diff options
author | Lars-Peter Clausen <lars@metafoo.de> | 2010-06-19 22:41:27 +0200 |
---|---|---|
committer | Lars-Peter Clausen <lars@metafoo.de> | 2010-06-19 22:41:27 +0200 |
commit | 5c40321e6e6a8ca8552c5c69d041d7f7f3205318 (patch) | |
tree | 1ee5916ea352a4dda4fbee92605b916b282a3ca9 | |
parent | 82af18df11de291fdfed385a510b90b3d6425d64 (diff) | |
parent | 87a15e3aa7538bfd187eb44b7580512c59b3fb86 (diff) |
Merge branch 'glamo-2.6.34' into om-gta02-2.6.34
-rw-r--r-- | drivers/mfd/glamo/glamo-core.c | 23 | ||||
-rw-r--r-- | drivers/mfd/glamo/glamo-fb.c | 2 | ||||
-rw-r--r-- | drivers/mfd/glamo/glamo-mci.c | 197 |
3 files changed, 115 insertions, 107 deletions
diff --git a/drivers/mfd/glamo/glamo-core.c b/drivers/mfd/glamo/glamo-core.c index 23073fedea2..3cb48b28570 100644 --- a/drivers/mfd/glamo/glamo-core.c +++ b/drivers/mfd/glamo/glamo-core.c @@ -743,14 +743,14 @@ EXPORT_SYMBOL_GPL(glamo_engine_reclock); * allows reversion to 2.6.24 settings */ static const uint16_t reg_0x200[] = { -0xe03, /* 0 waits on Async BB R & W, Use PLL 2 for mem bus */ -0xef0, /* 3 waits on Async BB R & W, Use PLL 1 for mem bus */ -0xea0, /* 2 waits on Async BB R & W, Use PLL 1 for mem bus */ -0xe50, /* 1 waits on Async BB R & W, Use PLL 1 for mem bus */ -0xe00, /* 0 waits on Async BB R & W, Use PLL 1 for mem bus */ -0xef3, /* 3 waits on Async BB R & W, Use PLL 2 for mem bus */ -0xea3, /* 2 waits on Async BB R & W, Use PLL 2 for mem bus */ -0xe53, /* 1 waits on Async BB R & W, Use PLL 2 for mem bus */ + 0xe03, /* 0 waits on Async BB R & W, Use PLL 2 for mem bus */ + 0xef0, /* 3 waits on Async BB R & W, Use PLL 1 for mem bus */ + 0xea0, /* 2 waits on Async BB R & W, Use PLL 1 for mem bus */ + 0xe50, /* 1 waits on Async BB R & W, Use PLL 1 for mem bus */ + 0xe00, /* 0 waits on Async BB R & W, Use PLL 1 for mem bus */ + 0xef3, /* 3 waits on Async BB R & W, Use PLL 2 for mem bus */ + 0xea3, /* 2 waits on Async BB R & W, Use PLL 2 for mem bus */ + 0xe53, /* 1 waits on Async BB R & W, Use PLL 2 for mem bus */ }; static int glamo_run_script(struct glamo_core *glamo, @@ -779,7 +779,7 @@ static int glamo_run_script(struct glamo_core *glamo, break; case 0x200: __reg_write(glamo, line->reg, - reg_0x200[slow_memory & 0x8]); + reg_0x200[slow_memory & 0x7]); break; default: __reg_write(glamo, line->reg, line->val); @@ -793,7 +793,7 @@ static int glamo_run_script(struct glamo_core *glamo, static const struct glamo_script glamo_init_script[] = { { GLAMO_REG_CLOCK_HOST, 0x1000 }, { GLAMO_SCRIPT_WAIT, 2 }, - { GLAMO_REG_CLOCK_MEMORY, 0x1000 }, + { GLAMO_REG_CLOCK_MEMORY, 0x1000 }, { GLAMO_REG_CLOCK_MEMORY, 0x2000 }, { GLAMO_REG_CLOCK_LCD, 0x1000 }, { GLAMO_REG_CLOCK_MMC, 0x1000 }, @@ -995,7 +995,6 @@ static int __devinit glamo_probe(struct platform_device *pdev) set_irq_type(glamo->irq, IRQ_TYPE_EDGE_FALLING); set_irq_data(glamo->irq, glamo); set_irq_chained_handler(glamo->irq, glamo_irq_demux_handler); - glamo->irq_works = 1; ret = mfd_add_devices(&pdev->dev, pdev->id, glamo_cells, ARRAY_SIZE(glamo_cells), mem, glamo->irq_base); @@ -1247,7 +1246,7 @@ static int glamo_resume(struct device *dev) return 0; } -static struct dev_pm_ops glamo_pm_ops = { +static const struct dev_pm_ops glamo_pm_ops = { .suspend = glamo_suspend, .resume = glamo_resume, .poweroff = glamo_suspend, diff --git a/drivers/mfd/glamo/glamo-fb.c b/drivers/mfd/glamo/glamo-fb.c index cffb18032fa..29c16f019aa 100644 --- a/drivers/mfd/glamo/glamo-fb.c +++ b/drivers/mfd/glamo/glamo-fb.c @@ -943,7 +943,7 @@ static int glamofb_resume(struct device *dev) return 0; } -static struct dev_pm_ops glamofb_pm_ops = { +static const struct dev_pm_ops glamofb_pm_ops = { .suspend = glamofb_suspend, .resume = glamofb_resume, }; diff --git a/drivers/mfd/glamo/glamo-mci.c b/drivers/mfd/glamo/glamo-mci.c index 1d3f9af6385..4c372e0f459 100644 --- a/drivers/mfd/glamo/glamo-mci.c +++ b/drivers/mfd/glamo/glamo-mci.c @@ -52,17 +52,14 @@ struct glamo_mci_host { unsigned char request_counter; - struct timer_list disable_timer; - + struct workqueue_struct *workqueue; struct work_struct read_work; - - unsigned clk_enabled:1; }; static void glamo_mci_send_request(struct mmc_host *mmc, - struct mmc_request *mrq); + struct mmc_request *mrq); static void glamo_mci_send_command(struct glamo_mci_host *host, - struct mmc_command *cmd); + struct mmc_command *cmd); /* * Max SD clock rate @@ -121,14 +118,14 @@ static inline void glamo_reg_write(struct glamo_mci_host *glamo, } static inline uint16_t glamo_reg_read(struct glamo_mci_host *glamo, - uint16_t reg) + uint16_t reg) { return readw(glamo->mmio_base + reg); } static void glamo_reg_set_bit_mask(struct glamo_mci_host *glamo, - uint16_t reg, uint16_t mask, - uint16_t val) + uint16_t reg, uint16_t mask, + uint16_t val) { uint16_t tmp; @@ -156,25 +153,22 @@ static void glamo_mci_reset(struct glamo_mci_host *host) } -static void glamo_mci_clock_disable(struct glamo_mci_host *host) +static int glamo_mci_clock_disable(struct mmc_host *mmc, int lazy) { - glamo_engine_suspend(host->core, GLAMO_ENGINE_MMC); -} - -static void glamo_mci_clock_enable(struct glamo_mci_host *host) -{ - del_timer_sync(&host->disable_timer); - - glamo_engine_enable(host->core, GLAMO_ENGINE_MMC); + struct glamo_mci_host *host = mmc_priv(mmc); + glamo_engine_suspend(host->core, GLAMO_ENGINE_MMC); + return 0; } -static void glamo_mci_disable_timer(unsigned long data) +static int glamo_mci_clock_enable(struct mmc_host *mmc) { - struct glamo_mci_host *host = (struct glamo_mci_host *)data; - glamo_mci_clock_disable(host); + struct glamo_mci_host *host = mmc_priv(mmc); + glamo_engine_enable(host->core, GLAMO_ENGINE_MMC); + return 0; } +#ifndef GLAMO_MCI_WORKER static void do_pio_read(struct glamo_mci_host *host, struct mmc_data *data) { struct sg_mapping_iter miter; @@ -196,6 +190,7 @@ static void do_pio_read(struct glamo_mci_host *host, struct mmc_data *data) dev_dbg(&host->pdev->dev, "pio_read(): " "complete (no more data).\n"); } +#endif static void do_pio_write(struct glamo_mci_host *host, struct mmc_data *data) { @@ -220,19 +215,15 @@ static int glamo_mci_set_card_clock(struct glamo_mci_host *host, int freq) { int real_rate = 0; - if (freq) { - glamo_mci_clock_enable(host); + if (freq) real_rate = glamo_engine_reclock(host->core, GLAMO_ENGINE_MMC, - freq); - } else { - glamo_mci_clock_disable(host); - } + freq); return real_rate; } static int glamo_mci_wait_idle(struct glamo_mci_host *host, - unsigned long timeout) + unsigned long timeout) { uint16_t status; do { @@ -248,10 +239,9 @@ static int glamo_mci_wait_idle(struct glamo_mci_host *host, return 0; } -static void glamo_mci_request_done(struct glamo_mci_host *host, struct -mmc_request *mrq) +static void glamo_mci_request_done(struct glamo_mci_host *host, + struct mmc_request *mrq) { - mod_timer(&host->disable_timer, jiffies + HZ / 16); mmc_request_done(host->mmc, mrq); } @@ -268,11 +258,6 @@ static irqreturn_t glamo_mci_irq(int irq, void *data) mrq = host->mrq; cmd = mrq->cmd; -#if 0 - if (cmd->data->flags & MMC_DATA_READ) - return; -#endif - status = glamo_reg_read(host, GLAMO_REG_MMC_RB_STAT1); dev_dbg(&host->pdev->dev, "status = 0x%04x\n", status); @@ -295,7 +280,11 @@ static irqreturn_t glamo_mci_irq(int irq, void *data) glamo_mci_send_command(host, mrq->stop); if (cmd->data->flags & MMC_DATA_READ) +#ifndef GLAMO_MCI_WORKER do_pio_read(host, cmd->data); +#else + flush_workqueue(host->workqueue); +#endif if (mrq->stop) mrq->stop->error = glamo_mci_wait_idle(host, jiffies + HZ); @@ -307,6 +296,7 @@ done: return IRQ_HANDLED; } +#ifdef GLAMO_MCI_WORKER static void glamo_mci_read_worker(struct work_struct *work) { struct glamo_mci_host *host = container_of(work, struct glamo_mci_host, @@ -324,6 +314,13 @@ static void glamo_mci_read_worker(struct work_struct *work) cmd = host->mrq->cmd; sg = cmd->data->sg; do { + /* + * TODO: How to get rid of that? + * Maybe just drop it... In fact, it is already handled in + * the IRQ handler, maybe we should only check cmd->error. + * But the question is: what happens between the moment + * the error occurs, and the moment the IRQ handler handles it? + */ status = glamo_reg_read(host, GLAMO_REG_MMC_RB_STAT1); if (status & (GLAMO_STAT1_MMC_RTOUT | GLAMO_STAT1_MMC_DTOUT)) @@ -333,7 +330,7 @@ static void glamo_mci_read_worker(struct work_struct *work) if (cmd->error) { dev_info(&host->pdev->dev, "Error after cmd: 0x%x\n", status); - goto done; + return; } blocks_ready = glamo_reg_read(host, GLAMO_REG_MMC_RB_BLKCNT); @@ -348,29 +345,17 @@ static void glamo_mci_read_worker(struct work_struct *work) from_ptr += sg->length >> 1; data_read += sg->length; + sg = sg_next(sg); } } while (sg); cmd->data->bytes_xfered = data_read; - - do { - status = glamo_reg_read(host, GLAMO_REG_MMC_RB_STAT1); - } while (!(status & GLAMO_STAT1_MMC_IDLE)); - - if (host->mrq->stop) - glamo_mci_send_command(host, host->mrq->stop); - - do { - status = glamo_reg_read(host, GLAMO_REG_MMC_RB_STAT1); - } while (!(status & GLAMO_STAT1_MMC_IDLE)); -done: - host->mrq = NULL; - glamo_mci_request_done(host, cmd->mrq); } +#endif static void glamo_mci_send_command(struct glamo_mci_host *host, - struct mmc_command *cmd) + struct mmc_command *cmd) { uint8_t u8a[6]; uint16_t fire = 0; @@ -520,14 +505,13 @@ static void glamo_mci_send_command(struct glamo_mci_host *host, status = glamo_reg_read(host, GLAMO_REG_MMC_RB_STAT1); while (((((status >> 15) & 1) != (host->request_counter & 1)) || (!(status & (GLAMO_STAT1_MMC_RB_RRDY | - GLAMO_STAT1_MMC_RTOUT | - GLAMO_STAT1_MMC_DTOUT | - GLAMO_STAT1_MMC_BWERR | - GLAMO_STAT1_MMC_BRERR)))) && (timeout--)); - - if ((status & (GLAMO_STAT1_MMC_RTOUT | - GLAMO_STAT1_MMC_DTOUT)) || - (timeout == 0)) { + GLAMO_STAT1_MMC_RTOUT | + GLAMO_STAT1_MMC_DTOUT | + GLAMO_STAT1_MMC_BWERR | + GLAMO_STAT1_MMC_BRERR)))) && (timeout--)); + + if ((status & (GLAMO_STAT1_MMC_RTOUT | GLAMO_STAT1_MMC_DTOUT)) || + (timeout == 0)) { cmd->error = -ETIMEDOUT; } else if (status & (GLAMO_STAT1_MMC_BWERR | GLAMO_STAT1_MMC_BRERR)) { cmd->error = -EILSEQ; @@ -550,12 +534,12 @@ static void glamo_mci_send_command(struct glamo_mci_host *host, } } -#if 0 +#ifdef GLAMO_MCI_WORKER /* We'll only get an interrupt when all data has been transfered. By starting to copy data when it's avaiable we can increase throughput by up to 30%. */ if (cmd->data && (cmd->data->flags & MMC_DATA_READ)) - schedule_work(&host->read_work); + queue_work(host->workqueue, &host->read_work); #endif } @@ -579,12 +563,11 @@ static int glamo_mci_prepare_pio(struct glamo_mci_host *host, } static void glamo_mci_send_request(struct mmc_host *mmc, - struct mmc_request *mrq) + struct mmc_request *mrq) { struct glamo_mci_host *host = mmc_priv(mmc); struct mmc_command *cmd = mrq->cmd; - glamo_mci_clock_enable(host); host->request_counter++; if (cmd->data) { if (glamo_mci_prepare_pio(host, cmd->data)) { @@ -622,7 +605,7 @@ done: } static void glamo_mci_set_power_mode(struct glamo_mci_host *host, - unsigned char power_mode) + unsigned char power_mode) { int ret; @@ -643,8 +626,6 @@ static void glamo_mci_set_power_mode(struct glamo_mci_host *host, break; case MMC_POWER_OFF: default: - glamo_engine_disable(host->core, GLAMO_ENGINE_MMC); - ret = regulator_disable(host->regulator); if (ret) dev_warn(&host->pdev->dev, @@ -663,6 +644,8 @@ static void glamo_mci_set_ios(struct mmc_host *mmc, struct mmc_ios *ios) int sd_drive; int ret; + mmc_host_enable(mmc); + /* Set power */ glamo_mci_set_power_mode(host, ios->power_mode); @@ -674,6 +657,7 @@ static void glamo_mci_set_ios(struct mmc_host *mmc, struct mmc_ios *ios) else host->vdd = ios->vdd; } + rate = glamo_mci_set_card_clock(host, ios->clock); if ((ios->power_mode == MMC_POWER_ON) || @@ -696,8 +680,13 @@ static void glamo_mci_set_ios(struct mmc_host *mmc, struct mmc_ios *ios) sd_drive = 3; glamo_reg_set_bit_mask(host, GLAMO_REG_MMC_BASIC, - GLAMO_BASIC_MMC_EN_4BIT_DATA | 0xb0, + GLAMO_BASIC_MMC_EN_4BIT_DATA | 0xc0, bus_width | sd_drive << 6); + + if (host->power_mode == MMC_POWER_OFF) + mmc_host_disable(host->mmc); + else + mmc_host_lazy_disable(host->mmc); } @@ -710,6 +699,8 @@ static int glamo_mci_get_ro(struct mmc_host *mmc) } static struct mmc_host_ops glamo_mci_ops = { + .enable = glamo_mci_clock_enable, + .disable = glamo_mci_clock_disable, .request = glamo_mci_send_request, .set_ios = glamo_mci_set_ios, .get_ro = glamo_mci_get_ro, @@ -736,11 +727,13 @@ static int glamo_mci_probe(struct platform_device *pdev) if (core->pdata) host->pdata = core->pdata->mmc_data; host->power_mode = MMC_POWER_OFF; - host->clk_enabled = 0; host->core = core; host->irq = platform_get_irq(pdev, 0); +#ifdef GLAMO_MCI_WORKER INIT_WORK(&host->read_work, glamo_mci_read_worker); + host->workqueue = create_singlethread_workqueue("glamo-mci-read"); +#endif host->regulator = regulator_get(pdev->dev.parent, "SD_3V3"); if (IS_ERR(host->regulator)) { @@ -758,8 +751,8 @@ static int glamo_mci_probe(struct platform_device *pdev) } host->mmio_mem = request_mem_region(host->mmio_mem->start, - resource_size(host->mmio_mem), - pdev->name); + resource_size(host->mmio_mem), + pdev->name); if (!host->mmio_mem) { dev_err(&pdev->dev, "failed to request io memory region.\n"); @@ -768,7 +761,7 @@ static int glamo_mci_probe(struct platform_device *pdev) } host->mmio_base = ioremap(host->mmio_mem->start, - resource_size(host->mmio_mem)); + resource_size(host->mmio_mem)); if (!host->mmio_base) { dev_err(&pdev->dev, "failed to ioremap() io memory region.\n"); ret = -EINVAL; @@ -786,8 +779,8 @@ static int glamo_mci_probe(struct platform_device *pdev) } host->data_mem = request_mem_region(host->data_mem->start, - resource_size(host->data_mem), - pdev->name); + resource_size(host->data_mem), + pdev->name); if (!host->data_mem) { dev_err(&pdev->dev, "failed to request io memory region.\n"); @@ -795,7 +788,7 @@ static int glamo_mci_probe(struct platform_device *pdev) goto probe_iounmap_mmio; } host->data_base = ioremap(host->data_mem->start, - resource_size(host->data_mem)); + resource_size(host->data_mem)); if (host->data_base == 0) { dev_err(&pdev->dev, "failed to ioremap() io memory region.\n"); @@ -804,7 +797,7 @@ static int glamo_mci_probe(struct platform_device *pdev) } ret = request_threaded_irq(host->irq, NULL, glamo_mci_irq, IRQF_SHARED, - pdev->name, host); + pdev->name, host); if (ret) { dev_err(&pdev->dev, "failed to register irq.\n"); goto probe_iounmap_data; @@ -839,36 +832,41 @@ static int glamo_mci_probe(struct platform_device *pdev) platform_set_drvdata(pdev, mmc); - glamo_engine_enable(host->core, GLAMO_ENGINE_MMC); - glamo_mci_reset(host); - glamo_engine_disable(host->core, GLAMO_ENGINE_MMC); + mmc->caps |= MMC_CAP_DISABLE; + mmc_set_disable_delay(mmc, 1000 / 16); - setup_timer(&host->disable_timer, glamo_mci_disable_timer, - (unsigned long)host); + mmc_host_enable(mmc); + glamo_mci_reset(host); ret = mmc_add_host(mmc); if (ret) { dev_err(&pdev->dev, "failed to add mmc host.\n"); - goto probe_freeirq; + goto probe_mmc_host_disable; } + mmc_host_lazy_disable(mmc); + return 0; -probe_freeirq: +probe_mmc_host_disable: + mmc_host_disable(mmc); free_irq(host->irq, host); probe_iounmap_data: iounmap(host->data_base); probe_free_mem_region_data: release_mem_region(host->data_mem->start, - resource_size(host->data_mem)); + resource_size(host->data_mem)); probe_iounmap_mmio: iounmap(host->mmio_base); probe_free_mem_region_mmio: release_mem_region(host->mmio_mem->start, - resource_size(host->mmio_mem)); + resource_size(host->mmio_mem)); probe_regulator_put: regulator_put(host->regulator); probe_free_host: +#ifdef GLAMO_MCI_WORKER + destroy_workqueue(host->workqueue); +#endif mmc_free_host(mmc); probe_out: return ret; @@ -879,9 +877,18 @@ static int glamo_mci_remove(struct platform_device *pdev) struct mmc_host *mmc = platform_get_drvdata(pdev); struct glamo_mci_host *host = mmc_priv(mmc); - free_irq(host->irq, host); +#ifdef GLAMO_MCI_WORKER + flush_workqueue(host->workqueue); + destroy_workqueue(host->workqueue); +#endif + mmc_host_enable(mmc); mmc_remove_host(mmc); + mmc_host_disable(mmc); + + synchronize_irq(host->irq); + free_irq(host->irq, host); + iounmap(host->mmio_base); iounmap(host->data_base); release_mem_region(host->mmio_mem->start, @@ -893,7 +900,6 @@ static int glamo_mci_remove(struct platform_device *pdev) mmc_free_host(mmc); - glamo_engine_disable(host->core, GLAMO_ENGINE_MMC); return 0; } @@ -908,7 +914,9 @@ static int glamo_mci_suspend(struct device *dev) disable_irq(host->irq); + mmc_host_enable(mmc); ret = mmc_suspend_host(mmc, PMSG_SUSPEND); + mmc_host_disable(mmc); return ret; } @@ -919,23 +927,24 @@ static int glamo_mci_resume(struct device *dev) struct glamo_mci_host *host = mmc_priv(mmc); int ret; - + mmc_host_enable(mmc); glamo_mci_reset(host); - glamo_engine_enable(host->core, GLAMO_ENGINE_MMC); - mdelay(10); + mdelay(10); ret = mmc_resume_host(host->mmc); enable_irq(host->irq); - return 0; + mmc_host_lazy_disable(host->mmc); + + return ret; } -static struct dev_pm_ops glamo_mci_pm_ops = { +static const struct dev_pm_ops glamo_mci_pm_ops = { .suspend = glamo_mci_suspend, .resume = glamo_mci_resume, - .freeze = glamo_mci_suspend, - .thaw = glamo_mci_resume, + .freeze = glamo_mci_suspend, + .thaw = glamo_mci_resume, }; #define GLAMO_MCI_PM_OPS (&glamo_mci_pm_ops) |