MMC: [PATCH 0/2] Two fixes for mmc system

MMC: [PATCH 0/2] Two fixes for mmc system

[Maxim Levitsky]

Hi,

Here is updated version of fixes I created for mmc subsystem.

First patch is unchanged, and I can say is very well tested.
The second patch that fixes nasty hang that happens if you have MMC card in reader
during suspend.
I updated it to address corner issues that I was told about.

Consider patch #1 for 2.6.36, and patch #2 for 2.6.36 and stable trees

Best regards,
	Maxim Levitsky

[PATCH 1/2] MMC: make sdhci work with ricoh mmc controller

[Maxim Levitsky]

The current way of disabling it is not well tested by vendor
and has all kinds of bugs that show up on resume from ram/disk.
A very good example is a dead SDHCI controller.

Old way of disabling is still supported by
continuing to use CONFIG_MMC_RICOH_MMC.

Based on 'http://list.drzeus.cx/pipermail/sdhci-devel/2007-December/002085.html'
Therefore most of the credit for this goes to Andrew de Quincey

CC: Andrew de Quincey <adq_dvb@lidskialf.net>

Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
Acked-by: Philip Langdale <philipl@overt.org>
---
 drivers/mmc/host/sdhci-pci.c |   41 +++++++++++++++++++++++++++++++++++++++++
 drivers/mmc/host/sdhci.c     |    3 ++-
 drivers/mmc/host/sdhci.h     |    4 ++++
 3 files changed, 47 insertions(+), 1 deletions(-)

diff --git a/drivers/mmc/host/sdhci-pci.c b/drivers/mmc/host/sdhci-pci.c
index 65483fd..e021431 100644
--- a/drivers/mmc/host/sdhci-pci.c
+++ b/drivers/mmc/host/sdhci-pci.c
@@ -17,6 +17,7 @@
 #include <linux/pci.h>
 #include <linux/dma-mapping.h>
 #include <linux/slab.h>
+#include <linux/device.h>
 
 #include <linux/mmc/host.h>
 
@@ -84,7 +85,30 @@ static int ricoh_probe(struct sdhci_pci_chip *chip)
 	if (chip->pdev->subsystem_vendor == PCI_VENDOR_ID_SAMSUNG ||
 	    chip->pdev->subsystem_vendor == PCI_VENDOR_ID_SONY)
 		chip->quirks |= SDHCI_QUIRK_NO_CARD_NO_RESET;
+	return 0;
+}
+
+static int ricoh_mmc_probe_slot(struct sdhci_pci_slot *slot)
+{
+	slot->host->caps =
+		((0x21 << SDHCI_TIMEOUT_CLK_SHIFT)
+			& SDHCI_TIMEOUT_CLK_MASK) |
+
+		((0x21 << SDHCI_CLOCK_BASE_SHIFT)
+			& SDHCI_CLOCK_BASE_MASK) |
 
+		SDHCI_TIMEOUT_CLK_UNIT |
+		SDHCI_CAN_VDD_330 |
+		SDHCI_CAN_DO_SDMA;
+	return 0;
+}
+
+static int ricoh_mmc_resume(struct sdhci_pci_chip *chip)
+{
+	/* Apply a delay to allow controller to settle */
+	/* Otherwise it becomes confused if card state changed
+		during suspend */
+	msleep(500);
 	return 0;
 }
 
@@ -95,6 +119,15 @@ static const struct sdhci_pci_fixes sdhci_ricoh = {
 			  SDHCI_QUIRK_CLOCK_BEFORE_RESET,
 };
 
+static const struct sdhci_pci_fixes sdhci_ricoh_mmc = {
+	.probe_slot	= ricoh_mmc_probe_slot,
+	.resume		= ricoh_mmc_resume,
+	.quirks		= SDHCI_QUIRK_32BIT_DMA_ADDR |
+			  SDHCI_QUIRK_CLOCK_BEFORE_RESET |
+			  SDHCI_QUIRK_NO_CARD_NO_RESET |
+			  SDHCI_QUIRK_MISSING_CAPS
+};
+
 static const struct sdhci_pci_fixes sdhci_ene_712 = {
 	.quirks		= SDHCI_QUIRK_SINGLE_POWER_WRITE |
 			  SDHCI_QUIRK_BROKEN_DMA,
@@ -374,6 +407,14 @@ static const struct pci_device_id pci_ids[] __devinitdata = {
 	},
 
 	{
+		.vendor         = PCI_VENDOR_ID_RICOH,
+		.device         = 0x843,
+		.subvendor      = PCI_ANY_ID,
+		.subdevice      = PCI_ANY_ID,
+		.driver_data    = (kernel_ulong_t)&sdhci_ricoh_mmc,
+	},
+
+	{
 		.vendor		= PCI_VENDOR_ID_ENE,
 		.device		= PCI_DEVICE_ID_ENE_CB712_SD,
 		.subvendor	= PCI_ANY_ID,
diff --git a/drivers/mmc/host/sdhci.c b/drivers/mmc/host/sdhci.c
index c6d1bd8..483b78e 100644
--- a/drivers/mmc/host/sdhci.c
+++ b/drivers/mmc/host/sdhci.c
@@ -1687,7 +1687,8 @@ int sdhci_add_host(struct sdhci_host *host)
 			host->version);
 	}
 
-	caps = sdhci_readl(host, SDHCI_CAPABILITIES);
+	caps = (host->quirks & SDHCI_QUIRK_MISSING_CAPS) ? host->caps :
+		sdhci_readl(host, SDHCI_CAPABILITIES);
 
 	if (host->quirks & SDHCI_QUIRK_FORCE_DMA)
 		host->flags |= SDHCI_USE_SDMA;
diff --git a/drivers/mmc/host/sdhci.h b/drivers/mmc/host/sdhci.h
index c846813..b1839a3 100644
--- a/drivers/mmc/host/sdhci.h
+++ b/drivers/mmc/host/sdhci.h
@@ -240,6 +240,8 @@ struct sdhci_host {
 #define SDHCI_QUIRK_CAP_CLOCK_BASE_BROKEN		(1<<25)
 /* Controller cannot support End Attribute in NOP ADMA descriptor */
 #define SDHCI_QUIRK_NO_ENDATTR_IN_NOPDESC		(1<<26)
+/* Controller is missing device caps. Use caps provided by host */
+#define SDHCI_QUIRK_MISSING_CAPS			(1<<27)
 
 	int			irq;		/* Device IRQ */
 	void __iomem *		ioaddr;		/* Mapped address */
@@ -292,6 +294,8 @@ struct sdhci_host {
 
 	struct timer_list	timer;		/* Timer for timeouts */
 
+	unsigned int		caps;		/* Alternative capabilities */
+
 	unsigned long		private[0] ____cacheline_aligned;
 };
 
-- 
1.7.0.4

[PATCH 2/2] MMC: fix all hangs related to mmc/sd card insert/removal during suspend/resume.

[Maxim Levitsky]

If you don't use CONFIG_MMC_UNSAFE_RESUME, as soon as you attempt to
suspend, the card will be removed, therefore this patch doesn't change
the behavior of this option.

However the removal will be done by pm notifier, which runs while
userspace is still not frozen and thus can freely use del_gendisk,
without the risk of deadlock which would happen otherwise.


Card detect workqueue is now disabled while userspace is frozen,
Therefore if you do use CONFIG_MMC_UNSAFE_RESUME,
and remove the card during suspend, the removal will be
detected as soon as userspace is unfrozen, again at the moment
it is safe to call del_gendisk.

Tested with and without CONFIG_MMC_UNSAFE_RESUME with suspend and hibernate.

Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
---
 drivers/mmc/core/core.c  |   82 ++++++++++++++++++++++++++++++++--------------
 drivers/mmc/core/host.c  |    5 +++
 include/linux/mmc/host.h |    4 ++
 3 files changed, 66 insertions(+), 25 deletions(-)

diff --git a/drivers/mmc/core/core.c b/drivers/mmc/core/core.c
index 569e94d..0366fe8 100644
--- a/drivers/mmc/core/core.c
+++ b/drivers/mmc/core/core.c
@@ -1057,6 +1057,17 @@ void mmc_rescan(struct work_struct *work)
 		container_of(work, struct mmc_host, detect.work);
 	u32 ocr;
 	int err;
+	unsigned long flags;
+
+	spin_lock_irqsave(&host->lock, flags);
+
+	if (host->rescan_disable) {
+		spin_unlock_irqrestore(&host->lock, flags);
+		return;
+	}
+
+	spin_unlock_irqrestore(&host->lock, flags);
+
 
 	mmc_bus_get(host);
 
@@ -1266,19 +1277,6 @@ int mmc_suspend_host(struct mmc_host *host)
 	if (host->bus_ops && !host->bus_dead) {
 		if (host->bus_ops->suspend)
 			err = host->bus_ops->suspend(host);
-		if (err == -ENOSYS || !host->bus_ops->resume) {
-			/*
-			 * We simply "remove" the card in this case.
-			 * It will be redetected on resume.
-			 */
-			if (host->bus_ops->remove)
-				host->bus_ops->remove(host);
-			mmc_claim_host(host);
-			mmc_detach_bus(host);
-			mmc_release_host(host);
-			host->pm_flags = 0;
-			err = 0;
-		}
 	}
 	mmc_bus_put(host);
 
@@ -1310,26 +1308,60 @@ int mmc_resume_host(struct mmc_host *host)
 			printk(KERN_WARNING "%s: error %d during resume "
 					    "(card was removed?)\n",
 					    mmc_hostname(host), err);
-			if (host->bus_ops->remove)
-				host->bus_ops->remove(host);
-			mmc_claim_host(host);
-			mmc_detach_bus(host);
-			mmc_release_host(host);
-			/* no need to bother upper layers */
 			err = 0;
 		}
 	}
 	mmc_bus_put(host);
 
-	/*
-	 * We add a slight delay here so that resume can progress
-	 * in parallel.
-	 */
-	mmc_detect_change(host, 1);
-
 	return err;
 }
 
+/* Do the card removal on suspend if card is assumed removeable
+ * Do that in pm notifier while userspace isn't yet frozen, so we will be able
+   to sync the card.
+*/
+int mmc_pm_notify(struct notifier_block *notify_block,
+					unsigned long mode, void *unused)
+{
+	struct mmc_host *host = container_of(
+		notify_block, struct mmc_host, pm_notify);
+	unsigned long flags;
+
+
+	switch (mode) {
+	case PM_HIBERNATION_PREPARE:
+	case PM_SUSPEND_PREPARE:
+
+		spin_lock_irqsave(&host->lock, flags);
+		host->rescan_disable = 1;
+		spin_unlock_irqrestore(&host->lock, flags);
+
+		if (!host->bus_ops || host->bus_ops->suspend)
+			break;
+
+		mmc_claim_host(host);
+
+		if (host->bus_ops->remove)
+			host->bus_ops->remove(host);
+
+		mmc_detach_bus(host);
+		mmc_release_host(host);
+		host->pm_flags = 0;
+		break;
+
+	case PM_POST_SUSPEND:
+	case PM_POST_HIBERNATION:
+
+		spin_lock_irqsave(&host->lock, flags);
+		host->rescan_disable = 0;
+		spin_unlock_irqrestore(&host->lock, flags);
+		mmc_detect_change(host, 0);
+
+	}
+
+	return 0;
+}
+
 EXPORT_SYMBOL(mmc_resume_host);
 
 #endif
diff --git a/drivers/mmc/core/host.c b/drivers/mmc/core/host.c
index 4735390..01646a1 100644
--- a/drivers/mmc/core/host.c
+++ b/drivers/mmc/core/host.c
@@ -17,6 +17,7 @@
 #include <linux/pagemap.h>
 #include <linux/leds.h>
 #include <linux/slab.h>
+#include <linux/suspend.h>
 
 #include <linux/mmc/host.h>
 
@@ -85,6 +86,8 @@ struct mmc_host *mmc_alloc_host(int extra, struct device *dev)
 	init_waitqueue_head(&host->wq);
 	INIT_DELAYED_WORK(&host->detect, mmc_rescan);
 	INIT_DELAYED_WORK_DEFERRABLE(&host->disable, mmc_host_deeper_disable);
+	host->pm_notify.notifier_call = mmc_pm_notify;
+
 
 	/*
 	 * By default, hosts do not support SGIO or large requests.
@@ -133,6 +136,7 @@ int mmc_add_host(struct mmc_host *host)
 #endif
 
 	mmc_start_host(host);
+	register_pm_notifier(&host->pm_notify);
 
 	return 0;
 }
@@ -149,6 +153,7 @@ EXPORT_SYMBOL(mmc_add_host);
  */
 void mmc_remove_host(struct mmc_host *host)
 {
+	unregister_pm_notifier(&host->pm_notify);
 	mmc_stop_host(host);
 
 #ifdef CONFIG_DEBUG_FS
diff --git a/include/linux/mmc/host.h b/include/linux/mmc/host.h
index f65913c..9063df5 100644
--- a/include/linux/mmc/host.h
+++ b/include/linux/mmc/host.h
@@ -124,6 +124,7 @@ struct mmc_host {
 	unsigned int		f_min;
 	unsigned int		f_max;
 	u32			ocr_avail;
+	struct notifier_block	pm_notify;
 
 #define MMC_VDD_165_195		0x00000080	/* VDD voltage 1.65 - 1.95 */
 #define MMC_VDD_20_21		0x00000100	/* VDD voltage 2.0 ~ 2.1 */
@@ -183,6 +184,7 @@ struct mmc_host {
 
 	/* Only used with MMC_CAP_DISABLE */
 	int			enabled;	/* host is enabled */
+	int			rescan_disable;	/* disable card detection */
 	int			nesting_cnt;	/* "enable" nesting count */
 	int			en_dis_recurs;	/* detect recursion */
 	unsigned int		disable_delay;	/* disable delay in msecs */
@@ -257,6 +259,8 @@ int mmc_card_can_sleep(struct mmc_host *host);
 int mmc_host_enable(struct mmc_host *host);
 int mmc_host_disable(struct mmc_host *host);
 int mmc_host_lazy_disable(struct mmc_host *host);
+int mmc_pm_notify(struct notifier_block *notify_block, unsigned long, void *);
+
 
 static inline void mmc_set_disable_delay(struct mmc_host *host,
 					 unsigned int disable_delay)
-- 
1.7.0.4

MMC: [PATCH 0/2 v2] Two fixes for mmc system

[Maxim Levitsky]

Hi,

This is resend of these patches.
Tell me if I don't need to resent the first patch, it doesn't much matter to me.

The second patch is slightly updated to plug last oportunety for a race.
Also fixed accident move of EXPORT_SYMBOL.

Best regards,
	Maxim Levitsky

[PATCH 2/2] MMC: fix all hangs related to mmc/sd card insert/removal during suspend/resume.

[Maxim Levitsky]

If you don't use CONFIG_MMC_UNSAFE_RESUME, as soon as you attempt to
suspend, the card will be removed, therefore this patch doesn't change
the behavior of this option.

However the removal will be done by pm notifier, which runs while
userspace is still not frozen and thus can freely use del_gendisk,
without the risk of deadlock which would happen otherwise.


Card detect workqueue is now disabled while userspace is frozen,
Therefore if you do use CONFIG_MMC_UNSAFE_RESUME,
and remove the card during suspend, the removal will be
detected as soon as userspace is unfrozen, again at the moment
it is safe to call del_gendisk.

Tested with and without CONFIG_MMC_UNSAFE_RESUME with suspend and hibernate.

Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
---
 drivers/mmc/core/core.c  |   83 +++++++++++++++++++++++++++++++--------------
 drivers/mmc/core/host.c  |    4 ++
 include/linux/mmc/host.h |    3 ++
 3 files changed, 64 insertions(+), 26 deletions(-)

diff --git a/drivers/mmc/core/core.c b/drivers/mmc/core/core.c
index 569e94d..83e7543 100644
--- a/drivers/mmc/core/core.c
+++ b/drivers/mmc/core/core.c
@@ -1057,6 +1057,17 @@ void mmc_rescan(struct work_struct *work)
 		container_of(work, struct mmc_host, detect.work);
 	u32 ocr;
 	int err;
+	unsigned long flags;
+
+	spin_lock_irqsave(&host->lock, flags);
+
+	if (host->rescan_disable) {
+		spin_unlock_irqrestore(&host->lock, flags);
+		return;
+	}
+
+	spin_unlock_irqrestore(&host->lock, flags);
+
 
 	mmc_bus_get(host);
 
@@ -1266,19 +1277,6 @@ int mmc_suspend_host(struct mmc_host *host)
 	if (host->bus_ops && !host->bus_dead) {
 		if (host->bus_ops->suspend)
 			err = host->bus_ops->suspend(host);
-		if (err == -ENOSYS || !host->bus_ops->resume) {
-			/*
-			 * We simply "remove" the card in this case.
-			 * It will be redetected on resume.
-			 */
-			if (host->bus_ops->remove)
-				host->bus_ops->remove(host);
-			mmc_claim_host(host);
-			mmc_detach_bus(host);
-			mmc_release_host(host);
-			host->pm_flags = 0;
-			err = 0;
-		}
 	}
 	mmc_bus_put(host);
 
@@ -1310,28 +1308,61 @@ int mmc_resume_host(struct mmc_host *host)
 			printk(KERN_WARNING "%s: error %d during resume "
 					    "(card was removed?)\n",
 					    mmc_hostname(host), err);
-			if (host->bus_ops->remove)
-				host->bus_ops->remove(host);
-			mmc_claim_host(host);
-			mmc_detach_bus(host);
-			mmc_release_host(host);
-			/* no need to bother upper layers */
 			err = 0;
 		}
 	}
 	mmc_bus_put(host);
 
-	/*
-	 * We add a slight delay here so that resume can progress
-	 * in parallel.
-	 */
-	mmc_detect_change(host, 1);
-
 	return err;
 }
-
 EXPORT_SYMBOL(mmc_resume_host);
 
+/* Do the card removal on suspend if card is assumed removeable
+ * Do that in pm notifier while userspace isn't yet frozen, so we will be able
+   to sync the card.
+*/
+int mmc_pm_notify(struct notifier_block *notify_block,
+					unsigned long mode, void *unused)
+{
+	struct mmc_host *host = container_of(
+		notify_block, struct mmc_host, pm_notify);
+	unsigned long flags;
+
+
+	switch (mode) {
+	case PM_HIBERNATION_PREPARE:
+	case PM_SUSPEND_PREPARE:
+
+		spin_lock_irqsave(&host->lock, flags);
+		host->rescan_disable = 1;
+		spin_unlock_irqrestore(&host->lock, flags);
+		cancel_delayed_work_sync(&host->detect);
+
+		if (!host->bus_ops || host->bus_ops->suspend)
+			break;
+
+		mmc_claim_host(host);
+
+		if (host->bus_ops->remove)
+			host->bus_ops->remove(host);
+
+		mmc_detach_bus(host);
+		mmc_release_host(host);
+		host->pm_flags = 0;
+		break;
+
+	case PM_POST_SUSPEND:
+	case PM_POST_HIBERNATION:
+
+		spin_lock_irqsave(&host->lock, flags);
+		host->rescan_disable = 0;
+		spin_unlock_irqrestore(&host->lock, flags);
+		mmc_detect_change(host, 0);
+
+	}
+
+	return 0;
+}
 #endif
 
 static int __init mmc_init(void)
diff --git a/drivers/mmc/core/host.c b/drivers/mmc/core/host.c
index 4735390..0efe631 100644
--- a/drivers/mmc/core/host.c
+++ b/drivers/mmc/core/host.c
@@ -17,6 +17,7 @@
 #include <linux/pagemap.h>
 #include <linux/leds.h>
 #include <linux/slab.h>
+#include <linux/suspend.h>
 
 #include <linux/mmc/host.h>
 
@@ -85,6 +86,7 @@ struct mmc_host *mmc_alloc_host(int extra, struct device *dev)
 	init_waitqueue_head(&host->wq);
 	INIT_DELAYED_WORK(&host->detect, mmc_rescan);
 	INIT_DELAYED_WORK_DEFERRABLE(&host->disable, mmc_host_deeper_disable);
+	host->pm_notify.notifier_call = mmc_pm_notify;
 
 	/*
 	 * By default, hosts do not support SGIO or large requests.
@@ -133,6 +135,7 @@ int mmc_add_host(struct mmc_host *host)
 #endif
 
 	mmc_start_host(host);
+	register_pm_notifier(&host->pm_notify);
 
 	return 0;
 }
@@ -149,6 +152,7 @@ EXPORT_SYMBOL(mmc_add_host);
  */
 void mmc_remove_host(struct mmc_host *host)
 {
+	unregister_pm_notifier(&host->pm_notify);
 	mmc_stop_host(host);
 
 #ifdef CONFIG_DEBUG_FS
diff --git a/include/linux/mmc/host.h b/include/linux/mmc/host.h
index f65913c..513ff03 100644
--- a/include/linux/mmc/host.h
+++ b/include/linux/mmc/host.h
@@ -124,6 +124,7 @@ struct mmc_host {
 	unsigned int		f_min;
 	unsigned int		f_max;
 	u32			ocr_avail;
+	struct notifier_block	pm_notify;
 
 #define MMC_VDD_165_195		0x00000080	/* VDD voltage 1.65 - 1.95 */
 #define MMC_VDD_20_21		0x00000100	/* VDD voltage 2.0 ~ 2.1 */
@@ -183,6 +184,7 @@ struct mmc_host {
 
 	/* Only used with MMC_CAP_DISABLE */
 	int			enabled;	/* host is enabled */
+	int			rescan_disable;	/* disable card detection */
 	int			nesting_cnt;	/* "enable" nesting count */
 	int			en_dis_recurs;	/* detect recursion */
 	unsigned int		disable_delay;	/* disable delay in msecs */
@@ -257,6 +259,7 @@ int mmc_card_can_sleep(struct mmc_host *host);
 int mmc_host_enable(struct mmc_host *host);
 int mmc_host_disable(struct mmc_host *host);
 int mmc_host_lazy_disable(struct mmc_host *host);
+int mmc_pm_notify(struct notifier_block *notify_block, unsigned long, void *);
 
 static inline void mmc_set_disable_delay(struct mmc_host *host,
 					 unsigned int disable_delay)
-- 
1.7.0.4