[PATCH 1/4] habanalabs: minor fixes to signals from graph

[PATCH 1/4] habanalabs: minor fixes to signals from graph

[Oded Gabbay]

From: farah kassabri <fkassabri@habana.ai>

- fix signals reservation flow when setting the next available signal
  value.
- fix how driver treats the wait offset, by decrement 1 from the user
  value before actually building the wait cs.
- fix debug logs to print in decimal the sob value instead of hex.

Signed-off-by: farah kassabri <fkassabri@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
---
 .../habanalabs/common/command_submission.c    | 30 +++++++++----------
 drivers/misc/habanalabs/common/hw_queue.c     | 16 ++++++----
 drivers/misc/habanalabs/gaudi/gaudi.c         |  7 +++--
 3 files changed, 29 insertions(+), 24 deletions(-)

diff --git a/drivers/misc/habanalabs/common/command_submission.c b/drivers/misc/habanalabs/common/command_submission.c
index 94b37ab08f90..fea0f05a4954 100644
--- a/drivers/misc/habanalabs/common/command_submission.c
+++ b/drivers/misc/habanalabs/common/command_submission.c
@@ -556,7 +556,7 @@ static inline void cs_release_sob_reset_handler(struct hl_device *hdev,
 			(hl_cs_cmpl->type == CS_TYPE_COLLECTIVE_WAIT) ||
 			(!!hl_cs_cmpl->encaps_signals)) {
 		dev_dbg(hdev->dev,
-				"CS 0x%llx type %d finished, sob_id: %d, sob_val: 0x%x\n",
+				"CS 0x%llx type %d finished, sob_id: %d, sob_val: %u\n",
 				hl_cs_cmpl->cs_seq,
 				hl_cs_cmpl->type,
 				hl_cs_cmpl->hw_sob->sob_id,
@@ -1694,7 +1694,15 @@ int hl_cs_signal_sob_wraparound_handler(struct hl_device *hdev, u32 q_idx,
 			return -EINVAL;
 		}
 
-		prop->next_sob_val = count;
+		/*
+		 * next_sob_val always points to the next available signal
+		 * in the sob, so in encaps signals it will be the next one
+		 * after reserving the required amount.
+		 */
+		if (encaps_sig)
+			prop->next_sob_val = count + 1;
+		else
+			prop->next_sob_val = count;
 
 		/* only two SOBs are currently in use */
 		prop->curr_sob_offset = other_sob_offset;
@@ -1710,11 +1718,11 @@ int hl_cs_signal_sob_wraparound_handler(struct hl_device *hdev, u32 q_idx,
 		 * in addition, if we have combination of cs signal and
 		 * encaps, and at the point we need to reset the sob there was
 		 * no more reservations and only signal cs keep coming,
-		 * in such case we need to signal_cs to put the refcount and
+		 * in such case we need signal_cs to put the refcount and
 		 * reset the sob.
 		 */
 		if (other_sob->need_reset)
-			kref_put(&other_sob->kref, hl_sob_reset);
+			hw_sob_put(other_sob);
 
 		if (encaps_sig) {
 			/* set reset indication for the sob */
@@ -1946,16 +1954,6 @@ static int cs_ioctl_reserve_signals(struct hl_fpriv *hpriv,
 	 */
 	handle->pre_sob_val = prop->next_sob_val - handle->count;
 
-	/* set the hw_sob to the handle after calling the sob wraparound handler
-	 * since sob could have changed.
-	 */
-	handle->hw_sob = hw_sob;
-
-	/* store the current sob value for unreserve validity check, and
-	 * signal offset support
-	 */
-	handle->pre_sob_val = prop->next_sob_val - handle->count;
-
 	*signals_count = prop->next_sob_val;
 	hdev->asic_funcs->hw_queues_unlock(hdev);
 
@@ -1963,9 +1961,9 @@ static int cs_ioctl_reserve_signals(struct hl_fpriv *hpriv,
 	*handle_id = hdl_id;
 
 	dev_dbg(hdev->dev,
-		"Signals reserved, sob_id: %d, sob addr: 0x%x, sob val: 0x%x, q_idx: %d, hdl_id: %d\n",
+		"Signals reserved, sob_id: %d, sob addr: 0x%x, last sob_val: %u, q_idx: %d, hdl_id: %d\n",
 			hw_sob->sob_id, handle->hw_sob->sob_addr,
-			prop->next_sob_val, q_idx, hdl_id);
+			prop->next_sob_val - 1, q_idx, hdl_id);
 	goto out;
 
 remove_idr:
diff --git a/drivers/misc/habanalabs/common/hw_queue.c b/drivers/misc/habanalabs/common/hw_queue.c
index 3d9b24ce2020..6d3beccad91b 100644
--- a/drivers/misc/habanalabs/common/hw_queue.c
+++ b/drivers/misc/habanalabs/common/hw_queue.c
@@ -416,7 +416,7 @@ static int init_signal_cs(struct hl_device *hdev,
 	cs_cmpl->sob_val = prop->next_sob_val;
 
 	dev_dbg(hdev->dev,
-		"generate signal CB, sob_id: %d, sob val: 0x%x, q_idx: %d, seq: %llu\n",
+		"generate signal CB, sob_id: %d, sob val: %u, q_idx: %d, seq: %llu\n",
 		cs_cmpl->hw_sob->sob_id, cs_cmpl->sob_val, q_idx,
 		cs_cmpl->cs_seq);
 
@@ -442,9 +442,13 @@ void hl_hw_queue_encaps_sig_set_sob_info(struct hl_device *hdev,
 
 	/* Note that encaps_sig_wait_offset was validated earlier in the flow
 	 * for offset value which exceeds the max reserved signal count.
+	 * always decrement 1 of the offset since when the user
+	 * set offset 1 for example he mean to wait only for the first
+	 * signal only, which will be pre_sob_val, and if he set offset 2
+	 * then the value required is (pre_sob_val + 1) and so on...
 	 */
 	cs_cmpl->sob_val = handle->pre_sob_val +
-			(job->encaps_sig_wait_offset);
+			(job->encaps_sig_wait_offset - 1);
 }
 
 static int init_wait_cs(struct hl_device *hdev, struct hl_cs *cs,
@@ -469,7 +473,7 @@ static int init_wait_cs(struct hl_device *hdev, struct hl_cs *cs,
 		 */
 		hl_hw_queue_encaps_sig_set_sob_info(hdev, cs, job, cs_cmpl);
 
-		dev_dbg(hdev->dev, "Wait for encaps signals handle, qidx(%u), CS sequence(%llu), sob val: %u, offset: %u\n",
+		dev_dbg(hdev->dev, "Wait for encaps signals handle, qidx(%u), CS sequence(%llu), sob val: 0x%x, offset: %u\n",
 				cs->encaps_sig_hdl->q_idx,
 				cs->encaps_sig_hdl->cs_seq,
 				cs_cmpl->sob_val,
@@ -587,10 +591,12 @@ static int encaps_sig_first_staged_cs_handler
 		cs_cmpl->sob_val = encaps_sig_hdl->pre_sob_val +
 						encaps_sig_hdl->count;
 
-		dev_dbg(hdev->dev, "CS seq (%llu) added to encaps signal handler id (%u), count(%u), qidx(%u)\n",
+		dev_dbg(hdev->dev, "CS seq (%llu) added to encaps signal handler id (%u), count(%u), qidx(%u), sob(%u), val(%u)\n",
 				cs->sequence, encaps_sig_hdl->id,
 				encaps_sig_hdl->count,
-				encaps_sig_hdl->q_idx);
+				encaps_sig_hdl->q_idx,
+				cs_cmpl->hw_sob->sob_id,
+				cs_cmpl->sob_val);
 
 	} else {
 		dev_err(hdev->dev, "encaps handle id(%u) wasn't found!\n",
diff --git a/drivers/misc/habanalabs/gaudi/gaudi.c b/drivers/misc/habanalabs/gaudi/gaudi.c
index 529ccd8b8911..f134cd7a6798 100644
--- a/drivers/misc/habanalabs/gaudi/gaudi.c
+++ b/drivers/misc/habanalabs/gaudi/gaudi.c
@@ -1223,9 +1223,10 @@ static void gaudi_collective_slave_init_job(struct hl_device *hdev,
 		hl_hw_queue_encaps_sig_set_sob_info(hdev, job->cs, job,
 						cs_cmpl);
 
-		dev_dbg(hdev->dev, "collective wait: Sequence %llu found, sob val incremented to: %u\n",
+		dev_dbg(hdev->dev, "collective wait: Sequence %llu found, sob_id: %u,  wait for sob_val: %u\n",
 				job->cs->sequence,
-				job->cs->encaps_sig_hdl->pre_sob_val);
+				cs_cmpl->hw_sob->sob_id,
+				cs_cmpl->sob_val);
 	}
 
 	/* Add to wait CBs using slave monitor */
@@ -1238,7 +1239,7 @@ static void gaudi_collective_slave_init_job(struct hl_device *hdev,
 	wait_prop.size = cb_size;
 
 	dev_dbg(hdev->dev,
-		"Generate slave wait CB, sob %d, val:0x%x, mon %d, q %d\n",
+		"Generate slave wait CB, sob %d, val:%x, mon %d, q %d\n",
 		cs_cmpl->hw_sob->sob_id, cs_cmpl->sob_val,
 		prop->collective_slave_mon_id, queue_id);
 
-- 
2.25.1

[PATCH 2/4] habanalabs: remove redundant warning message

[Oded Gabbay]

This warning is redundant as we will print a notice in case the device
is still in use after the FD was closed. No need to print the same
message per context.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
---
 drivers/misc/habanalabs/common/context.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/drivers/misc/habanalabs/common/context.c b/drivers/misc/habanalabs/common/context.c
index abbba4194d3b..cd06852aa17c 100644
--- a/drivers/misc/habanalabs/common/context.c
+++ b/drivers/misc/habanalabs/common/context.c
@@ -184,9 +184,6 @@ void hl_ctx_free(struct hl_device *hdev, struct hl_ctx *ctx)
 {
 	if (kref_put(&ctx->refcount, hl_ctx_do_release) == 1)
 		return;
-
-	dev_warn(hdev->dev,
-		"user process released device but its command submissions are still executing\n");
 }
 
 int hl_ctx_init(struct hl_device *hdev, struct hl_ctx *ctx, bool is_kernel_ctx)
-- 
2.25.1

[PATCH 3/4] habanalabs: expose server type in INFO IOCTL

[Oded Gabbay]

Add the server type property to the hl_info_hw_ip_info structure
that is exposed to the user via the INFO IOCTL.

This is needed by the userspace s/w stack to know the connections map
of the internal links that connect the ASIC among themselves inside the
server.

The F/W will tell us, as part of the NIC information, the server type
that the GAUDI is located in.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
---
 drivers/misc/habanalabs/common/firmware_if.c  |  2 +-
 drivers/misc/habanalabs/common/habanalabs.h   |  3 ++
 .../misc/habanalabs/common/habanalabs_ioctl.c |  2 +
 drivers/misc/habanalabs/gaudi/gaudi.c         |  2 +
 drivers/misc/habanalabs/goya/goya.c           |  2 +
 .../misc/habanalabs/include/common/cpucp_if.h | 11 +++++
 include/uapi/misc/habanalabs.h                | 49 +++++++++++++++++--
 7 files changed, 66 insertions(+), 5 deletions(-)

diff --git a/drivers/misc/habanalabs/common/firmware_if.c b/drivers/misc/habanalabs/common/firmware_if.c
index 869c6057ae31..0be3f5414f0b 100644
--- a/drivers/misc/habanalabs/common/firmware_if.c
+++ b/drivers/misc/habanalabs/common/firmware_if.c
@@ -1,7 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0
 
 /*
- * Copyright 2016-2019 HabanaLabs, Ltd.
+ * Copyright 2016-2021 HabanaLabs, Ltd.
  * All Rights Reserved.
  */
 
diff --git a/drivers/misc/habanalabs/common/habanalabs.h b/drivers/misc/habanalabs/common/habanalabs.h
index 1ea8e70a5a34..6fc2234576de 100644
--- a/drivers/misc/habanalabs/common/habanalabs.h
+++ b/drivers/misc/habanalabs/common/habanalabs.h
@@ -489,6 +489,8 @@ struct hl_hints_range {
  *                                       reserved for the user
  * @first_available_cq: first available CQ for the user.
  * @user_interrupt_count: number of user interrupts.
+ * @server_type: Server type that the ASIC is currently installed in.
+ *               The value is according to enum hl_server_type in uapi file.
  * @tpc_enabled_mask: which TPCs are enabled.
  * @completion_queues_count: number of completion queues.
  * @fw_security_enabled: true if security measures are enabled in firmware,
@@ -570,6 +572,7 @@ struct asic_fixed_properties {
 	u16				first_available_user_msix_interrupt;
 	u16				first_available_cq[HL_MAX_DCORES];
 	u16				user_interrupt_count;
+	u16				server_type;
 	u8				tpc_enabled_mask;
 	u8				completion_queues_count;
 	u8				fw_security_enabled;
diff --git a/drivers/misc/habanalabs/common/habanalabs_ioctl.c b/drivers/misc/habanalabs/common/habanalabs_ioctl.c
index f4dda7b4acdd..86c3257d9ae1 100644
--- a/drivers/misc/habanalabs/common/habanalabs_ioctl.c
+++ b/drivers/misc/habanalabs/common/habanalabs_ioctl.c
@@ -94,6 +94,8 @@ static int hw_ip_info(struct hl_device *hdev, struct hl_info_args *args)
 
 	hw_ip.first_available_interrupt_id =
 			prop->first_available_user_msix_interrupt;
+	hw_ip.server_type = prop->server_type;
+
 	return copy_to_user(out, &hw_ip,
 		min((size_t) size, sizeof(hw_ip))) ? -EFAULT : 0;
 }
diff --git a/drivers/misc/habanalabs/gaudi/gaudi.c b/drivers/misc/habanalabs/gaudi/gaudi.c
index f134cd7a6798..4bb9970db46a 100644
--- a/drivers/misc/habanalabs/gaudi/gaudi.c
+++ b/drivers/misc/habanalabs/gaudi/gaudi.c
@@ -645,6 +645,8 @@ static int gaudi_set_fixed_properties(struct hl_device *hdev)
 	prop->hard_reset_done_by_fw = false;
 	prop->gic_interrupts_enable = true;
 
+	prop->server_type = HL_SERVER_TYPE_UNKNOWN;
+
 	return 0;
 }
 
diff --git a/drivers/misc/habanalabs/goya/goya.c b/drivers/misc/habanalabs/goya/goya.c
index c070cd14753e..8d890c7cce08 100644
--- a/drivers/misc/habanalabs/goya/goya.c
+++ b/drivers/misc/habanalabs/goya/goya.c
@@ -469,6 +469,8 @@ int goya_set_fixed_properties(struct hl_device *hdev)
 	prop->hard_reset_done_by_fw = false;
 	prop->gic_interrupts_enable = true;
 
+	prop->server_type = HL_SERVER_TYPE_UNKNOWN;
+
 	return 0;
 }
 
diff --git a/drivers/misc/habanalabs/include/common/cpucp_if.h b/drivers/misc/habanalabs/include/common/cpucp_if.h
index 80b1d5a9d9f1..2d6f8ea35375 100644
--- a/drivers/misc/habanalabs/include/common/cpucp_if.h
+++ b/drivers/misc/habanalabs/include/common/cpucp_if.h
@@ -700,6 +700,15 @@ struct cpucp_mac_addr {
 	__u8 mac_addr[ETH_ALEN];
 };
 
+enum cpucp_serdes_type {
+	TYPE_1_SERDES_TYPE,
+	TYPE_2_SERDES_TYPE,
+	HLS1_SERDES_TYPE,
+	HLS1H_SERDES_TYPE,
+	UNKNOWN_SERDES_TYPE,
+	MAX_NUM_SERDES_TYPE = UNKNOWN_SERDES_TYPE
+};
+
 struct cpucp_nic_info {
 	struct cpucp_mac_addr mac_addrs[CPUCP_MAX_NICS];
 	__le64 link_mask[CPUCP_NIC_MASK_ARR_LEN];
@@ -708,6 +717,8 @@ struct cpucp_nic_info {
 	__le64 link_ext_mask[CPUCP_NIC_MASK_ARR_LEN];
 	__u8 qsfp_eeprom[CPUCP_NIC_QSFP_EEPROM_MAX_LEN];
 	__le64 auto_neg_mask[CPUCP_NIC_MASK_ARR_LEN];
+	__le16 serdes_type; /* enum cpucp_serdes_type */
+	__u8 reserved[6];
 };
 
 #endif /* CPUCP_IF_H */
diff --git a/include/uapi/misc/habanalabs.h b/include/uapi/misc/habanalabs.h
index eca86c545916..b877a5b7ee74 100644
--- a/include/uapi/misc/habanalabs.h
+++ b/include/uapi/misc/habanalabs.h
@@ -279,6 +279,14 @@ enum hl_device_status {
 	HL_DEVICE_STATUS_NEEDS_RESET
 };
 
+enum hl_server_type {
+	HL_SERVER_TYPE_UNKNOWN = 0,
+	HL_SERVER_GAUDI_HLS1 = 1,
+	HL_SERVER_GAUDI_HLS1H = 2,
+	HL_SERVER_GAUDI_TYPE1 = 3,
+	HL_SERVER_GAUDI_TYPE2 = 4
+};
+
 /* Opcode for management ioctl
  *
  * HW_IP_INFO            - Receive information about different IP blocks in the
@@ -337,17 +345,50 @@ enum hl_device_status {
 #define HL_INFO_VERSION_MAX_LEN	128
 #define HL_INFO_CARD_NAME_MAX_LEN	16
 
+/**
+ * struct hl_info_hw_ip_info - hardware information on various IPs in the ASIC
+ * @sram_base_address: The first SRAM physical base address that is free to be
+ *                     used by the user.
+ * @dram_base_address: The first DRAM virtual or physical base address that is
+ *                     free to be used by the user.
+ * @dram_size: The DRAM size that is available to the user.
+ * @sram_size: The SRAM size that is available to the user.
+ * @num_of_events: The number of events that can be received from the f/w. This
+ *                 is needed so the user can what is the size of the h/w events
+ *                 array he needs to pass to the kernel when he wants to fetch
+ *                 the event counters.
+ * @device_id: PCI device ID of the ASIC.
+ * @module_id: Module ID of the ASIC for mezzanine cards in servers
+ *             (From OCP spec).
+ * @first_available_interrupt_id: The first available interrupt ID for the user
+ *                                to be used when it works with user interrupts.
+ *                                Relevant for Gaudi2 and later.
+ * @server_type: Server type that the Gaudi ASIC is currently installed in.
+ *               The value is according to enum hl_server_type
+ * @cpld_version: CPLD version on the board.
+ * @psoc_pci_pll_nr: PCI PLL NR value. Needed by the profiler in some ASICs.
+ * @psoc_pci_pll_nf: PCI PLL NF value. Needed by the profiler in some ASICs.
+ * @psoc_pci_pll_od: PCI PLL OD value. Needed by the profiler in some ASICs.
+ * @psoc_pci_pll_div_factor: PCI PLL DIV factor value. Needed by the profiler
+ *                           in some ASICs.
+ * @tpc_enabled_mask: Bit-mask that represents which TPCs are enabled. Relevant
+ *                    for Goya/Gaudi only.
+ * @dram_enabled: Whether the DRAM is enabled.
+ * @cpucp_version: The CPUCP f/w version.
+ * @card_name: The card name as passed by the f/w.
+ * @dram_page_size: The DRAM physical page size.
+ */
 struct hl_info_hw_ip_info {
 	__u64 sram_base_address;
 	__u64 dram_base_address;
 	__u64 dram_size;
 	__u32 sram_size;
 	__u32 num_of_events;
-	__u32 device_id; /* PCI Device ID */
-	__u32 module_id; /* For mezzanine cards in servers (From OCP spec.) */
+	__u32 device_id;
+	__u32 module_id;
 	__u32 reserved;
 	__u16 first_available_interrupt_id;
-	__u16 reserved2;
+	__u16 server_type;
 	__u32 cpld_version;
 	__u32 psoc_pci_pll_nr;
 	__u32 psoc_pci_pll_nf;
@@ -358,7 +399,7 @@ struct hl_info_hw_ip_info {
 	__u8 pad[2];
 	__u8 cpucp_version[HL_INFO_VERSION_MAX_LEN];
 	__u8 card_name[HL_INFO_CARD_NAME_MAX_LEN];
-	__u64 reserved3;
+	__u64 reserved2;
 	__u64 dram_page_size;
 };
 
-- 
2.25.1

[PATCH 4/4] habanalabs: convert PCI BAR offset to u64

[Oded Gabbay]

From: Ohad Sharabi <osharabi@habana.ai>

Done as the bar size can exceed 4GB.

Signed-off-by: Ohad Sharabi <osharabi@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
---
 drivers/misc/habanalabs/common/habanalabs.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/misc/habanalabs/common/habanalabs.h b/drivers/misc/habanalabs/common/habanalabs.h
index 6fc2234576de..815d7e1a7e6d 100644
--- a/drivers/misc/habanalabs/common/habanalabs.h
+++ b/drivers/misc/habanalabs/common/habanalabs.h
@@ -931,7 +931,7 @@ struct pci_mem_region {
 	u64 region_base;
 	u64 region_size;
 	u64 bar_size;
-	u32 offset_in_bar;
+	u64 offset_in_bar;
 	u8 bar_id;
 	u8 used;
 };
-- 
2.25.1