scsi: hisi_sas: Fix phyup timeout on FPGA

The OOB interrupt and phyup interrupt handlers may run out-of-order in high
CPU usage scenarios. Since the hisi_sas_phy.timer is added in
hisi_sas_phy_oob_ready() and disarmed in phy_up_v3_hw(), this out-of-order
execution will cause hisi_sas_phy.timer timeout to trigger.

To solve, protect hisi_sas_phy.timer and .attached with a lock, and ensure
that the timer won't be added after phyup handler completes.

Link: https://lore.kernel.org/r/1639579061-179473-8-git-send-email-john.garry@huawei.com
Signed-off-by: Qi Liu <liuqi115@huawei.com>
Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

diff --git a/drivers/scsi/hisi_sas/hisi_sas_main.c b/drivers/scsi/hisi_sas/hisi_sas_main.c
index 0e14f90..66e63a3 100644 (file)
--- a/drivers/scsi/hisi_sas/hisi_sas_main.c
+++ b/drivers/scsi/hisi_sas/hisi_sas_main.c
@@ -909,10 +909,14 @@ void hisi_sas_phy_oob_ready(struct hisi_hba *hisi_hba, int phy_no)
 {
        struct hisi_sas_phy *phy = &hisi_hba->phy[phy_no];
        struct device *dev = hisi_hba->dev;
+       unsigned long flags;
 
        dev_dbg(dev, "phy%d OOB ready\n", phy_no);
-       if (phy->phy_attached)
+       spin_lock_irqsave(&phy->lock, flags);
+       if (phy->phy_attached) {
+               spin_unlock_irqrestore(&phy->lock, flags);
                return;
+       }
 
        if (!timer_pending(&phy->timer)) {
                if (phy->wait_phyup_cnt < HISI_SAS_WAIT_PHYUP_RETRIES) {
@@ -920,13 +924,17 @@ void hisi_sas_phy_oob_ready(struct hisi_hba *hisi_hba, int phy_no)
                        phy->timer.expires = jiffies +
                                             HISI_SAS_WAIT_PHYUP_TIMEOUT;
                        add_timer(&phy->timer);
-               } else {
-                       dev_warn(dev, "phy%d failed to come up %d times, giving up\n",
-                                phy_no, phy->wait_phyup_cnt);
-                       phy->wait_phyup_cnt = 0;
+                       spin_unlock_irqrestore(&phy->lock, flags);
+                       return;
                }
+
+               dev_warn(dev, "phy%d failed to come up %d times, giving up\n",
+                        phy_no, phy->wait_phyup_cnt);
+               phy->wait_phyup_cnt = 0;
        }
+       spin_unlock_irqrestore(&phy->lock, flags);
 }
+
 EXPORT_SYMBOL_GPL(hisi_sas_phy_oob_ready);
 
 static void hisi_sas_phy_init(struct hisi_hba *hisi_hba, int phy_no)

diff --git a/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c b/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c
index 11a44d9..0239e2b 100644 (file)
--- a/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c
+++ b/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c
@@ -1484,7 +1484,6 @@ static irqreturn_t phy_up_v3_hw(int phy_no, struct hisi_hba *hisi_hba)
        struct asd_sas_phy *sas_phy = &phy->sas_phy;
        struct device *dev = hisi_hba->dev;
 
-       del_timer(&phy->timer);
        hisi_sas_phy_write32(hisi_hba, phy_no, PHYCTRL_PHY_ENA_MSK, 1);
 
        port_id = hisi_sas_read32(hisi_hba, PHY_PORT_NUM_MA);
@@ -1561,9 +1560,16 @@ static irqreturn_t phy_up_v3_hw(int phy_no, struct hisi_hba *hisi_hba)
        }
 
        phy->port_id = port_id;
-       phy->phy_attached = 1;
+
        hisi_sas_notify_phy_event(phy, HISI_PHYE_PHY_UP);
+
        res = IRQ_HANDLED;
+
+       spin_lock(&phy->lock);
+       /* Delete timer and set phy_attached atomically */
+       del_timer(&phy->timer);
+       phy->phy_attached = 1;
+       spin_unlock(&phy->lock);
 end:
        if (phy->reset_completion)
                complete(phy->reset_completion);