* [PATCH net-next 0/7] qed: load/unload mfw series
@ 2017-03-28 12:12 Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 1/7] qed: Correct HW stop flow Yuval Mintz
` (7 more replies)
0 siblings, 8 replies; 9+ messages in thread
From: Yuval Mintz @ 2017-03-28 12:12 UTC (permalink / raw)
To: davem, netdev; +Cc: Yuval Mintz
This series correct the unload flow and greatly enhances its
initialization flow in regard to interactions between driver
and management firmware.
Patch #1 makes sure unloading is done under management-firmware's
'criticial section' protection.
Patches #2 - #4 move driver into using a newer scheme for loading
in regard to the MFW; This newer scheme would help cleaning the device
in case a previous instance has dirtied it [preboot, PDA, etc.].
Patches #5 - #6 let driver inform management-firmware on number of
resources which are dependent on the non-management firmware used.
Patch #7 then uses a new resource [BDQ] instead of some set value.
Dave,
Please consider applying this series to 'net-next'.
Thanks,
Yuval
Tomer Tayar (4):
qed: Correct HW stop flow
qed: Move to new load request scheme
qed: Support management-based resource locking
qed: Utilize resource-lock based scheme
Yuval Mintz (3):
qed: hw_init() to receive parameter-struct
qed: Send pf-flr as part of initialization
qed: Use BDQ resource for storage protocols
drivers/net/ethernet/qlogic/qed/qed.h | 33 +-
drivers/net/ethernet/qlogic/qed/qed_dcbx.h | 3 -
drivers/net/ethernet/qlogic/qed/qed_dev.c | 579 ++++++++++--------
drivers/net/ethernet/qlogic/qed/qed_dev_api.h | 73 ++-
drivers/net/ethernet/qlogic/qed/qed_fcoe.c | 30 +-
drivers/net/ethernet/qlogic/qed/qed_hsi.h | 87 ++-
drivers/net/ethernet/qlogic/qed/qed_iscsi.c | 32 +-
drivers/net/ethernet/qlogic/qed/qed_main.c | 50 +-
drivers/net/ethernet/qlogic/qed/qed_mcp.c | 819 +++++++++++++++++++++++---
drivers/net/ethernet/qlogic/qed/qed_mcp.h | 179 +++++-
10 files changed, 1480 insertions(+), 405 deletions(-)
--
1.9.3
^ permalink raw reply [flat|nested] 9+ messages in thread
* [PATCH net-next 1/7] qed: Correct HW stop flow
2017-03-28 12:12 [PATCH net-next 0/7] qed: load/unload mfw series Yuval Mintz
@ 2017-03-28 12:12 ` Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 2/7] qed: hw_init() to receive parameter-struct Yuval Mintz
` (6 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: Yuval Mintz @ 2017-03-28 12:12 UTC (permalink / raw)
To: davem, netdev; +Cc: Tomer Tayar, Yuval Mintz
From: Tomer Tayar <Tomer.Tayar@cavium.com>
Management firmware is used as arbiter between different PFs
which are loading/unloading, but in order to use the synchronization
it offers the contending configurations need to be applied either
between their LOAD_REQ <-> LOAD_DONE or UNLOAD_REQ <-> UNLOAD_DONE
management firmware commands.
Existing HW stop flow utilizes 2 different functions: qed_hw_stop() and
qed_hw_reset() which don't abide this requirement; Most of the closure
is doing outside the scope of the unload request.
This patch removes qed_hw_reset() and places the relevant stop
functionality underneath the management firmware protection.
Signed-off-by: Tomer Tayar <Tomer.Tayar@cavium.com>
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
---
drivers/net/ethernet/qlogic/qed/qed.h | 1 +
drivers/net/ethernet/qlogic/qed/qed_dev.c | 146 ++++++++++----------------
drivers/net/ethernet/qlogic/qed/qed_dev_api.h | 8 --
drivers/net/ethernet/qlogic/qed/qed_main.c | 29 ++---
drivers/net/ethernet/qlogic/qed/qed_mcp.c | 73 +++++++++----
drivers/net/ethernet/qlogic/qed/qed_mcp.h | 20 ++++
6 files changed, 142 insertions(+), 135 deletions(-)
diff --git a/drivers/net/ethernet/qlogic/qed/qed.h b/drivers/net/ethernet/qlogic/qed/qed.h
index ca30a27..90a1988 100644
--- a/drivers/net/ethernet/qlogic/qed/qed.h
+++ b/drivers/net/ethernet/qlogic/qed/qed.h
@@ -732,5 +732,6 @@ void qed_get_protocol_stats(struct qed_dev *cdev,
enum qed_mcp_protocol_type type,
union qed_mcp_protocol_stats *stats);
int qed_slowpath_irq_req(struct qed_hwfn *hwfn);
+void qed_slowpath_irq_sync(struct qed_hwfn *p_hwfn);
#endif /* _QED_H */
diff --git a/drivers/net/ethernet/qlogic/qed/qed_dev.c b/drivers/net/ethernet/qlogic/qed/qed_dev.c
index 11e45f0..cf95da5 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_dev.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_dev.c
@@ -1303,27 +1303,53 @@ void qed_hw_timers_stop_all(struct qed_dev *cdev)
int qed_hw_stop(struct qed_dev *cdev)
{
- int rc = 0, t_rc;
+ struct qed_hwfn *p_hwfn;
+ struct qed_ptt *p_ptt;
+ int rc, rc2 = 0;
int j;
for_each_hwfn(cdev, j) {
- struct qed_hwfn *p_hwfn = &cdev->hwfns[j];
- struct qed_ptt *p_ptt = p_hwfn->p_main_ptt;
+ p_hwfn = &cdev->hwfns[j];
+ p_ptt = p_hwfn->p_main_ptt;
DP_VERBOSE(p_hwfn, NETIF_MSG_IFDOWN, "Stopping hw/fw\n");
if (IS_VF(cdev)) {
qed_vf_pf_int_cleanup(p_hwfn);
+ rc = qed_vf_pf_reset(p_hwfn);
+ if (rc) {
+ DP_NOTICE(p_hwfn,
+ "qed_vf_pf_reset failed. rc = %d.\n",
+ rc);
+ rc2 = -EINVAL;
+ }
continue;
}
/* mark the hw as uninitialized... */
p_hwfn->hw_init_done = false;
+ /* Send unload command to MCP */
+ rc = qed_mcp_unload_req(p_hwfn, p_ptt);
+ if (rc) {
+ DP_NOTICE(p_hwfn,
+ "Failed sending a UNLOAD_REQ command. rc = %d.\n",
+ rc);
+ rc2 = -EINVAL;
+ }
+
+ qed_slowpath_irq_sync(p_hwfn);
+
+ /* After this point no MFW attentions are expected, e.g. prevent
+ * race between pf stop and dcbx pf update.
+ */
rc = qed_sp_pf_stop(p_hwfn);
- if (rc)
+ if (rc) {
DP_NOTICE(p_hwfn,
- "Failed to close PF against FW. Continue to stop HW to prevent illegal host access by the device\n");
+ "Failed to close PF against FW [rc = %d]. Continue to stop HW to prevent illegal host access by the device.\n",
+ rc);
+ rc2 = -EINVAL;
+ }
qed_wr(p_hwfn, p_ptt,
NIG_REG_RX_LLH_BRB_GATE_DNTFWD_PERPF, 0x1);
@@ -1346,20 +1372,37 @@ int qed_hw_stop(struct qed_dev *cdev)
/* Need to wait 1ms to guarantee SBs are cleared */
usleep_range(1000, 2000);
+
+ /* Disable PF in HW blocks */
+ qed_wr(p_hwfn, p_ptt, DORQ_REG_PF_DB_ENABLE, 0);
+ qed_wr(p_hwfn, p_ptt, QM_REG_PF_EN, 0);
+
+ qed_mcp_unload_done(p_hwfn, p_ptt);
+ if (rc) {
+ DP_NOTICE(p_hwfn,
+ "Failed sending a UNLOAD_DONE command. rc = %d.\n",
+ rc);
+ rc2 = -EINVAL;
+ }
}
if (IS_PF(cdev)) {
+ p_hwfn = QED_LEADING_HWFN(cdev);
+ p_ptt = QED_LEADING_HWFN(cdev)->p_main_ptt;
+
/* Disable DMAE in PXP - in CMT, this should only be done for
* first hw-function, and only after all transactions have
* stopped for all active hw-functions.
*/
- t_rc = qed_change_pci_hwfn(&cdev->hwfns[0],
- cdev->hwfns[0].p_main_ptt, false);
- if (t_rc != 0)
- rc = t_rc;
+ rc = qed_change_pci_hwfn(p_hwfn, p_ptt, false);
+ if (rc) {
+ DP_NOTICE(p_hwfn,
+ "qed_change_pci_hwfn failed. rc = %d.\n", rc);
+ rc2 = -EINVAL;
+ }
}
- return rc;
+ return rc2;
}
void qed_hw_stop_fastpath(struct qed_dev *cdev)
@@ -1404,89 +1447,6 @@ void qed_hw_start_fastpath(struct qed_hwfn *p_hwfn)
NIG_REG_RX_LLH_BRB_GATE_DNTFWD_PERPF, 0x0);
}
-static int qed_reg_assert(struct qed_hwfn *p_hwfn,
- struct qed_ptt *p_ptt, u32 reg, bool expected)
-{
- u32 assert_val = qed_rd(p_hwfn, p_ptt, reg);
-
- if (assert_val != expected) {
- DP_NOTICE(p_hwfn, "Value at address 0x%08x != 0x%08x\n",
- reg, expected);
- return -EINVAL;
- }
-
- return 0;
-}
-
-int qed_hw_reset(struct qed_dev *cdev)
-{
- int rc = 0;
- u32 unload_resp, unload_param;
- u32 wol_param;
- int i;
-
- switch (cdev->wol_config) {
- case QED_OV_WOL_DISABLED:
- wol_param = DRV_MB_PARAM_UNLOAD_WOL_DISABLED;
- break;
- case QED_OV_WOL_ENABLED:
- wol_param = DRV_MB_PARAM_UNLOAD_WOL_ENABLED;
- break;
- default:
- DP_NOTICE(cdev,
- "Unknown WoL configuration %02x\n", cdev->wol_config);
- /* Fallthrough */
- case QED_OV_WOL_DEFAULT:
- wol_param = DRV_MB_PARAM_UNLOAD_WOL_MCP;
- }
-
- for_each_hwfn(cdev, i) {
- struct qed_hwfn *p_hwfn = &cdev->hwfns[i];
-
- if (IS_VF(cdev)) {
- rc = qed_vf_pf_reset(p_hwfn);
- if (rc)
- return rc;
- continue;
- }
-
- DP_VERBOSE(p_hwfn, NETIF_MSG_IFDOWN, "Resetting hw/fw\n");
-
- /* Check for incorrect states */
- qed_reg_assert(p_hwfn, p_hwfn->p_main_ptt,
- QM_REG_USG_CNT_PF_TX, 0);
- qed_reg_assert(p_hwfn, p_hwfn->p_main_ptt,
- QM_REG_USG_CNT_PF_OTHER, 0);
-
- /* Disable PF in HW blocks */
- qed_wr(p_hwfn, p_hwfn->p_main_ptt, DORQ_REG_PF_DB_ENABLE, 0);
- qed_wr(p_hwfn, p_hwfn->p_main_ptt, QM_REG_PF_EN, 0);
- qed_wr(p_hwfn, p_hwfn->p_main_ptt,
- TCFC_REG_STRONG_ENABLE_PF, 0);
- qed_wr(p_hwfn, p_hwfn->p_main_ptt,
- CCFC_REG_STRONG_ENABLE_PF, 0);
-
- /* Send unload command to MCP */
- rc = qed_mcp_cmd(p_hwfn, p_hwfn->p_main_ptt,
- DRV_MSG_CODE_UNLOAD_REQ, wol_param,
- &unload_resp, &unload_param);
- if (rc) {
- DP_NOTICE(p_hwfn, "qed_hw_reset: UNLOAD_REQ failed\n");
- unload_resp = FW_MSG_CODE_DRV_UNLOAD_ENGINE;
- }
-
- rc = qed_mcp_cmd(p_hwfn, p_hwfn->p_main_ptt,
- DRV_MSG_CODE_UNLOAD_DONE,
- 0, &unload_resp, &unload_param);
- if (rc) {
- DP_NOTICE(p_hwfn, "qed_hw_reset: UNLOAD_DONE failed\n");
- return rc;
- }
- }
-
- return rc;
-}
-
/* Free hwfn memory and resources acquired in hw_hwfn_prepare */
static void qed_hw_hwfn_free(struct qed_hwfn *p_hwfn)
{
diff --git a/drivers/net/ethernet/qlogic/qed/qed_dev_api.h b/drivers/net/ethernet/qlogic/qed/qed_dev_api.h
index 6812003..e96eccf 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_dev_api.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_dev_api.h
@@ -140,14 +140,6 @@ int qed_hw_init(struct qed_dev *cdev,
*/
void qed_hw_start_fastpath(struct qed_hwfn *p_hwfn);
-/**
- * @brief qed_hw_reset -
- *
- * @param cdev
- *
- * @return int
- */
-int qed_hw_reset(struct qed_dev *cdev);
/**
* @brief qed_hw_prepare -
diff --git a/drivers/net/ethernet/qlogic/qed/qed_main.c b/drivers/net/ethernet/qlogic/qed/qed_main.c
index 766c6f3..f2ae90a 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_main.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_main.c
@@ -589,6 +589,19 @@ int qed_slowpath_irq_req(struct qed_hwfn *hwfn)
return rc;
}
+void qed_slowpath_irq_sync(struct qed_hwfn *p_hwfn)
+{
+ struct qed_dev *cdev = p_hwfn->cdev;
+ u8 id = p_hwfn->my_id;
+ u32 int_mode;
+
+ int_mode = cdev->int_params.out.int_mode;
+ if (int_mode == QED_INT_MODE_MSIX)
+ synchronize_irq(cdev->int_params.msix_table[id].vector);
+ else
+ synchronize_irq(cdev->pdev->irq);
+}
+
static void qed_slowpath_irq_free(struct qed_dev *cdev)
{
int i;
@@ -631,19 +644,6 @@ static int qed_nic_stop(struct qed_dev *cdev)
return rc;
}
-static int qed_nic_reset(struct qed_dev *cdev)
-{
- int rc;
-
- rc = qed_hw_reset(cdev);
- if (rc)
- return rc;
-
- qed_resc_free(cdev);
-
- return 0;
-}
-
static int qed_nic_setup(struct qed_dev *cdev)
{
int rc, i;
@@ -1043,7 +1043,8 @@ static int qed_slowpath_stop(struct qed_dev *cdev)
}
qed_disable_msix(cdev);
- qed_nic_reset(cdev);
+
+ qed_resc_free(cdev);
qed_iov_wq_stop(cdev, true);
diff --git a/drivers/net/ethernet/qlogic/qed/qed_mcp.c b/drivers/net/ethernet/qlogic/qed/qed_mcp.c
index ccea0ea..f2e5ab9 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_mcp.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_mcp.c
@@ -550,32 +550,12 @@ int qed_mcp_cmd(struct qed_hwfn *p_hwfn,
u32 *o_mcp_param)
{
struct qed_mcp_mb_params mb_params;
- struct mcp_mac wol_mac;
int rc;
memset(&mb_params, 0, sizeof(mb_params));
mb_params.cmd = cmd;
mb_params.param = param;
- /* In case of UNLOAD_DONE, set the primary MAC */
- if ((cmd == DRV_MSG_CODE_UNLOAD_DONE) &&
- (p_hwfn->cdev->wol_config == QED_OV_WOL_ENABLED)) {
- u8 *p_mac = p_hwfn->cdev->wol_mac;
-
- memset(&wol_mac, 0, sizeof(wol_mac));
- wol_mac.mac_upper = p_mac[0] << 8 | p_mac[1];
- wol_mac.mac_lower = p_mac[2] << 24 | p_mac[3] << 16 |
- p_mac[4] << 8 | p_mac[5];
-
- DP_VERBOSE(p_hwfn,
- (QED_MSG_SP | NETIF_MSG_IFDOWN),
- "Setting WoL MAC: %pM --> [%08x,%08x]\n",
- p_mac, wol_mac.mac_upper, wol_mac.mac_lower);
-
- mb_params.p_data_src = &wol_mac;
- mb_params.data_src_size = sizeof(wol_mac);
- }
-
rc = qed_mcp_cmd_and_union(p_hwfn, p_ptt, &mb_params);
if (rc)
return rc;
@@ -663,6 +643,59 @@ int qed_mcp_load_req(struct qed_hwfn *p_hwfn,
return 0;
}
+int qed_mcp_unload_req(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt)
+{
+ u32 wol_param, mcp_resp, mcp_param;
+
+ switch (p_hwfn->cdev->wol_config) {
+ case QED_OV_WOL_DISABLED:
+ wol_param = DRV_MB_PARAM_UNLOAD_WOL_DISABLED;
+ break;
+ case QED_OV_WOL_ENABLED:
+ wol_param = DRV_MB_PARAM_UNLOAD_WOL_ENABLED;
+ break;
+ default:
+ DP_NOTICE(p_hwfn,
+ "Unknown WoL configuration %02x\n",
+ p_hwfn->cdev->wol_config);
+ /* Fallthrough */
+ case QED_OV_WOL_DEFAULT:
+ wol_param = DRV_MB_PARAM_UNLOAD_WOL_MCP;
+ }
+
+ return qed_mcp_cmd(p_hwfn, p_ptt, DRV_MSG_CODE_UNLOAD_REQ, wol_param,
+ &mcp_resp, &mcp_param);
+}
+
+int qed_mcp_unload_done(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt)
+{
+ struct qed_mcp_mb_params mb_params;
+ struct mcp_mac wol_mac;
+
+ memset(&mb_params, 0, sizeof(mb_params));
+ mb_params.cmd = DRV_MSG_CODE_UNLOAD_DONE;
+
+ /* Set the primary MAC if WoL is enabled */
+ if (p_hwfn->cdev->wol_config == QED_OV_WOL_ENABLED) {
+ u8 *p_mac = p_hwfn->cdev->wol_mac;
+
+ memset(&wol_mac, 0, sizeof(wol_mac));
+ wol_mac.mac_upper = p_mac[0] << 8 | p_mac[1];
+ wol_mac.mac_lower = p_mac[2] << 24 | p_mac[3] << 16 |
+ p_mac[4] << 8 | p_mac[5];
+
+ DP_VERBOSE(p_hwfn,
+ (QED_MSG_SP | NETIF_MSG_IFDOWN),
+ "Setting WoL MAC: %pM --> [%08x,%08x]\n",
+ p_mac, wol_mac.mac_upper, wol_mac.mac_lower);
+
+ mb_params.p_data_src = &wol_mac;
+ mb_params.data_src_size = sizeof(wol_mac);
+ }
+
+ return qed_mcp_cmd_and_union(p_hwfn, p_ptt, &mb_params);
+}
+
static void qed_mcp_handle_vf_flr(struct qed_hwfn *p_hwfn,
struct qed_ptt *p_ptt)
{
diff --git a/drivers/net/ethernet/qlogic/qed/qed_mcp.h b/drivers/net/ethernet/qlogic/qed/qed_mcp.h
index f63693d..7d067d0 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_mcp.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_mcp.h
@@ -593,6 +593,26 @@ int qed_mcp_load_req(struct qed_hwfn *p_hwfn,
u32 *p_load_code);
/**
+ * @brief Sends a UNLOAD_REQ message to the MFW
+ *
+ * @param p_hwfn
+ * @param p_ptt
+ *
+ * @return int - 0 - Operation was successful.
+ */
+int qed_mcp_unload_req(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt);
+
+/**
+ * @brief Sends a UNLOAD_DONE message to the MFW
+ *
+ * @param p_hwfn
+ * @param p_ptt
+ *
+ * @return int - 0 - Operation was successful.
+ */
+int qed_mcp_unload_done(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt);
+
+/**
* @brief Read the MFW mailbox into Current buffer.
*
* @param p_hwfn
--
1.9.3
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH net-next 2/7] qed: hw_init() to receive parameter-struct
2017-03-28 12:12 [PATCH net-next 0/7] qed: load/unload mfw series Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 1/7] qed: Correct HW stop flow Yuval Mintz
@ 2017-03-28 12:12 ` Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 3/7] qed: Move to new load request scheme Yuval Mintz
` (5 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: Yuval Mintz @ 2017-03-28 12:12 UTC (permalink / raw)
To: davem, netdev; +Cc: Yuval Mintz
We'll soon need additional information, so start by changing
the infrastructure to receive the initializing variables
via a parameter struct.
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
---
drivers/net/ethernet/qlogic/qed/qed_dev.c | 21 +++++++++---------
drivers/net/ethernet/qlogic/qed/qed_dev_api.h | 31 ++++++++++++++++-----------
drivers/net/ethernet/qlogic/qed/qed_main.c | 12 ++++++++---
3 files changed, 37 insertions(+), 27 deletions(-)
diff --git a/drivers/net/ethernet/qlogic/qed/qed_dev.c b/drivers/net/ethernet/qlogic/qed/qed_dev.c
index cf95da5..1092998 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_dev.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_dev.c
@@ -1106,25 +1106,20 @@ static void qed_reset_mb_shadow(struct qed_hwfn *p_hwfn,
p_hwfn->mcp_info->mfw_mb_cur, p_hwfn->mcp_info->mfw_mb_length);
}
-int qed_hw_init(struct qed_dev *cdev,
- struct qed_tunn_start_params *p_tunn,
- bool b_hw_start,
- enum qed_int_mode int_mode,
- bool allow_npar_tx_switch,
- const u8 *bin_fw_data)
+int qed_hw_init(struct qed_dev *cdev, struct qed_hw_init_params *p_params)
{
u32 load_code, param, drv_mb_param;
bool b_default_mtu = true;
struct qed_hwfn *p_hwfn;
int rc = 0, mfw_rc, i;
- if ((int_mode == QED_INT_MODE_MSI) && (cdev->num_hwfns > 1)) {
+ if ((p_params->int_mode == QED_INT_MODE_MSI) && (cdev->num_hwfns > 1)) {
DP_NOTICE(cdev, "MSI mode is not supported for CMT devices\n");
return -EINVAL;
}
if (IS_PF(cdev)) {
- rc = qed_init_fw_data(cdev, bin_fw_data);
+ rc = qed_init_fw_data(cdev, p_params->bin_fw_data);
if (rc)
return rc;
}
@@ -1181,11 +1176,15 @@ int qed_hw_init(struct qed_dev *cdev,
/* Fall into */
case FW_MSG_CODE_DRV_LOAD_FUNCTION:
rc = qed_hw_init_pf(p_hwfn, p_hwfn->p_main_ptt,
- p_tunn, p_hwfn->hw_info.hw_mode,
- b_hw_start, int_mode,
- allow_npar_tx_switch);
+ p_params->p_tunn,
+ p_hwfn->hw_info.hw_mode,
+ p_params->b_hw_start,
+ p_params->int_mode,
+ p_params->allow_npar_tx_switch);
break;
default:
+ DP_NOTICE(p_hwfn,
+ "Unexpected load code [0x%08x]", load_code);
rc = -EINVAL;
break;
}
diff --git a/drivers/net/ethernet/qlogic/qed/qed_dev_api.h b/drivers/net/ethernet/qlogic/qed/qed_dev_api.h
index e96eccf..8f45f46 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_dev_api.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_dev_api.h
@@ -82,26 +82,31 @@ void qed_init_dp(struct qed_dev *cdev,
*/
void qed_resc_setup(struct qed_dev *cdev);
+struct qed_hw_init_params {
+ /* Tunneling parameters */
+ struct qed_tunn_start_params *p_tunn;
+
+ bool b_hw_start;
+
+ /* Interrupt mode [msix, inta, etc.] to use */
+ enum qed_int_mode int_mode;
+
+ /* NPAR tx switching to be used for vports for tx-switching */
+ bool allow_npar_tx_switch;
+
+ /* Binary fw data pointer in binary fw file */
+ const u8 *bin_fw_data;
+};
+
/**
* @brief qed_hw_init -
*
* @param cdev
- * @param p_tunn
- * @param b_hw_start
- * @param int_mode - interrupt mode [msix, inta, etc.] to use.
- * @param allow_npar_tx_switch - npar tx switching to be used
- * for vports configured for tx-switching.
- * @param bin_fw_data - binary fw data pointer in binary fw file.
- * Pass NULL if not using binary fw file.
+ * @param p_params
*
* @return int
*/
-int qed_hw_init(struct qed_dev *cdev,
- struct qed_tunn_start_params *p_tunn,
- bool b_hw_start,
- enum qed_int_mode int_mode,
- bool allow_npar_tx_switch,
- const u8 *bin_fw_data);
+int qed_hw_init(struct qed_dev *cdev, struct qed_hw_init_params *p_params);
/**
* @brief qed_hw_timers_stop_all - stop the timers HW block
diff --git a/drivers/net/ethernet/qlogic/qed/qed_main.c b/drivers/net/ethernet/qlogic/qed/qed_main.c
index f2ae90a..ae0ab3b 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_main.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_main.c
@@ -901,6 +901,7 @@ static void qed_update_pf_params(struct qed_dev *cdev,
static int qed_slowpath_start(struct qed_dev *cdev,
struct qed_slowpath_params *params)
{
+ struct qed_hw_init_params hw_init_params;
struct qed_tunn_start_params tunn_info;
struct qed_mcp_drv_version drv_version;
const u8 *data = NULL;
@@ -966,9 +967,14 @@ static int qed_slowpath_start(struct qed_dev *cdev,
tunn_info.tunn_clss_ipgre = QED_TUNN_CLSS_MAC_VLAN;
/* Start the slowpath */
- rc = qed_hw_init(cdev, &tunn_info, true,
- cdev->int_params.out.int_mode,
- true, data);
+ memset(&hw_init_params, 0, sizeof(hw_init_params));
+ hw_init_params.p_tunn = &tunn_info;
+ hw_init_params.b_hw_start = true;
+ hw_init_params.int_mode = cdev->int_params.out.int_mode;
+ hw_init_params.allow_npar_tx_switch = true;
+ hw_init_params.bin_fw_data = data;
+
+ rc = qed_hw_init(cdev, &hw_init_params);
if (rc)
goto err2;
--
1.9.3
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH net-next 3/7] qed: Move to new load request scheme
2017-03-28 12:12 [PATCH net-next 0/7] qed: load/unload mfw series Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 1/7] qed: Correct HW stop flow Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 2/7] qed: hw_init() to receive parameter-struct Yuval Mintz
@ 2017-03-28 12:12 ` Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 4/7] qed: Send pf-flr as part of initialization Yuval Mintz
` (4 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: Yuval Mintz @ 2017-03-28 12:12 UTC (permalink / raw)
To: davem, netdev; +Cc: Tomer Tayar, Yuval Mintz
From: Tomer Tayar <Tomer.Tayar@cavium.com>
Management firmware is used as an arbiter between the various PFs
in regard to loading - it causes the various PFs to load/unload
sequentially and informs each of its appropriate rule in the init.
But the existing flow is too weak to handle some scenarios where
PFs aren't properly cleaned prior to loading.
The significant scenarios falling under this criteria:
a. Preboot drivers in some environment can't properly unload.
b. Unexpected driver replacement [kdump, PDA].
Modern management firmware supports a more intricate loading flow,
where the driver has the ability to overcome previous limitations.
This moves qed into using this newer scheme.
Notice new scheme is backward compatible, so new drivers would
still be able to load properly on top of older management firmwares
and vice versa.
Signed-off-by: Tomer Tayar <Tomer.Tayar@cavium.com>
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
---
drivers/net/ethernet/qlogic/qed/qed.h | 29 ++-
drivers/net/ethernet/qlogic/qed/qed_dcbx.h | 3 -
drivers/net/ethernet/qlogic/qed/qed_dev.c | 35 ++-
drivers/net/ethernet/qlogic/qed/qed_dev_api.h | 32 +++
drivers/net/ethernet/qlogic/qed/qed_hsi.h | 50 +++-
drivers/net/ethernet/qlogic/qed/qed_main.c | 9 +
drivers/net/ethernet/qlogic/qed/qed_mcp.c | 356 ++++++++++++++++++++++++--
drivers/net/ethernet/qlogic/qed/qed_mcp.h | 41 +--
8 files changed, 497 insertions(+), 58 deletions(-)
diff --git a/drivers/net/ethernet/qlogic/qed/qed.h b/drivers/net/ethernet/qlogic/qed/qed.h
index 90a1988..187a84b 100644
--- a/drivers/net/ethernet/qlogic/qed/qed.h
+++ b/drivers/net/ethernet/qlogic/qed/qed.h
@@ -51,7 +51,19 @@
#include "qed_hsi.h"
extern const struct qed_common_ops qed_common_ops_pass;
-#define DRV_MODULE_VERSION "8.10.10.21"
+
+#define QED_MAJOR_VERSION 8
+#define QED_MINOR_VERSION 10
+#define QED_REVISION_VERSION 10
+#define QED_ENGINEERING_VERSION 21
+
+#define QED_VERSION \
+ ((QED_MAJOR_VERSION << 24) | (QED_MINOR_VERSION << 16) | \
+ (QED_REVISION_VERSION << 8) | QED_ENGINEERING_VERSION)
+
+#define STORM_FW_VERSION \
+ ((FW_MAJOR_VERSION << 24) | (FW_MINOR_VERSION << 16) | \
+ (FW_REVISION_VERSION << 8) | FW_ENGINEERING_VERSION)
#define MAX_HWFNS_PER_DEVICE (4)
#define NAME_SIZE 16
@@ -76,6 +88,15 @@ enum qed_coalescing_mode {
enum qed_mcp_protocol_type;
/* helpers */
+#define QED_MFW_GET_FIELD(name, field) \
+ (((name) & (field ## _MASK)) >> (field ## _SHIFT))
+
+#define QED_MFW_SET_FIELD(name, field, value) \
+ do { \
+ (name) &= ~((field ## _MASK) << (field ## _SHIFT)); \
+ (name) |= (((value) << (field ## _SHIFT)) & (field ## _MASK));\
+ } while (0)
+
static inline u32 qed_db_addr(u32 cid, u32 DEMS)
{
u32 db_addr = FIELD_VALUE(DB_LEGACY_ADDR_DEMS, DEMS) |
@@ -355,6 +376,12 @@ struct qed_fw_data {
u32 init_ops_size;
};
+#define DRV_MODULE_VERSION \
+ __stringify(QED_MAJOR_VERSION) "." \
+ __stringify(QED_MINOR_VERSION) "." \
+ __stringify(QED_REVISION_VERSION) "." \
+ __stringify(QED_ENGINEERING_VERSION)
+
struct qed_simd_fp_handler {
void *token;
void (*func)(void *);
diff --git a/drivers/net/ethernet/qlogic/qed/qed_dcbx.h b/drivers/net/ethernet/qlogic/qed/qed_dcbx.h
index 0fabe97..2eb988f 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_dcbx.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_dcbx.h
@@ -85,9 +85,6 @@ struct qed_dcbx_app_metadata {
enum qed_pci_personality personality;
};
-#define QED_MFW_GET_FIELD(name, field) \
- (((name) & (field ## _MASK)) >> (field ## _SHIFT))
-
struct qed_dcbx_info {
struct lldp_status_params_s lldp_remote[LLDP_MAX_LLDP_AGENTS];
struct lldp_config_params_s lldp_local[LLDP_MAX_LLDP_AGENTS];
diff --git a/drivers/net/ethernet/qlogic/qed/qed_dev.c b/drivers/net/ethernet/qlogic/qed/qed_dev.c
index 1092998..759b819 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_dev.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_dev.c
@@ -1106,8 +1106,22 @@ static void qed_reset_mb_shadow(struct qed_hwfn *p_hwfn,
p_hwfn->mcp_info->mfw_mb_cur, p_hwfn->mcp_info->mfw_mb_length);
}
+static void
+qed_fill_load_req_params(struct qed_load_req_params *p_load_req,
+ struct qed_drv_load_params *p_drv_load)
+{
+ memset(p_load_req, 0, sizeof(*p_load_req));
+
+ p_load_req->drv_role = p_drv_load->is_crash_kernel ?
+ QED_DRV_ROLE_KDUMP : QED_DRV_ROLE_OS;
+ p_load_req->timeout_val = p_drv_load->mfw_timeout_val;
+ p_load_req->avoid_eng_reset = p_drv_load->avoid_eng_reset;
+ p_load_req->override_force_load = p_drv_load->override_force_load;
+}
+
int qed_hw_init(struct qed_dev *cdev, struct qed_hw_init_params *p_params)
{
+ struct qed_load_req_params load_req_params;
u32 load_code, param, drv_mb_param;
bool b_default_mtu = true;
struct qed_hwfn *p_hwfn;
@@ -1145,17 +1159,21 @@ int qed_hw_init(struct qed_dev *cdev, struct qed_hw_init_params *p_params)
if (rc)
return rc;
- rc = qed_mcp_load_req(p_hwfn, p_hwfn->p_main_ptt, &load_code);
+ qed_fill_load_req_params(&load_req_params,
+ p_params->p_drv_load_params);
+ rc = qed_mcp_load_req(p_hwfn, p_hwfn->p_main_ptt,
+ &load_req_params);
if (rc) {
- DP_NOTICE(p_hwfn, "Failed sending LOAD_REQ command\n");
+ DP_NOTICE(p_hwfn, "Failed sending a LOAD_REQ command\n");
return rc;
}
- qed_reset_mb_shadow(p_hwfn, p_hwfn->p_main_ptt);
-
+ load_code = load_req_params.load_code;
DP_VERBOSE(p_hwfn, QED_MSG_SP,
- "Load request was sent. Resp:0x%x, Load code: 0x%x\n",
- rc, load_code);
+ "Load request was sent. Load code: 0x%x\n",
+ load_code);
+
+ qed_reset_mb_shadow(p_hwfn, p_hwfn->p_main_ptt);
p_hwfn->first_on_engine = (load_code ==
FW_MSG_CODE_DRV_LOAD_ENGINE);
@@ -1224,10 +1242,7 @@ int qed_hw_init(struct qed_dev *cdev, struct qed_hw_init_params *p_params)
if (IS_PF(cdev)) {
p_hwfn = QED_LEADING_HWFN(cdev);
- drv_mb_param = (FW_MAJOR_VERSION << 24) |
- (FW_MINOR_VERSION << 16) |
- (FW_REVISION_VERSION << 8) |
- (FW_ENGINEERING_VERSION);
+ drv_mb_param = STORM_FW_VERSION;
rc = qed_mcp_cmd(p_hwfn, p_hwfn->p_main_ptt,
DRV_MSG_CODE_OV_UPDATE_STORM_FW_VER,
drv_mb_param, &load_code, ¶m);
diff --git a/drivers/net/ethernet/qlogic/qed/qed_dev_api.h b/drivers/net/ethernet/qlogic/qed/qed_dev_api.h
index 8f45f46..cb97349 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_dev_api.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_dev_api.h
@@ -82,6 +82,35 @@ void qed_init_dp(struct qed_dev *cdev,
*/
void qed_resc_setup(struct qed_dev *cdev);
+enum qed_override_force_load {
+ QED_OVERRIDE_FORCE_LOAD_NONE,
+ QED_OVERRIDE_FORCE_LOAD_ALWAYS,
+ QED_OVERRIDE_FORCE_LOAD_NEVER,
+};
+
+struct qed_drv_load_params {
+ /* Indicates whether the driver is running over a crash kernel.
+ * As part of the load request, this will be used for providing the
+ * driver role to the MFW.
+ * In case of a crash kernel over PDA - this should be set to false.
+ */
+ bool is_crash_kernel;
+
+ /* The timeout value that the MFW should use when locking the engine for
+ * the driver load process.
+ * A value of '0' means the default value, and '255' means no timeout.
+ */
+ u8 mfw_timeout_val;
+#define QED_LOAD_REQ_LOCK_TO_DEFAULT 0
+#define QED_LOAD_REQ_LOCK_TO_NONE 255
+
+ /* Avoid engine reset when first PF loads on it */
+ bool avoid_eng_reset;
+
+ /* Allow overriding the default force load behavior */
+ enum qed_override_force_load override_force_load;
+};
+
struct qed_hw_init_params {
/* Tunneling parameters */
struct qed_tunn_start_params *p_tunn;
@@ -96,6 +125,9 @@ struct qed_hw_init_params {
/* Binary fw data pointer in binary fw file */
const u8 *bin_fw_data;
+
+ /* Driver load parameters */
+ struct qed_drv_load_params *p_drv_load_params;
};
/**
diff --git a/drivers/net/ethernet/qlogic/qed/qed_hsi.h b/drivers/net/ethernet/qlogic/qed/qed_hsi.h
index e9acdc9..bf2b201 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_hsi.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_hsi.h
@@ -9887,9 +9887,11 @@ struct public_func {
#define DRV_ID_PDA_COMP_VER_MASK 0x0000ffff
#define DRV_ID_PDA_COMP_VER_SHIFT 0
+#define LOAD_REQ_HSI_VERSION 2
#define DRV_ID_MCP_HSI_VER_MASK 0x00ff0000
#define DRV_ID_MCP_HSI_VER_SHIFT 16
-#define DRV_ID_MCP_HSI_VER_CURRENT (1 << DRV_ID_MCP_HSI_VER_SHIFT)
+#define DRV_ID_MCP_HSI_VER_CURRENT (LOAD_REQ_HSI_VERSION << \
+ DRV_ID_MCP_HSI_VER_SHIFT)
#define DRV_ID_DRV_TYPE_MASK 0x7f000000
#define DRV_ID_DRV_TYPE_SHIFT 24
@@ -10001,6 +10003,46 @@ struct resource_info {
#define RESOURCE_ELEMENT_STRICT (1 << 0)
};
+#define DRV_ROLE_NONE 0
+#define DRV_ROLE_PREBOOT 1
+#define DRV_ROLE_OS 2
+#define DRV_ROLE_KDUMP 3
+
+struct load_req_stc {
+ u32 drv_ver_0;
+ u32 drv_ver_1;
+ u32 fw_ver;
+ u32 misc0;
+#define LOAD_REQ_ROLE_MASK 0x000000FF
+#define LOAD_REQ_ROLE_SHIFT 0
+#define LOAD_REQ_LOCK_TO_MASK 0x0000FF00
+#define LOAD_REQ_LOCK_TO_SHIFT 8
+#define LOAD_REQ_LOCK_TO_DEFAULT 0
+#define LOAD_REQ_LOCK_TO_NONE 255
+#define LOAD_REQ_FORCE_MASK 0x000F0000
+#define LOAD_REQ_FORCE_SHIFT 16
+#define LOAD_REQ_FORCE_NONE 0
+#define LOAD_REQ_FORCE_PF 1
+#define LOAD_REQ_FORCE_ALL 2
+#define LOAD_REQ_FLAGS0_MASK 0x00F00000
+#define LOAD_REQ_FLAGS0_SHIFT 20
+#define LOAD_REQ_FLAGS0_AVOID_RESET (0x1 << 0)
+};
+
+struct load_rsp_stc {
+ u32 drv_ver_0;
+ u32 drv_ver_1;
+ u32 fw_ver;
+ u32 misc0;
+#define LOAD_RSP_ROLE_MASK 0x000000FF
+#define LOAD_RSP_ROLE_SHIFT 0
+#define LOAD_RSP_HSI_MASK 0x0000FF00
+#define LOAD_RSP_HSI_SHIFT 8
+#define LOAD_RSP_FLAGS0_MASK 0x000F0000
+#define LOAD_RSP_FLAGS0_SHIFT 16
+#define LOAD_RSP_FLAGS0_DRV_EXISTS (0x1 << 0)
+};
+
union drv_union_data {
u32 ver_str[MCP_DRV_VER_STR_SIZE_DWORD];
struct mcp_mac wol_mac;
@@ -10032,6 +10074,7 @@ struct public_drv_mb {
#define DRV_MSG_CODE_LOAD_REQ 0x10000000
#define DRV_MSG_CODE_LOAD_DONE 0x11000000
#define DRV_MSG_CODE_INIT_HW 0x12000000
+#define DRV_MSG_CODE_CANCEL_LOAD_REQ 0x13000000
#define DRV_MSG_CODE_UNLOAD_REQ 0x20000000
#define DRV_MSG_CODE_UNLOAD_DONE 0x21000000
#define DRV_MSG_CODE_INIT_PHY 0x22000000
@@ -10167,8 +10210,11 @@ struct public_drv_mb {
#define FW_MSG_CODE_DRV_LOAD_PORT 0x10110000
#define FW_MSG_CODE_DRV_LOAD_FUNCTION 0x10120000
#define FW_MSG_CODE_DRV_LOAD_REFUSED_PDA 0x10200000
-#define FW_MSG_CODE_DRV_LOAD_REFUSED_HSI 0x10210000
+#define FW_MSG_CODE_DRV_LOAD_REFUSED_HSI_1 0x10210000
#define FW_MSG_CODE_DRV_LOAD_REFUSED_DIAG 0x10220000
+#define FW_MSG_CODE_DRV_LOAD_REFUSED_HSI 0x10230000
+#define FW_MSG_CODE_DRV_LOAD_REFUSED_REQUIRES_FORCE 0x10300000
+#define FW_MSG_CODE_DRV_LOAD_REFUSED_REJECT 0x10310000
#define FW_MSG_CODE_DRV_LOAD_DONE 0x11100000
#define FW_MSG_CODE_DRV_UNLOAD_ENGINE 0x20110000
#define FW_MSG_CODE_DRV_UNLOAD_PORT 0x20120000
diff --git a/drivers/net/ethernet/qlogic/qed/qed_main.c b/drivers/net/ethernet/qlogic/qed/qed_main.c
index ae0ab3b..d4edb99 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_main.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_main.c
@@ -45,6 +45,7 @@
#include <linux/ethtool.h>
#include <linux/etherdevice.h>
#include <linux/vmalloc.h>
+#include <linux/crash_dump.h>
#include <linux/qed/qed_if.h>
#include <linux/qed/qed_ll2_if.h>
@@ -901,6 +902,7 @@ static void qed_update_pf_params(struct qed_dev *cdev,
static int qed_slowpath_start(struct qed_dev *cdev,
struct qed_slowpath_params *params)
{
+ struct qed_drv_load_params drv_load_params;
struct qed_hw_init_params hw_init_params;
struct qed_tunn_start_params tunn_info;
struct qed_mcp_drv_version drv_version;
@@ -974,6 +976,13 @@ static int qed_slowpath_start(struct qed_dev *cdev,
hw_init_params.allow_npar_tx_switch = true;
hw_init_params.bin_fw_data = data;
+ memset(&drv_load_params, 0, sizeof(drv_load_params));
+ drv_load_params.is_crash_kernel = is_kdump_kernel();
+ drv_load_params.mfw_timeout_val = QED_LOAD_REQ_LOCK_TO_DEFAULT;
+ drv_load_params.avoid_eng_reset = false;
+ drv_load_params.override_force_load = QED_OVERRIDE_FORCE_LOAD_NONE;
+ hw_init_params.p_drv_load_params = &drv_load_params;
+
rc = qed_hw_init(cdev, &hw_init_params);
if (rc)
goto err2;
diff --git a/drivers/net/ethernet/qlogic/qed/qed_mcp.c b/drivers/net/ethernet/qlogic/qed/qed_mcp.c
index f2e5ab9..1a820b4 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_mcp.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_mcp.c
@@ -598,48 +598,352 @@ int qed_mcp_nvm_rd_cmd(struct qed_hwfn *p_hwfn,
return 0;
}
-int qed_mcp_load_req(struct qed_hwfn *p_hwfn,
- struct qed_ptt *p_ptt, u32 *p_load_code)
+static bool
+qed_mcp_can_force_load(u8 drv_role,
+ u8 exist_drv_role,
+ enum qed_override_force_load override_force_load)
+{
+ bool can_force_load = false;
+
+ switch (override_force_load) {
+ case QED_OVERRIDE_FORCE_LOAD_ALWAYS:
+ can_force_load = true;
+ break;
+ case QED_OVERRIDE_FORCE_LOAD_NEVER:
+ can_force_load = false;
+ break;
+ default:
+ can_force_load = (drv_role == DRV_ROLE_OS &&
+ exist_drv_role == DRV_ROLE_PREBOOT) ||
+ (drv_role == DRV_ROLE_KDUMP &&
+ exist_drv_role == DRV_ROLE_OS);
+ break;
+ }
+
+ return can_force_load;
+}
+
+static int qed_mcp_cancel_load_req(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt)
+{
+ u32 resp = 0, param = 0;
+ int rc;
+
+ rc = qed_mcp_cmd(p_hwfn, p_ptt, DRV_MSG_CODE_CANCEL_LOAD_REQ, 0,
+ &resp, ¶m);
+ if (rc)
+ DP_NOTICE(p_hwfn,
+ "Failed to send cancel load request, rc = %d\n", rc);
+
+ return rc;
+}
+
+#define CONFIG_QEDE_BITMAP_IDX BIT(0)
+#define CONFIG_QED_SRIOV_BITMAP_IDX BIT(1)
+#define CONFIG_QEDR_BITMAP_IDX BIT(2)
+#define CONFIG_QEDF_BITMAP_IDX BIT(4)
+#define CONFIG_QEDI_BITMAP_IDX BIT(5)
+#define CONFIG_QED_LL2_BITMAP_IDX BIT(6)
+
+static u32 qed_get_config_bitmap(void)
+{
+ u32 config_bitmap = 0x0;
+
+ if (IS_ENABLED(CONFIG_QEDE))
+ config_bitmap |= CONFIG_QEDE_BITMAP_IDX;
+
+ if (IS_ENABLED(CONFIG_QED_SRIOV))
+ config_bitmap |= CONFIG_QED_SRIOV_BITMAP_IDX;
+
+ if (IS_ENABLED(CONFIG_QED_RDMA))
+ config_bitmap |= CONFIG_QEDR_BITMAP_IDX;
+
+ if (IS_ENABLED(CONFIG_QED_FCOE))
+ config_bitmap |= CONFIG_QEDF_BITMAP_IDX;
+
+ if (IS_ENABLED(CONFIG_QED_ISCSI))
+ config_bitmap |= CONFIG_QEDI_BITMAP_IDX;
+
+ if (IS_ENABLED(CONFIG_QED_LL2))
+ config_bitmap |= CONFIG_QED_LL2_BITMAP_IDX;
+
+ return config_bitmap;
+}
+
+struct qed_load_req_in_params {
+ u8 hsi_ver;
+#define QED_LOAD_REQ_HSI_VER_DEFAULT 0
+#define QED_LOAD_REQ_HSI_VER_1 1
+ u32 drv_ver_0;
+ u32 drv_ver_1;
+ u32 fw_ver;
+ u8 drv_role;
+ u8 timeout_val;
+ u8 force_cmd;
+ bool avoid_eng_reset;
+};
+
+struct qed_load_req_out_params {
+ u32 load_code;
+ u32 exist_drv_ver_0;
+ u32 exist_drv_ver_1;
+ u32 exist_fw_ver;
+ u8 exist_drv_role;
+ u8 mfw_hsi_ver;
+ bool drv_exists;
+};
+
+static int
+__qed_mcp_load_req(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt,
+ struct qed_load_req_in_params *p_in_params,
+ struct qed_load_req_out_params *p_out_params)
{
- struct qed_dev *cdev = p_hwfn->cdev;
struct qed_mcp_mb_params mb_params;
- union drv_union_data union_data;
+ struct load_req_stc load_req;
+ struct load_rsp_stc load_rsp;
+ u32 hsi_ver;
int rc;
+ memset(&load_req, 0, sizeof(load_req));
+ load_req.drv_ver_0 = p_in_params->drv_ver_0;
+ load_req.drv_ver_1 = p_in_params->drv_ver_1;
+ load_req.fw_ver = p_in_params->fw_ver;
+ QED_MFW_SET_FIELD(load_req.misc0, LOAD_REQ_ROLE, p_in_params->drv_role);
+ QED_MFW_SET_FIELD(load_req.misc0, LOAD_REQ_LOCK_TO,
+ p_in_params->timeout_val);
+ QED_MFW_SET_FIELD(load_req.misc0, LOAD_REQ_FORCE,
+ p_in_params->force_cmd);
+ QED_MFW_SET_FIELD(load_req.misc0, LOAD_REQ_FLAGS0,
+ p_in_params->avoid_eng_reset);
+
+ hsi_ver = (p_in_params->hsi_ver == QED_LOAD_REQ_HSI_VER_DEFAULT) ?
+ DRV_ID_MCP_HSI_VER_CURRENT :
+ (p_in_params->hsi_ver << DRV_ID_MCP_HSI_VER_SHIFT);
+
memset(&mb_params, 0, sizeof(mb_params));
- /* Load Request */
mb_params.cmd = DRV_MSG_CODE_LOAD_REQ;
- mb_params.param = PDA_COMP | DRV_ID_MCP_HSI_VER_CURRENT |
- cdev->drv_type;
- memcpy(&union_data.ver_str, cdev->ver_str, MCP_DRV_VER_STR_SIZE);
- mb_params.p_data_src = &union_data;
- mb_params.data_src_size = sizeof(union_data.ver_str);
- rc = qed_mcp_cmd_and_union(p_hwfn, p_ptt, &mb_params);
+ mb_params.param = PDA_COMP | hsi_ver | p_hwfn->cdev->drv_type;
+ mb_params.p_data_src = &load_req;
+ mb_params.data_src_size = sizeof(load_req);
+ mb_params.p_data_dst = &load_rsp;
+ mb_params.data_dst_size = sizeof(load_rsp);
- /* if mcp fails to respond we must abort */
+ DP_VERBOSE(p_hwfn, QED_MSG_SP,
+ "Load Request: param 0x%08x [init_hw %d, drv_type %d, hsi_ver %d, pda 0x%04x]\n",
+ mb_params.param,
+ QED_MFW_GET_FIELD(mb_params.param, DRV_ID_DRV_INIT_HW),
+ QED_MFW_GET_FIELD(mb_params.param, DRV_ID_DRV_TYPE),
+ QED_MFW_GET_FIELD(mb_params.param, DRV_ID_MCP_HSI_VER),
+ QED_MFW_GET_FIELD(mb_params.param, DRV_ID_PDA_COMP_VER));
+
+ if (p_in_params->hsi_ver != QED_LOAD_REQ_HSI_VER_1) {
+ DP_VERBOSE(p_hwfn, QED_MSG_SP,
+ "Load Request: drv_ver 0x%08x_0x%08x, fw_ver 0x%08x, misc0 0x%08x [role %d, timeout %d, force %d, flags0 0x%x]\n",
+ load_req.drv_ver_0,
+ load_req.drv_ver_1,
+ load_req.fw_ver,
+ load_req.misc0,
+ QED_MFW_GET_FIELD(load_req.misc0, LOAD_REQ_ROLE),
+ QED_MFW_GET_FIELD(load_req.misc0,
+ LOAD_REQ_LOCK_TO),
+ QED_MFW_GET_FIELD(load_req.misc0, LOAD_REQ_FORCE),
+ QED_MFW_GET_FIELD(load_req.misc0, LOAD_REQ_FLAGS0));
+ }
+
+ rc = qed_mcp_cmd_and_union(p_hwfn, p_ptt, &mb_params);
if (rc) {
- DP_ERR(p_hwfn, "MCP response failure, aborting\n");
+ DP_NOTICE(p_hwfn, "Failed to send load request, rc = %d\n", rc);
return rc;
}
- *p_load_code = mb_params.mcp_resp;
+ DP_VERBOSE(p_hwfn, QED_MSG_SP,
+ "Load Response: resp 0x%08x\n", mb_params.mcp_resp);
+ p_out_params->load_code = mb_params.mcp_resp;
- /* If MFW refused (e.g. other port is in diagnostic mode) we
- * must abort. This can happen in the following cases:
- * - Other port is in diagnostic mode
- * - Previously loaded function on the engine is not compliant with
- * the requester.
- * - MFW cannot cope with the requester's DRV_MFW_HSI_VERSION.
- * -
+ if (p_in_params->hsi_ver != QED_LOAD_REQ_HSI_VER_1 &&
+ p_out_params->load_code != FW_MSG_CODE_DRV_LOAD_REFUSED_HSI_1) {
+ DP_VERBOSE(p_hwfn,
+ QED_MSG_SP,
+ "Load Response: exist_drv_ver 0x%08x_0x%08x, exist_fw_ver 0x%08x, misc0 0x%08x [exist_role %d, mfw_hsi %d, flags0 0x%x]\n",
+ load_rsp.drv_ver_0,
+ load_rsp.drv_ver_1,
+ load_rsp.fw_ver,
+ load_rsp.misc0,
+ QED_MFW_GET_FIELD(load_rsp.misc0, LOAD_RSP_ROLE),
+ QED_MFW_GET_FIELD(load_rsp.misc0, LOAD_RSP_HSI),
+ QED_MFW_GET_FIELD(load_rsp.misc0, LOAD_RSP_FLAGS0));
+
+ p_out_params->exist_drv_ver_0 = load_rsp.drv_ver_0;
+ p_out_params->exist_drv_ver_1 = load_rsp.drv_ver_1;
+ p_out_params->exist_fw_ver = load_rsp.fw_ver;
+ p_out_params->exist_drv_role =
+ QED_MFW_GET_FIELD(load_rsp.misc0, LOAD_RSP_ROLE);
+ p_out_params->mfw_hsi_ver =
+ QED_MFW_GET_FIELD(load_rsp.misc0, LOAD_RSP_HSI);
+ p_out_params->drv_exists =
+ QED_MFW_GET_FIELD(load_rsp.misc0, LOAD_RSP_FLAGS0) &
+ LOAD_RSP_FLAGS0_DRV_EXISTS;
+ }
+
+ return 0;
+}
+
+static int eocre_get_mfw_drv_role(struct qed_hwfn *p_hwfn,
+ enum qed_drv_role drv_role,
+ u8 *p_mfw_drv_role)
+{
+ switch (drv_role) {
+ case QED_DRV_ROLE_OS:
+ *p_mfw_drv_role = DRV_ROLE_OS;
+ break;
+ case QED_DRV_ROLE_KDUMP:
+ *p_mfw_drv_role = DRV_ROLE_KDUMP;
+ break;
+ default:
+ DP_ERR(p_hwfn, "Unexpected driver role %d\n", drv_role);
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+enum qed_load_req_force {
+ QED_LOAD_REQ_FORCE_NONE,
+ QED_LOAD_REQ_FORCE_PF,
+ QED_LOAD_REQ_FORCE_ALL,
+};
+
+static void qed_get_mfw_force_cmd(struct qed_hwfn *p_hwfn,
+
+ enum qed_load_req_force force_cmd,
+ u8 *p_mfw_force_cmd)
+{
+ switch (force_cmd) {
+ case QED_LOAD_REQ_FORCE_NONE:
+ *p_mfw_force_cmd = LOAD_REQ_FORCE_NONE;
+ break;
+ case QED_LOAD_REQ_FORCE_PF:
+ *p_mfw_force_cmd = LOAD_REQ_FORCE_PF;
+ break;
+ case QED_LOAD_REQ_FORCE_ALL:
+ *p_mfw_force_cmd = LOAD_REQ_FORCE_ALL;
+ break;
+ }
+}
+
+int qed_mcp_load_req(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt,
+ struct qed_load_req_params *p_params)
+{
+ struct qed_load_req_out_params out_params;
+ struct qed_load_req_in_params in_params;
+ u8 mfw_drv_role, mfw_force_cmd;
+ int rc;
+
+ memset(&in_params, 0, sizeof(in_params));
+ in_params.hsi_ver = QED_LOAD_REQ_HSI_VER_DEFAULT;
+ in_params.drv_ver_0 = QED_VERSION;
+ in_params.drv_ver_1 = qed_get_config_bitmap();
+ in_params.fw_ver = STORM_FW_VERSION;
+ rc = eocre_get_mfw_drv_role(p_hwfn, p_params->drv_role, &mfw_drv_role);
+ if (rc)
+ return rc;
+
+ in_params.drv_role = mfw_drv_role;
+ in_params.timeout_val = p_params->timeout_val;
+ qed_get_mfw_force_cmd(p_hwfn,
+ QED_LOAD_REQ_FORCE_NONE, &mfw_force_cmd);
+
+ in_params.force_cmd = mfw_force_cmd;
+ in_params.avoid_eng_reset = p_params->avoid_eng_reset;
+
+ memset(&out_params, 0, sizeof(out_params));
+ rc = __qed_mcp_load_req(p_hwfn, p_ptt, &in_params, &out_params);
+ if (rc)
+ return rc;
+
+ /* First handle cases where another load request should/might be sent:
+ * - MFW expects the old interface [HSI version = 1]
+ * - MFW responds that a force load request is required
*/
- if (!(*p_load_code) ||
- ((*p_load_code) == FW_MSG_CODE_DRV_LOAD_REFUSED_HSI) ||
- ((*p_load_code) == FW_MSG_CODE_DRV_LOAD_REFUSED_PDA) ||
- ((*p_load_code) == FW_MSG_CODE_DRV_LOAD_REFUSED_DIAG)) {
- DP_ERR(p_hwfn, "MCP refused load request, aborting\n");
+ if (out_params.load_code == FW_MSG_CODE_DRV_LOAD_REFUSED_HSI_1) {
+ DP_INFO(p_hwfn,
+ "MFW refused a load request due to HSI > 1. Resending with HSI = 1\n");
+
+ in_params.hsi_ver = QED_LOAD_REQ_HSI_VER_1;
+ memset(&out_params, 0, sizeof(out_params));
+ rc = __qed_mcp_load_req(p_hwfn, p_ptt, &in_params, &out_params);
+ if (rc)
+ return rc;
+ } else if (out_params.load_code ==
+ FW_MSG_CODE_DRV_LOAD_REFUSED_REQUIRES_FORCE) {
+ if (qed_mcp_can_force_load(in_params.drv_role,
+ out_params.exist_drv_role,
+ p_params->override_force_load)) {
+ DP_INFO(p_hwfn,
+ "A force load is required [{role, fw_ver, drv_ver}: loading={%d, 0x%08x, x%08x_0x%08x}, existing={%d, 0x%08x, 0x%08x_0x%08x}]\n",
+ in_params.drv_role, in_params.fw_ver,
+ in_params.drv_ver_0, in_params.drv_ver_1,
+ out_params.exist_drv_role,
+ out_params.exist_fw_ver,
+ out_params.exist_drv_ver_0,
+ out_params.exist_drv_ver_1);
+
+ qed_get_mfw_force_cmd(p_hwfn,
+ QED_LOAD_REQ_FORCE_ALL,
+ &mfw_force_cmd);
+
+ in_params.force_cmd = mfw_force_cmd;
+ memset(&out_params, 0, sizeof(out_params));
+ rc = __qed_mcp_load_req(p_hwfn, p_ptt, &in_params,
+ &out_params);
+ if (rc)
+ return rc;
+ } else {
+ DP_NOTICE(p_hwfn,
+ "A force load is required [{role, fw_ver, drv_ver}: loading={%d, 0x%08x, x%08x_0x%08x}, existing={%d, 0x%08x, 0x%08x_0x%08x}] - Avoid\n",
+ in_params.drv_role, in_params.fw_ver,
+ in_params.drv_ver_0, in_params.drv_ver_1,
+ out_params.exist_drv_role,
+ out_params.exist_fw_ver,
+ out_params.exist_drv_ver_0,
+ out_params.exist_drv_ver_1);
+ DP_NOTICE(p_hwfn,
+ "Avoid sending a force load request to prevent disruption of active PFs\n");
+
+ qed_mcp_cancel_load_req(p_hwfn, p_ptt);
+ return -EBUSY;
+ }
+ }
+
+ /* Now handle the other types of responses.
+ * The "REFUSED_HSI_1" and "REFUSED_REQUIRES_FORCE" responses are not
+ * expected here after the additional revised load requests were sent.
+ */
+ switch (out_params.load_code) {
+ case FW_MSG_CODE_DRV_LOAD_ENGINE:
+ case FW_MSG_CODE_DRV_LOAD_PORT:
+ case FW_MSG_CODE_DRV_LOAD_FUNCTION:
+ if (out_params.mfw_hsi_ver != QED_LOAD_REQ_HSI_VER_1 &&
+ out_params.drv_exists) {
+ /* The role and fw/driver version match, but the PF is
+ * already loaded and has not been unloaded gracefully.
+ */
+ DP_NOTICE(p_hwfn,
+ "PF is already loaded\n");
+ return -EINVAL;
+ }
+ break;
+ default:
+ DP_NOTICE(p_hwfn,
+ "Unexpected refusal to load request [resp 0x%08x]. Aborting.\n",
+ out_params.load_code);
return -EBUSY;
}
+ p_params->load_code = out_params.load_code;
+
return 0;
}
diff --git a/drivers/net/ethernet/qlogic/qed/qed_mcp.h b/drivers/net/ethernet/qlogic/qed/qed_mcp.h
index 7d067d0..f76afec 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_mcp.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_mcp.h
@@ -39,6 +39,7 @@
#include <linux/spinlock.h>
#include <linux/qed/qed_fcoe_if.h>
#include "qed_hsi.h"
+#include "qed_dev_api.h"
struct qed_mcp_link_speed_params {
bool autoneg;
@@ -570,27 +571,35 @@ void qed_mcp_cmd_port_init(struct qed_hwfn *p_hwfn,
int qed_mcp_handle_events(struct qed_hwfn *p_hwfn,
struct qed_ptt *p_ptt);
+enum qed_drv_role {
+ QED_DRV_ROLE_OS,
+ QED_DRV_ROLE_KDUMP,
+};
+
+struct qed_load_req_params {
+ /* Input params */
+ enum qed_drv_role drv_role;
+ u8 timeout_val;
+ bool avoid_eng_reset;
+ enum qed_override_force_load override_force_load;
+
+ /* Output params */
+ u32 load_code;
+};
+
/**
- * @brief Sends a LOAD_REQ to the MFW, and in case operation
- * succeed, returns whether this PF is the first on the
- * chip/engine/port or function. This function should be
- * called when driver is ready to accept MFW events after
- * Storms initializations are done.
+ * @brief Sends a LOAD_REQ to the MFW, and in case the operation succeeds,
+ * returns whether this PF is the first on the engine/port or function.
*
- * @param p_hwfn - hw function
- * @param p_ptt - PTT required for register access
- * @param p_load_code - The MCP response param containing one
- * of the following:
- * FW_MSG_CODE_DRV_LOAD_ENGINE
- * FW_MSG_CODE_DRV_LOAD_PORT
- * FW_MSG_CODE_DRV_LOAD_FUNCTION
- * @return int -
- * 0 - Operation was successul.
- * -EBUSY - Operation failed
+ * @param p_hwfn
+ * @param p_ptt
+ * @param p_params
+ *
+ * @return int - 0 - Operation was successful.
*/
int qed_mcp_load_req(struct qed_hwfn *p_hwfn,
struct qed_ptt *p_ptt,
- u32 *p_load_code);
+ struct qed_load_req_params *p_params);
/**
* @brief Sends a UNLOAD_REQ message to the MFW
--
1.9.3
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH net-next 4/7] qed: Send pf-flr as part of initialization
2017-03-28 12:12 [PATCH net-next 0/7] qed: load/unload mfw series Yuval Mintz
` (2 preceding siblings ...)
2017-03-28 12:12 ` [PATCH net-next 3/7] qed: Move to new load request scheme Yuval Mintz
@ 2017-03-28 12:12 ` Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 5/7] qed: Support management-based resource locking Yuval Mintz
` (3 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: Yuval Mintz @ 2017-03-28 12:12 UTC (permalink / raw)
To: davem, netdev; +Cc: Yuval Mintz
During HW initialization, driver would set various registers to their
needed values - but it assumes all registers start at their reset-value,
so there's no need to re-configure a register's default value.
This assumption might be incorrect, e.g., in case of preboot driver
running and initializing the driver prior to our driver.
To overcome this, we now ask management firmware to initiate a PF-flr
early during the initialization sequence. That would return everything
in the PF's scope back to default and prevent previous configurations
from still being applied.
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
---
drivers/net/ethernet/qlogic/qed/qed_dev.c | 9 +++++++++
drivers/net/ethernet/qlogic/qed/qed_hsi.h | 1 +
drivers/net/ethernet/qlogic/qed/qed_mcp.c | 8 ++++++++
drivers/net/ethernet/qlogic/qed/qed_mcp.h | 10 ++++++++++
4 files changed, 28 insertions(+)
diff --git a/drivers/net/ethernet/qlogic/qed/qed_dev.c b/drivers/net/ethernet/qlogic/qed/qed_dev.c
index 759b819..e4b1450 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_dev.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_dev.c
@@ -2280,6 +2280,15 @@ static int qed_hw_prepare_single(struct qed_hwfn *p_hwfn,
goto err2;
}
+ /* Sending a mailbox to the MFW should be done after qed_get_hw_info()
+ * is called as it sets the ports number in an engine.
+ */
+ if (IS_LEAD_HWFN(p_hwfn)) {
+ rc = qed_mcp_initiate_pf_flr(p_hwfn, p_hwfn->p_main_ptt);
+ if (rc)
+ DP_NOTICE(p_hwfn, "Failed to initiate PF FLR\n");
+ }
+
/* Allocate the init RT array and initialize the init-ops engine */
rc = qed_init_alloc(p_hwfn);
if (rc)
diff --git a/drivers/net/ethernet/qlogic/qed/qed_hsi.h b/drivers/net/ethernet/qlogic/qed/qed_hsi.h
index bf2b201..2de7aa1 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_hsi.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_hsi.h
@@ -10093,6 +10093,7 @@ struct public_drv_mb {
#define DRV_MSG_CODE_BW_UPDATE_ACK 0x32000000
#define DRV_MSG_CODE_NIG_DRAIN 0x30000000
#define DRV_MSG_GET_RESOURCE_ALLOC_MSG 0x34000000
+#define DRV_MSG_CODE_INITIATE_PF_FLR 0x02010000
#define DRV_MSG_CODE_VF_DISABLED_DONE 0xc0000000
#define DRV_MSG_CODE_CFG_VF_MSIX 0xc0010000
#define DRV_MSG_CODE_NVM_GET_FILE_ATT 0x00030000
diff --git a/drivers/net/ethernet/qlogic/qed/qed_mcp.c b/drivers/net/ethernet/qlogic/qed/qed_mcp.c
index 1a820b4..490619f 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_mcp.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_mcp.c
@@ -2263,3 +2263,11 @@ int qed_mcp_get_resc_info(struct qed_hwfn *p_hwfn,
return 0;
}
+
+int qed_mcp_initiate_pf_flr(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt)
+{
+ u32 mcp_resp, mcp_param;
+
+ return qed_mcp_cmd(p_hwfn, p_ptt, DRV_MSG_CODE_INITIATE_PF_FLR, 0,
+ &mcp_resp, &mcp_param);
+}
diff --git a/drivers/net/ethernet/qlogic/qed/qed_mcp.h b/drivers/net/ethernet/qlogic/qed/qed_mcp.h
index f76afec..0056cfe 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_mcp.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_mcp.h
@@ -770,4 +770,14 @@ int qed_mcp_get_resc_info(struct qed_hwfn *p_hwfn,
struct qed_ptt *p_ptt,
struct resource_info *p_resc_info,
u32 *p_mcp_resp, u32 *p_mcp_param);
+
+/**
+ * @brief - Initiates PF FLR
+ *
+ * @param p_hwfn
+ * @param p_ptt
+ *
+ * @return int - 0 - operation was successful.
+ */
+int qed_mcp_initiate_pf_flr(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt);
#endif
--
1.9.3
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH net-next 5/7] qed: Support management-based resource locking
2017-03-28 12:12 [PATCH net-next 0/7] qed: load/unload mfw series Yuval Mintz
` (3 preceding siblings ...)
2017-03-28 12:12 ` [PATCH net-next 4/7] qed: Send pf-flr as part of initialization Yuval Mintz
@ 2017-03-28 12:12 ` Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 6/7] qed: Utilize resource-lock based scheme Yuval Mintz
` (2 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: Yuval Mintz @ 2017-03-28 12:12 UTC (permalink / raw)
To: davem, netdev; +Cc: Tomer Tayar, Yuval Mintz
From: Tomer Tayar <Tomer.Tayar@cavium.com>
Global locking can't properly be used to synchronize between different
PFs in all scenarios, as those instances might reside in different
logical partitions [e.g., when a PF is assigned via PDA to some VM].
The management firmware provides a generic infrastructure for
device locks. For each 'resource', it's guaranteed it could be acquired
by at most a single PF at any given time [or by management firmware].
This patch adds the necessary logic in qed for utilizing said
infrastructure, implementing lock/unlock internal APIs.
Signed-off-by: Tomer Tayar <Tomer.Tayar@cavium.com>
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
---
drivers/net/ethernet/qlogic/qed/qed_hsi.h | 28 +++++
drivers/net/ethernet/qlogic/qed/qed_mcp.c | 176 ++++++++++++++++++++++++++++++
drivers/net/ethernet/qlogic/qed/qed_mcp.h | 65 +++++++++++
3 files changed, 269 insertions(+)
diff --git a/drivers/net/ethernet/qlogic/qed/qed_hsi.h b/drivers/net/ethernet/qlogic/qed/qed_hsi.h
index 2de7aa1..4346399 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_hsi.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_hsi.h
@@ -10119,6 +10119,33 @@ struct public_drv_mb {
#define DRV_MSG_CODE_BIST_TEST 0x001e0000
#define DRV_MSG_CODE_SET_LED_MODE 0x00200000
+#define DRV_MSG_CODE_RESOURCE_CMD 0x00230000
+
+#define RESOURCE_CMD_REQ_RESC_MASK 0x0000001F
+#define RESOURCE_CMD_REQ_RESC_SHIFT 0
+#define RESOURCE_CMD_REQ_OPCODE_MASK 0x000000E0
+#define RESOURCE_CMD_REQ_OPCODE_SHIFT 5
+#define RESOURCE_OPCODE_REQ 1
+#define RESOURCE_OPCODE_REQ_WO_AGING 2
+#define RESOURCE_OPCODE_REQ_W_AGING 3
+#define RESOURCE_OPCODE_RELEASE 4
+#define RESOURCE_OPCODE_FORCE_RELEASE 5
+#define RESOURCE_CMD_REQ_AGE_MASK 0x0000FF00
+#define RESOURCE_CMD_REQ_AGE_SHIFT 8
+
+#define RESOURCE_CMD_RSP_OWNER_MASK 0x000000FF
+#define RESOURCE_CMD_RSP_OWNER_SHIFT 0
+#define RESOURCE_CMD_RSP_OPCODE_MASK 0x00000700
+#define RESOURCE_CMD_RSP_OPCODE_SHIFT 8
+#define RESOURCE_OPCODE_GNT 1
+#define RESOURCE_OPCODE_BUSY 2
+#define RESOURCE_OPCODE_RELEASED 3
+#define RESOURCE_OPCODE_RELEASED_PREVIOUS 4
+#define RESOURCE_OPCODE_WRONG_OWNER 5
+#define RESOURCE_OPCODE_UNKNOWN_CMD 255
+
+#define RESOURCE_DUMP 0
+
#define DRV_MSG_CODE_GET_PF_RDMA_PROTOCOL 0x002b0000
#define DRV_MSG_CODE_OS_WOL 0x002e0000
@@ -10207,6 +10234,7 @@ struct public_drv_mb {
u32 fw_mb_header;
#define FW_MSG_CODE_MASK 0xffff0000
+#define FW_MSG_CODE_UNSUPPORTED 0x00000000
#define FW_MSG_CODE_DRV_LOAD_ENGINE 0x10100000
#define FW_MSG_CODE_DRV_LOAD_PORT 0x10110000
#define FW_MSG_CODE_DRV_LOAD_FUNCTION 0x10120000
diff --git a/drivers/net/ethernet/qlogic/qed/qed_mcp.c b/drivers/net/ethernet/qlogic/qed/qed_mcp.c
index 490619f..ddcbc24 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_mcp.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_mcp.c
@@ -2271,3 +2271,179 @@ int qed_mcp_initiate_pf_flr(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt)
return qed_mcp_cmd(p_hwfn, p_ptt, DRV_MSG_CODE_INITIATE_PF_FLR, 0,
&mcp_resp, &mcp_param);
}
+
+static int qed_mcp_resource_cmd(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt,
+ u32 param, u32 *p_mcp_resp, u32 *p_mcp_param)
+{
+ int rc;
+
+ rc = qed_mcp_cmd(p_hwfn, p_ptt, DRV_MSG_CODE_RESOURCE_CMD, param,
+ p_mcp_resp, p_mcp_param);
+ if (rc)
+ return rc;
+
+ if (*p_mcp_resp == FW_MSG_CODE_UNSUPPORTED) {
+ DP_INFO(p_hwfn,
+ "The resource command is unsupported by the MFW\n");
+ return -EINVAL;
+ }
+
+ if (*p_mcp_param == RESOURCE_OPCODE_UNKNOWN_CMD) {
+ u8 opcode = QED_MFW_GET_FIELD(param, RESOURCE_CMD_REQ_OPCODE);
+
+ DP_NOTICE(p_hwfn,
+ "The resource command is unknown to the MFW [param 0x%08x, opcode %d]\n",
+ param, opcode);
+ return -EINVAL;
+ }
+
+ return rc;
+}
+
+int
+__qed_mcp_resc_lock(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt,
+ struct qed_resc_lock_params *p_params)
+{
+ u32 param = 0, mcp_resp, mcp_param;
+ u8 opcode;
+ int rc;
+
+ switch (p_params->timeout) {
+ case QED_MCP_RESC_LOCK_TO_DEFAULT:
+ opcode = RESOURCE_OPCODE_REQ;
+ p_params->timeout = 0;
+ break;
+ case QED_MCP_RESC_LOCK_TO_NONE:
+ opcode = RESOURCE_OPCODE_REQ_WO_AGING;
+ p_params->timeout = 0;
+ break;
+ default:
+ opcode = RESOURCE_OPCODE_REQ_W_AGING;
+ break;
+ }
+
+ QED_MFW_SET_FIELD(param, RESOURCE_CMD_REQ_RESC, p_params->resource);
+ QED_MFW_SET_FIELD(param, RESOURCE_CMD_REQ_OPCODE, opcode);
+ QED_MFW_SET_FIELD(param, RESOURCE_CMD_REQ_AGE, p_params->timeout);
+
+ DP_VERBOSE(p_hwfn,
+ QED_MSG_SP,
+ "Resource lock request: param 0x%08x [age %d, opcode %d, resource %d]\n",
+ param, p_params->timeout, opcode, p_params->resource);
+
+ /* Attempt to acquire the resource */
+ rc = qed_mcp_resource_cmd(p_hwfn, p_ptt, param, &mcp_resp, &mcp_param);
+ if (rc)
+ return rc;
+
+ /* Analyze the response */
+ p_params->owner = QED_MFW_GET_FIELD(mcp_param, RESOURCE_CMD_RSP_OWNER);
+ opcode = QED_MFW_GET_FIELD(mcp_param, RESOURCE_CMD_RSP_OPCODE);
+
+ DP_VERBOSE(p_hwfn,
+ QED_MSG_SP,
+ "Resource lock response: mcp_param 0x%08x [opcode %d, owner %d]\n",
+ mcp_param, opcode, p_params->owner);
+
+ switch (opcode) {
+ case RESOURCE_OPCODE_GNT:
+ p_params->b_granted = true;
+ break;
+ case RESOURCE_OPCODE_BUSY:
+ p_params->b_granted = false;
+ break;
+ default:
+ DP_NOTICE(p_hwfn,
+ "Unexpected opcode in resource lock response [mcp_param 0x%08x, opcode %d]\n",
+ mcp_param, opcode);
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+int
+qed_mcp_resc_lock(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt, struct qed_resc_lock_params *p_params)
+{
+ u32 retry_cnt = 0;
+ int rc;
+
+ do {
+ /* No need for an interval before the first iteration */
+ if (retry_cnt) {
+ if (p_params->sleep_b4_retry) {
+ u16 retry_interval_in_ms =
+ DIV_ROUND_UP(p_params->retry_interval,
+ 1000);
+
+ msleep(retry_interval_in_ms);
+ } else {
+ udelay(p_params->retry_interval);
+ }
+ }
+
+ rc = __qed_mcp_resc_lock(p_hwfn, p_ptt, p_params);
+ if (rc)
+ return rc;
+
+ if (p_params->b_granted)
+ break;
+ } while (retry_cnt++ < p_params->retry_num);
+
+ return 0;
+}
+
+int
+qed_mcp_resc_unlock(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt,
+ struct qed_resc_unlock_params *p_params)
+{
+ u32 param = 0, mcp_resp, mcp_param;
+ u8 opcode;
+ int rc;
+
+ opcode = p_params->b_force ? RESOURCE_OPCODE_FORCE_RELEASE
+ : RESOURCE_OPCODE_RELEASE;
+ QED_MFW_SET_FIELD(param, RESOURCE_CMD_REQ_RESC, p_params->resource);
+ QED_MFW_SET_FIELD(param, RESOURCE_CMD_REQ_OPCODE, opcode);
+
+ DP_VERBOSE(p_hwfn, QED_MSG_SP,
+ "Resource unlock request: param 0x%08x [opcode %d, resource %d]\n",
+ param, opcode, p_params->resource);
+
+ /* Attempt to release the resource */
+ rc = qed_mcp_resource_cmd(p_hwfn, p_ptt, param, &mcp_resp, &mcp_param);
+ if (rc)
+ return rc;
+
+ /* Analyze the response */
+ opcode = QED_MFW_GET_FIELD(mcp_param, RESOURCE_CMD_RSP_OPCODE);
+
+ DP_VERBOSE(p_hwfn, QED_MSG_SP,
+ "Resource unlock response: mcp_param 0x%08x [opcode %d]\n",
+ mcp_param, opcode);
+
+ switch (opcode) {
+ case RESOURCE_OPCODE_RELEASED_PREVIOUS:
+ DP_INFO(p_hwfn,
+ "Resource unlock request for an already released resource [%d]\n",
+ p_params->resource);
+ /* Fallthrough */
+ case RESOURCE_OPCODE_RELEASED:
+ p_params->b_released = true;
+ break;
+ case RESOURCE_OPCODE_WRONG_OWNER:
+ p_params->b_released = false;
+ break;
+ default:
+ DP_NOTICE(p_hwfn,
+ "Unexpected opcode in resource unlock response [mcp_param 0x%08x, opcode %d]\n",
+ mcp_param, opcode);
+ return -EINVAL;
+ }
+
+ return 0;
+}
diff --git a/drivers/net/ethernet/qlogic/qed/qed_mcp.h b/drivers/net/ethernet/qlogic/qed/qed_mcp.h
index 0056cfe..8673ac1 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_mcp.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_mcp.h
@@ -780,4 +780,69 @@ int qed_mcp_get_resc_info(struct qed_hwfn *p_hwfn,
* @return int - 0 - operation was successful.
*/
int qed_mcp_initiate_pf_flr(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt);
+struct qed_resc_lock_params {
+ /* Resource number [valid values are 0..31] */
+ u8 resource;
+
+ /* Lock timeout value in seconds [default, none or 1..254] */
+ u8 timeout;
+#define QED_MCP_RESC_LOCK_TO_DEFAULT 0
+#define QED_MCP_RESC_LOCK_TO_NONE 255
+
+ /* Number of times to retry locking */
+ u8 retry_num;
+
+ /* The interval in usec between retries */
+ u16 retry_interval;
+
+ /* Use sleep or delay between retries */
+ bool sleep_b4_retry;
+
+ /* Will be set as true if the resource is free and granted */
+ bool b_granted;
+
+ /* Will be filled with the resource owner.
+ * [0..15 = PF0-15, 16 = MFW]
+ */
+ u8 owner;
+};
+
+/**
+ * @brief Acquires MFW generic resource lock
+ *
+ * @param p_hwfn
+ * @param p_ptt
+ * @param p_params
+ *
+ * @return int - 0 - operation was successful.
+ */
+int
+qed_mcp_resc_lock(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt, struct qed_resc_lock_params *p_params);
+
+struct qed_resc_unlock_params {
+ /* Resource number [valid values are 0..31] */
+ u8 resource;
+
+ /* Allow to release a resource even if belongs to another PF */
+ bool b_force;
+
+ /* Will be set as true if the resource is released */
+ bool b_released;
+};
+
+/**
+ * @brief Releases MFW generic resource lock
+ *
+ * @param p_hwfn
+ * @param p_ptt
+ * @param p_params
+ *
+ * @return int - 0 - operation was successful.
+ */
+int
+qed_mcp_resc_unlock(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt,
+ struct qed_resc_unlock_params *p_params);
+
#endif
--
1.9.3
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH net-next 6/7] qed: Utilize resource-lock based scheme
2017-03-28 12:12 [PATCH net-next 0/7] qed: load/unload mfw series Yuval Mintz
` (4 preceding siblings ...)
2017-03-28 12:12 ` [PATCH net-next 5/7] qed: Support management-based resource locking Yuval Mintz
@ 2017-03-28 12:12 ` Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 7/7] qed: Use BDQ resource for storage protocols Yuval Mintz
2017-03-29 1:06 ` [PATCH net-next 0/7] qed: load/unload mfw series David Miller
7 siblings, 0 replies; 9+ messages in thread
From: Yuval Mintz @ 2017-03-28 12:12 UTC (permalink / raw)
To: davem, netdev; +Cc: Tomer Tayar, Yuval Mintz
From: Tomer Tayar <Tomer.Tayar@cavium.com>
Management firmware is used as an arbiter between the various PFs
in matters of resources, but some of the resources that need to
be divided are dependent on the non-management firmware used,
so management firmware first needs to be told how many resources
there are before trying to divide them.
As part of the initialization sequence, driver would first inform
the management firmware of the available resources under
a dedicated resource lock, and afterwards request for various
resources which might be based on the previous set values.
Signed-off-by: Tomer Tayar <Tomer.Tayar@cavium.com>
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
---
drivers/net/ethernet/qlogic/qed/qed.h | 1 +
drivers/net/ethernet/qlogic/qed/qed_dev.c | 368 ++++++++++++++++----------
drivers/net/ethernet/qlogic/qed/qed_dev_api.h | 2 +
drivers/net/ethernet/qlogic/qed/qed_hsi.h | 8 +-
drivers/net/ethernet/qlogic/qed/qed_mcp.c | 206 ++++++++++++--
drivers/net/ethernet/qlogic/qed/qed_mcp.h | 49 +++-
6 files changed, 468 insertions(+), 166 deletions(-)
diff --git a/drivers/net/ethernet/qlogic/qed/qed.h b/drivers/net/ethernet/qlogic/qed/qed.h
index 187a84b..cf71295 100644
--- a/drivers/net/ethernet/qlogic/qed/qed.h
+++ b/drivers/net/ethernet/qlogic/qed/qed.h
@@ -219,6 +219,7 @@ enum qed_resources {
QED_LL2_QUEUE,
QED_CMDQS_CQS,
QED_RDMA_STATS_QUEUE,
+ QED_BDQ,
QED_MAX_RESC,
};
diff --git a/drivers/net/ethernet/qlogic/qed/qed_dev.c b/drivers/net/ethernet/qlogic/qed/qed_dev.c
index e4b1450..e75c833 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_dev.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_dev.c
@@ -1565,187 +1565,222 @@ static void qed_hw_set_feat(struct qed_hwfn *p_hwfn)
RESC_NUM(p_hwfn, QED_SB));
}
-static enum resource_id_enum qed_hw_get_mfw_res_id(enum qed_resources res_id)
+const char *qed_hw_get_resc_name(enum qed_resources res_id)
{
- enum resource_id_enum mfw_res_id = RESOURCE_NUM_INVALID;
-
switch (res_id) {
- case QED_SB:
- mfw_res_id = RESOURCE_NUM_SB_E;
- break;
case QED_L2_QUEUE:
- mfw_res_id = RESOURCE_NUM_L2_QUEUE_E;
- break;
+ return "L2_QUEUE";
case QED_VPORT:
- mfw_res_id = RESOURCE_NUM_VPORT_E;
- break;
+ return "VPORT";
case QED_RSS_ENG:
- mfw_res_id = RESOURCE_NUM_RSS_ENGINES_E;
- break;
+ return "RSS_ENG";
case QED_PQ:
- mfw_res_id = RESOURCE_NUM_PQ_E;
- break;
+ return "PQ";
case QED_RL:
- mfw_res_id = RESOURCE_NUM_RL_E;
- break;
+ return "RL";
case QED_MAC:
+ return "MAC";
case QED_VLAN:
- /* Each VFC resource can accommodate both a MAC and a VLAN */
- mfw_res_id = RESOURCE_VFC_FILTER_E;
- break;
+ return "VLAN";
+ case QED_RDMA_CNQ_RAM:
+ return "RDMA_CNQ_RAM";
case QED_ILT:
- mfw_res_id = RESOURCE_ILT_E;
- break;
+ return "ILT";
case QED_LL2_QUEUE:
- mfw_res_id = RESOURCE_LL2_QUEUE_E;
- break;
- case QED_RDMA_CNQ_RAM:
+ return "LL2_QUEUE";
case QED_CMDQS_CQS:
- /* CNQ/CMDQS are the same resource */
- mfw_res_id = RESOURCE_CQS_E;
- break;
+ return "CMDQS_CQS";
case QED_RDMA_STATS_QUEUE:
- mfw_res_id = RESOURCE_RDMA_STATS_QUEUE_E;
- break;
+ return "RDMA_STATS_QUEUE";
+ case QED_BDQ:
+ return "BDQ";
+ case QED_SB:
+ return "SB";
default:
- break;
+ return "UNKNOWN_RESOURCE";
+ }
+}
+
+static int
+__qed_hw_set_soft_resc_size(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt,
+ enum qed_resources res_id,
+ u32 resc_max_val, u32 *p_mcp_resp)
+{
+ int rc;
+
+ rc = qed_mcp_set_resc_max_val(p_hwfn, p_ptt, res_id,
+ resc_max_val, p_mcp_resp);
+ if (rc) {
+ DP_NOTICE(p_hwfn,
+ "MFW response failure for a max value setting of resource %d [%s]\n",
+ res_id, qed_hw_get_resc_name(res_id));
+ return rc;
}
- return mfw_res_id;
+ if (*p_mcp_resp != FW_MSG_CODE_RESOURCE_ALLOC_OK)
+ DP_INFO(p_hwfn,
+ "Failed to set the max value of resource %d [%s]. mcp_resp = 0x%08x.\n",
+ res_id, qed_hw_get_resc_name(res_id), *p_mcp_resp);
+
+ return 0;
}
-static u32 qed_hw_get_dflt_resc_num(struct qed_hwfn *p_hwfn,
- enum qed_resources res_id)
+static int
+qed_hw_set_soft_resc_size(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt)
+{
+ bool b_ah = QED_IS_AH(p_hwfn->cdev);
+ u32 resc_max_val, mcp_resp;
+ u8 res_id;
+ int rc;
+
+ for (res_id = 0; res_id < QED_MAX_RESC; res_id++) {
+ switch (res_id) {
+ case QED_LL2_QUEUE:
+ resc_max_val = MAX_NUM_LL2_RX_QUEUES;
+ break;
+ case QED_RDMA_CNQ_RAM:
+ /* No need for a case for QED_CMDQS_CQS since
+ * CNQ/CMDQS are the same resource.
+ */
+ resc_max_val = NUM_OF_CMDQS_CQS;
+ break;
+ case QED_RDMA_STATS_QUEUE:
+ resc_max_val = b_ah ? RDMA_NUM_STATISTIC_COUNTERS_K2
+ : RDMA_NUM_STATISTIC_COUNTERS_BB;
+ break;
+ case QED_BDQ:
+ resc_max_val = BDQ_NUM_RESOURCES;
+ break;
+ default:
+ continue;
+ }
+
+ rc = __qed_hw_set_soft_resc_size(p_hwfn, p_ptt, res_id,
+ resc_max_val, &mcp_resp);
+ if (rc)
+ return rc;
+
+ /* There's no point to continue to the next resource if the
+ * command is not supported by the MFW.
+ * We do continue if the command is supported but the resource
+ * is unknown to the MFW. Such a resource will be later
+ * configured with the default allocation values.
+ */
+ if (mcp_resp == FW_MSG_CODE_UNSUPPORTED)
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static
+int qed_hw_get_dflt_resc(struct qed_hwfn *p_hwfn,
+ enum qed_resources res_id,
+ u32 *p_resc_num, u32 *p_resc_start)
{
u8 num_funcs = p_hwfn->num_funcs_on_engine;
bool b_ah = QED_IS_AH(p_hwfn->cdev);
struct qed_sb_cnt_info sb_cnt_info;
- u32 dflt_resc_num = 0;
switch (res_id) {
- case QED_SB:
- memset(&sb_cnt_info, 0, sizeof(sb_cnt_info));
- qed_int_get_num_sbs(p_hwfn, &sb_cnt_info);
- dflt_resc_num = sb_cnt_info.sb_cnt;
- break;
case QED_L2_QUEUE:
- dflt_resc_num = (b_ah ? MAX_NUM_L2_QUEUES_K2
- : MAX_NUM_L2_QUEUES_BB) / num_funcs;
+ *p_resc_num = (b_ah ? MAX_NUM_L2_QUEUES_K2 :
+ MAX_NUM_L2_QUEUES_BB) / num_funcs;
break;
case QED_VPORT:
- dflt_resc_num = MAX_NUM_VPORTS_BB / num_funcs;
- dflt_resc_num = (b_ah ? MAX_NUM_VPORTS_K2
- : MAX_NUM_VPORTS_BB) / num_funcs;
+ *p_resc_num = (b_ah ? MAX_NUM_VPORTS_K2 :
+ MAX_NUM_VPORTS_BB) / num_funcs;
break;
case QED_RSS_ENG:
- dflt_resc_num = (b_ah ? ETH_RSS_ENGINE_NUM_K2
- : ETH_RSS_ENGINE_NUM_BB) / num_funcs;
+ *p_resc_num = (b_ah ? ETH_RSS_ENGINE_NUM_K2 :
+ ETH_RSS_ENGINE_NUM_BB) / num_funcs;
break;
case QED_PQ:
- /* The granularity of the PQs is 8 */
- dflt_resc_num = (b_ah ? MAX_QM_TX_QUEUES_K2
- : MAX_QM_TX_QUEUES_BB) / num_funcs;
- dflt_resc_num &= ~0x7;
+ *p_resc_num = (b_ah ? MAX_QM_TX_QUEUES_K2 :
+ MAX_QM_TX_QUEUES_BB) / num_funcs;
+ *p_resc_num &= ~0x7; /* The granularity of the PQs is 8 */
break;
case QED_RL:
- dflt_resc_num = MAX_QM_GLOBAL_RLS / num_funcs;
+ *p_resc_num = MAX_QM_GLOBAL_RLS / num_funcs;
break;
case QED_MAC:
case QED_VLAN:
/* Each VFC resource can accommodate both a MAC and a VLAN */
- dflt_resc_num = ETH_NUM_MAC_FILTERS / num_funcs;
+ *p_resc_num = ETH_NUM_MAC_FILTERS / num_funcs;
break;
case QED_ILT:
- dflt_resc_num = (b_ah ? PXP_NUM_ILT_RECORDS_K2
- : PXP_NUM_ILT_RECORDS_BB) / num_funcs;
+ *p_resc_num = (b_ah ? PXP_NUM_ILT_RECORDS_K2 :
+ PXP_NUM_ILT_RECORDS_BB) / num_funcs;
break;
case QED_LL2_QUEUE:
- dflt_resc_num = MAX_NUM_LL2_RX_QUEUES / num_funcs;
+ *p_resc_num = MAX_NUM_LL2_RX_QUEUES / num_funcs;
break;
case QED_RDMA_CNQ_RAM:
case QED_CMDQS_CQS:
/* CNQ/CMDQS are the same resource */
- dflt_resc_num = NUM_OF_CMDQS_CQS / num_funcs;
+ *p_resc_num = NUM_OF_CMDQS_CQS / num_funcs;
break;
case QED_RDMA_STATS_QUEUE:
- dflt_resc_num = (b_ah ? RDMA_NUM_STATISTIC_COUNTERS_K2
- : RDMA_NUM_STATISTIC_COUNTERS_BB) /
- num_funcs;
-
+ *p_resc_num = (b_ah ? RDMA_NUM_STATISTIC_COUNTERS_K2 :
+ RDMA_NUM_STATISTIC_COUNTERS_BB) / num_funcs;
break;
- default:
+ case QED_BDQ:
+ if (p_hwfn->hw_info.personality != QED_PCI_ISCSI &&
+ p_hwfn->hw_info.personality != QED_PCI_FCOE)
+ *p_resc_num = 0;
+ else
+ *p_resc_num = 1;
break;
+ case QED_SB:
+ memset(&sb_cnt_info, 0, sizeof(sb_cnt_info));
+ qed_int_get_num_sbs(p_hwfn, &sb_cnt_info);
+ *p_resc_num = sb_cnt_info.sb_cnt;
+ break;
+ default:
+ return -EINVAL;
}
- return dflt_resc_num;
-}
-
-static const char *qed_hw_get_resc_name(enum qed_resources res_id)
-{
switch (res_id) {
- case QED_SB:
- return "SB";
- case QED_L2_QUEUE:
- return "L2_QUEUE";
- case QED_VPORT:
- return "VPORT";
- case QED_RSS_ENG:
- return "RSS_ENG";
- case QED_PQ:
- return "PQ";
- case QED_RL:
- return "RL";
- case QED_MAC:
- return "MAC";
- case QED_VLAN:
- return "VLAN";
- case QED_RDMA_CNQ_RAM:
- return "RDMA_CNQ_RAM";
- case QED_ILT:
- return "ILT";
- case QED_LL2_QUEUE:
- return "LL2_QUEUE";
- case QED_CMDQS_CQS:
- return "CMDQS_CQS";
- case QED_RDMA_STATS_QUEUE:
- return "RDMA_STATS_QUEUE";
+ case QED_BDQ:
+ if (!*p_resc_num)
+ *p_resc_start = 0;
+ else if (p_hwfn->cdev->num_ports_in_engines == 4)
+ *p_resc_start = p_hwfn->port_id;
+ else if (p_hwfn->hw_info.personality == QED_PCI_ISCSI)
+ *p_resc_start = p_hwfn->port_id;
+ else if (p_hwfn->hw_info.personality == QED_PCI_FCOE)
+ *p_resc_start = p_hwfn->port_id + 2;
+ break;
default:
- return "UNKNOWN_RESOURCE";
+ *p_resc_start = *p_resc_num * p_hwfn->enabled_func_idx;
+ break;
}
+
+ return 0;
}
-static int qed_hw_set_resc_info(struct qed_hwfn *p_hwfn,
- enum qed_resources res_id)
+static int __qed_hw_set_resc_info(struct qed_hwfn *p_hwfn,
+ enum qed_resources res_id)
{
- u32 dflt_resc_num = 0, dflt_resc_start = 0, mcp_resp, mcp_param;
- u32 *p_resc_num, *p_resc_start;
- struct resource_info resc_info;
+ u32 dflt_resc_num = 0, dflt_resc_start = 0;
+ u32 mcp_resp, *p_resc_num, *p_resc_start;
int rc;
p_resc_num = &RESC_NUM(p_hwfn, res_id);
p_resc_start = &RESC_START(p_hwfn, res_id);
- /* Default values assumes that each function received equal share */
- dflt_resc_num = qed_hw_get_dflt_resc_num(p_hwfn, res_id);
- if (!dflt_resc_num) {
+ rc = qed_hw_get_dflt_resc(p_hwfn, res_id, &dflt_resc_num,
+ &dflt_resc_start);
+ if (rc) {
DP_ERR(p_hwfn,
"Failed to get default amount for resource %d [%s]\n",
res_id, qed_hw_get_resc_name(res_id));
- return -EINVAL;
- }
- dflt_resc_start = dflt_resc_num * p_hwfn->enabled_func_idx;
-
- memset(&resc_info, 0, sizeof(resc_info));
- resc_info.res_id = qed_hw_get_mfw_res_id(res_id);
- if (resc_info.res_id == RESOURCE_NUM_INVALID) {
- DP_ERR(p_hwfn,
- "Failed to match resource %d [%s] with the MFW resources\n",
- res_id, qed_hw_get_resc_name(res_id));
- return -EINVAL;
+ return rc;
}
- rc = qed_mcp_get_resc_info(p_hwfn, p_hwfn->p_main_ptt, &resc_info,
- &mcp_resp, &mcp_param);
+ rc = qed_mcp_get_resc_info(p_hwfn, p_hwfn->p_main_ptt, res_id,
+ &mcp_resp, p_resc_num, p_resc_start);
if (rc) {
DP_NOTICE(p_hwfn,
"MFW response failure for an allocation request for resource %d [%s]\n",
@@ -1758,13 +1793,12 @@ static int qed_hw_set_resc_info(struct qed_hwfn *p_hwfn,
* - There is an internal error in the MFW while processing the request
* - The resource ID is unknown to the MFW
*/
- if (mcp_resp != FW_MSG_CODE_RESOURCE_ALLOC_OK &&
- mcp_resp != FW_MSG_CODE_RESOURCE_ALLOC_DEPRECATED) {
- DP_NOTICE(p_hwfn,
- "Resource %d [%s]: No allocation info was received [mcp_resp 0x%x]. Applying default values [num %d, start %d].\n",
- res_id,
- qed_hw_get_resc_name(res_id),
- mcp_resp, dflt_resc_num, dflt_resc_start);
+ if (mcp_resp != FW_MSG_CODE_RESOURCE_ALLOC_OK) {
+ DP_INFO(p_hwfn,
+ "Failed to receive allocation info for resource %d [%s]. mcp_resp = 0x%x. Applying default values [%d,%d].\n",
+ res_id,
+ qed_hw_get_resc_name(res_id),
+ mcp_resp, dflt_resc_num, dflt_resc_start);
*p_resc_num = dflt_resc_num;
*p_resc_start = dflt_resc_start;
goto out;
@@ -1772,13 +1806,9 @@ static int qed_hw_set_resc_info(struct qed_hwfn *p_hwfn,
/* Special handling for status blocks; Would be revised in future */
if (res_id == QED_SB) {
- resc_info.size -= 1;
- resc_info.offset -= p_hwfn->enabled_func_idx;
+ *p_resc_num -= 1;
+ *p_resc_start -= p_hwfn->enabled_func_idx;
}
-
- *p_resc_num = resc_info.size;
- *p_resc_start = resc_info.offset;
-
out:
/* PQs have to divide by 8 [that's the HW granularity].
* Reduce number so it would fit.
@@ -1796,18 +1826,85 @@ static int qed_hw_set_resc_info(struct qed_hwfn *p_hwfn,
return 0;
}
-static int qed_hw_get_resc(struct qed_hwfn *p_hwfn)
+static int qed_hw_set_resc_info(struct qed_hwfn *p_hwfn)
{
- bool b_ah = QED_IS_AH(p_hwfn->cdev);
- u8 res_id;
int rc;
+ u8 res_id;
for (res_id = 0; res_id < QED_MAX_RESC; res_id++) {
- rc = qed_hw_set_resc_info(p_hwfn, res_id);
+ rc = __qed_hw_set_resc_info(p_hwfn, res_id);
if (rc)
return rc;
}
+ return 0;
+}
+
+#define QED_RESC_ALLOC_LOCK_RETRY_CNT 10
+#define QED_RESC_ALLOC_LOCK_RETRY_INTVL_US 10000 /* 10 msec */
+
+static int qed_hw_get_resc(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt)
+{
+ struct qed_resc_unlock_params resc_unlock_params;
+ struct qed_resc_lock_params resc_lock_params;
+ bool b_ah = QED_IS_AH(p_hwfn->cdev);
+ u8 res_id;
+ int rc;
+
+ /* Setting the max values of the soft resources and the following
+ * resources allocation queries should be atomic. Since several PFs can
+ * run in parallel - a resource lock is needed.
+ * If either the resource lock or resource set value commands are not
+ * supported - skip the the max values setting, release the lock if
+ * needed, and proceed to the queries. Other failures, including a
+ * failure to acquire the lock, will cause this function to fail.
+ */
+ memset(&resc_lock_params, 0, sizeof(resc_lock_params));
+ resc_lock_params.resource = QED_RESC_LOCK_RESC_ALLOC;
+ resc_lock_params.retry_num = QED_RESC_ALLOC_LOCK_RETRY_CNT;
+ resc_lock_params.retry_interval = QED_RESC_ALLOC_LOCK_RETRY_INTVL_US;
+ resc_lock_params.sleep_b4_retry = true;
+ memset(&resc_unlock_params, 0, sizeof(resc_unlock_params));
+ resc_unlock_params.resource = QED_RESC_LOCK_RESC_ALLOC;
+
+ rc = qed_mcp_resc_lock(p_hwfn, p_ptt, &resc_lock_params);
+ if (rc && rc != -EINVAL) {
+ return rc;
+ } else if (rc == -EINVAL) {
+ DP_INFO(p_hwfn,
+ "Skip the max values setting of the soft resources since the resource lock is not supported by the MFW\n");
+ } else if (!rc && !resc_lock_params.b_granted) {
+ DP_NOTICE(p_hwfn,
+ "Failed to acquire the resource lock for the resource allocation commands\n");
+ return -EBUSY;
+ } else {
+ rc = qed_hw_set_soft_resc_size(p_hwfn, p_ptt);
+ if (rc && rc != -EINVAL) {
+ DP_NOTICE(p_hwfn,
+ "Failed to set the max values of the soft resources\n");
+ goto unlock_and_exit;
+ } else if (rc == -EINVAL) {
+ DP_INFO(p_hwfn,
+ "Skip the max values setting of the soft resources since it is not supported by the MFW\n");
+ rc = qed_mcp_resc_unlock(p_hwfn, p_ptt,
+ &resc_unlock_params);
+ if (rc)
+ DP_INFO(p_hwfn,
+ "Failed to release the resource lock for the resource allocation commands\n");
+ }
+ }
+
+ rc = qed_hw_set_resc_info(p_hwfn);
+ if (rc)
+ goto unlock_and_exit;
+
+ if (resc_lock_params.b_granted && !resc_unlock_params.b_released) {
+ rc = qed_mcp_resc_unlock(p_hwfn, p_ptt, &resc_unlock_params);
+ if (rc)
+ DP_INFO(p_hwfn,
+ "Failed to release the resource lock for the resource allocation commands\n");
+ }
+
/* Sanity for ILT */
if ((b_ah && (RESC_END(p_hwfn, QED_ILT) > PXP_NUM_ILT_RECORDS_K2)) ||
(!b_ah && (RESC_END(p_hwfn, QED_ILT) > PXP_NUM_ILT_RECORDS_BB))) {
@@ -1819,8 +1916,6 @@ static int qed_hw_get_resc(struct qed_hwfn *p_hwfn)
qed_hw_set_feat(p_hwfn);
- DP_VERBOSE(p_hwfn, NETIF_MSG_PROBE,
- "The numbers for each resource are:\n");
for (res_id = 0; res_id < QED_MAX_RESC; res_id++)
DP_VERBOSE(p_hwfn, NETIF_MSG_PROBE, "%s = %d start = %d\n",
qed_hw_get_resc_name(res_id),
@@ -1828,6 +1923,11 @@ static int qed_hw_get_resc(struct qed_hwfn *p_hwfn)
RESC_START(p_hwfn, res_id));
return 0;
+
+unlock_and_exit:
+ if (resc_lock_params.b_granted && !resc_unlock_params.b_released)
+ qed_mcp_resc_unlock(p_hwfn, p_ptt, &resc_unlock_params);
+ return rc;
}
static int qed_hw_get_nvm_info(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt)
@@ -2158,7 +2258,7 @@ static void qed_hw_info_port_num(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt)
if (qed_mcp_is_init(p_hwfn))
p_hwfn->hw_info.mtu = p_hwfn->mcp_info->func_info.mtu;
- return qed_hw_get_resc(p_hwfn);
+ return qed_hw_get_resc(p_hwfn, p_ptt);
}
static int qed_get_dev_info(struct qed_dev *cdev)
diff --git a/drivers/net/ethernet/qlogic/qed/qed_dev_api.h b/drivers/net/ethernet/qlogic/qed/qed_dev_api.h
index cb97349..2c6637f 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_dev_api.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_dev_api.h
@@ -470,4 +470,6 @@ int qed_set_rxq_coalesce(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt,
*/
int qed_set_txq_coalesce(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt,
u16 coalesce, u8 qid, u16 sb_id);
+
+const char *qed_hw_get_resc_name(enum qed_resources res_id);
#endif
diff --git a/drivers/net/ethernet/qlogic/qed/qed_hsi.h b/drivers/net/ethernet/qlogic/qed/qed_hsi.h
index 4346399..c6b9a3f 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_hsi.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_hsi.h
@@ -9986,6 +9986,7 @@ enum resource_id_enum {
RESOURCE_NUM_RSS_ENGINES_E = 14,
RESOURCE_LL2_QUEUE_E = 15,
RESOURCE_RDMA_STATS_QUEUE_E = 16,
+ RESOURCE_BDQ_E = 17,
RESOURCE_MAX_NUM,
RESOURCE_NUM_INVALID = 0xFFFFFFFF
};
@@ -10087,12 +10088,13 @@ struct public_drv_mb {
#define DRV_MSG_CODE_OV_UPDATE_DRIVER_STATE 0x31000000
#define DRV_MSG_CODE_BW_UPDATE_ACK 0x32000000
#define DRV_MSG_CODE_OV_UPDATE_MTU 0x33000000
+#define DRV_MSG_GET_RESOURCE_ALLOC_MSG 0x34000000
+#define DRV_MSG_SET_RESOURCE_VALUE_MSG 0x35000000
#define DRV_MSG_CODE_OV_UPDATE_WOL 0x38000000
#define DRV_MSG_CODE_OV_UPDATE_ESWITCH_MODE 0x39000000
#define DRV_MSG_CODE_BW_UPDATE_ACK 0x32000000
#define DRV_MSG_CODE_NIG_DRAIN 0x30000000
-#define DRV_MSG_GET_RESOURCE_ALLOC_MSG 0x34000000
#define DRV_MSG_CODE_INITIATE_PF_FLR 0x02010000
#define DRV_MSG_CODE_VF_DISABLED_DONE 0xc0000000
#define DRV_MSG_CODE_CFG_VF_MSIX 0xc0010000
@@ -10263,6 +10265,10 @@ struct public_drv_mb {
#define FW_MSG_SEQ_NUMBER_MASK 0x0000ffff
u32 fw_mb_param;
+#define FW_MB_PARAM_RESOURCE_ALLOC_VERSION_MAJOR_MASK 0xFFFF0000
+#define FW_MB_PARAM_RESOURCE_ALLOC_VERSION_MAJOR_SHIFT 16
+#define FW_MB_PARAM_RESOURCE_ALLOC_VERSION_MINOR_MASK 0x0000FFFF
+#define FW_MB_PARAM_RESOURCE_ALLOC_VERSION_MINOR_SHIFT 0
/* get pf rdma protocol command responce */
#define FW_MB_PARAM_GET_PF_RDMA_NONE 0x0
diff --git a/drivers/net/ethernet/qlogic/qed/qed_mcp.c b/drivers/net/ethernet/qlogic/qed/qed_mcp.c
index ddcbc24..619eac8 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_mcp.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_mcp.c
@@ -2220,46 +2220,212 @@ int qed_mcp_bist_nvm_test_get_image_att(struct qed_hwfn *p_hwfn,
return rc;
}
-#define QED_RESC_ALLOC_VERSION_MAJOR 1
+static enum resource_id_enum qed_mcp_get_mfw_res_id(enum qed_resources res_id)
+{
+ enum resource_id_enum mfw_res_id = RESOURCE_NUM_INVALID;
+
+ switch (res_id) {
+ case QED_SB:
+ mfw_res_id = RESOURCE_NUM_SB_E;
+ break;
+ case QED_L2_QUEUE:
+ mfw_res_id = RESOURCE_NUM_L2_QUEUE_E;
+ break;
+ case QED_VPORT:
+ mfw_res_id = RESOURCE_NUM_VPORT_E;
+ break;
+ case QED_RSS_ENG:
+ mfw_res_id = RESOURCE_NUM_RSS_ENGINES_E;
+ break;
+ case QED_PQ:
+ mfw_res_id = RESOURCE_NUM_PQ_E;
+ break;
+ case QED_RL:
+ mfw_res_id = RESOURCE_NUM_RL_E;
+ break;
+ case QED_MAC:
+ case QED_VLAN:
+ /* Each VFC resource can accommodate both a MAC and a VLAN */
+ mfw_res_id = RESOURCE_VFC_FILTER_E;
+ break;
+ case QED_ILT:
+ mfw_res_id = RESOURCE_ILT_E;
+ break;
+ case QED_LL2_QUEUE:
+ mfw_res_id = RESOURCE_LL2_QUEUE_E;
+ break;
+ case QED_RDMA_CNQ_RAM:
+ case QED_CMDQS_CQS:
+ /* CNQ/CMDQS are the same resource */
+ mfw_res_id = RESOURCE_CQS_E;
+ break;
+ case QED_RDMA_STATS_QUEUE:
+ mfw_res_id = RESOURCE_RDMA_STATS_QUEUE_E;
+ break;
+ case QED_BDQ:
+ mfw_res_id = RESOURCE_BDQ_E;
+ break;
+ default:
+ break;
+ }
+
+ return mfw_res_id;
+}
+
+#define QED_RESC_ALLOC_VERSION_MAJOR 2
#define QED_RESC_ALLOC_VERSION_MINOR 0
#define QED_RESC_ALLOC_VERSION \
((QED_RESC_ALLOC_VERSION_MAJOR << \
DRV_MB_PARAM_RESOURCE_ALLOC_VERSION_MAJOR_SHIFT) | \
(QED_RESC_ALLOC_VERSION_MINOR << \
DRV_MB_PARAM_RESOURCE_ALLOC_VERSION_MINOR_SHIFT))
-int qed_mcp_get_resc_info(struct qed_hwfn *p_hwfn,
- struct qed_ptt *p_ptt,
- struct resource_info *p_resc_info,
- u32 *p_mcp_resp, u32 *p_mcp_param)
+
+struct qed_resc_alloc_in_params {
+ u32 cmd;
+ enum qed_resources res_id;
+ u32 resc_max_val;
+};
+
+struct qed_resc_alloc_out_params {
+ u32 mcp_resp;
+ u32 mcp_param;
+ u32 resc_num;
+ u32 resc_start;
+ u32 vf_resc_num;
+ u32 vf_resc_start;
+ u32 flags;
+};
+
+static int
+qed_mcp_resc_allocation_msg(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt,
+ struct qed_resc_alloc_in_params *p_in_params,
+ struct qed_resc_alloc_out_params *p_out_params)
{
struct qed_mcp_mb_params mb_params;
+ struct resource_info mfw_resc_info;
int rc;
+ memset(&mfw_resc_info, 0, sizeof(mfw_resc_info));
+
+ mfw_resc_info.res_id = qed_mcp_get_mfw_res_id(p_in_params->res_id);
+ if (mfw_resc_info.res_id == RESOURCE_NUM_INVALID) {
+ DP_ERR(p_hwfn,
+ "Failed to match resource %d [%s] with the MFW resources\n",
+ p_in_params->res_id,
+ qed_hw_get_resc_name(p_in_params->res_id));
+ return -EINVAL;
+ }
+
+ switch (p_in_params->cmd) {
+ case DRV_MSG_SET_RESOURCE_VALUE_MSG:
+ mfw_resc_info.size = p_in_params->resc_max_val;
+ /* Fallthrough */
+ case DRV_MSG_GET_RESOURCE_ALLOC_MSG:
+ break;
+ default:
+ DP_ERR(p_hwfn, "Unexpected resource alloc command [0x%08x]\n",
+ p_in_params->cmd);
+ return -EINVAL;
+ }
+
memset(&mb_params, 0, sizeof(mb_params));
- mb_params.cmd = DRV_MSG_GET_RESOURCE_ALLOC_MSG;
+ mb_params.cmd = p_in_params->cmd;
mb_params.param = QED_RESC_ALLOC_VERSION;
+ mb_params.p_data_src = &mfw_resc_info;
+ mb_params.data_src_size = sizeof(mfw_resc_info);
+ mb_params.p_data_dst = mb_params.p_data_src;
+ mb_params.data_dst_size = mb_params.data_src_size;
+
+ DP_VERBOSE(p_hwfn,
+ QED_MSG_SP,
+ "Resource message request: cmd 0x%08x, res_id %d [%s], hsi_version %d.%d, val 0x%x\n",
+ p_in_params->cmd,
+ p_in_params->res_id,
+ qed_hw_get_resc_name(p_in_params->res_id),
+ QED_MFW_GET_FIELD(mb_params.param,
+ DRV_MB_PARAM_RESOURCE_ALLOC_VERSION_MAJOR),
+ QED_MFW_GET_FIELD(mb_params.param,
+ DRV_MB_PARAM_RESOURCE_ALLOC_VERSION_MINOR),
+ p_in_params->resc_max_val);
- mb_params.p_data_src = p_resc_info;
- mb_params.data_src_size = sizeof(*p_resc_info);
- mb_params.p_data_dst = p_resc_info;
- mb_params.data_dst_size = sizeof(*p_resc_info);
rc = qed_mcp_cmd_and_union(p_hwfn, p_ptt, &mb_params);
if (rc)
return rc;
- /* Copy the data back */
- *p_mcp_resp = mb_params.mcp_resp;
- *p_mcp_param = mb_params.mcp_param;
+ p_out_params->mcp_resp = mb_params.mcp_resp;
+ p_out_params->mcp_param = mb_params.mcp_param;
+ p_out_params->resc_num = mfw_resc_info.size;
+ p_out_params->resc_start = mfw_resc_info.offset;
+ p_out_params->vf_resc_num = mfw_resc_info.vf_size;
+ p_out_params->vf_resc_start = mfw_resc_info.vf_offset;
+ p_out_params->flags = mfw_resc_info.flags;
DP_VERBOSE(p_hwfn,
QED_MSG_SP,
- "MFW resource_info: version 0x%x, res_id 0x%x, size 0x%x, offset 0x%x, vf_size 0x%x, vf_offset 0x%x, flags 0x%x\n",
- *p_mcp_param,
- p_resc_info->res_id,
- p_resc_info->size,
- p_resc_info->offset,
- p_resc_info->vf_size,
- p_resc_info->vf_offset, p_resc_info->flags);
+ "Resource message response: mfw_hsi_version %d.%d, num 0x%x, start 0x%x, vf_num 0x%x, vf_start 0x%x, flags 0x%08x\n",
+ QED_MFW_GET_FIELD(p_out_params->mcp_param,
+ FW_MB_PARAM_RESOURCE_ALLOC_VERSION_MAJOR),
+ QED_MFW_GET_FIELD(p_out_params->mcp_param,
+ FW_MB_PARAM_RESOURCE_ALLOC_VERSION_MINOR),
+ p_out_params->resc_num,
+ p_out_params->resc_start,
+ p_out_params->vf_resc_num,
+ p_out_params->vf_resc_start, p_out_params->flags);
+
+ return 0;
+}
+
+int
+qed_mcp_set_resc_max_val(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt,
+ enum qed_resources res_id,
+ u32 resc_max_val, u32 *p_mcp_resp)
+{
+ struct qed_resc_alloc_out_params out_params;
+ struct qed_resc_alloc_in_params in_params;
+ int rc;
+
+ memset(&in_params, 0, sizeof(in_params));
+ in_params.cmd = DRV_MSG_SET_RESOURCE_VALUE_MSG;
+ in_params.res_id = res_id;
+ in_params.resc_max_val = resc_max_val;
+ memset(&out_params, 0, sizeof(out_params));
+ rc = qed_mcp_resc_allocation_msg(p_hwfn, p_ptt, &in_params,
+ &out_params);
+ if (rc)
+ return rc;
+
+ *p_mcp_resp = out_params.mcp_resp;
+
+ return 0;
+}
+
+int
+qed_mcp_get_resc_info(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt,
+ enum qed_resources res_id,
+ u32 *p_mcp_resp, u32 *p_resc_num, u32 *p_resc_start)
+{
+ struct qed_resc_alloc_out_params out_params;
+ struct qed_resc_alloc_in_params in_params;
+ int rc;
+
+ memset(&in_params, 0, sizeof(in_params));
+ in_params.cmd = DRV_MSG_GET_RESOURCE_ALLOC_MSG;
+ in_params.res_id = res_id;
+ memset(&out_params, 0, sizeof(out_params));
+ rc = qed_mcp_resc_allocation_msg(p_hwfn, p_ptt, &in_params,
+ &out_params);
+ if (rc)
+ return rc;
+
+ *p_mcp_resp = out_params.mcp_resp;
+
+ if (*p_mcp_resp == FW_MSG_CODE_RESOURCE_ALLOC_OK) {
+ *p_resc_num = out_params.resc_num;
+ *p_resc_start = out_params.resc_start;
+ }
return 0;
}
diff --git a/drivers/net/ethernet/qlogic/qed/qed_mcp.h b/drivers/net/ethernet/qlogic/qed/qed_mcp.h
index 8673ac1..ac7d406 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_mcp.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_mcp.h
@@ -743,33 +743,60 @@ int qed_mcp_mask_parities(struct qed_hwfn *p_hwfn,
struct qed_ptt *p_ptt, u32 mask_parities);
/**
- * @brief Send eswitch mode to MFW
+ * @brief - Sets the MFW's max value for the given resource
*
* @param p_hwfn
* @param p_ptt
- * @param eswitch - eswitch mode
+ * @param res_id
+ * @param resc_max_val
+ * @param p_mcp_resp
*
* @return int - 0 - operation was successful.
*/
-int qed_mcp_ov_update_eswitch(struct qed_hwfn *p_hwfn,
- struct qed_ptt *p_ptt,
- enum qed_ov_eswitch eswitch);
+int
+qed_mcp_set_resc_max_val(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt,
+ enum qed_resources res_id,
+ u32 resc_max_val, u32 *p_mcp_resp);
/**
* @brief - Gets the MFW allocation info for the given resource
*
* @param p_hwfn
* @param p_ptt
- * @param p_resc_info - descriptor of requested resource
+ * @param res_id
* @param p_mcp_resp
- * @param p_mcp_param
+ * @param p_resc_num
+ * @param p_resc_start
*
* @return int - 0 - operation was successful.
*/
-int qed_mcp_get_resc_info(struct qed_hwfn *p_hwfn,
- struct qed_ptt *p_ptt,
- struct resource_info *p_resc_info,
- u32 *p_mcp_resp, u32 *p_mcp_param);
+int
+qed_mcp_get_resc_info(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt,
+ enum qed_resources res_id,
+ u32 *p_mcp_resp, u32 *p_resc_num, u32 *p_resc_start);
+
+/**
+ * @brief Send eswitch mode to MFW
+ *
+ * @param p_hwfn
+ * @param p_ptt
+ * @param eswitch - eswitch mode
+ *
+ * @return int - 0 - operation was successful.
+ */
+int qed_mcp_ov_update_eswitch(struct qed_hwfn *p_hwfn,
+ struct qed_ptt *p_ptt,
+ enum qed_ov_eswitch eswitch);
+
+#define QED_MCP_RESC_LOCK_MIN_VAL RESOURCE_DUMP
+#define QED_MCP_RESC_LOCK_MAX_VAL 31
+
+enum qed_resc_lock {
+ QED_RESC_LOCK_DBG_DUMP = QED_MCP_RESC_LOCK_MIN_VAL,
+ QED_RESC_LOCK_RESC_ALLOC = QED_MCP_RESC_LOCK_MAX_VAL
+};
/**
* @brief - Initiates PF FLR
--
1.9.3
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH net-next 7/7] qed: Use BDQ resource for storage protocols
2017-03-28 12:12 [PATCH net-next 0/7] qed: load/unload mfw series Yuval Mintz
` (5 preceding siblings ...)
2017-03-28 12:12 ` [PATCH net-next 6/7] qed: Utilize resource-lock based scheme Yuval Mintz
@ 2017-03-28 12:12 ` Yuval Mintz
2017-03-29 1:06 ` [PATCH net-next 0/7] qed: load/unload mfw series David Miller
7 siblings, 0 replies; 9+ messages in thread
From: Yuval Mintz @ 2017-03-28 12:12 UTC (permalink / raw)
To: davem, netdev; +Cc: Yuval Mintz
Until now, qed used some port-defined value as BDQ index for both iSCSI
and FCoE.
As management firmware now treats BDQ as a resource and tells each PF
its BDQ-range, start using a valure from that range instead.
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
---
drivers/net/ethernet/qlogic/qed/qed.h | 2 --
drivers/net/ethernet/qlogic/qed/qed_fcoe.c | 30 +++++++++++++++++++--------
drivers/net/ethernet/qlogic/qed/qed_iscsi.c | 32 +++++++++++++++++++----------
3 files changed, 42 insertions(+), 22 deletions(-)
diff --git a/drivers/net/ethernet/qlogic/qed/qed.h b/drivers/net/ethernet/qlogic/qed/qed.h
index cf71295..8c85fd2 100644
--- a/drivers/net/ethernet/qlogic/qed/qed.h
+++ b/drivers/net/ethernet/qlogic/qed/qed.h
@@ -71,8 +71,6 @@
#define QED_WFQ_UNIT 100
-#define ISCSI_BDQ_ID(_port_id) (_port_id)
-#define FCOE_BDQ_ID(_port_id) ((_port_id) + 2)
#define QED_WID_SIZE (1024)
#define QED_PF_DEMS_SIZE (4)
diff --git a/drivers/net/ethernet/qlogic/qed/qed_fcoe.c b/drivers/net/ethernet/qlogic/qed/qed_fcoe.c
index cbc8141..60921b7 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_fcoe.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_fcoe.c
@@ -191,7 +191,7 @@ struct qed_fcoe_conn {
p_data->q_params.cq_sb_pi = fcoe_pf_params->gl_rq_pi;
p_data->q_params.cmdq_sb_pi = fcoe_pf_params->gl_cmd_pi;
- p_data->q_params.bdq_resource_id = FCOE_BDQ_ID(p_hwfn->port_id);
+ p_data->q_params.bdq_resource_id = (u8)RESC_START(p_hwfn, QED_BDQ);
DMA_REGPAIR_LE(p_data->q_params.bdq_pbl_base_address[BDQ_ID_RQ],
fcoe_pf_params->bdq_pbl_base_addr[BDQ_ID_RQ]);
@@ -512,19 +512,31 @@ static void __iomem *qed_fcoe_get_db_addr(struct qed_hwfn *p_hwfn, u32 cid)
static void __iomem *qed_fcoe_get_primary_bdq_prod(struct qed_hwfn *p_hwfn,
u8 bdq_id)
{
- u8 bdq_function_id = FCOE_BDQ_ID(p_hwfn->port_id);
-
- return (u8 __iomem *)p_hwfn->regview + GTT_BAR0_MAP_REG_MSDM_RAM +
- MSTORM_SCSI_BDQ_EXT_PROD_OFFSET(bdq_function_id, bdq_id);
+ if (RESC_NUM(p_hwfn, QED_BDQ)) {
+ return (u8 __iomem *)p_hwfn->regview +
+ GTT_BAR0_MAP_REG_MSDM_RAM +
+ MSTORM_SCSI_BDQ_EXT_PROD_OFFSET(RESC_START(p_hwfn,
+ QED_BDQ),
+ bdq_id);
+ } else {
+ DP_NOTICE(p_hwfn, "BDQ is not allocated!\n");
+ return NULL;
+ }
}
static void __iomem *qed_fcoe_get_secondary_bdq_prod(struct qed_hwfn *p_hwfn,
u8 bdq_id)
{
- u8 bdq_function_id = FCOE_BDQ_ID(p_hwfn->port_id);
-
- return (u8 __iomem *)p_hwfn->regview + GTT_BAR0_MAP_REG_TSDM_RAM +
- TSTORM_SCSI_BDQ_EXT_PROD_OFFSET(bdq_function_id, bdq_id);
+ if (RESC_NUM(p_hwfn, QED_BDQ)) {
+ return (u8 __iomem *)p_hwfn->regview +
+ GTT_BAR0_MAP_REG_TSDM_RAM +
+ TSTORM_SCSI_BDQ_EXT_PROD_OFFSET(RESC_START(p_hwfn,
+ QED_BDQ),
+ bdq_id);
+ } else {
+ DP_NOTICE(p_hwfn, "BDQ is not allocated!\n");
+ return NULL;
+ }
}
struct qed_fcoe_info *qed_fcoe_alloc(struct qed_hwfn *p_hwfn)
diff --git a/drivers/net/ethernet/qlogic/qed/qed_iscsi.c b/drivers/net/ethernet/qlogic/qed/qed_iscsi.c
index 098766f..2f8ac75 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_iscsi.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_iscsi.c
@@ -216,7 +216,7 @@ struct qed_iscsi_conn {
p_queue->cq_cmdq_sb_num_arr[i] = cpu_to_le16(val);
}
- p_queue->bdq_resource_id = ISCSI_BDQ_ID(p_hwfn->port_id);
+ p_queue->bdq_resource_id = (u8)RESC_START(p_hwfn, QED_BDQ);
DMA_REGPAIR_LE(p_queue->bdq_pbl_base_address[BDQ_ID_RQ],
p_params->bdq_pbl_base_addr[BDQ_ID_RQ]);
@@ -593,21 +593,31 @@ static void __iomem *qed_iscsi_get_db_addr(struct qed_hwfn *p_hwfn, u32 cid)
static void __iomem *qed_iscsi_get_primary_bdq_prod(struct qed_hwfn *p_hwfn,
u8 bdq_id)
{
- u8 bdq_function_id = ISCSI_BDQ_ID(p_hwfn->port_id);
-
- return (u8 __iomem *)p_hwfn->regview + GTT_BAR0_MAP_REG_MSDM_RAM +
- MSTORM_SCSI_BDQ_EXT_PROD_OFFSET(bdq_function_id,
- bdq_id);
+ if (RESC_NUM(p_hwfn, QED_BDQ)) {
+ return (u8 __iomem *)p_hwfn->regview +
+ GTT_BAR0_MAP_REG_MSDM_RAM +
+ MSTORM_SCSI_BDQ_EXT_PROD_OFFSET(RESC_START(p_hwfn,
+ QED_BDQ),
+ bdq_id);
+ } else {
+ DP_NOTICE(p_hwfn, "BDQ is not allocated!\n");
+ return NULL;
+ }
}
static void __iomem *qed_iscsi_get_secondary_bdq_prod(struct qed_hwfn *p_hwfn,
u8 bdq_id)
{
- u8 bdq_function_id = ISCSI_BDQ_ID(p_hwfn->port_id);
-
- return (u8 __iomem *)p_hwfn->regview + GTT_BAR0_MAP_REG_TSDM_RAM +
- TSTORM_SCSI_BDQ_EXT_PROD_OFFSET(bdq_function_id,
- bdq_id);
+ if (RESC_NUM(p_hwfn, QED_BDQ)) {
+ return (u8 __iomem *)p_hwfn->regview +
+ GTT_BAR0_MAP_REG_TSDM_RAM +
+ TSTORM_SCSI_BDQ_EXT_PROD_OFFSET(RESC_START(p_hwfn,
+ QED_BDQ),
+ bdq_id);
+ } else {
+ DP_NOTICE(p_hwfn, "BDQ is not allocated!\n");
+ return NULL;
+ }
}
static int qed_iscsi_setup_connection(struct qed_hwfn *p_hwfn,
--
1.9.3
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH net-next 0/7] qed: load/unload mfw series
2017-03-28 12:12 [PATCH net-next 0/7] qed: load/unload mfw series Yuval Mintz
` (6 preceding siblings ...)
2017-03-28 12:12 ` [PATCH net-next 7/7] qed: Use BDQ resource for storage protocols Yuval Mintz
@ 2017-03-29 1:06 ` David Miller
7 siblings, 0 replies; 9+ messages in thread
From: David Miller @ 2017-03-29 1:06 UTC (permalink / raw)
To: Yuval.Mintz; +Cc: netdev
From: Yuval Mintz <Yuval.Mintz@cavium.com>
Date: Tue, 28 Mar 2017 15:12:49 +0300
> This series correct the unload flow and greatly enhances its
> initialization flow in regard to interactions between driver
> and management firmware.
>
> Patch #1 makes sure unloading is done under management-firmware's
> 'criticial section' protection.
>
> Patches #2 - #4 move driver into using a newer scheme for loading
> in regard to the MFW; This newer scheme would help cleaning the device
> in case a previous instance has dirtied it [preboot, PDA, etc.].
>
> Patches #5 - #6 let driver inform management-firmware on number of
> resources which are dependent on the non-management firmware used.
> Patch #7 then uses a new resource [BDQ] instead of some set value.
Series applied, thank you.
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2017-03-29 1:06 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-03-28 12:12 [PATCH net-next 0/7] qed: load/unload mfw series Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 1/7] qed: Correct HW stop flow Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 2/7] qed: hw_init() to receive parameter-struct Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 3/7] qed: Move to new load request scheme Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 4/7] qed: Send pf-flr as part of initialization Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 5/7] qed: Support management-based resource locking Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 6/7] qed: Utilize resource-lock based scheme Yuval Mintz
2017-03-28 12:12 ` [PATCH net-next 7/7] qed: Use BDQ resource for storage protocols Yuval Mintz
2017-03-29 1:06 ` [PATCH net-next 0/7] qed: load/unload mfw series David Miller
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.