mmc: sdhci-pci: Respect PM flags when enabling card detect GPIO IRQ wakeup
Adrian Hunter
maintainercover mmc: sdhci-pci: Respect PM flags when enabling card detect GPIO IRQ wakeup
Hi
Commit 03dbaa04a2e5 ("mmc: slot-gpio: Add support to enable irq wake on cd_irq") enabled wakeup irrespective of the host controller's PM flags. However, users also want to control it from sysfs power/wakeup attribute. That means the driver needs to check the PM flags before enabling it in the suspend callback. Patch 9 adds support for that in sdhci-pci, which is the only driver presently using the MMC_CAP_CD_WAKE flag. Patches 1 - 7 tidy up aspects of sdhci and sdhci-pci wakeup handling, and patch 8 adds a helper function to make it easy for drivers.
There are 2 contentious aspects to this patch set:
1) An existing problem with the SDIO API which does not let the host controller driver know that the SDIO function driver has requested SDIO card interrupt wakeup until the suspend callback - which results in the host controller driver having to enable or disable wakeup in the suspend callback. Fixing the SDIO API is a separate issue IMHO.
2) In order to use the sysfs power/wakeup attribute, the driver must set the device as wake capable even when it is really a GPIO that wakes the system.
Adrian Hunter (9):
mmc: sdhci-pci: Stop calling sdhci_enable_irq_wakeups()
mmc: sdhci-pci: Use device wakeup capability to determine MMC_PM_WAKE_SDIO_IRQ capability
mmc: sdhci: Stop exporting sdhci_enable_irq_wakeups()
mmc: sdhci: Handle failure of enable_irq_wake()
mmc: sdhci: Rework sdhci_enable_irq_wakeups()
mmc: sdhci: Do not unnecessarily enable wakeup for card detect interrupt
mmc: sdhci: Do not unnecessarily enable wakeup for SDIO card interrupt
mmc: slot-gpio: Add a function to enable/disable card detect IRQ wakeup
mmc: sdhci-pci: Respect PM flags when enabling card detect GPIO IRQ wakeup
drivers/mmc/core/core.c | 3 +- drivers/mmc/core/slot-gpio.c | 25 ++++++++++++-- drivers/mmc/host/sdhci-pci-core.c | 73 +++++++++++++++++++-------------------- drivers/mmc/host/sdhci.c | 67 ++++++++++++++++++++++++----------- drivers/mmc/host/sdhci.h | 2 +- include/linux/mmc/slot-gpio.h | 1 + 6 files changed, 108 insertions(+), 63 deletions(-)
Regards
Adrian
Adrian Hunter
maintainerPATCH RESEND 1/9 mmc: sdhci-pci: Stop calling sdhci_enable_irq_wakeups()
sdhci_enable_irq_wakeups() is already called by sdhci_suspend_host() so sdhci-pci should not need to call it. However sdhci_suspend_host() only calls it if wakeups are enabled, and sdhci-pci does not enable them until after calling sdhci_suspend_host(). So move the calls to sdhci_pci_init_wakeup() before calling sdhci_suspend_host(), and stop calling sdhci_enable_irq_wakeups(). That results in some simplification because sdhci_pci_suspend_host() and __sdhci_pci_suspend_host() no longer need to be separate functions.
--- drivers/mmc/host/sdhci-pci-core.c | 58 ++++++++++++++------------------------- 1 file changed, 21 insertions(+), 37 deletions(-)
drivers/mmc/host/sdhci-pci-core.c +21 -37
@@ -38,10 +38,29 @@static void sdhci_pci_hw_reset (struct sdhci_host *host);#ifdef CONFIG_PM_SLEEP -static int __sdhci_pci_suspend_host (struct sdhci_pci_chip *chip) +static int sdhci_pci_init_wakeup (struct sdhci_pci_chip *chip) +{ +mmc_pm_flag_t pm_flags =0 ; +int i; + +for (i =0 ; i < chip->num_slots; i++) { +struct sdhci_pci_slot *slot = chip->slots[i]; + +if (slot) + pm_flags |= slot->host->mmc->pm_flags; + } + +return device_init_wakeup (&chip->pdev->dev, + (pm_flags &MMC_PM_KEEP_POWER ) && + (pm_flags &MMC_PM_WAKE_SDIO_IRQ )); +} + +static int sdhci_pci_suspend_host (struct sdhci_pci_chip *chip) {int i, ret; +sdhci_pci_init_wakeup (chip); +for (i =0 ; i < chip->num_slots; i++) {struct sdhci_pci_slot *slot = chip->slots[i];struct sdhci_host *host;
@@ -57,9 +76,6 @@ static int __sdhci_pci_suspend_host(struct sdhci_pci_chip *chip) ret =sdhci_suspend_host (host);if (ret) goto err_pci_suspend; - -if (host->mmc->pm_flags &MMC_PM_WAKE_SDIO_IRQ ) -sdhci_enable_irq_wakeups (host); }return 0 ;
@@ -70,36 +86,6 @@ static int __sdhci_pci_suspend_host(struct sdhci_pci_chip *chip)return ret; } -static int sdhci_pci_init_wakeup (struct sdhci_pci_chip *chip) -{ -mmc_pm_flag_t pm_flags =0 ; -int i; - -for (i =0 ; i < chip->num_slots; i++) { -struct sdhci_pci_slot *slot = chip->slots[i]; - -if (slot) - pm_flags |= slot->host->mmc->pm_flags; - } - -return device_init_wakeup (&chip->pdev->dev, - (pm_flags &MMC_PM_KEEP_POWER ) && - (pm_flags &MMC_PM_WAKE_SDIO_IRQ )); -} - -static int sdhci_pci_suspend_host (struct sdhci_pci_chip *chip) -{ -int ret; - - ret =__sdhci_pci_suspend_host (chip); -if (ret) -return ret; - -sdhci_pci_init_wakeup (chip); - -return 0 ; -} -int sdhci_pci_resume_host (struct sdhci_pci_chip *chip) {struct sdhci_pci_slot *slot;
@@ -1107,7 +1093,7 @@ static int jmicron_suspend(struct sdhci_pci_chip *chip) {int i, ret; - ret =__sdhci_pci_suspend_host (chip); + ret =sdhci_pci_suspend_host (chip);if (ret)return ret;
@@ -1117,8 +1103,6 @@ static int jmicron_suspend(struct sdhci_pci_chip *chip)jmicron_enable_mmc (chip->slots[i]->host,0 ); } -sdhci_pci_init_wakeup (chip); -return 0 ; }
Signature · 1 line
1.9.1
Ulf Hansson
AppliedRe: [PATCH RESEND 1/9] mmc: sdhci-pci: Stop calling sdhci_enable_irq_wakeups()
Quoted 10 lines · Adrian Hunter
sdhci_enable_irq_wakeups()is already called bysdhci_suspend_host()so sdhci-pci should not need to call it. Howeversdhci_suspend_host()only calls it if wakeups are enabled, and sdhci-pci does not enable them until after callingsdhci_suspend_host(). So move the calls tosdhci_pci_init_wakeup()before callingsdhci_suspend_host(), and stop callingsdhci_enable_irq_wakeups(). That results in some simplification becausesdhci_pci_suspend_host()and__sdhci_pci_suspend_host()no longer need to be separate functions.Signed-off-by Adrian Hunter <[email protected]>
Thanks, applied for next!
Kind regards
Uffe
Quoted 107 lines
--- drivers/mmc/host/sdhci-pci-core.c | 58 ++++++++++++++------------------------- 1 file changed, 21 insertions(+), 37 deletions(-)drivers/mmc/host/sdhci-pci-core.c +21
-38@@ -38,10 +38,29 @@static void sdhci_pci_hw_reset (struct sdhci_host *host);#ifdef CONFIG_PM_SLEEP -static int __sdhci_pci_suspend_host (struct sdhci_pci_chip *chip) +static int sdhci_pci_init_wakeup (struct sdhci_pci_chip *chip) +{ +mmc_pm_flag_t pm_flags =0 ; +int i; + +for (i =0 ; i < chip->num_slots; i++) { +struct sdhci_pci_slot *slot = chip->slots[i]; + +if (slot) + pm_flags |= slot->host->mmc->pm_flags; + } + +return device_init_wakeup (&chip->pdev->dev, + (pm_flags &MMC_PM_KEEP_POWER ) && + (pm_flags &MMC_PM_WAKE_SDIO_IRQ )); +} + +static int sdhci_pci_suspend_host (struct sdhci_pci_chip *chip) {int i, ret; +sdhci_pci_init_wakeup (chip); +for (i =0 ; i < chip->num_slots; i++) {struct sdhci_pci_slot *slot = chip->slots[i];struct sdhci_host *host;@@ -57,9 +76,6 @@ static int __sdhci_pci_suspend_host(struct sdhci_pci_chip *chip) ret =sdhci_suspend_host (host);if (ret) goto err_pci_suspend; - -if (host->mmc->pm_flags &MMC_PM_WAKE_SDIO_IRQ ) -sdhci_enable_irq_wakeups (host); }return 0 ;@@ -70,36 +86,6 @@ static int __sdhci_pci_suspend_host(struct sdhci_pci_chip *chip)return ret; } -static int sdhci_pci_init_wakeup (struct sdhci_pci_chip *chip) -{ -mmc_pm_flag_t pm_flags =0 ; -int i; - -for (i =0 ; i < chip->num_slots; i++) { -struct sdhci_pci_slot *slot = chip->slots[i]; - -if (slot) - pm_flags |= slot->host->mmc->pm_flags; - } - -return device_init_wakeup (&chip->pdev->dev, - (pm_flags &MMC_PM_KEEP_POWER ) && - (pm_flags &MMC_PM_WAKE_SDIO_IRQ )); -} - -static int sdhci_pci_suspend_host (struct sdhci_pci_chip *chip) -{ -int ret; - - ret =__sdhci_pci_suspend_host (chip); -if (ret) -return ret; - -sdhci_pci_init_wakeup (chip); - -return 0 ; -} -int sdhci_pci_resume_host (struct sdhci_pci_chip *chip) {struct sdhci_pci_slot *slot;@@ -1107,7 +1093,7 @@ static int jmicron_suspend(struct sdhci_pci_chip *chip) {int i, ret; - ret =__sdhci_pci_suspend_host (chip); + ret =sdhci_pci_suspend_host (chip);if (ret)return ret;@@ -1117,8 +1103,6 @@ static int jmicron_suspend(struct sdhci_pci_chip *chip)jmicron_enable_mmc (chip->slots[i]->host,0 ); } -sdhci_pci_init_wakeup (chip); -return 0 ; } --1.9.1
Adrian Hunter
maintainerPATCH RESEND 2/9 mmc: sdhci-pci: Use device wakeup capability to determine MMC_PM_WAKE_SDIO_IRQ capability
In reply to Adrian HunterPCI and ACPI determine if a device is wakeup capable, so use that to determine the MMC_PM_WAKE_SDIO_IRQ capability correctly.
--- drivers/mmc/host/sdhci-pci-core.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)
drivers/mmc/host/sdhci-pci-core.c +7 -4
@@ -50,9 +50,9 @@ static int sdhci_pci_init_wakeup(struct sdhci_pci_chip *chip) pm_flags |= slot->host->mmc->pm_flags; } -return device_init_wakeup (&chip->pdev->dev, - (pm_flags &MMC_PM_KEEP_POWER ) && - (pm_flags &MMC_PM_WAKE_SDIO_IRQ )); +return device_set_wakeup_enable (&chip->pdev->dev, + (pm_flags &MMC_PM_KEEP_POWER ) && + (pm_flags &MMC_PM_WAKE_SDIO_IRQ )); }static int sdhci_pci_suspend_host (struct sdhci_pci_chip *chip)
@@ -1680,10 +1680,13 @@ static struct sdhci_pci_slot *sdhci_pci_probe_slot( } } - host->mmc->pm_caps =MMC_PM_KEEP_POWER |MMC_PM_WAKE_SDIO_IRQ ; + host->mmc->pm_caps =MMC_PM_KEEP_POWER ; host->mmc->slotno = slotno; host->mmc->caps2 |=MMC_CAP2_NO_PRESCAN_POWERUP ; +if (device_can_wakeup (&pdev->dev)) + host->mmc->pm_caps |=MMC_PM_WAKE_SDIO_IRQ ; +if (slot->cd_idx >=0 ) { ret =mmc_gpiod_request_cd (host->mmc,NULL , slot->cd_idx, slot->cd_override_level,0 ,NULL );
Signature · 1 line
1.9.1
Ulf Hansson
AppliedRe: [PATCH RESEND 2/9] mmc: sdhci-pci: Use device wakeup capability to determine MMC_PM_WAKE_SDIO_IRQ capability
Quoted 4 lines · Adrian Hunter
PCI and ACPI determine if a device is wakeup capable, so use that to determine the
MMC_PM_WAKE_SDIO_IRQcapability correctly.Signed-off-by Adrian Hunter <[email protected]>
Thanks, applied for next!
Kind regards
Uffe
Quoted 39 lines
--- drivers/mmc/host/sdhci-pci-core.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)drivers/mmc/host/sdhci-pci-core.c +7
-5@@ -50,9 +50,9 @@ static int sdhci_pci_init_wakeup(struct sdhci_pci_chip *chip) pm_flags |= slot->host->mmc->pm_flags; } -return device_init_wakeup (&chip->pdev->dev, - (pm_flags &MMC_PM_KEEP_POWER ) && - (pm_flags &MMC_PM_WAKE_SDIO_IRQ )); +return device_set_wakeup_enable (&chip->pdev->dev, + (pm_flags &MMC_PM_KEEP_POWER ) && + (pm_flags &MMC_PM_WAKE_SDIO_IRQ )); }static int sdhci_pci_suspend_host (struct sdhci_pci_chip *chip)@@ -1680,10 +1680,13 @@ static struct sdhci_pci_slot *sdhci_pci_probe_slot( } } - host->mmc->pm_caps =MMC_PM_KEEP_POWER |MMC_PM_WAKE_SDIO_IRQ ; + host->mmc->pm_caps =MMC_PM_KEEP_POWER ; host->mmc->slotno = slotno; host->mmc->caps2 |=MMC_CAP2_NO_PRESCAN_POWERUP ; +if (device_can_wakeup (&pdev->dev)) + host->mmc->pm_caps |=MMC_PM_WAKE_SDIO_IRQ ; +if (slot->cd_idx >=0 ) { ret =mmc_gpiod_request_cd (host->mmc,NULL , slot->cd_idx, slot->cd_override_level,0 ,NULL ); --1.9.1
Adrian Hunter
maintainerPATCH RESEND 3/9 mmc: sdhci: Stop exporting sdhci_enable_irq_wakeups()
In reply to Adrian HunterNow that it is not being used by any drivers, stop exporting it.
--- drivers/mmc/host/sdhci.c | 3 +-- drivers/mmc/host/sdhci.h | 1 - 2 files changed, 1 insertion(+), 3 deletions(-)
drivers/mmc/host/sdhci.c +1 -2
@@ -2821,7 +2821,7 @@ static irqreturn_t sdhci_thread_irq(int irq, void *dev_id)* sdhci_disable_irq_wakeups() since it will be set by * sdhci_enable_card_detection() or sdhci_init(). */ -void sdhci_enable_irq_wakeups (struct sdhci_host *host) +static void sdhci_enable_irq_wakeups (struct sdhci_host *host) {u8 val;u8 mask =SDHCI_WAKE_ON_INSERT |SDHCI_WAKE_ON_REMOVE
@@ -2839,7 +2839,6 @@ void sdhci_enable_irq_wakeups(struct sdhci_host *host)sdhci_writeb (host, val,SDHCI_WAKE_UP_CONTROL );sdhci_writel (host, irq_val,SDHCI_INT_ENABLE ); } -EXPORT_SYMBOL_GPL (sdhci_enable_irq_wakeups);static void sdhci_disable_irq_wakeups (struct sdhci_host *host) {
drivers/mmc/host/sdhci.h +0 -1
@@ -718,7 +718,6 @@ int sdhci_start_signal_voltage_switch(struct mmc_host *mmc,#ifdef CONFIG_PM int sdhci_suspend_host (struct sdhci_host *host);int sdhci_resume_host (struct sdhci_host *host); -void sdhci_enable_irq_wakeups (struct sdhci_host *host);int sdhci_runtime_suspend_host (struct sdhci_host *host);int sdhci_runtime_resume_host (struct sdhci_host *host);#endif
Signature · 1 line
1.9.1
Ulf Hansson
AppliedRe: [PATCH RESEND 3/9] mmc: sdhci: Stop exporting sdhci_enable_irq_wakeups()
Quoted 3 lines · Adrian Hunter
Now that it is not being used by any drivers, stop exporting it.
Signed-off-by Adrian Hunter <[email protected]>
Thanks, applied for next!
Kind regards
Uffe
Quoted 41 lines
--- drivers/mmc/host/sdhci.c | 3 +-- drivers/mmc/host/sdhci.h | 1 - 2 files changed, 1 insertion(+), 3 deletions(-)drivers/mmc/host/sdhci.c +1
-2@@ -2821,7 +2821,7 @@ static irqreturn_t sdhci_thread_irq(int irq, void *dev_id)* sdhci_disable_irq_wakeups() since it will be set by * sdhci_enable_card_detection() or sdhci_init(). */ -void sdhci_enable_irq_wakeups (struct sdhci_host *host) +static void sdhci_enable_irq_wakeups (struct sdhci_host *host) {u8 val;u8 mask =SDHCI_WAKE_ON_INSERT |SDHCI_WAKE_ON_REMOVE @@ -2839,7 +2839,6 @@ void sdhci_enable_irq_wakeups(struct sdhci_host *host)sdhci_writeb (host, val,SDHCI_WAKE_UP_CONTROL );sdhci_writel (host, irq_val,SDHCI_INT_ENABLE ); } -EXPORT_SYMBOL_GPL (sdhci_enable_irq_wakeups);static void sdhci_disable_irq_wakeups (struct sdhci_host *host) {drivers/mmc/host/sdhci.h +0
-2@@ -718,7 +718,6 @@ int sdhci_start_signal_voltage_switch(struct mmc_host *mmc,#ifdef CONFIG_PM int sdhci_suspend_host (struct sdhci_host *host);int sdhci_resume_host (struct sdhci_host *host); -void sdhci_enable_irq_wakeups (struct sdhci_host *host);int sdhci_runtime_suspend_host (struct sdhci_host *host);int sdhci_runtime_resume_host (struct sdhci_host *host);#endif --1.9.1
Adrian Hunter
maintainerPATCH RESEND 4/9 mmc: sdhci: Handle failure of enable_irq_wake()
In reply to Adrian HunterNow that sdhci_enable_irq_wakeups() is a local function, change it to return whether the IRQ wakeup was successfully enabled. This is in preparation for adding more conditions for whether IRQ wakeup is enabled.
Note it is assumed, for SDHCI devices, that suspend is more important than wakeup, so we continue to suspend regardless.
--- drivers/mmc/host/sdhci.c | 24 +++++++++++++++--------- drivers/mmc/host/sdhci.h | 1 + 2 files changed, 16 insertions(+), 9 deletions(-)
drivers/mmc/host/sdhci.c +15 -9
@@ -2821,7 +2821,7 @@ static irqreturn_t sdhci_thread_irq(int irq, void *dev_id)* sdhci_disable_irq_wakeups() since it will be set by * sdhci_enable_card_detection() or sdhci_init(). */ -static void sdhci_enable_irq_wakeups (struct sdhci_host *host) +static bool sdhci_enable_irq_wakeups (struct sdhci_host *host) {u8 val;u8 mask =SDHCI_WAKE_ON_INSERT |SDHCI_WAKE_ON_REMOVE
@@ -2838,6 +2838,10 @@ static void sdhci_enable_irq_wakeups(struct sdhci_host *host) }sdhci_writeb (host, val,SDHCI_WAKE_UP_CONTROL );sdhci_writel (host, irq_val,SDHCI_INT_ENABLE ); + + host->irq_wake_enabled = !enable_irq_wake (host->irq); + +return host->irq_wake_enabled; }static void sdhci_disable_irq_wakeups (struct sdhci_host *host)
@@ -2849,6 +2853,10 @@ static void sdhci_disable_irq_wakeups(struct sdhci_host *host) val =sdhci_readb (host,SDHCI_WAKE_UP_CONTROL ); val &= ~mask;sdhci_writeb (host, val,SDHCI_WAKE_UP_CONTROL ); + +disable_irq_wake (host->irq); + + host->irq_wake_enabled = false; }int sdhci_suspend_host (struct sdhci_host *host)
@@ -2857,15 +2865,14 @@ int sdhci_suspend_host(struct sdhci_host *host)mmc_retune_timer_stop (host->mmc); -if (!device_may_wakeup (mmc_dev (host->mmc))) { +if (!device_may_wakeup (mmc_dev (host->mmc)) || + !sdhci_enable_irq_wakeups (host)) { host->ier =0 ;sdhci_writel (host,0 ,SDHCI_INT_ENABLE );sdhci_writel (host,0 ,SDHCI_SIGNAL_ENABLE );free_irq (host->irq, host); - }else { -sdhci_enable_irq_wakeups (host); -enable_irq_wake (host->irq); } +return 0 ; }
@@ -2893,15 +2900,14 @@ int sdhci_resume_host(struct sdhci_host *host)mmiowb (); } -if (!device_may_wakeup (mmc_dev (host->mmc))) { +if (host->irq_wake_enabled) { +sdhci_disable_irq_wakeups (host); + }else { ret =request_threaded_irq (host->irq, sdhci_irq, sdhci_thread_irq,IRQF_SHARED ,mmc_hostname (host->mmc), host);if (ret)return ret; - }else { -sdhci_disable_irq_wakeups (host); -disable_irq_wake (host->irq); }sdhci_enable_card_detection (host);
drivers/mmc/host/sdhci.h +1 -0
@@ -484,6 +484,7 @@ struct sdhci_host {bool bus_on;/* Bus power prevents runtime suspend */ bool preset_enabled;/* Preset is enabled */ bool pending_reset;/* Cmd/data reset is pending */ +bool irq_wake_enabled;/* IRQ wakeup is enabled */ struct mmc_request *mrqs_done[SDHCI_MAX_MRQS ];/* Requests done */ struct mmc_command *cmd;/* Current command */
Signature · 1 line
1.9.1
Ulf Hansson
AppliedRe: [PATCH RESEND 4/9] mmc: sdhci: Handle failure of enable_irq_wake()
Quoted 8 lines · Adrian Hunter
Now that
sdhci_enable_irq_wakeups()is a local function, change it to return whether the IRQ wakeup was successfully enabled. This is in preparation for adding more conditions for whether IRQ wakeup is enabled.Note it is assumed, for SDHCI devices, that suspend is more important than wakeup, so we continue to suspend regardless.
Signed-off-by Adrian Hunter <[email protected]>
Thanks, applied for next!
Kind regards
Uffe
Quoted 93 lines
--- drivers/mmc/host/sdhci.c | 24 +++++++++++++++--------- drivers/mmc/host/sdhci.h | 1 + 2 files changed, 16 insertions(+), 9 deletions(-)drivers/mmc/host/sdhci.c +15
-9@@ -2821,7 +2821,7 @@ static irqreturn_t sdhci_thread_irq(int irq, void *dev_id)* sdhci_disable_irq_wakeups() since it will be set by * sdhci_enable_card_detection() or sdhci_init(). */ -static void sdhci_enable_irq_wakeups (struct sdhci_host *host) +static bool sdhci_enable_irq_wakeups (struct sdhci_host *host) {u8 val;u8 mask =SDHCI_WAKE_ON_INSERT |SDHCI_WAKE_ON_REMOVE @@ -2838,6 +2838,10 @@ static void sdhci_enable_irq_wakeups(struct sdhci_host *host) }sdhci_writeb (host, val,SDHCI_WAKE_UP_CONTROL );sdhci_writel (host, irq_val,SDHCI_INT_ENABLE ); + + host->irq_wake_enabled = !enable_irq_wake (host->irq); + +return host->irq_wake_enabled; }static void sdhci_disable_irq_wakeups (struct sdhci_host *host)@@ -2849,6 +2853,10 @@ static void sdhci_disable_irq_wakeups(struct sdhci_host *host) val =sdhci_readb (host,SDHCI_WAKE_UP_CONTROL ); val &= ~mask;sdhci_writeb (host, val,SDHCI_WAKE_UP_CONTROL ); + +disable_irq_wake (host->irq); + + host->irq_wake_enabled = false; }int sdhci_suspend_host (struct sdhci_host *host)@@ -2857,15 +2865,14 @@ int sdhci_suspend_host(struct sdhci_host *host)mmc_retune_timer_stop (host->mmc); -if (!device_may_wakeup (mmc_dev (host->mmc))) { +if (!device_may_wakeup (mmc_dev (host->mmc)) || + !sdhci_enable_irq_wakeups (host)) { host->ier =0 ;sdhci_writel (host,0 ,SDHCI_INT_ENABLE );sdhci_writel (host,0 ,SDHCI_SIGNAL_ENABLE );free_irq (host->irq, host); - }else { -sdhci_enable_irq_wakeups (host); -enable_irq_wake (host->irq); } +return 0 ; }@@ -2893,15 +2900,14 @@ int sdhci_resume_host(struct sdhci_host *host)mmiowb (); } -if (!device_may_wakeup (mmc_dev (host->mmc))) { +if (host->irq_wake_enabled) { +sdhci_disable_irq_wakeups (host); + }else { ret =request_threaded_irq (host->irq, sdhci_irq, sdhci_thread_irq,IRQF_SHARED ,mmc_hostname (host->mmc), host);if (ret)return ret; - }else { -sdhci_disable_irq_wakeups (host); -disable_irq_wake (host->irq); }sdhci_enable_card_detection (host);drivers/mmc/host/sdhci.h +1
-1@@ -484,6 +484,7 @@ struct sdhci_host {bool bus_on;/* Bus power prevents runtime suspend */ bool preset_enabled;/* Preset is enabled */ bool pending_reset;/* Cmd/data reset is pending */ +bool irq_wake_enabled;/* IRQ wakeup is enabled */ struct mmc_request *mrqs_done[SDHCI_MAX_MRQS ];/* Requests done */ struct mmc_command *cmd;/* Current command */ --1.9.1
Adrian Hunter
maintainerPATCH RESEND 5/9 mmc: sdhci: Rework sdhci_enable_irq_wakeups()
In reply to Adrian HunterIn preparation for adding more conditions for whether IRQ wakeup is enabled, rework sdhci_enable_irq_wakeups() so that needed bits are added instead of adding them all and then removing the unneeded bits.
--- drivers/mmc/host/sdhci.c | 25 +++++++++++++++---------- 1 file changed, 15 insertions(+), 10 deletions(-)
drivers/mmc/host/sdhci.c +15 -10
@@ -2823,20 +2823,25 @@ static irqreturn_t sdhci_thread_irq(int irq, void *dev_id)*/ static bool sdhci_enable_irq_wakeups (struct sdhci_host *host) { +u8 mask =SDHCI_WAKE_ON_INSERT |SDHCI_WAKE_ON_REMOVE | +SDHCI_WAKE_ON_INT ; +u32 irq_val =0 ; +u8 wake_val =0 ;u8 val; -u8 mask =SDHCI_WAKE_ON_INSERT |SDHCI_WAKE_ON_REMOVE - |SDHCI_WAKE_ON_INT ; -u32 irq_val =SDHCI_INT_CARD_INSERT |SDHCI_INT_CARD_REMOVE | -SDHCI_INT_CARD_INT ; - val =sdhci_readb (host,SDHCI_WAKE_UP_CONTROL ); - val |= mask ; -/* Avoid fake wake up */ -if (host->quirks &SDHCI_QUIRK_BROKEN_CARD_DETECTION ) { - val &= ~(SDHCI_WAKE_ON_INSERT |SDHCI_WAKE_ON_REMOVE ); - irq_val &= ~(SDHCI_INT_CARD_INSERT |SDHCI_INT_CARD_REMOVE ); +if (!(host->quirks &SDHCI_QUIRK_BROKEN_CARD_DETECTION )) { + wake_val |=SDHCI_WAKE_ON_INSERT |SDHCI_WAKE_ON_REMOVE ; + irq_val |=SDHCI_INT_CARD_INSERT |SDHCI_INT_CARD_REMOVE ; } + + wake_val |=SDHCI_WAKE_ON_INT ; + irq_val |=SDHCI_INT_CARD_INT ; + + val =sdhci_readb (host,SDHCI_WAKE_UP_CONTROL ); + val &= ~mask; + val |= wake_val;sdhci_writeb (host, val,SDHCI_WAKE_UP_CONTROL ); +sdhci_writel (host, irq_val,SDHCI_INT_ENABLE ); host->irq_wake_enabled = !enable_irq_wake (host->irq);
Signature · 1 line
1.9.1
Ulf Hansson
AppliedRe: [PATCH RESEND 5/9] mmc: sdhci: Rework sdhci_enable_irq_wakeups()
Quoted 5 lines · Adrian Hunter
In preparation for adding more conditions for whether IRQ wakeup is enabled, rework
sdhci_enable_irq_wakeups()so that needed bits are added instead of adding them all and then removing the unneeded bits.Signed-off-by Adrian Hunter <[email protected]>
Thanks, applied for next!
Kind regards
Uffe
Quoted 47 lines
--- drivers/mmc/host/sdhci.c | 25 +++++++++++++++---------- 1 file changed, 15 insertions(+), 10 deletions(-)drivers/mmc/host/sdhci.c +15
-11@@ -2823,20 +2823,25 @@ static irqreturn_t sdhci_thread_irq(int irq, void *dev_id)*/ static bool sdhci_enable_irq_wakeups (struct sdhci_host *host) { +u8 mask =SDHCI_WAKE_ON_INSERT |SDHCI_WAKE_ON_REMOVE | +SDHCI_WAKE_ON_INT ; +u32 irq_val =0 ; +u8 wake_val =0 ;u8 val; -u8 mask =SDHCI_WAKE_ON_INSERT |SDHCI_WAKE_ON_REMOVE - |SDHCI_WAKE_ON_INT ; -u32 irq_val =SDHCI_INT_CARD_INSERT |SDHCI_INT_CARD_REMOVE | -SDHCI_INT_CARD_INT ; - val =sdhci_readb (host,SDHCI_WAKE_UP_CONTROL ); - val |= mask ; -/* Avoid fake wake up */ -if (host->quirks &SDHCI_QUIRK_BROKEN_CARD_DETECTION ) { - val &= ~(SDHCI_WAKE_ON_INSERT |SDHCI_WAKE_ON_REMOVE ); - irq_val &= ~(SDHCI_INT_CARD_INSERT |SDHCI_INT_CARD_REMOVE ); +if (!(host->quirks &SDHCI_QUIRK_BROKEN_CARD_DETECTION )) { + wake_val |=SDHCI_WAKE_ON_INSERT |SDHCI_WAKE_ON_REMOVE ; + irq_val |=SDHCI_INT_CARD_INSERT |SDHCI_INT_CARD_REMOVE ; } + + wake_val |=SDHCI_WAKE_ON_INT ; + irq_val |=SDHCI_INT_CARD_INT ; + + val =sdhci_readb (host,SDHCI_WAKE_UP_CONTROL ); + val &= ~mask; + val |= wake_val;sdhci_writeb (host, val,SDHCI_WAKE_UP_CONTROL ); +sdhci_writel (host, irq_val,SDHCI_INT_ENABLE ); host->irq_wake_enabled = !enable_irq_wake (host->irq); --1.9.1
Adrian Hunter
maintainerPATCH RESEND 6/9 mmc: sdhci: Do not unnecessarily enable wakeup for card detect interrupt
In reply to Adrian HunterDo not unnecessarily enable card detect wakeup in the cases that the card is not removable or a GPIO is used for card detect.
--- drivers/mmc/host/sdhci.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)
drivers/mmc/host/sdhci.c +9 -1
@@ -2813,6 +2813,14 @@ static irqreturn_t sdhci_thread_irq(int irq, void *dev_id)\*****************************************************************************/ #ifdef CONFIG_PM + +static bool sdhci_cd_irq_can_wakeup (struct sdhci_host *host) +{ +return mmc_card_is_removable (host->mmc) && + !(host->quirks &SDHCI_QUIRK_BROKEN_CARD_DETECTION ) && + host->mmc->slot.cd_irq <0 ; +} +/* * To enable wakeup events, the corresponding events have to be enabled in * the Interrupt Status Enable register too. See 'Table 1-6: Wakeup Signal
@@ -2829,7 +2837,7 @@ static bool sdhci_enable_irq_wakeups(struct sdhci_host *host)u8 wake_val =0 ;u8 val; -if (!(host->quirks &SDHCI_QUIRK_BROKEN_CARD_DETECTION )) { +if (sdhci_cd_irq_can_wakeup (host)) { wake_val |=SDHCI_WAKE_ON_INSERT |SDHCI_WAKE_ON_REMOVE ; irq_val |=SDHCI_INT_CARD_INSERT |SDHCI_INT_CARD_REMOVE ; }
Signature · 1 line
1.9.1
Adrian Hunter
maintainerPATCH RESEND 7/9 mmc: sdhci: Do not unnecessarily enable wakeup for SDIO card interrupt
In reply to Adrian HunterDo not enable wakeup for SDIO card interrupt unless the SDIO function driver has requested it which is indicated by mmc_card_wake_sdio_irq().
--- drivers/mmc/host/sdhci.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-)
drivers/mmc/host/sdhci.c +7 -2
@@ -2842,8 +2842,13 @@ static bool sdhci_enable_irq_wakeups(struct sdhci_host *host) irq_val |=SDHCI_INT_CARD_INSERT |SDHCI_INT_CARD_REMOVE ; } - wake_val |=SDHCI_WAKE_ON_INT ; - irq_val |=SDHCI_INT_CARD_INT ; +if (mmc_card_wake_sdio_irq (host->mmc)) { + wake_val |=SDHCI_WAKE_ON_INT ; + irq_val |=SDHCI_INT_CARD_INT ; + } + +if (!irq_val) +return false; val =sdhci_readb (host,SDHCI_WAKE_UP_CONTROL ); val &= ~mask;
Signature · 1 line
1.9.1
Adrian Hunter
maintainerPATCH RESEND 8/9 mmc: slot-gpio: Add a function to enable/disable card detect IRQ wakeup
In reply to Adrian HunterCommit 03dbaa04a2e5 ("mmc: slot-gpio: Add support to enable irq wake on cd_irq") enabled wakeup irrespective of the host controller's PM flags. However, users also want to control it from sysfs power/wakeup attribute. That means the driver needs to check the PM flags before enabling it in the suspend callback. Add helper function mmc_gpio_set_cd_wake() to make it easy for drivers to do that.
--- drivers/mmc/core/core.c | 3 +-- drivers/mmc/core/slot-gpio.c | 23 +++++++++++++++++++++++ include/linux/mmc/slot-gpio.h | 1 + 3 files changed, 25 insertions(+), 2 deletions(-)
drivers/mmc/core/core.c +1 -2
@@ -2655,8 +2655,7 @@ void mmc_start_host(struct mmc_host *host)void mmc_stop_host (struct mmc_host *host) {if (host->slot.cd_irq >=0 ) { -if (host->slot.cd_wake_enabled) -disable_irq_wake (host->slot.cd_irq); +mmc_gpio_set_cd_wake (host, false);disable_irq (host->slot.cd_irq); }
drivers/mmc/core/slot-gpio.c +23 -0
@@ -154,6 +154,29 @@ void mmc_gpiod_request_cd_irq(struct mmc_host *host) }EXPORT_SYMBOL (mmc_gpiod_request_cd_irq); +int mmc_gpio_set_cd_wake (struct mmc_host *host,bool on) +{ +int ret =0 ; + +if (!(host->caps &MMC_CAP_CD_WAKE ) || + host->slot.cd_irq <0 || + on == host->slot.cd_wake_enabled) +return 0 ; + +if (on) { +if (device_may_wakeup (mmc_dev (host))) { + ret =enable_irq_wake (host->slot.cd_irq); + host->slot.cd_wake_enabled = !ret; + } + }else { +disable_irq_wake (host->slot.cd_irq); + host->slot.cd_wake_enabled = false; + } + +return ret; +} +EXPORT_SYMBOL (mmc_gpio_set_cd_wake); +/* Register an alternate interrupt service routine for * the card-detect GPIO. */
include/linux/mmc/slot-gpio.h +1 -0
@@ -33,5 +33,6 @@ void mmc_gpio_set_cd_isr(struct mmc_host *host, irqreturn_t (*isr)(int irq, void *dev_id));void mmc_gpiod_request_cd_irq (struct mmc_host *host);bool mmc_can_gpio_cd (struct mmc_host *host); +int mmc_gpio_set_cd_wake (struct mmc_host *host,bool on);#endif
Signature · 1 line
1.9.1
Adrian Hunter
maintainerPATCH RESEND 9/9 mmc: sdhci-pci: Respect PM flags when enabling card detect GPIO IRQ wakeup
In reply to Adrian HunterCommit 03dbaa04a2e5 ("mmc: slot-gpio: Add support to enable irq wake on cd_irq") enabled wakeup irrespective of the host controller's PM flags. However, users also want to control it from sysfs power/wakeup attribute. That means the driver needs to check the PM flags before enabling it in the suspend callback. Add support for that in sdhci-pci, which is the only driver presently using the MMC_CAP_CD_WAKE flag, and remove the enabling in mmc_gpiod_request_cd_irq().
--- drivers/mmc/core/slot-gpio.c | 2 -- drivers/mmc/host/sdhci-pci-core.c | 18 ++++++++++++++---- drivers/mmc/host/sdhci.c | 4 ++++ 3 files changed, 18 insertions(+), 6 deletions(-)
drivers/mmc/core/slot-gpio.c +0 -2
@@ -149,8 +149,6 @@ void mmc_gpiod_request_cd_irq(struct mmc_host *host)if (irq <0 ) host->caps |=MMC_CAP_NEEDS_POLL ; -else if ((host->caps &MMC_CAP_CD_WAKE ) && !enable_irq_wake (irq)) - host->slot.cd_wake_enabled = true; }EXPORT_SYMBOL (mmc_gpiod_request_cd_irq);
drivers/mmc/host/sdhci-pci-core.c +14 -4
@@ -41,18 +41,25 @@static int sdhci_pci_init_wakeup (struct sdhci_pci_chip *chip) {mmc_pm_flag_t pm_flags =0 ; +bool cap_cd_wake = false;int i;for (i =0 ; i < chip->num_slots; i++) {struct sdhci_pci_slot *slot = chip->slots[i]; -if (slot) +if (slot) { pm_flags |= slot->host->mmc->pm_flags; +if (slot->host->mmc->caps &MMC_CAP_CD_WAKE ) + cap_cd_wake = true; + } } -return device_set_wakeup_enable (&chip->pdev->dev, - (pm_flags &MMC_PM_KEEP_POWER ) && - (pm_flags &MMC_PM_WAKE_SDIO_IRQ )); +if ((pm_flags &MMC_PM_KEEP_POWER ) && (pm_flags &MMC_PM_WAKE_SDIO_IRQ )) +return device_wakeup_enable (&chip->pdev->dev); +else if (!cap_cd_wake) +return device_wakeup_disable (&chip->pdev->dev); + +return 0 ; }static int sdhci_pci_suspend_host (struct sdhci_pci_chip *chip)
@@ -1687,6 +1694,9 @@ static struct sdhci_pci_slot *sdhci_pci_probe_slot(if (device_can_wakeup (&pdev->dev)) host->mmc->pm_caps |=MMC_PM_WAKE_SDIO_IRQ ; +if (host->mmc->caps &MMC_CAP_CD_WAKE ) +device_init_wakeup (&pdev->dev, true); +if (slot->cd_idx >=0 ) { ret =mmc_gpiod_request_cd (host->mmc,NULL , slot->cd_idx, slot->cd_override_level,0 ,NULL );
drivers/mmc/host/sdhci.c +4 -0
@@ -2891,6 +2891,8 @@ int sdhci_suspend_host(struct sdhci_host *host)free_irq (host->irq, host); } +mmc_gpio_set_cd_wake (host->mmc, true); +return 0 ; }
@@ -2918,6 +2920,8 @@ int sdhci_resume_host(struct sdhci_host *host)mmiowb (); } +mmc_gpio_set_cd_wake (host->mmc, false); +if (host->irq_wake_enabled) {sdhci_disable_irq_wakeups (host); }else {
Signature · 1 line
1.9.1
Adrian Hunter
maintainerQuoted 13 lines · Adrian Hunter
Hi
Commit
03dbaa04a2e5("mmc: slot-gpio: Add support to enable irq wake oncd_irq") enabled wakeup irrespective of the host controller's PM flags. However, users also want to control it from sysfs power/wakeup attribute. That means the driver needs to check the PM flags before enabling it in the suspend callback. Patch 9 adds support for that in sdhci-pci, which is the only driver presently using theMMC_CAP_CD_WAKEflag. Patches 1 - 7 tidy up aspects of sdhci and sdhci-pci wakeup handling, and patch 8 adds a helper function to make it easy for drivers.There are 2 contentious aspects to this patch set:
Quoted 9 lines
1) An existing problem with the SDIO API which does not let the host controller driver know that the SDIO function driver has requested SDIO card interrupt wakeup until the suspend callback - which results in the host controller driver having to enable or disable wakeup in the suspend callback. Fixing the SDIO API is a separate issue IMHO.
2) In order to use the sysfs power/wakeup attribute, the driver must set the device as wake capable even when it is really a GPIO that wakes the system.
Given that we have more-or-less agreed to use sysfs power/wakeup attribute, I don't see a problem with going ahead with these patches, and adjusting the API later. Thoughts?
Quoted 20 lines · Adrian Hunter
Quoted 19 lines · Adrian Hunter
Hi
Commit
03dbaa04a2e5("mmc: slot-gpio: Add support to enable irq wake oncd_irq") enabled wakeup irrespective of the host controller's PM flags. However, users also want to control it from sysfs power/wakeup attribute. That means the driver needs to check the PM flags before enabling it in the suspend callback. Patch 9 adds support for that in sdhci-pci, which is the only driver presently using theMMC_CAP_CD_WAKEflag. Patches 1 - 7 tidy up aspects of sdhci and sdhci-pci wakeup handling, and patch 8 adds a helper function to make it easy for drivers.There are 2 contentious aspects to this patch set:
1) An existing problem with the SDIO API which does not let the host controller driver know that the SDIO function driver has requested SDIO card interrupt wakeup until the suspend callback - which results in the host controller driver having to enable or disable wakeup in the suspend callback. Fixing the SDIO API is a separate issue IMHO.
Quoted 7 lines
Quoted 3 lines
2) In order to use the sysfs power/wakeup attribute, the driver must set the device as wake capable even when it is really a GPIO that wakes the system.
Given that we have more-or-less agreed to use sysfs power/wakeup attribute, I don't see a problem with going ahead with these patches, and adjusting the API later. Thoughts?
Let me have look again, I might be able to pick up at least some of the changes in $subject series. No matter what, please re-post once 4.16 rc1 is out.
The reason for being conservative, is that I would prefer the discussion [1] to conclude on how to handle out/in band wakeups, as that would help to understand the final goal of $subject series.
Kind regards
Uffe
Quoted 25 lines · Ulf Hansson
Quoted 24 lines · Adrian Hunter
Quoted 22 lines · Adrian Hunter
Hi
Commit
03dbaa04a2e5("mmc: slot-gpio: Add support to enable irq wake oncd_irq") enabled wakeup irrespective of the host controller's PM flags. However, users also want to control it from sysfs power/wakeup attribute. That means the driver needs to check the PM flags before enabling it in the suspend callback. Patch 9 adds support for that in sdhci-pci, which is the only driver presently using theMMC_CAP_CD_WAKEflag. Patches 1 - 7 tidy up aspects of sdhci and sdhci-pci wakeup handling, and patch 8 adds a helper function to make it easy for drivers.There are 2 contentious aspects to this patch set:
1) An existing problem with the SDIO API which does not let the host controller driver know that the SDIO function driver has requested SDIO card interrupt wakeup until the suspend callback - which results in the host controller driver having to enable or disable wakeup in the suspend callback. Fixing the SDIO API is a separate issue IMHO.
2) In order to use the sysfs power/wakeup attribute, the driver must set the device as wake capable even when it is really a GPIO that wakes the system.
Quoted 7 lines
Quoted 3 lines
Given that we have more-or-less agreed to use sysfs power/wakeup attribute, I don't see a problem with going ahead with these patches, and adjusting the API later. Thoughts?
Let me have look again, I might be able to pick up at least some of the changes in $subject series. No matter what, please re-post once 4.16 rc1 is out.
I have picked up patch 1 -> patch 5, the rest seems like something we need to defer until we know how to deal with in/out-band wakeups. However, please re-post them once 4.16rc1 is out.
[...]
Kind regards
Uffe