[SCSI] lpfc: add PCI error recovery support

This patch adds PCI Error recovery support to the Emulex Lightpulse Fibrechannel (lpfc) SCSI device driver. Lightly tested at this point, works. Signed-off-by: Linas Vepstas <linas@austin.ibm.com> Acked-by: Bino.Sebastian@Emulex.Com Acked-by: James Smart <james.smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] lpfc: add PCI error recovery support
This patch adds PCI Error recovery support to the Emulex Lightpulse Fibrechannel (lpfc) SCSI device driver. Lightly tested at this point, works. Signed-off-by: Linas Vepstas <linas@austin.ibm.com> Acked-by: Bino.Sebastian@Emulex.Com Acked-by: James Smart <james.smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Linas Vepstas · James Bottomley
1 parent 4520b0089b
Showing 2 changed files with 109 additions and 0 deletions Side-by-side Diff
drivers/scsi/lpfc/lpfc_init.c
drivers/scsi/lpfc/lpfc_sli.c
@@ -518,6 +518,10 @@
 	struct lpfc_sli *psli = &phba->sli;
 	struct lpfc_sli_ring  *pring;
 	uint32_t event_data;
+	/* If the pci channel is offline, ignore possible errors,
+	 * since we cannot communicate with the pci card anyway. */
+	if (pci_channel_offline(phba->pcidev))
+		return;
  
 	if (phba->work_hs & HS_FFER6 ||
 	    phba->work_hs & HS_FFER5) {
@@ -1797,6 +1801,92 @@
 	pci_set_drvdata(pdev, NULL);
 }
  
+/**
+ * lpfc_io_error_detected - called when PCI error is detected
+ * @pdev: Pointer to PCI device
+ * @state: The current pci conneection state
+ *
+ * This function is called after a PCI bus error affecting
+ * this device has been detected.
+ */
+static pci_ers_result_t lpfc_io_error_detected(struct pci_dev *pdev,
+				pci_channel_state_t state)
+{
+	struct Scsi_Host *host = pci_get_drvdata(pdev);
+	struct lpfc_hba *phba = (struct lpfc_hba *)host->hostdata;
+	struct lpfc_sli *psli = &phba->sli;
+	struct lpfc_sli_ring  *pring;
+
+	if (state == pci_channel_io_perm_failure) {
+		lpfc_pci_remove_one(pdev);
+		return PCI_ERS_RESULT_DISCONNECT;
+	}
+	pci_disable_device(pdev);
+	/*
+	 * There may be I/Os dropped by the firmware.
+	 * Error iocb (I/O) on txcmplq and let the SCSI layer
+	 * retry it after re-establishing link.
+	 */
+	pring = &psli->ring[psli->fcp_ring];
+	lpfc_sli_abort_iocb_ring(phba, pring);
+
+	/* Request a slot reset. */
+	return PCI_ERS_RESULT_NEED_RESET;
+}
+
+/**
+ * lpfc_io_slot_reset - called after the pci bus has been reset.
+ * @pdev: Pointer to PCI device
+ *
+ * Restart the card from scratch, as if from a cold-boot.
+ */
+static pci_ers_result_t lpfc_io_slot_reset(struct pci_dev *pdev)
+{
+	struct Scsi_Host *host = pci_get_drvdata(pdev);
+	struct lpfc_hba *phba = (struct lpfc_hba *)host->hostdata;
+	struct lpfc_sli *psli = &phba->sli;
+	int bars = pci_select_bars(pdev, IORESOURCE_MEM);
+
+	dev_printk(KERN_INFO, &pdev->dev, "recovering from a slot reset.\n");
+	if (pci_enable_device_bars(pdev, bars)) {
+		printk(KERN_ERR "lpfc: Cannot re-enable "
+			"PCI device after reset.\n");
+		return PCI_ERS_RESULT_DISCONNECT;
+	}
+
+	pci_set_master(pdev);
+
+	/* Re-establishing Link */
+	spin_lock_irq(phba->host->host_lock);
+	phba->fc_flag |= FC_ESTABLISH_LINK;
+	psli->sli_flag &= ~LPFC_SLI2_ACTIVE;
+	spin_unlock_irq(phba->host->host_lock);
+
+
+	/* Take device offline; this will perform cleanup */
+	lpfc_offline(phba);
+	lpfc_sli_brdrestart(phba);
+
+	return PCI_ERS_RESULT_RECOVERED;
+}
+
+/**
+ * lpfc_io_resume - called when traffic can start flowing again.
+ * @pdev: Pointer to PCI device
+ *
+ * This callback is called when the error recovery driver tells us that
+ * its OK to resume normal operation.
+ */
+static void lpfc_io_resume(struct pci_dev *pdev)
+{
+	struct Scsi_Host *host = pci_get_drvdata(pdev);
+	struct lpfc_hba *phba = (struct lpfc_hba *)host->hostdata;
+
+	if (lpfc_online(phba) == 0) {
+		mod_timer(&phba->fc_estabtmo, jiffies + HZ * 60);
+	}
+}
+
 static struct pci_device_id lpfc_id_table[] = {
 	{PCI_VENDOR_ID_EMULEX, PCI_DEVICE_ID_VIPER,
 		PCI_ANY_ID, PCI_ANY_ID, },
  
@@ -1857,11 +1947,18 @@
  
 MODULE_DEVICE_TABLE(pci, lpfc_id_table);
  
+static struct pci_error_handlers lpfc_err_handler = {
+	.error_detected = lpfc_io_error_detected,
+	.slot_reset = lpfc_io_slot_reset,
+	.resume = lpfc_io_resume,
+};
+
 static struct pci_driver lpfc_driver = {
 	.name		= LPFC_DRIVER_NAME,
 	.id_table	= lpfc_id_table,
 	.probe		= lpfc_pci_probe_one,
 	.remove		= __devexit_p(lpfc_pci_remove_one),
+	.err_handler = &lpfc_err_handler,
 };
  
 static int __init
@@ -2104,6 +2104,10 @@
 	volatile uint32_t word0, ldata;
 	void __iomem *to_slim;
  
+	/* If the PCI channel is in offline state, do not post mbox. */
+	if (unlikely(pci_channel_offline(phba->pcidev)))
+		return MBX_NOT_FINISHED;
+
 	psli = &phba->sli;
  
 	spin_lock_irqsave(phba->host->host_lock, drvr_flag);
@@ -2407,6 +2411,10 @@
 	struct lpfc_iocbq *nextiocb;
 	IOCB_t *iocb;
  
+	/* If the PCI channel is in offline state, do not post iocbs. */
+	if (unlikely(pci_channel_offline(phba->pcidev)))
+		return IOCB_ERROR;
+
 	/*
 	 * We should never get an IOCB if we are in a < LINK_DOWN state
 	 */
@@ -3152,6 +3160,10 @@
 	phba = (struct lpfc_hba *) dev_id;
  
 	if (unlikely(!phba))
+		return IRQ_NONE;
+
+	/* If the pci channel is offline, ignore all the interrupts. */
+	if (unlikely(pci_channel_offline(phba->pcidev)))
 		return IRQ_NONE;
  
 	phba->sli.slistat.sli_intr++;
...	...	@@ -518,6 +518,10 @@
518	518	struct lpfc_sli *psli = &phba->sli;
519	519	struct lpfc_sli_ring *pring;
520	520	uint32_t event_data;
	521	+ /* If the pci channel is offline, ignore possible errors,
	522	+ * since we cannot communicate with the pci card anyway. */
	523	+ if (pci_channel_offline(phba->pcidev))
	524	+ return;
521	525
522	526	if (phba->work_hs & HS_FFER6 \|\|
523	527	phba->work_hs & HS_FFER5) {
...	...	@@ -1797,6 +1801,92 @@
1797	1801	pci_set_drvdata(pdev, NULL);
1798	1802	}
1799	1803
	1804	+/**
	1805	+ * lpfc_io_error_detected - called when PCI error is detected
	1806	+ * @pdev: Pointer to PCI device
	1807	+ * @state: The current pci conneection state
	1808	+ *
	1809	+ * This function is called after a PCI bus error affecting
	1810	+ * this device has been detected.
	1811	+ */
	1812	+static pci_ers_result_t lpfc_io_error_detected(struct pci_dev *pdev,
	1813	+ pci_channel_state_t state)
	1814	+{
	1815	+ struct Scsi_Host *host = pci_get_drvdata(pdev);
	1816	+ struct lpfc_hba phba = (struct lpfc_hba )host->hostdata;
	1817	+ struct lpfc_sli *psli = &phba->sli;
	1818	+ struct lpfc_sli_ring *pring;
	1819	+
	1820	+ if (state == pci_channel_io_perm_failure) {
	1821	+ lpfc_pci_remove_one(pdev);
	1822	+ return PCI_ERS_RESULT_DISCONNECT;
	1823	+ }
	1824	+ pci_disable_device(pdev);
	1825	+ /*
	1826	+ * There may be I/Os dropped by the firmware.
	1827	+ * Error iocb (I/O) on txcmplq and let the SCSI layer
	1828	+ * retry it after re-establishing link.
	1829	+ */
	1830	+ pring = &psli->ring[psli->fcp_ring];
	1831	+ lpfc_sli_abort_iocb_ring(phba, pring);
	1832	+
	1833	+ /* Request a slot reset. */
	1834	+ return PCI_ERS_RESULT_NEED_RESET;
	1835	+}
	1836	+
	1837	+/**
	1838	+ * lpfc_io_slot_reset - called after the pci bus has been reset.
	1839	+ * @pdev: Pointer to PCI device
	1840	+ *
	1841	+ * Restart the card from scratch, as if from a cold-boot.
	1842	+ */
	1843	+static pci_ers_result_t lpfc_io_slot_reset(struct pci_dev *pdev)
	1844	+{
	1845	+ struct Scsi_Host *host = pci_get_drvdata(pdev);
	1846	+ struct lpfc_hba phba = (struct lpfc_hba )host->hostdata;
	1847	+ struct lpfc_sli *psli = &phba->sli;
	1848	+ int bars = pci_select_bars(pdev, IORESOURCE_MEM);
	1849	+
	1850	+ dev_printk(KERN_INFO, &pdev->dev, "recovering from a slot reset.\n");
	1851	+ if (pci_enable_device_bars(pdev, bars)) {
	1852	+ printk(KERN_ERR "lpfc: Cannot re-enable "
	1853	+ "PCI device after reset.\n");
	1854	+ return PCI_ERS_RESULT_DISCONNECT;
	1855	+ }
	1856	+
	1857	+ pci_set_master(pdev);
	1858	+
	1859	+ /* Re-establishing Link */
	1860	+ spin_lock_irq(phba->host->host_lock);
	1861	+ phba->fc_flag \|= FC_ESTABLISH_LINK;
	1862	+ psli->sli_flag &= ~LPFC_SLI2_ACTIVE;
	1863	+ spin_unlock_irq(phba->host->host_lock);
	1864	+
	1865	+
	1866	+ /* Take device offline; this will perform cleanup */
	1867	+ lpfc_offline(phba);
	1868	+ lpfc_sli_brdrestart(phba);
	1869	+
	1870	+ return PCI_ERS_RESULT_RECOVERED;
	1871	+}
	1872	+
	1873	+/**
	1874	+ * lpfc_io_resume - called when traffic can start flowing again.
	1875	+ * @pdev: Pointer to PCI device
	1876	+ *
	1877	+ * This callback is called when the error recovery driver tells us that
	1878	+ * its OK to resume normal operation.
	1879	+ */
	1880	+static void lpfc_io_resume(struct pci_dev *pdev)
	1881	+{
	1882	+ struct Scsi_Host *host = pci_get_drvdata(pdev);
	1883	+ struct lpfc_hba phba = (struct lpfc_hba )host->hostdata;
	1884	+
	1885	+ if (lpfc_online(phba) == 0) {
	1886	+ mod_timer(&phba->fc_estabtmo, jiffies + HZ * 60);
	1887	+ }
	1888	+}
	1889	+
1800	1890	static struct pci_device_id lpfc_id_table[] = {
1801	1891	{PCI_VENDOR_ID_EMULEX, PCI_DEVICE_ID_VIPER,
1802	1892	PCI_ANY_ID, PCI_ANY_ID, },
1803	1893
...	...	@@ -1857,11 +1947,18 @@
1857	1947
1858	1948	MODULE_DEVICE_TABLE(pci, lpfc_id_table);
1859	1949
	1950	+static struct pci_error_handlers lpfc_err_handler = {
	1951	+ .error_detected = lpfc_io_error_detected,
	1952	+ .slot_reset = lpfc_io_slot_reset,
	1953	+ .resume = lpfc_io_resume,
	1954	+};
	1955	+
1860	1956	static struct pci_driver lpfc_driver = {
1861	1957	.name = LPFC_DRIVER_NAME,
1862	1958	.id_table = lpfc_id_table,
1863	1959	.probe = lpfc_pci_probe_one,
1864	1960	.remove = __devexit_p(lpfc_pci_remove_one),
	1961	+ .err_handler = &lpfc_err_handler,
1865	1962	};
1866	1963
1867	1964	static int __init
...	...	@@ -2104,6 +2104,10 @@
2104	2104	volatile uint32_t word0, ldata;
2105	2105	void __iomem *to_slim;
2106	2106
	2107	+ /* If the PCI channel is in offline state, do not post mbox. */
	2108	+ if (unlikely(pci_channel_offline(phba->pcidev)))
	2109	+ return MBX_NOT_FINISHED;
	2110	+
2107	2111	psli = &phba->sli;
2108	2112
2109	2113	spin_lock_irqsave(phba->host->host_lock, drvr_flag);
...	...	@@ -2407,6 +2411,10 @@
2407	2411	struct lpfc_iocbq *nextiocb;
2408	2412	IOCB_t *iocb;
2409	2413
	2414	+ /* If the PCI channel is in offline state, do not post iocbs. */
	2415	+ if (unlikely(pci_channel_offline(phba->pcidev)))
	2416	+ return IOCB_ERROR;
	2417	+
2410	2418	/*
2411	2419	* We should never get an IOCB if we are in a < LINK_DOWN state
2412	2420	*/
...	...	@@ -3152,6 +3160,10 @@
3152	3160	phba = (struct lpfc_hba *) dev_id;
3153	3161
3154	3162	if (unlikely(!phba))
	3163	+ return IRQ_NONE;
	3164	+
	3165	+ /* If the pci channel is offline, ignore all the interrupts. */
	3166	+ if (unlikely(pci_channel_offline(phba->pcidev)))
3155	3167	return IRQ_NONE;
3156	3168
3157	3169	phba->sli.slistat.sli_intr++;