hisi_sas: Some misc updates
Subsystems (2)
cover hisi_sas: Some misc updates
Hi Martin,
This series includes some updates as following:
- better management of freeing PCI vectors
- hopefully the following series can make it soon so we may remove this
new code:
https://lore.kernel.org/linux-pci/[email protected]/T/#md27a594ac53b8e05abb2f22ec350137b9023ff18- Fix prot_mask module param permissions
- Change phyup timeout to handle some special SATA disks
- modify underflow handling to quickly identify faulty disks
- debugfs hardening
Please consider for 5.18 - hopefully not too late
Thanks!
Qi Liu (2):
scsi: hisi_sas: Free irq vectors in order for v3 HW
scsi: hisi_sas: Rename error labels in hisi_sas_v3_probe()
Xiang Chen (3):
scsi: hisi_sas: Change permission of parameter prot_mask
scsi: hisi_sas: Change hisi_sas_control_phy() phyup timeout
scsi: hisi_sas: Limit users changing debugfs BIST count value
Xingui Yang (1):
scsi: hisi_sas: Modify v3 HW SSP underflow error processing
drivers/scsi/hisi_sas/hisi_sas.h | 2 +- drivers/scsi/hisi_sas/hisi_sas_main.c | 3 +- drivers/scsi/hisi_sas/hisi_sas_v3_hw.c | 127 +++++++++++++++++++------ 3 files changed, 100 insertions(+), 32 deletions(-)
Signature · 1 line
2.26.2
PATCH 1/6 scsi: hisi_sas: Change permission of parameter prot_mask
From: Xiang Chen <[email protected]>
Currently the permission of parameter prot_mask is 0x0, which means that the member does not appear in sysfs. Change it as other module parameters to 0444 for world-readable.
--- drivers/scsi/hisi_sas/hisi_sas_v3_hw.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/scsi/hisi_sas/hisi_sas_v3_hw.c +1 -1
@@ -530,7 +530,7 @@ MODULE_PARM_DESC(intr_conv, "interrupt converge enable (0-1)");/* permit overriding the host protection capabilities mask (EEDP/T10 PI) */ static int prot_mask; -module_param (prot_mask, int,0 ); +module_param (prot_mask, int,0444 );MODULE_PARM_DESC (prot_mask," host protection capabilities mask, def=0x0 " );static void debugfs_work_handler_v3_hw (struct work_struct *work);
Signature · 1 line
2.26.2
PATCH 2/6 scsi: hisi_sas: Change hisi_sas_control_phy() phyup timeout
In reply to John GarryFrom: Xiang Chen <[email protected]>
The time of phyup not only depends on the controller but also the type of disk connected. As an example, from experience, for some SATA disks the amount of time from reset/power-on to receive the D2H FIS for phyup can take upto and more than 10s sometimes. According to the specification of some SATA disks such as ST14000NM0018, the max time from power-on to ready is 30s.
Based on this the current timeout of phyup at 2s which is not enough. So set the value as HISI_SAS_WAIT_PHYUP_TIMEOUT (30s) in hisi_sas_control_phy().
For v3 hw there is a pre-existing workaround for a HW bug, being that we issue a link reset when the OOB occurs but the phyup does not. The current phyup timeout is HISI_SAS_WAIT_PHYUP_TIMEOUT. So if this does occur from when issuing a phy enable or similar via hisi_sas_control_phy(), the subsequent HW workaround linkreset processing calls hisi_sas_control_phy(), but this will pend the original phy reset timing out, so it is safe.
--- drivers/scsi/hisi_sas/hisi_sas.h | 2 +- drivers/scsi/hisi_sas/hisi_sas_main.c | 3 ++- 2 files changed, 3 insertions(+), 2 deletions(-)
drivers/scsi/hisi_sas/hisi_sas.h +1 -1
@@ -91,7 +91,7 @@#define HISI_SAS_PROT_MASK (HISI_SAS_DIF_PROT_MASK | HISI_SAS_DIX_PROT_MASK) -#define HISI_SAS_WAIT_PHYUP_TIMEOUT (20 * HZ) +#define HISI_SAS_WAIT_PHYUP_TIMEOUT (30 * HZ)#define HISI_SAS_CLEAR_ITCT_TIMEOUT (20 * HZ)struct hisi_hba ;
drivers/scsi/hisi_sas/hisi_sas_main.c +2 -1
@@ -1201,7 +1201,8 @@ static int hisi_sas_control_phy(struct asd_sas_phy *sas_phy, enum phy_func func, goto out; } -if (sts && !wait_for_completion_timeout (&completion,2 *HZ )) { +if (sts && !wait_for_completion_timeout (&completion, +HISI_SAS_WAIT_PHYUP_TIMEOUT )) {dev_warn (dev,"phy%d wait phyup timed out for func %d\n" , phy_no, func);if (phy->in_reset)
Signature · 1 line
2.26.2
PATCH 3/6 scsi: hisi_sas: Free irq vectors in order for v3 HW
In reply to John GarryFrom: Qi Liu <[email protected]>
If the driver probe fails to request the channel IRQ or fatal IRQ, the driver will free the IRQ vectors before freeing the IRQs in free_irq(), and this will cause a kernel BUG like this:
------------[ cut here ]------------ kernel BUG at drivers/pci/msi.c:369! Internal error: Oops - BUG: 0 [#1] PREEMPT SMP Call trace: free_msi_irqs+0x118/0x13c pci_disable_msi+0xfc/0x120 pci_free_irq_vectors+0x24/0x3c hisi_sas_v3_probe+0x360/0x9d0 [hisi_sas_v3_hw] local_pci_probe+0x44/0xb0 work_for_cpu_fn+0x20/0x34 process_one_work+0x1d0/0x340 worker_thread+0x2e0/0x460 kthread+0x180/0x190 ret_from_fork+0x10/0x20 ---[ end trace b88990335b610c11 ]---
So we use devm_add_action() to control the order in which we free the vectors.
--- drivers/scsi/hisi_sas/hisi_sas_v3_hw.c | 16 +++++++++++----- 1 file changed, 11 insertions(+), 5 deletions(-)
drivers/scsi/hisi_sas/hisi_sas_v3_hw.c +11 -5
@@ -2397,17 +2397,25 @@ static irqreturn_t cq_interrupt_v3_hw(int irq_no, void *p)return IRQ_WAKE_THREAD ; } +static void hisi_sas_v3_free_vectors (void *data) +{ +struct pci_dev *pdev = data; + +pci_free_irq_vectors (pdev); +} +static int interrupt_preinit_v3_hw (struct hisi_hba *hisi_hba) {int vectors;int max_msi =HISI_SAS_MSI_COUNT_V3_HW , min_msi;struct Scsi_Host *shost = hisi_hba->shost; +struct pci_dev *pdev = hisi_hba->pci_dev;struct irq_affinity desc = { .pre_vectors =BASE_VECTORS_V3_HW , }; min_msi =MIN_AFFINE_VECTORS_V3_HW ; - vectors = pci_alloc_irq_vectors_affinity(hisi_hba->pci_dev, + vectors = pci_alloc_irq_vectors_affinity(pdev, min_msi, max_msi, PCI_IRQ_MSI | PCI_IRQ_AFFINITY,
@@ -2419,6 +2427,7 @@ static int interrupt_preinit_v3_hw(struct hisi_hba *hisi_hba) hisi_hba->cq_nvecs = vectors -BASE_VECTORS_V3_HW ; shost->nr_hw_queues = hisi_hba->cq_nvecs; +devm_add_action (&pdev->dev, hisi_sas_v3_free_vectors, pdev);return 0 ; }
@@ -4768,7 +4777,7 @@ hisi_sas_v3_probe(struct pci_dev *pdev, const struct pci_device_id *id) rc =scsi_add_host (shost, dev);if (rc) - goto err_out_free_irq_vectors; + goto err_out_debugfs; rc =sas_register_ha (sha);if (rc)
@@ -4799,8 +4808,6 @@ hisi_sas_v3_probe(struct pci_dev *pdev, const struct pci_device_id *id)sas_unregister_ha (sha); err_out_register_ha:scsi_remove_host (shost); -err_out_free_irq_vectors: -pci_free_irq_vectors (pdev); err_out_debugfs:debugfs_exit_v3_hw (hisi_hba); err_out_ha:
@@ -4824,7 +4831,6 @@ hisi_sas_v3_destroy_irqs(struct pci_dev *pdev, struct hisi_hba *hisi_hba)devm_free_irq (&pdev->dev,pci_irq_vector (pdev, nr), cq); } -pci_free_irq_vectors (pdev); }static void hisi_sas_v3_remove (struct pci_dev *pdev)
Signature · 1 line
2.26.2
PATCH 4/6 scsi: hisi_sas: Rename error labels in hisi_sas_v3_probe()
In reply to John GarryFrom: Qi Liu <[email protected]>
To avoid doubt, rename the error labels to indicate the action they will take.
--- drivers/scsi/hisi_sas/hisi_sas_v3_hw.c | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-)
drivers/scsi/hisi_sas/hisi_sas_v3_hw.c +10 -10
@@ -4723,7 +4723,7 @@ hisi_sas_v3_probe(struct pci_dev *pdev, const struct pci_device_id *id)if (!hisi_hba->regs) {dev_err (dev,"cannot map register\n" ); rc = -ENOMEM ; - goto err_out_ha; + goto err_out_free_host; } phy_nr = port_nr = hisi_hba->n_phy;
@@ -4732,7 +4732,7 @@ hisi_sas_v3_probe(struct pci_dev *pdev, const struct pci_device_id *id) arr_port =devm_kcalloc (dev, port_nr,sizeof (void *),GFP_KERNEL );if (!arr_phy || !arr_port) { rc = -ENOMEM ; - goto err_out_ha; + goto err_out_free_host; } sha->sas_phy = arr_phy;
@@ -4773,19 +4773,19 @@ hisi_sas_v3_probe(struct pci_dev *pdev, const struct pci_device_id *id) rc =interrupt_preinit_v3_hw (hisi_hba);if (rc) - goto err_out_debugfs; + goto err_out_undo_debugfs; rc =scsi_add_host (shost, dev);if (rc) - goto err_out_debugfs; + goto err_out_undo_debugfs; rc =sas_register_ha (sha);if (rc) - goto err_out_register_ha; + goto err_out_remove_host; rc =hisi_sas_v3_init (hisi_hba);if (rc) - goto err_out_hw_init; + goto err_out_unregister_ha;scsi_scan_host (shost);
@@ -4804,13 +4804,13 @@ hisi_sas_v3_probe(struct pci_dev *pdev, const struct pci_device_id *id)return 0 ; -err_out_hw_init: +err_out_unregister_ha:sas_unregister_ha (sha); -err_out_register_ha: +err_out_remove_host:scsi_remove_host (shost); -err_out_debugfs: +err_out_undo_debugfs:debugfs_exit_v3_hw (hisi_hba); -err_out_ha: +err_out_free_host:hisi_sas_free (hisi_hba);scsi_host_put (shost); err_out:
Signature · 1 line
2.26.2
PATCH 5/6 scsi: hisi_sas: Limit users changing debugfs BIST count value
In reply to John GarryFrom: Xiang Chen <[email protected]>
Add a file operation for "cnt" file under bist directory, so users could only read "cnt" or clear "cnt" to zero, but cannot randomly modify.
--- drivers/scsi/hisi_sas/hisi_sas_v3_hw.c | 52 +++++++++++++++++++++++++- 1 file changed, 50 insertions(+), 2 deletions(-)
drivers/scsi/hisi_sas/hisi_sas_v3_hw.c +50 -2
@@ -3975,6 +3975,54 @@ static const struct file_operations debugfs_bist_phy_v3_hw_fops = { .owner = THIS_MODULE, }; +static ssize_t debugfs_bist_cnt_v3_hw_write (struct file *filp, +const char __user *buf, +size_t count,loff_t *ppos) +{ +struct seq_file *m = filp->private_data; +struct hisi_hba *hisi_hba = m->private; +unsigned int cnt; +int val; + +if (hisi_hba->debugfs_bist_enable) +return -EPERM ; + + val =kstrtouint_from_user (buf, count,0 , &cnt); +if (val) +return val; + +if (cnt) +return -EINVAL ; + + hisi_hba->debugfs_bist_cnt =0 ; +return count; +} + +static int debugfs_bist_cnt_v3_hw_show (struct seq_file *s,void *p) +{ +struct hisi_hba *hisi_hba = s->private; + +seq_printf (s,"%u\n" , hisi_hba->debugfs_bist_cnt); + +return 0 ; +} + +static int debugfs_bist_cnt_v3_hw_open (struct inode *inode, +struct file *filp) +{ +return single_open (filp, debugfs_bist_cnt_v3_hw_show, + inode->i_private); +} + +static const struct file_operations debugfs_bist_cnt_v3_hw_ops = { + .open = debugfs_bist_cnt_v3_hw_open, + .read = seq_read, + .write = debugfs_bist_cnt_v3_hw_write, + .llseek = seq_lseek, + .release = single_release, + .owner =THIS_MODULE , +}; + static const struct { int value; char *name;
@@ -4612,8 +4660,8 @@ static void debugfs_bist_init_v3_hw(struct hisi_hba *hisi_hba)debugfs_create_file ("phy_id" ,0600 , hisi_hba->debugfs_bist_dentry, hisi_hba, &debugfs_bist_phy_v3_hw_fops); -debugfs_create_u32 ("cnt" ,0600 , hisi_hba->debugfs_bist_dentry, - &hisi_hba->debugfs_bist_cnt); +debugfs_create_file ("cnt" ,0600 , hisi_hba->debugfs_bist_dentry, + hisi_hba, &debugfs_bist_cnt_v3_hw_ops); debugfs_create_file("loopback_mode", 0600, hisi_hba->debugfs_bist_dentry,
Signature · 1 line
2.26.2
PATCH 6/6 scsi: hisi_sas: Modify v3 HW SSP underflow error processing
In reply to John GarryFrom: Xingui Yang <[email protected]>
In case of SSP underflow allow the response frame IU to be examined for setting the response stat value rather than always setting SAS_DATA_UNDERRUN.
This will mean that we call sas_ssp_task_response() in those scenarios and may send sense data to upper layer.
Such a condition would be for bad blocks were we just reporting an underflow error to upper layer, but now the sense data will tell immediately that the media is faulty.
--- drivers/scsi/hisi_sas/hisi_sas_v3_hw.c | 39 +++++++++++++++++--------- 1 file changed, 26 insertions(+), 13 deletions(-)
drivers/scsi/hisi_sas/hisi_sas_v3_hw.c +26 -13
@@ -406,6 +406,8 @@#define CMPLT_HDR_ERROR_PHASE_MSK (0xff << CMPLT_HDR_ERROR_PHASE_OFF)#define CMPLT_HDR_RSPNS_XFRD_OFF 10#define CMPLT_HDR_RSPNS_XFRD_MSK (0x1 << CMPLT_HDR_RSPNS_XFRD_OFF) +#define CMPLT_HDR_RSPNS_GOOD_OFF 11 +#define CMPLT_HDR_RSPNS_GOOD_MSK (0x1 << CMPLT_HDR_RSPNS_GOOD_OFF)#define CMPLT_HDR_ERX_OFF 12#define CMPLT_HDR_ERX_MSK (0x1 << CMPLT_HDR_ERX_OFF)#define CMPLT_HDR_ABORT_STAT_OFF 13
@@ -2140,7 +2142,7 @@ static irqreturn_t fatal_axi_int_v3_hw(int irq_no, void *p)return IRQ_HANDLED ; } -static void +static bool slot_err_v3_hw (struct hisi_hba *hisi_hba,struct sas_task *task,struct hisi_sas_slot *slot) {
@@ -2158,6 +2160,15 @@ slot_err_v3_hw(struct hisi_hba *hisi_hba, struct sas_task *task,switch (task->task_proto) {case SAS_PROTOCOL_SSP :if (dma_rx_err_type &RX_DATA_LEN_UNDERFLOW_MSK ) { +/* +* If returned response frame is incorrect because of data underflow, +* but I/O information has been written to the host memory, we examine +* response IU. +*/ +if (!(complete_hdr->dw0 &CMPLT_HDR_RSPNS_GOOD_MSK ) && + (complete_hdr->dw0 &CMPLT_HDR_RSPNS_XFRD_MSK )) +return false; + ts->residual = trans_tx_fail_type; ts->stat =SAS_DATA_UNDERRUN ; }else if (dw3 &CMPLT_HDR_IO_IN_TARGET_MSK ) {
@@ -2189,6 +2200,7 @@ slot_err_v3_hw(struct hisi_hba *hisi_hba, struct sas_task *task,default :break ; } +return true; } static void slot_complete_v3_hw(struct hisi_hba *hisi_hba,
@@ -2262,19 +2274,20 @@ static void slot_complete_v3_hw(struct hisi_hba *hisi_hba,if ((dw0 &CMPLT_HDR_CMPLT_MSK ) ==0x3 ) {u32 *error_info =hisi_sas_status_buf_addr_mem (slot); -slot_err_v3_hw (hisi_hba, task, slot); -if (ts->stat !=SAS_DATA_UNDERRUN ) -dev_info (dev,"erroneous completion iptt=%d task=%pK dev id=%d addr=%016llx CQ hdr: 0x%x 0x%x 0x%x 0x%x Error info: 0x%x 0x%x 0x%x 0x%x\n" , - slot->idx, task, sas_dev->device_id, -SAS_ADDR (device->sas_addr), - dw0, dw1, complete_hdr->act, dw3, - error_info[0 ], error_info[1 ], - error_info[2 ], error_info[3 ]); -if (unlikely (slot->abort)) { -sas_task_abort (task); -return ; +if (slot_err_v3_hw (hisi_hba, task, slot)) { +if (ts->stat !=SAS_DATA_UNDERRUN ) +dev_info (dev,"erroneous completion iptt=%d task=%pK dev id=%d addr=%016llx CQ hdr: 0x%x 0x%x 0x%x 0x%x Error info: 0x%x 0x%x 0x%x 0x%x\n" , + slot->idx, task, sas_dev->device_id, +SAS_ADDR (device->sas_addr), + dw0, dw1, complete_hdr->act, dw3, + error_info[0 ], error_info[1 ], + error_info[2 ], error_info[3 ]); +if (unlikely (slot->abort)) { +sas_task_abort (task); +return ; + } + goto out; } - goto out; }switch (task->task_proto) {
Signature · 1 line
2.26.2
Martin K. Petersen
AppliedJohn,
Quoted 1 line
This series includes some updates as following:
Applied to 5.18/scsi-staging, thanks!
Signature · 1 line
Martin K. Petersen Oracle Linux Engineering
Martin K. Petersen
AppliedQuoted 11 lines · John Garry
This series includes some updates as following:
- better management of freeing PCI vectors
- hopefully the following series can make it soon so we may remove thisnew code: https://lore.kernel.org/linux-pci/[email protected]/T/#md27a594ac53b8e05abb2f22ec350137b9023ff18- Fix
prot_maskmodule param permissions
- Change phyup timeout to handle some special SATA disks
- modify underflow handling to quickly identify faulty disks
- debugfs hardening[...]
Applied to 5.18/scsi-queue, thanks!
[1/6] scsi: hisi_sas: Change permission of parameter prot_mask
https://git.kernel.org/mkp/scsi/c/c4e070457a93 [2/6] scsi: hisi_sas: Change hisi_sas_control_phy() phyup timeout https://git.kernel.org/mkp/scsi/c/512623de5239 [3/6] scsi: hisi_sas: Free irq vectors in order for v3 HW https://git.kernel.org/mkp/scsi/c/554fb72ee34f [4/6] scsi: hisi_sas: Rename error labels in hisi_sas_v3_probe() https://git.kernel.org/mkp/scsi/c/86287065fac2 [5/6] scsi: hisi_sas: Limit users changing debugfs BIST count value https://git.kernel.org/mkp/scsi/c/286ce4c65fbd [6/6] scsi: hisi_sas: Modify v3 HW SSP underflow error processing https://git.kernel.org/mkp/scsi/c/62413199cd6d
Signature · 1 line
Martin K. Petersen Oracle Linux Engineering