Commit fb145456 authored by Kleber Sacilotto de Souza's avatar Kleber Sacilotto de Souza Committed by Greg Kroah-Hartman
Browse files

GenWQE: Add support for EEH error recovery 



This patch implements the callbacks and functions necessary to have EEH
recovery support.

It adds a config option to enable or disable explicit calls to trigger
platform specific mechanisms on error recovery paths. This option is
enabled by default only on PPC64 systems and can be overritten via
debugfs. If this option is enabled, on the error recovery path the
driver will call pci_channel_offline() to check for error condition and
issue non-raw MMIO reads to trigger early EEH detection in case of
hardware failures. This is necessary since the driver MMIO helper
funtions use raw accessors.

Signed-off-by: default avatarKleber Sacilotto de Souza <klebers@linux.vnet.ibm.com>
Acked-by: default avatarFrank Haverkamp <haver@linux.vnet.ibm.com>
Signed-off-by: default avatarGreg Kroah-Hartman <gregkh@linuxfoundation.org>
parent c1f732ad

drivers/misc/genwqe/Kconfig

+6 −0
Original line number Diff line number Diff line
@@ -11,3 +11,9 @@ menuconfig GENWQE 
	  Enables PCIe card driver for IBM GenWQE accelerators.

	  The user-space interface is described in

	  include/linux/genwqe/genwqe_card.h.



config GENWQE_PLATFORM_ERROR_RECOVERY

	int "Use platform recovery procedures (0=off, 1=on)"

	depends on GENWQE

	default 1 if PPC64

	default 0

drivers/misc/genwqe/card_base.c

+66 −13
Original line number Diff line number Diff line
@@ -140,6 +140,12 @@ static struct genwqe_dev *genwqe_dev_alloc(void) 
	cd->class_genwqe = class_genwqe;

	cd->debugfs_genwqe = debugfs_genwqe;



	/*

	 * This comes from kernel config option and can be overritten via

	 * debugfs.

	 */

	cd->use_platform_recovery = CONFIG_GENWQE_PLATFORM_ERROR_RECOVERY;



	init_waitqueue_head(&cd->queue_waitq);



	spin_lock_init(&cd->file_lock);
@@ -943,6 +949,19 @@ static int genwqe_health_thread(void *data) 
	return 0;



 fatal_error:

	if (cd->use_platform_recovery) {

		/*

		 * Since we use raw accessors, EEH errors won't be detected

		 * by the platform until we do a non-raw MMIO or config space

		 * read

		 */

		readq(cd->mmio + IO_SLC_CFGREG_GFIR);



		/* We do nothing if the card is going over PCI recovery */

		if (pci_channel_offline(pci_dev))

			return -EIO;

	}



	dev_err(&pci_dev->dev,

		"[%s] card unusable. Please trigger unbind!\n", __func__);
@@ -1048,6 +1067,9 @@ static int genwqe_pci_setup(struct genwqe_dev *cd) 
	pci_set_master(pci_dev);

	pci_enable_pcie_error_reporting(pci_dev);



	/* EEH recovery requires PCIe fundamental reset */

	pci_dev->needs_freset = 1;



	/* request complete BAR-0 space (length = 0) */

	cd->mmio_len = pci_resource_len(pci_dev, 0);

	cd->mmio = pci_iomap(pci_dev, 0, 0);
@@ -1186,23 +1208,40 @@ static pci_ers_result_t genwqe_err_error_detected(struct pci_dev *pci_dev, 


	dev_err(&pci_dev->dev, "[%s] state=%d\n", __func__, state);



	if (pci_dev == NULL)

		return PCI_ERS_RESULT_NEED_RESET;



	cd = dev_get_drvdata(&pci_dev->dev);

	if (cd == NULL)

		return PCI_ERS_RESULT_NEED_RESET;

		return PCI_ERS_RESULT_DISCONNECT;



	switch (state) {

	case pci_channel_io_normal:

		return PCI_ERS_RESULT_CAN_RECOVER;

	case pci_channel_io_frozen:

		return PCI_ERS_RESULT_NEED_RESET;

	case pci_channel_io_perm_failure:

	/* Stop the card */

	genwqe_health_check_stop(cd);

	genwqe_stop(cd);



	/*

	 * On permanent failure, the PCI code will call device remove

	 * after the return of this function.

	 * genwqe_stop() can be called twice.

	 */

	if (state == pci_channel_io_perm_failure) {

		return PCI_ERS_RESULT_DISCONNECT;

	} else {

		genwqe_pci_remove(cd);

		return PCI_ERS_RESULT_NEED_RESET;

	}

}



	return PCI_ERS_RESULT_NEED_RESET;

static pci_ers_result_t genwqe_err_slot_reset(struct pci_dev *pci_dev)

{

	int rc;

	struct genwqe_dev *cd = dev_get_drvdata(&pci_dev->dev);



	rc = genwqe_pci_setup(cd);

	if (!rc) {

		return PCI_ERS_RESULT_RECOVERED;

	} else {

		dev_err(&pci_dev->dev,

			"err: problems with PCI setup (err=%d)\n", rc);

		return PCI_ERS_RESULT_DISCONNECT;

	}

}



static pci_ers_result_t genwqe_err_result_none(struct pci_dev *dev)
@@ -1210,8 +1249,22 @@ static pci_ers_result_t genwqe_err_result_none(struct pci_dev *dev) 
	return PCI_ERS_RESULT_NONE;

}



static void genwqe_err_resume(struct pci_dev *dev)

static void genwqe_err_resume(struct pci_dev *pci_dev)

{

	int rc;

	struct genwqe_dev *cd = dev_get_drvdata(&pci_dev->dev);



	rc = genwqe_start(cd);

	if (!rc) {

		rc = genwqe_health_check_start(cd);

		if (rc)

			dev_err(&pci_dev->dev,

				"err: cannot start health checking! (err=%d)\n",

				rc);

	} else {

		dev_err(&pci_dev->dev,

			"err: cannot start card services! (err=%d)\n", rc);

	}

}



static int genwqe_sriov_configure(struct pci_dev *dev, int numvfs)
@@ -1234,7 +1287,7 @@ static struct pci_error_handlers genwqe_err_handler = { 
	.error_detected = genwqe_err_error_detected,

	.mmio_enabled	= genwqe_err_result_none,

	.link_reset	= genwqe_err_result_none,

	.slot_reset	= genwqe_err_result_none,

	.slot_reset	= genwqe_err_slot_reset,

	.resume		= genwqe_err_resume,

};

drivers/misc/genwqe/card_base.h

+2 −0
Original line number Diff line number Diff line
@@ -291,6 +291,8 @@ struct genwqe_dev { 
	struct task_struct *health_thread;

	wait_queue_head_t health_waitq;



	int use_platform_recovery;	/* use platform recovery mechanisms */



	/* char device */

	dev_t  devnum_genwqe;		/* major/minor num card */

	struct class *class_genwqe;	/* reference to class object */

drivers/misc/genwqe/card_ddcb.c

+19 −5
Original line number Diff line number Diff line
@@ -1118,7 +1118,21 @@ static irqreturn_t genwqe_pf_isr(int irq, void *dev_id) 
	 * safer, but slower for the good-case ... See above.

	 */

	gfir = __genwqe_readq(cd, IO_SLC_CFGREG_GFIR);

	if ((gfir & GFIR_ERR_TRIGGER) != 0x0) {

	if (((gfir & GFIR_ERR_TRIGGER) != 0x0) &&

	    !pci_channel_offline(pci_dev)) {



		if (cd->use_platform_recovery) {

			/*

			 * Since we use raw accessors, EEH errors won't be

			 * detected by the platform until we do a non-raw

			 * MMIO or config space read

			 */

			readq(cd->mmio + IO_SLC_CFGREG_GFIR);



			/* Don't do anything if the PCI channel is frozen */

			if (pci_channel_offline(pci_dev))

				goto exit;

		}



		wake_up_interruptible(&cd->health_waitq);
@@ -1126,12 +1140,12 @@ static irqreturn_t genwqe_pf_isr(int irq, void *dev_id) 
		 * By default GFIRs causes recovery actions. This

		 * count is just for debug when recovery is masked.

		 */

		printk_ratelimited(KERN_ERR

				   "%s %s: [%s] GFIR=%016llx\n",

				   GENWQE_DEVNAME, dev_name(&pci_dev->dev),

		dev_err_ratelimited(&pci_dev->dev,

				    "[%s] GFIR=%016llx\n",

				    __func__, gfir);

	}



 exit:

	return IRQ_HANDLED;

}

drivers/misc/genwqe/card_debugfs.c

+7 −0
Original line number Diff line number Diff line
@@ -485,6 +485,13 @@ int genwqe_init_debugfs(struct genwqe_dev *cd) 
		goto err1;

	}



	file = debugfs_create_u32("use_platform_recovery", 0666, root,

				  &cd->use_platform_recovery);

	if (!file) {

		ret = -ENOMEM;

		goto err1;

	}



	cd->debugfs_root = root;

	return 0;

err1:

CRA Git | Maintained and supported by SUSTech CRA and CCSE