Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
scsi: lpfc: Trigger SLI4 firmware dump before doing driver cleanup
Extraneous teardown routines are present in the firmware dump path causing
altered states in firmware captures.
When a firmware dump is requested via sysfs, trigger the dump immediately
without tearing down structures and changing adapter state.
The driver shall rely on pre-existing firmware error state clean up
handlers to restore the adapter.
Link: https://lore.kernel.org/r/20211204002644.116455-6-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
authored by
James Smart
and committed by
Martin K. Petersen
7dd2e2a9
8ed190a9
+51
-33
+1
-1
drivers/scsi/lpfc/lpfc.h
+1
-1
drivers/scsi/lpfc/lpfc.h
···
1021
1021
#define HBA_DEVLOSS_TMO 0x2000 /* HBA in devloss timeout */
1022
1022
#define HBA_RRQ_ACTIVE 0x4000 /* process the rrq active list */
1023
1023
#define HBA_IOQ_FLUSH 0x8000 /* FCP/NVME I/O queues being flushed */
1024
-
#define HBA_FW_DUMP_OP 0x10000 /* Skips fn reset before FW dump */
1025
1024
#define HBA_RECOVERABLE_UE 0x20000 /* Firmware supports recoverable UE */
1026
1025
#define HBA_FORCED_LINK_SPEED 0x40000 /*
1027
1026
* Firmware supports Forced Link Speed
···
1037
1038
#define HBA_HBEAT_TMO 0x8000000 /* HBEAT initiated after timeout */
1038
1039
#define HBA_FLOGI_OUTSTANDING 0x10000000 /* FLOGI is outstanding */
1039
1040
1041
+
struct completion *fw_dump_cmpl; /* cmpl event tracker for fw_dump */
1040
1042
uint32_t fcp_ring_in_use; /* When polling test if intr-hndlr active*/
1041
1043
struct lpfc_dmabuf slim2p;
1042
1044
+43
-25
drivers/scsi/lpfc/lpfc_attr.c
+43
-25
drivers/scsi/lpfc/lpfc_attr.c
···
1709
1709
before_fc_flag = phba->pport->fc_flag;
1710
1710
sriov_nr_virtfn = phba->cfg_sriov_nr_virtfn;
1711
1711
1712
-
/* Disable SR-IOV virtual functions if enabled */
1713
-
if (phba->cfg_sriov_nr_virtfn) {
1714
-
pci_disable_sriov(pdev);
1715
-
phba->cfg_sriov_nr_virtfn = 0;
1712
+
if (opcode == LPFC_FW_DUMP) {
1713
+
init_completion(&online_compl);
1714
+
phba->fw_dump_cmpl = &online_compl;
1715
+
} else {
1716
+
/* Disable SR-IOV virtual functions if enabled */
1717
+
if (phba->cfg_sriov_nr_virtfn) {
1718
+
pci_disable_sriov(pdev);
1719
+
phba->cfg_sriov_nr_virtfn = 0;
1720
+
}
1721
+
1722
+
status = lpfc_do_offline(phba, LPFC_EVT_OFFLINE);
1723
+
1724
+
if (status != 0)
1725
+
return status;
1726
+
1727
+
/* wait for the device to be quiesced before firmware reset */
1728
+
msleep(100);
1716
1729
}
1717
-
1718
-
if (opcode == LPFC_FW_DUMP)
1719
-
phba->hba_flag |= HBA_FW_DUMP_OP;
1720
-
1721
-
status = lpfc_do_offline(phba, LPFC_EVT_OFFLINE);
1722
-
1723
-
if (status != 0) {
1724
-
phba->hba_flag &= ~HBA_FW_DUMP_OP;
1725
-
return status;
1726
-
}
1727
-
1728
-
/* wait for the device to be quiesced before firmware reset */
1729
-
msleep(100);
1730
1730
1731
1731
reg_val = readl(phba->sli4_hba.conf_regs_memmap_p +
1732
1732
LPFC_CTL_PDEV_CTL_OFFSET);
···
1756
1756
lpfc_printf_log(phba, KERN_ERR, LOG_SLI,
1757
1757
"3153 Fail to perform the requested "
1758
1758
"access: x%x\n", reg_val);
1759
+
if (phba->fw_dump_cmpl)
1760
+
phba->fw_dump_cmpl = NULL;
1759
1761
return rc;
1760
1762
}
1761
1763
1762
1764
/* keep the original port state */
1763
-
if (before_fc_flag & FC_OFFLINE_MODE)
1765
+
if (before_fc_flag & FC_OFFLINE_MODE) {
1766
+
if (phba->fw_dump_cmpl)
1767
+
phba->fw_dump_cmpl = NULL;
1764
1768
goto out;
1769
+
}
1765
1770
1766
-
init_completion(&online_compl);
1767
-
job_posted = lpfc_workq_post_event(phba, &status, &online_compl,
1768
-
LPFC_EVT_ONLINE);
1769
-
if (!job_posted)
1770
-
goto out;
1771
+
/* Firmware dump will trigger an HA_ERATT event, and
1772
+
* lpfc_handle_eratt_s4 routine already handles bringing the port back
1773
+
* online.
1774
+
*/
1775
+
if (opcode == LPFC_FW_DUMP) {
1776
+
wait_for_completion(phba->fw_dump_cmpl);
1777
+
} else {
1778
+
init_completion(&online_compl);
1779
+
job_posted = lpfc_workq_post_event(phba, &status, &online_compl,
1780
+
LPFC_EVT_ONLINE);
1781
+
if (!job_posted)
1782
+
goto out;
1771
1783
1772
-
wait_for_completion(&online_compl);
1773
-
1784
+
wait_for_completion(&online_compl);
1785
+
}
1774
1786
out:
1775
1787
/* in any case, restore the virtual functions enabled as before */
1776
1788
if (sriov_nr_virtfn) {
1789
+
/* If fw_dump was performed, first disable to clean up */
1790
+
if (opcode == LPFC_FW_DUMP) {
1791
+
pci_disable_sriov(pdev);
1792
+
phba->cfg_sriov_nr_virtfn = 0;
1793
+
}
1794
+
1777
1795
sriov_err =
1778
1796
lpfc_sli_probe_sriov_nr_virtfn(phba, sriov_nr_virtfn);
1779
1797
if (!sriov_err)
+7
-1
drivers/scsi/lpfc/lpfc_hbadisc.c
+7
-1
drivers/scsi/lpfc/lpfc_hbadisc.c
···
869
869
if (phba->pci_dev_grp == LPFC_PCI_DEV_OC)
870
870
lpfc_sli4_post_async_mbox(phba);
871
871
872
-
if (ha_copy & HA_ERATT)
872
+
if (ha_copy & HA_ERATT) {
873
873
/* Handle the error attention event */
874
874
lpfc_handle_eratt(phba);
875
+
876
+
if (phba->fw_dump_cmpl) {
877
+
complete(phba->fw_dump_cmpl);
878
+
phba->fw_dump_cmpl = NULL;
879
+
}
880
+
}
875
881
876
882
if (ha_copy & HA_MBATT)
877
883
lpfc_sli_handle_mb_event(phba);
-6
drivers/scsi/lpfc/lpfc_sli.c
-6
drivers/scsi/lpfc/lpfc_sli.c
···
5046
5046
phba->fcf.fcf_flag = 0;
5047
5047
spin_unlock_irq(&phba->hbalock);
5048
5048
5049
-
/* SLI4 INTF 2: if FW dump is being taken skip INIT_PORT */
5050
-
if (phba->hba_flag & HBA_FW_DUMP_OP) {
5051
-
phba->hba_flag &= ~HBA_FW_DUMP_OP;
5052
-
return rc;
5053
-
}
5054
-
5055
5049
/* Now physically reset the device */
5056
5050
lpfc_printf_log(phba, KERN_INFO, LOG_INIT,
5057
5051
"0389 Performing PCI function reset!\n");