diff --git a/drivers/net/ethernet/chelsio/cxgb4/cudbg_entity.h b/drivers/net/ethernet/chelsio/cxgb4/cudbg_entity.h
index f99db7b283fcb75365cea229e8bfca13efdd856d..6056899574963812598ec2eb69b8738b3a04e8a4 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cudbg_entity.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/cudbg_entity.h
@@ -145,6 +145,14 @@ struct cudbg_tid_info_region_rev1 {
 	u32 reserved[16];
 };
 
+#define CUDBG_MAX_FL_QIDS 1024
+
+struct cudbg_ch_cntxt {
+	u32 cntxt_type;
+	u32 cntxt_id;
+	u32 data[SGE_CTXT_SIZE / 4];
+};
+
 #define CUDBG_MAX_RPLC_SIZE 128
 
 struct cudbg_mps_tcam {
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cudbg_if.h b/drivers/net/ethernet/chelsio/cxgb4/cudbg_if.h
index 4e5d189eae62dd05ead0081ae2fb0bf6ddf92672..e10ff1ee62c5341c1ae122e69b183665c272ddde 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cudbg_if.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/cudbg_if.h
@@ -63,6 +63,7 @@ enum cudbg_dbg_entity_type {
 	CUDBG_PCIE_INDIRECT = 50,
 	CUDBG_PM_INDIRECT = 51,
 	CUDBG_TID_INFO = 54,
+	CUDBG_DUMP_CONTEXT = 56,
 	CUDBG_MPS_TCAM = 57,
 	CUDBG_VPD_DATA = 58,
 	CUDBG_LE_TCAM = 59,
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib.c b/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib.c
index dd7e26be98cfc2d3c23fc9a5486c0c04268710ad..d699bf88d18f8ebbfe84186fe5036fc2b719e5f1 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib.c
@@ -1115,6 +1115,84 @@ int cudbg_collect_tid(struct cudbg_init *pdbg_init,
 	return rc;
 }
 
+int cudbg_dump_context_size(struct adapter *padap)
+{
+	u32 value, size;
+	u8 flq;
+
+	value = t4_read_reg(padap, SGE_FLM_CFG_A);
+
+	/* Get number of data freelist queues */
+	flq = HDRSTARTFLQ_G(value);
+	size = CUDBG_MAX_FL_QIDS >> flq;
+
+	/* Add extra space for congestion manager contexts.
+	 * The number of CONM contexts are same as number of freelist
+	 * queues.
+	 */
+	size += size;
+	return size * sizeof(struct cudbg_ch_cntxt);
+}
+
+static void cudbg_read_sge_ctxt(struct cudbg_init *pdbg_init, u32 cid,
+				enum ctxt_type ctype, u32 *data)
+{
+	struct adapter *padap = pdbg_init->adap;
+	int rc = -1;
+
+	/* Under heavy traffic, the SGE Queue contexts registers will be
+	 * frequently accessed by firmware.
+	 *
+	 * To avoid conflicts with firmware, always ask firmware to fetch
+	 * the SGE Queue contexts via mailbox. On failure, fallback to
+	 * accessing hardware registers directly.
+	 */
+	if (is_fw_attached(pdbg_init))
+		rc = t4_sge_ctxt_rd(padap, padap->mbox, cid, ctype, data);
+	if (rc)
+		t4_sge_ctxt_rd_bd(padap, cid, ctype, data);
+}
+
+int cudbg_collect_dump_context(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err)
+{
+	struct adapter *padap = pdbg_init->adap;
+	struct cudbg_buffer temp_buff = { 0 };
+	struct cudbg_ch_cntxt *buff;
+	u32 size, i = 0;
+	int rc;
+
+	rc = cudbg_dump_context_size(padap);
+	if (rc <= 0)
+		return CUDBG_STATUS_ENTITY_NOT_FOUND;
+
+	size = rc;
+	rc = cudbg_get_buff(dbg_buff, size, &temp_buff);
+	if (rc)
+		return rc;
+
+	buff = (struct cudbg_ch_cntxt *)temp_buff.data;
+	while (size > 0) {
+		buff->cntxt_type = CTXT_FLM;
+		buff->cntxt_id = i;
+		cudbg_read_sge_ctxt(pdbg_init, i, CTXT_FLM, buff->data);
+		buff++;
+		size -= sizeof(struct cudbg_ch_cntxt);
+
+		buff->cntxt_type = CTXT_CNM;
+		buff->cntxt_id = i;
+		cudbg_read_sge_ctxt(pdbg_init, i, CTXT_CNM, buff->data);
+		buff++;
+		size -= sizeof(struct cudbg_ch_cntxt);
+
+		i++;
+	}
+
+	cudbg_write_and_release_buff(&temp_buff, dbg_buff);
+	return rc;
+}
+
 static inline void cudbg_tcamxy2valmask(u64 x, u64 y, u8 *addr, u64 *mask)
 {
 	*mask = x | y;
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib.h b/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib.h
index ebb2d9907fc9bfebdd74805aba83414221d352f2..caeee8e33e867dfc60bb01e97e14e24de22d5285 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib.h
@@ -123,6 +123,9 @@ int cudbg_collect_pm_indirect(struct cudbg_init *pdbg_init,
 int cudbg_collect_tid(struct cudbg_init *pdbg_init,
 		      struct cudbg_buffer *dbg_buff,
 		      struct cudbg_error *cudbg_err);
+int cudbg_collect_dump_context(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err);
 int cudbg_collect_mps_tcam(struct cudbg_init *pdbg_init,
 			   struct cudbg_buffer *dbg_buff,
 			   struct cudbg_error *cudbg_err);
@@ -158,6 +161,7 @@ struct cudbg_entity_hdr *cudbg_get_entity_hdr(void *outbuf, int i);
 void cudbg_align_debug_buffer(struct cudbg_buffer *dbg_buff,
 			      struct cudbg_entity_hdr *entity_hdr);
 u32 cudbg_cim_obq_size(struct adapter *padap, int qid);
+int cudbg_dump_context_size(struct adapter *padap);
 
 struct cudbg_tcam;
 void cudbg_fill_le_tcam_info(struct adapter *padap,
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
index 0de1a4b2223e07bf67cec5166ac17c877a0f8a67..6f9fa6e3c42a0cc1988e82ae976cce17d45ca0e5 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
@@ -1670,6 +1670,10 @@ int t4_fwaddrspace_write(struct adapter *adap, unsigned int mbox,
 void t4_read_pace_tbl(struct adapter *adap, unsigned int pace_vals[NTX_SCHED]);
 void t4_get_tx_sched(struct adapter *adap, unsigned int sched,
 		     unsigned int *kbps, unsigned int *ipg, bool sleep_ok);
+int t4_sge_ctxt_rd(struct adapter *adap, unsigned int mbox, unsigned int cid,
+		   enum ctxt_type ctype, u32 *data);
+int t4_sge_ctxt_rd_bd(struct adapter *adap, unsigned int cid,
+		      enum ctxt_type ctype, u32 *data);
 int t4_sched_params(struct adapter *adapter, int type, int level, int mode,
 		    int rateunit, int ratemode, int channel, int class,
 		    int minrate, int maxrate, int weight, int pktsize);
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_cudbg.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_cudbg.c
index 05eb2d2ef5920502d63c1cc86b1fdcfe591e2935..29cc625e9833a456b14d61224580da187afc0972 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_cudbg.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_cudbg.c
@@ -60,6 +60,7 @@ static const struct cxgb4_collect_entity cxgb4_collect_hw_dump[] = {
 	{ CUDBG_PCIE_INDIRECT, cudbg_collect_pcie_indirect },
 	{ CUDBG_PM_INDIRECT, cudbg_collect_pm_indirect },
 	{ CUDBG_TID_INFO, cudbg_collect_tid },
+	{ CUDBG_DUMP_CONTEXT, cudbg_collect_dump_context },
 	{ CUDBG_MPS_TCAM, cudbg_collect_mps_tcam },
 	{ CUDBG_VPD_DATA, cudbg_collect_vpd_data },
 	{ CUDBG_LE_TCAM, cudbg_collect_le_tcam },
@@ -218,6 +219,9 @@ static u32 cxgb4_get_entity_length(struct adapter *adap, u32 entity)
 	case CUDBG_TID_INFO:
 		len = sizeof(struct cudbg_tid_info_region_rev1);
 		break;
+	case CUDBG_DUMP_CONTEXT:
+		len = cudbg_dump_context_size(adap);
+		break;
 	case CUDBG_MPS_TCAM:
 		len = sizeof(struct cudbg_mps_tcam) *
 		      adap->params.arch.mps_tcam_size;
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
index b4fad081ac782e25b9cf2c6e3ba4dce98f7ba3ed..f63210f155796c9809415d7da54f59086ebdbf7c 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
@@ -9647,6 +9647,68 @@ void t4_get_tx_sched(struct adapter *adap, unsigned int sched,
 	}
 }
 
+/* t4_sge_ctxt_rd - read an SGE context through FW
+ * @adap: the adapter
+ * @mbox: mailbox to use for the FW command
+ * @cid: the context id
+ * @ctype: the context type
+ * @data: where to store the context data
+ *
+ * Issues a FW command through the given mailbox to read an SGE context.
+ */
+int t4_sge_ctxt_rd(struct adapter *adap, unsigned int mbox, unsigned int cid,
+		   enum ctxt_type ctype, u32 *data)
+{
+	struct fw_ldst_cmd c;
+	int ret;
+
+	if (ctype == CTXT_FLM)
+		ret = FW_LDST_ADDRSPC_SGE_FLMC;
+	else
+		ret = FW_LDST_ADDRSPC_SGE_CONMC;
+
+	memset(&c, 0, sizeof(c));
+	c.op_to_addrspace = cpu_to_be32(FW_CMD_OP_V(FW_LDST_CMD) |
+					FW_CMD_REQUEST_F | FW_CMD_READ_F |
+					FW_LDST_CMD_ADDRSPACE_V(ret));
+	c.cycles_to_len16 = cpu_to_be32(FW_LEN16(c));
+	c.u.idctxt.physid = cpu_to_be32(cid);
+
+	ret = t4_wr_mbox(adap, mbox, &c, sizeof(c), &c);
+	if (ret == 0) {
+		data[0] = be32_to_cpu(c.u.idctxt.ctxt_data0);
+		data[1] = be32_to_cpu(c.u.idctxt.ctxt_data1);
+		data[2] = be32_to_cpu(c.u.idctxt.ctxt_data2);
+		data[3] = be32_to_cpu(c.u.idctxt.ctxt_data3);
+		data[4] = be32_to_cpu(c.u.idctxt.ctxt_data4);
+		data[5] = be32_to_cpu(c.u.idctxt.ctxt_data5);
+	}
+	return ret;
+}
+
+/**
+ * t4_sge_ctxt_rd_bd - read an SGE context bypassing FW
+ * @adap: the adapter
+ * @cid: the context id
+ * @ctype: the context type
+ * @data: where to store the context data
+ *
+ * Reads an SGE context directly, bypassing FW.  This is only for
+ * debugging when FW is unavailable.
+ */
+int t4_sge_ctxt_rd_bd(struct adapter *adap, unsigned int cid,
+		      enum ctxt_type ctype, u32 *data)
+{
+	int i, ret;
+
+	t4_write_reg(adap, SGE_CTXT_CMD_A, CTXTQID_V(cid) | CTXTTYPE_V(ctype));
+	ret = t4_wait_op_done(adap, SGE_CTXT_CMD_A, BUSY_F, 0, 3, 1);
+	if (!ret)
+		for (i = SGE_CTXT_DATA0_A; i <= SGE_CTXT_DATA5_A; i += 4)
+			*data++ = t4_read_reg(adap, i);
+	return ret;
+}
+
 int t4_sched_params(struct adapter *adapter, int type, int level, int mode,
 		    int rateunit, int ratemode, int channel, int class,
 		    int minrate, int maxrate, int weight, int pktsize)
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.h b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.h
index 7c6af14905c2dd3257d29f67a2cf8b211cd3c7d0..a964ed18435615b0bba94974601ce8f736560f1f 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.h
@@ -68,6 +68,12 @@ enum {
 	ULPRX_LA_SIZE  = 512,   /* # of 256-bit words in ULP_RX LA */
 };
 
+/* SGE context types */
+enum ctxt_type {
+	CTXT_FLM = 2,
+	CTXT_CNM,
+};
+
 enum {
 	SF_PAGE_SIZE = 256,           /* serial flash page size */
 	SF_SEC_SIZE = 64 * 1024,      /* serial flash sector size */
@@ -79,6 +85,7 @@ enum { MBOX_OWNER_NONE, MBOX_OWNER_FW, MBOX_OWNER_DRV };    /* mailbox owners */
 
 enum {
 	SGE_MAX_WR_LEN = 512,     /* max WR size in bytes */
+	SGE_CTXT_SIZE = 24,       /* size of SGE context */
 	SGE_NTIMERS = 6,          /* # of interrupt holdoff timer values */
 	SGE_NCOUNTERS = 4,        /* # of interrupt packet counter values */
 	SGE_MAX_IQ_SIZE = 65520,
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h b/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h
index f5576ce004fab0aec2c394ebbc4db8258f709d37..a7cfece728282ca96db0aac65266b3146f32d563 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h
@@ -153,6 +153,23 @@
 #define T6_DBVFIFO_SIZE_M    0x1fffU
 #define T6_DBVFIFO_SIZE_G(x) (((x) >> T6_DBVFIFO_SIZE_S) & T6_DBVFIFO_SIZE_M)
 
+#define SGE_CTXT_CMD_A 0x11fc
+
+#define BUSY_S    31
+#define BUSY_V(x) ((x) << BUSY_S)
+#define BUSY_F    BUSY_V(1U)
+
+#define CTXTTYPE_S    24
+#define CTXTTYPE_M    0x3U
+#define CTXTTYPE_V(x) ((x) << CTXTTYPE_S)
+
+#define CTXTQID_S    0
+#define CTXTQID_M    0x1ffffU
+#define CTXTQID_V(x) ((x) << CTXTQID_S)
+
+#define SGE_CTXT_DATA0_A 0x1200
+#define SGE_CTXT_DATA5_A 0x1214
+
 #define GLOBALENABLE_S    0
 #define GLOBALENABLE_V(x) ((x) << GLOBALENABLE_S)
 #define GLOBALENABLE_F    GLOBALENABLE_V(1U)
@@ -322,6 +339,16 @@
 
 #define SGE_IMSG_CTXT_BADDR_A 0x1088
 #define SGE_FLM_CACHE_BADDR_A 0x108c
+#define SGE_FLM_CFG_A 0x1090
+
+#define NOHDR_S    18
+#define NOHDR_V(x) ((x) << NOHDR_S)
+#define NOHDR_F    NOHDR_V(1U)
+
+#define HDRSTARTFLQ_S    11
+#define HDRSTARTFLQ_M    0x7U
+#define HDRSTARTFLQ_G(x) (((x) >> HDRSTARTFLQ_S) & HDRSTARTFLQ_M)
+
 #define SGE_INGRESS_RX_THRESHOLD_A 0x10a0
 
 #define THRESHOLD_0_S    24