6.1-stable review patch. If anyone has any objections, please let me know.
------------------
From: Yihang Li liyihang9@huawei.com
[ Upstream commit ce26497c745d0541aec930d5211b431a1c26af97 ]
Currently in directly attached scenario, the phyup event HISI_PHYE_PHY_UP_PM is notified before .phy_attached is set - this may cause the phyup work hisi_sas_bytes_dmaed() execution failed and the attached device will not be found.
To fix it, set .phy_attached before notifing phyup event.
Signed-off-by: Yihang Li liyihang9@huawei.com Signed-off-by: Xiang Chen chenxiang66@hisilicon.com Link: https://lore.kernel.org/r/1702525516-51258-2-git-send-email-chenxiang66@hisi... Signed-off-by: Martin K. Petersen martin.petersen@oracle.com Signed-off-by: Sasha Levin sashal@kernel.org --- drivers/scsi/hisi_sas/hisi_sas_v3_hw.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c b/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c index 0c80ff9affa3..7ae56a2fe232 100644 --- a/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c +++ b/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c @@ -1565,6 +1565,11 @@ static irqreturn_t phy_up_v3_hw(int phy_no, struct hisi_hba *hisi_hba) }
phy->port_id = port_id; + spin_lock(&phy->lock); + /* Delete timer and set phy_attached atomically */ + del_timer(&phy->timer); + phy->phy_attached = 1; + spin_unlock(&phy->lock);
/* * Call pm_runtime_get_noresume() which pairs with @@ -1578,11 +1583,6 @@ static irqreturn_t phy_up_v3_hw(int phy_no, struct hisi_hba *hisi_hba)
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);