore(struct host1x_syncpt *sp) { + u32 min = host1x_syncpt_read_min(sp); struct host1x *host = sp->host; - int min = host1x_syncpt_read_min(sp); + host1x_sync_writel(host, min, HOST1X_SYNC_SYNCPT(sp->id)); } @@ -37,6 +38,7 @@ static void syncpt_restore(struct host1x_syncpt *sp) static void syncpt_restore_wait_base(struct host1x_syncpt *sp) { struct host1x *host = sp->host; + host1x_sync_writel(host, sp->base_val, HOST1X_SYNC_SYNCPT_BASE(sp->id)); } @@ -47,6 +49,7 @@ static void syncpt_restore_wait_base(struct host1x_syncpt *sp) static void syncpt_read_wait_base(struct host1x_syncpt *sp) { struct host1x *host = sp->host; + sp->base_val = host1x_sync_readl(host, HOST1X_SYNC_SYNCPT_BASE(sp->id)); } @@ -85,6 +88,7 @@ static int syncpt_cpu_incr(struct host1x_syncpt *sp) if (!host1x_syncpt_client_managed(sp) && host1x_syncpt_idle(sp)) return -EINVAL; + host1x_sync_writel(host, BIT_MASK(sp->id), HOST1X_SYNC_SYNCPT_CPU_INCR(reg_offset)); wmb(); @@ -95,10 +99,10 @@ static int syncpt_cpu_incr(struct host1x_syncpt *sp) /* 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 override = host1x_class_host_wait_syncpt(HOST1X_SYNCPT_RESERVED, 0); *((u32 *)patch_addr) = override; + return 0; } diff --git a/drivers/gpu/host1x/intr.c b/drivers/gpu/host1x/intr.c index 2491bf82e30c..8b4fad0ab35d 100644 --- a/drivers/gpu/host1x/intr.c +++ b/drivers/gpu/host1x/intr.c @@ -122,18 +122,20 @@ static void action_submit_complete(struct host1x_waitlist *waiter) static void action_wakeup(struct host1x_waitlist *waiter) { wait_queue_head_t *wq = waiter->data; + wake_up(wq); } static void action_wakeup_interruptible(struct host1x_waitlist *waiter) { wait_queue_head_t *wq = waiter->data; + wake_up_interruptible(wq); } typedef void (*action_handler)(struct host1x_waitlist *waiter); -static action_handler action_handlers[HOST1X_INTR_ACTION_COUNT] = { +static const action_handler action_handlers[HOST1X_INTR_ACTION_COUNT] = { action_submit_complete, action_wakeup, action_wakeup_interruptible, @@ -209,7 +211,7 @@ static void syncpt_thresh_work(struct work_struct *work) host1x_syncpt_load(host->syncpt + id)); } -int host1x_intr_add_action(struct host1x *host, u32 id, u32 thresh, +int host1x_intr_add_action(struct host1x *host, unsigned int id, u32 thresh, enum host1x_intr_action action, void *data, struct host1x_waitlist *waiter, void **ref) { @@ -254,7 +256,7 @@ int host1x_intr_add_action(struct host1x *host, u32 id, u32 thresh, return 0; } -void host1x_intr_put_ref(struct host1x *host, u32 id, void *ref) +void host1x_intr_put_ref(struct host1x *host, unsigned int id, void *ref) { struct host1x_waitlist *waiter = ref; struct host1x_syncpt *syncpt; @@ -277,9 +279,6 @@ int host1x_intr_init(struct host1x *host, unsigned int irq_sync) mutex_init(&host->intr_mutex); host->intr_syncpt_irq = irq_sync; - host->intr_wq = create_workqueue("host_syncpt"); - if (!host->intr_wq) - return -ENOMEM; for (id = 0; id < nb_pts; ++id) { struct host1x_syncpt *syncpt = host->syncpt + id; @@ -288,7 +287,7 @@ int host1x_intr_init(struct host1x *host, unsigned int irq_sync) INIT_LIST_HEAD(&syncpt->intr.wait_head); snprintf(syncpt->intr.thresh_irq_name, sizeof(syncpt->intr.thresh_irq_name), - "host1x_sp_%02d", id); + "host1x_sp_%02u", id); } host1x_intr_start(host); @@ -299,7 +298,6 @@ int host1x_intr_init(struct host1x *host, unsigned int irq_sync) void host1x_intr_deinit(struct host1x *host) { host1x_intr_stop(host); - destroy_workqueue(host->intr_wq); } void host1x_intr_start(struct host1x *host) @@ -342,7 +340,7 @@ void host1x_intr_stop(struct host1x *host) if (!list_empty(&syncpt[id].intr.wait_head)) { /* output diagnostics */ mutex_unlock(&host->intr_mutex); - pr_warn("%s cannot stop syncpt intr id=%d\n", + pr_warn("%s cannot stop syncpt intr id=%u\n", __func__, id); return; } diff --git a/drivers/gpu/host1x/intr.h b/drivers/gpu/host1x/intr.h index 2b8adf016a05..1370c2bb75b8 100644 --- a/drivers/gpu/host1x/intr.h +++ b/drivers/gpu/host1x/intr.h @@ -75,7 +75,7 @@ struct host1x_waitlist { * * This is a non-blocking api. */ -int host1x_intr_add_action(struct host1x *host, u32 id, u32 thresh, +int host1x_intr_add_action(struct host1x *host, unsigned int id, u32 thresh, enum host1x_intr_action action, void *data, struct host1x_waitlist *waiter, void **ref); @@ -84,7 +84,7 @@ int host1x_intr_add_action(struct host1x *host, u32 id, u32 thresh, * You must call this if you passed non-NULL as ref. * @ref the ref returned from host1x_intr_add_action() */ -void host1x_intr_put_ref(struct host1x *host, u32 id, void *ref); +void host1x_intr_put_ref(struct host1x *host, unsigned int id, void *ref); /* Initialize host1x sync point interrupt */ int host1x_intr_init(struct host1x *host, unsigned int irq_sync); diff --git a/drivers/gpu/host1x/job.c b/drivers/gpu/host1x/job.c index b4515d544039..a91b7c4a6110 100644 --- a/drivers/gpu/host1x/job.c +++ b/drivers/gpu/host1x/job.c @@ -161,7 +161,7 @@ static int do_waitchks(struct host1x_job *job, struct host1x *host, if (host1x_syncpt_is_expired(sp, wait->thresh)) { dev_dbg(host->dev, - "drop WAIT id %d (%s) thresh 0x%x, min 0x%x\n", + "drop WAIT id %u (%s) thresh 0x%x, min 0x%x\n", wait->syncpt_id, sp->name, wait->thresh, host1x_syncpt_read_min(sp)); @@ -464,6 +464,7 @@ static inline int copy_gathers(struct host1x_job *job, struct device *dev) for (i = 0; i < job->num_gathers; i++) { struct host1x_job_gather *g = &job->gathers[i]; + size += g->words * sizeof(u32); } @@ -514,6 +515,7 @@ int host1x_job_pin(struct host1x_job *job, struct device *dev) 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); } @@ -571,14 +573,16 @@ void host1x_job_unpin(struct host1x_job *job) for (i = 0; i < job->num_unpins; i++) { struct host1x_job_unpin_data *unpin = &job->unpins[i]; + host1x_bo_unpin(unpin->bo, unpin->sgt); host1x_bo_put(unpin->bo); } + job->num_unpins = 0; if (job->gather_copy_size) dma_free_wc(job->channel->dev, job->gather_copy_size, - job->gather_copy_mapped, job->gather_copy); + job->gather_copy_mapped, job->gather_copy); } EXPORT_SYMBOL(host1x_job_unpin); diff --git a/drivers/gpu/host1x/syncpt.c b/drivers/gpu/host1x/syncpt.c index 6b7fdc1e2ed0..95589328ad52 100644 --- a/drivers/gpu/host1x/syncpt.c +++ b/drivers/gpu/host1x/syncpt.c @@ -73,7 +73,7 @@ static struct host1x_syncpt *host1x_syncpt_alloc(struct host1x *host, return NULL; } - name = kasprintf(GFP_KERNEL, "%02d-%s", sp->id, + name = kasprintf(GFP_KERNEL, "%02u-%s", sp->id, dev ? dev_name(dev) : NULL); if (!name) return NULL; @@ -110,12 +110,14 @@ EXPORT_SYMBOL(host1x_syncpt_incr_max); void host1x_syncpt_restore(struct host1x *host) { struct host1x_syncpt *sp_base = host->syncpt; - u32 i; + unsigned int i; for (i = 0; i < host1x_syncpt_nb_pts(host); i++) host1x_hw_syncpt_restore(host, sp_base + i); + for (i = 0; i < host1x_syncpt_nb_bases(host); i++) host1x_hw_syncpt_restore_wait_base(host, sp_base + i); + wmb(); } @@ -126,7 +128,7 @@ void host1x_syncpt_restore(struct host1x *host) void host1x_syncpt_save(struct host1x *host) { struct host1x_syncpt *sp_base = host->syncpt; - u32 i; + unsigned int i; for (i = 0; i < host1x_syncpt_nb_pts(host); i++) { if (host1x_syncpt_client_managed(sp_base + i)) @@ -146,6 +148,7 @@ void host1x_syncpt_save(struct host1x *host) u32 host1x_syncpt_load(struct host1x_syncpt *sp) { u32 val; + val = host1x_hw_syncpt_load(sp->host, sp); trace_host1x_syncpt_load_min(sp->id, val); @@ -157,10 +160,9 @@ u32 host1x_syncpt_load(struct host1x_syncpt *sp) */ u32 host1x_syncpt_load_wait_base(struct host1x_syncpt *sp) { - u32 val; host1x_hw_syncpt_load_wait_base(sp->host, sp); - val = sp->base_val; - return val; + + return sp->base_val; } /* @@ -179,6 +181,7 @@ EXPORT_SYMBOL(host1x_syncpt_incr); static bool syncpt_load_min_is_expired(struct host1x_syncpt *sp, u32 thresh) { host1x_hw_syncpt_load(sp->host, sp); + return host1x_syncpt_is_expired(sp, thresh); } @@ -186,7 +189,7 @@ static bool syncpt_load_min_is_expired(struct host1x_syncpt *sp, u32 thresh) * Main entrypoint for syncpoint value waits. */ int host1x_syncpt_wait(struct host1x_syncpt *sp, u32 thresh, long timeout, - u32 *value) + u32 *value) { DECLARE_WAIT_QUEUE_HEAD_ONSTACK(wq); void *ref; @@ -201,6 +204,7 @@ int host1x_syncpt_wait(struct host1x_syncpt *sp, u32 thresh, long timeout, if (host1x_syncpt_is_expired(sp, thresh)) { if (value) *value = host1x_syncpt_load(sp); + return 0; } @@ -209,6 +213,7 @@ int host1x_syncpt_wait(struct host1x_syncpt *sp, u32 thresh, long timeout, if (host1x_syncpt_is_expired(sp, thresh)) { if (value) *value = val; + goto done; } @@ -239,32 +244,42 @@ int host1x_syncpt_wait(struct host1x_syncpt *sp, u32 thresh, long timeout, /* wait for the syncpoint, or timeout, or signal */ while (timeout) { long check = min_t(long, SYNCPT_CHECK_PERIOD, timeout); - int remain = wait_event_interruptible_timeout(wq, + int remain; + + remain = wait_event_interruptible_timeout(wq, syncpt_load_min_is_expired(sp, thresh), check); if (remain > 0 || host1x_syncpt_is_expired(sp, thresh)) { if (value) *value = host1x_syncpt_load(sp); + err = 0; + break; } + if (remain < 0) { err = remain; break; } + timeout -= check; + if (timeout && check_count <= MAX_STUCK_CHECK_COUNT) { dev_warn(sp->host->dev, - "%s: syncpoint id %d (%s) stuck waiting %d, timeout=%ld\n", + "%s: syncpoint id %u (%s) stuck waiting %d, timeout=%ld\n", current->comm, sp->id, sp->name, thresh, timeout); host1x_debug_dump_syncpts(sp->host); + if (check_count == MAX_STUCK_CHECK_COUNT) host1x_debug_dump(sp->host); + check_count++; } } + host1x_intr_put_ref(sp->host, sp->id, ref); done: @@ -279,7 +294,9 @@ bool host1x_syncpt_is_expired(struct host1x_syncpt *sp, u32 thresh) { u32 current_val; u32 future_val; + smp_rmb(); + current_val = (u32)atomic_read(&sp->min_val); future_val = (u32)atomic_read(&sp->max_val); @@ -341,14 +358,14 @@ int host1x_syncpt_init(struct host1x *host) { struct host1x_syncpt_base *bases; struct host1x_syncpt *syncpt; - int i; + unsigned int i; - syncpt = devm_kzalloc(host->dev, sizeof(*syncpt) * host->info->nb_pts, + syncpt = devm_kcalloc(host->dev, host->info->nb_pts, sizeof(*syncpt), GFP_KERNEL); if (!syncpt) return -ENOMEM; - bases = devm_kzalloc(host->dev, sizeof(*bases) * host->info->nb_bases, + bases = devm_kcalloc(host->dev, host->info->nb_bases, sizeof(*bases), GFP_KERNEL); if (!bases) return -ENOMEM; @@ -378,6 +395,7 @@ struct host1x_syncpt *host1x_syncpt_request(struct device *dev, unsigned long flags) { struct host1x *host = dev_get_drvdata(dev->parent); + return host1x_syncpt_alloc(host, dev, flags); } EXPORT_SYMBOL(host1x_syncpt_request); @@ -398,8 +416,9 @@ EXPORT_SYMBOL(host1x_syncpt_free); void host1x_syncpt_deinit(struct host1x *host) { - int i; struct host1x_syncpt *sp = host->syncpt; + unsigned int i; + for (i = 0; i < host->info->nb_pts; i++, sp++) kfree(sp->name); } @@ -407,10 +426,11 @@ void host1x_syncpt_deinit(struct host1x *host) /* * Read max. It indicates how many operations there are in queue, either in * channel or in a software thread. - * */ + */ u32 host1x_syncpt_read_max(struct host1x_syncpt *sp) { smp_rmb(); + return (u32)atomic_read(&sp->max_val); } EXPORT_SYMBOL(host1x_syncpt_read_max); @@ -421,6 +441,7 @@ EXPORT_SYMBOL(host1x_syncpt_read_max); u32 host1x_syncpt_read_min(struct host1x_syncpt *sp) { smp_rmb(); + return (u32)atomic_read(&sp->min_val); } EXPORT_SYMBOL(host1x_syncpt_read_min); @@ -431,25 +452,26 @@ u32 host1x_syncpt_read(struct host1x_syncpt *sp) } EXPORT_SYMBOL(host1x_syncpt_read); -int host1x_syncpt_nb_pts(struct host1x *host) +unsigned int host1x_syncpt_nb_pts(struct host1x *host) { return host->info->nb_pts; } -int host1x_syncpt_nb_bases(struct host1x *host) +unsigned int host1x_syncpt_nb_bases(struct host1x *host) { return host->info->nb_bases; } -int host1x_syncpt_nb_mlocks(struct host1x *host) +unsigned int host1x_syncpt_nb_mlocks(struct host1x *host) { return host->info->nb_mlocks; } -struct host1x_syncpt *host1x_syncpt_get(struct host1x *host, u32 id) +struct host1x_syncpt *host1x_syncpt_get(struct host1x *host, unsigned int id) { if (host->info->nb_pts < id) return NULL; + return host->syncpt + id; } EXPORT_SYMBOL(host1x_syncpt_get); diff --git a/drivers/gpu/host1x/syncpt.h b/drivers/gpu/host1x/syncpt.h index 9056465ecd3f..f719205105ac 100644 --- a/drivers/gpu/host1x/syncpt.h +++ b/drivers/gpu/host1x/syncpt.h @@ -37,7 +37,7 @@ struct host1x_syncpt_base { }; struct host1x_syncpt { - int id; + unsigned int id; atomic_t min_val; atomic_t max_val; u32 base_val; @@ -58,13 +58,13 @@ int host1x_syncpt_init(struct host1x *host); void host1x_syncpt_deinit(struct host1x *host); /* Return number of sync point supported. */ -int host1x_syncpt_nb_pts(struct host1x *host); +unsigned int host1x_syncpt_nb_pts(struct host1x *host); /* Return number of wait bases supported. */ -int host1x_syncpt_nb_bases(struct host1x *host); +unsigned int host1x_syncpt_nb_bases(struct host1x *host); /* Return number of mlocks supported. */ -int host1x_syncpt_nb_mlocks(struct host1x *host); +unsigned int host1x_syncpt_nb_mlocks(struct host1x *host); /* * Check sync point sanity. If max is larger than min, there have too many |