[06/17] net/hns3: avoid accessing nonexistent VF reg when PF in FLR

Message ID 20200922085401.12272-7-huwei013@chinasoftinc.com (mailing list archive)
State Superseded, archived
Delegated to: Ferruh Yigit
Headers
Series updates for hns3 PMD driver |

Checks

Context Check Description
ci/checkpatch success coding style OK

Commit Message

Wei Hu (Xavier) Sept. 22, 2020, 8:53 a.m. UTC
  From: Hongbo Zheng <zhenghongbo3@huawei.com>

According to the protocol of PCIe, FLR to a PF device resets the PF state
as well as the SR-IOV extended capability including VF Enable which means
that VFs no longer exist.

When PF device is in FLR reset stage, at this time, the register state
of VF device is not reliable, so VF device's register state detection
is not carried out in PF FLR.

In this case, we just ignore the register states to avoid accessing
nonexistent register and return false in the internal function named
hns3vf_is_reset_pending to indicate that there are no other reset states
that need to be processed by PMD driver.

Fixes: 2790c6464725 ("net/hns3: support device reset")
Cc: stable@dpdk.org

Signed-off-by: Hongbo Zheng <zhenghongbo3@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
---
 drivers/net/hns3/hns3_ethdev.h    |  7 +++++++
 drivers/net/hns3/hns3_ethdev_vf.c | 15 +++++++++++++++
 2 files changed, 22 insertions(+)
  

Patch

diff --git a/drivers/net/hns3/hns3_ethdev.h b/drivers/net/hns3/hns3_ethdev.h
index f170df9..3f3f973 100644
--- a/drivers/net/hns3/hns3_ethdev.h
+++ b/drivers/net/hns3/hns3_ethdev.h
@@ -275,6 +275,13 @@  enum hns3_reset_level {
 	 * Kernel PF driver use mailbox to inform DPDK VF to do reset, the value
 	 * of the reset level and the one defined in kernel driver should be
 	 * same.
+	 *
+	 * According to the protocol of PCIe, FLR to a PF resets the PF state as
+	 * well as the SR-IOV extended capability including VF Enable which
+	 * means that VFs no longer exist.
+	 *
+	 * In PF FLR, the register state of VF is not reliable, VF's driver
+	 * should not access the registers of the VF device.
 	 */
 	HNS3_VF_FULL_RESET = 3,
 	HNS3_FLR_RESET,     /* A VF perform FLR reset */
diff --git a/drivers/net/hns3/hns3_ethdev_vf.c b/drivers/net/hns3/hns3_ethdev_vf.c
index 565cf60..cb2747b 100644
--- a/drivers/net/hns3/hns3_ethdev_vf.c
+++ b/drivers/net/hns3/hns3_ethdev_vf.c
@@ -2191,6 +2191,21 @@  hns3vf_is_reset_pending(struct hns3_adapter *hns)
 	struct hns3_hw *hw = &hns->hw;
 	enum hns3_reset_level reset;
 
+	/*
+	 * According to the protocol of PCIe, FLR to a PF device resets the PF
+	 * state as well as the SR-IOV extended capability including VF Enable
+	 * which means that VFs no longer exist.
+	 *
+	 * HNS3_VF_FULL_RESET means PF device is in FLR reset. when PF device
+	 * is in FLR stage, the register state of VF device is not reliable,
+	 * so register states detection can not be carried out. In this case,
+	 * we just ignore the register states and return false to indicate that
+	 * there are no other reset states that need to be processed by driver.
+	 */
+	if (hw->reset.level == HNS3_VF_FULL_RESET)
+		return false;
+
+	/* Check the registers to confirm whether there is reset pending */
 	hns3vf_check_event_cause(hns, NULL);
 	reset = hns3vf_get_reset_level(hw, &hw->reset.pending);
 	if (hw->reset.level != HNS3_NONE_RESET && hw->reset.level < reset) {