1 files changed, 149 insertions, 24 deletions

diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npa.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npa.c
index c0e165dfc403..e2a33e46b48a 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npa.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npa.c
@@ -1,13 +1,10 @@
 // SPDX-License-Identifier: GPL-2.0
-/* Marvell OcteonTx2 RVU Admin Function driver
+/* Marvell RVU Admin Function driver
  *
- * Copyright (C) 2018 Marvell International Ltd.
+ * Copyright (C) 2018 Marvell.
  *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License version 2 as
- * published by the Free Software Foundation.
  */
-
+#include <linux/bitfield.h>
 #include <linux/module.h>
 #include <linux/pci.h>
 
@@ -45,15 +42,24 @@ static int npa_aq_enqueue_wait(struct rvu *rvu, struct rvu_block *block,
 			return -EBUSY;
 	}
 
-	if (result->compcode != NPA_AQ_COMP_GOOD)
+	if (result->compcode != NPA_AQ_COMP_GOOD) {
 		/* TODO: Replace this with some error code */
+		if (result->compcode == NPA_AQ_COMP_CTX_FAULT ||
+		    result->compcode == NPA_AQ_COMP_LOCKERR ||
+		    result->compcode == NPA_AQ_COMP_CTX_POISON) {
+			if (rvu_ndc_fix_locked_cacheline(rvu, BLKADDR_NDC_NPA0))
+				dev_err(rvu->dev,
+					"%s: Not able to unlock cachelines\n", __func__);
+		}
+
 		return -EBUSY;
+	}
 
 	return 0;
 }
 
-static int rvu_npa_aq_enq_inst(struct rvu *rvu, struct npa_aq_enq_req *req,
-			       struct npa_aq_enq_rsp *rsp)
+int rvu_npa_aq_enq_inst(struct rvu *rvu, struct npa_aq_enq_req *req,
+			struct npa_aq_enq_rsp *rsp)
 {
 	struct rvu_hwinfo *hw = rvu->hw;
 	u16 pcifunc = req->hdr.pcifunc;
@@ -94,6 +100,11 @@ static int rvu_npa_aq_enq_inst(struct rvu *rvu, struct npa_aq_enq_req *req,
 	 */
 	inst.res_addr = (u64)aq->res->iova;
 
+	/* Hardware uses same aq->res->base for updating result of
+	 * previous instruction hence wait here till it is done.
+	 */
+	spin_lock(&aq->lock);
+
 	/* Clean result + context memory */
 	memset(aq->res->base, 0, aq->res->entry_sz);
 	/* Context needs to be written at RES_ADDR + 128 */
@@ -138,10 +149,10 @@ static int rvu_npa_aq_enq_inst(struct rvu *rvu, struct npa_aq_enq_req *req,
 		break;
 	}
 
-	if (rc)
+	if (rc) {
+		spin_unlock(&aq->lock);
 		return rc;
-
-	spin_lock(&aq->lock);
+	}
 
 	/* Submit the instruction to AQ */
 	rc = npa_aq_enqueue_wait(rvu, block, &inst);
@@ -218,6 +229,8 @@ static int npa_lf_hwctx_disable(struct rvu *rvu, struct hwctx_disable_req *req)
 	} else if (req->ctype == NPA_AQ_CTYPE_AURA) {
 		aq_req.aura.ena = 0;
 		aq_req.aura_mask.ena = 1;
+		aq_req.aura.bp_ena = 0;
+		aq_req.aura_mask.bp_ena = 1;
 		cnt = pfvf->aura_ctx->qsize;
 		bmap = pfvf->aura_bmap;
 	}
@@ -241,12 +254,50 @@ static int npa_lf_hwctx_disable(struct rvu *rvu, struct hwctx_disable_req *req)
 	return err;
 }
 
+#ifdef CONFIG_NDC_DIS_DYNAMIC_CACHING
+static int npa_lf_hwctx_lockdown(struct rvu *rvu, struct npa_aq_enq_req *req)
+{
+	struct npa_aq_enq_req lock_ctx_req;
+	int err;
+
+	if (req->op != NPA_AQ_INSTOP_INIT)
+		return 0;
+
+	memset(&lock_ctx_req, 0, sizeof(struct npa_aq_enq_req));
+	lock_ctx_req.hdr.pcifunc = req->hdr.pcifunc;
+	lock_ctx_req.ctype = req->ctype;
+	lock_ctx_req.op = NPA_AQ_INSTOP_LOCK;
+	lock_ctx_req.aura_id = req->aura_id;
+	err = rvu_npa_aq_enq_inst(rvu, &lock_ctx_req, NULL);
+	if (err)
+		dev_err(rvu->dev,
+			"PFUNC 0x%x: Failed to lock NPA context %s:%d\n",
+			req->hdr.pcifunc,
+			(req->ctype == NPA_AQ_CTYPE_AURA) ?
+			"Aura" : "Pool", req->aura_id);
+	return err;
+}
+
+int rvu_mbox_handler_npa_aq_enq(struct rvu *rvu,
+				struct npa_aq_enq_req *req,
+				struct npa_aq_enq_rsp *rsp)
+{
+	int err;
+
+	err = rvu_npa_aq_enq_inst(rvu, req, rsp);
+	if (!err)
+		err = npa_lf_hwctx_lockdown(rvu, req);
+	return err;
+}
+#else
+
 int rvu_mbox_handler_npa_aq_enq(struct rvu *rvu,
 				struct npa_aq_enq_req *req,
 				struct npa_aq_enq_rsp *rsp)
 {
 	return rvu_npa_aq_enq_inst(rvu, req, rsp);
 }
+#endif
 
 int rvu_mbox_handler_npa_hwctx_disable(struct rvu *rvu,
 				       struct hwctx_disable_req *req,
@@ -289,6 +340,9 @@ int rvu_mbox_handler_npa_lf_alloc(struct rvu *rvu,
 	    req->aura_sz == NPA_AURA_SZ_0 || !req->nr_pools)
 		return NPA_AF_ERR_PARAM;
 
+	if (req->way_mask)
+		req->way_mask &= 0xFFFF;
+
 	pfvf = rvu_get_pfvf(rvu, pcifunc);
 	blkaddr = rvu_get_blkaddr(rvu, BLKTYPE_NPA, pcifunc);
 	if (!pfvf->npalf || blkaddr < 0)
@@ -345,7 +399,8 @@ int rvu_mbox_handler_npa_lf_alloc(struct rvu *rvu,
 	/* Clear way partition mask and set aura offset to '0' */
 	cfg &= ~(BIT_ULL(34) - 1);
 	/* Set aura size & enable caching of contexts */
-	cfg |= (req->aura_sz << 16) | BIT_ULL(34);
+	cfg |= (req->aura_sz << 16) | BIT_ULL(34) | req->way_mask;
+
 	rvu_write64(rvu, blkaddr, NPA_AF_LFX_AURAS_CFG(npalf), cfg);
 
 	/* Configure aura HW context's base */
@@ -353,7 +408,8 @@ int rvu_mbox_handler_npa_lf_alloc(struct rvu *rvu,
 		    (u64)pfvf->aura_ctx->iova);
 
 	/* Enable caching of qints hw context */
-	rvu_write64(rvu, blkaddr, NPA_AF_LFX_QINTS_CFG(npalf), BIT_ULL(36));
+	rvu_write64(rvu, blkaddr, NPA_AF_LFX_QINTS_CFG(npalf),
+		    BIT_ULL(36) | req->way_mask << 20);
 	rvu_write64(rvu, blkaddr, NPA_AF_LFX_QINTS_BASE(npalf),
 		    (u64)pfvf->npa_qints_ctx->iova);
 
@@ -369,6 +425,10 @@ exit:
 	rsp->stack_pg_ptrs = (cfg >> 8) & 0xFF;
 	rsp->stack_pg_bytes = cfg & 0xFF;
 	rsp->qints = (cfg >> 28) & 0xFFF;
+	if (!is_rvu_otx2(rvu)) {
+		cfg = rvu_read64(rvu, block->addr, NPA_AF_BATCH_CTL);
+		rsp->cache_lines = (cfg >> 1) & 0x3F;
+	}
 	return rc;
 }
 
@@ -404,6 +464,23 @@ int rvu_mbox_handler_npa_lf_free(struct rvu *rvu, struct msg_req *req,
 	return 0;
 }
 
+static void npa_aq_ndc_config(struct rvu *rvu, struct rvu_block *block)
+{
+	u64 cfg;
+
+	if (is_cn20k(rvu->pdev)) /* NDC not applicable to cn20k */
+		return;
+
+	/* Do not bypass NDC cache */
+	cfg = rvu_read64(rvu, block->addr, NPA_AF_NDC_CFG);
+	cfg &= ~0x03DULL;
+#ifdef CONFIG_NDC_DIS_DYNAMIC_CACHING
+	/* Disable caching of stack pages */
+	cfg |= 0x10ULL;
+#endif
+	rvu_write64(rvu, block->addr, NPA_AF_NDC_CFG, cfg);
+}
+
 static int npa_aq_init(struct rvu *rvu, struct rvu_block *block)
 {
 	u64 cfg;
@@ -419,11 +496,15 @@ static int npa_aq_init(struct rvu *rvu, struct rvu_block *block)
 	rvu_write64(rvu, block->addr, NPA_AF_GEN_CFG, cfg);
 #endif
 
-	/* Do not bypass NDC cache */
-	cfg = rvu_read64(rvu, block->addr, NPA_AF_NDC_CFG);
-	cfg &= ~0x03DULL;
-	rvu_write64(rvu, block->addr, NPA_AF_NDC_CFG, cfg);
+	npa_aq_ndc_config(rvu, block);
 
+	/* For CN10K NPA BATCH DMA set 35 cache lines */
+	if (!is_rvu_otx2(rvu)) {
+		cfg = rvu_read64(rvu, block->addr, NPA_AF_BATCH_CTL);
+		cfg &= ~0x7EULL;
+		cfg |= BIT_ULL(6) | BIT_ULL(2) | BIT_ULL(1);
+		rvu_write64(rvu, block->addr, NPA_AF_BATCH_CTL, cfg);
+	}
 	/* Result structure can be followed by Aura/Pool context at
 	 * RES + 128bytes and a write mask at RES + 256 bytes, depending on
 	 * operation type. Alloc sufficient result memory for all operations.
@@ -443,18 +524,14 @@ static int npa_aq_init(struct rvu *rvu, struct rvu_block *block)
 int rvu_npa_init(struct rvu *rvu)
 {
 	struct rvu_hwinfo *hw = rvu->hw;
-	int blkaddr, err;
+	int blkaddr;
 
 	blkaddr = rvu_get_blkaddr(rvu, BLKTYPE_NPA, 0);
 	if (blkaddr < 0)
 		return 0;
 
 	/* Initialize admin queue */
-	err = npa_aq_init(rvu, &hw->block[blkaddr]);
-	if (err)
-		return err;
-
-	return 0;
+	return npa_aq_init(rvu, &hw->block[blkaddr]);
 }
 
 void rvu_npa_freemem(struct rvu *rvu)
@@ -487,3 +564,51 @@ void rvu_npa_lf_teardown(struct rvu *rvu, u16 pcifunc, int npalf)
 
 	npa_ctx_free(rvu, pfvf);
 }
+
+/* Due to an Hardware errata, in some corner cases, AQ context lock
+ * operations can result in a NDC way getting into an illegal state
+ * of not valid but locked.
+ *
+ * This API solves the problem by clearing the lock bit of the NDC block.
+ * The operation needs to be done for each line of all the NDC banks.
+ */
+int rvu_ndc_fix_locked_cacheline(struct rvu *rvu, int blkaddr)
+{
+	int bank, max_bank, line, max_line, err;
+	u64 reg, ndc_af_const;
+
+	if (is_cn20k(rvu->pdev)) /* NDC not applicable to cn20k */
+		return 0;
+
+	/* Set the ENABLE bit(63) to '0' */
+	reg = rvu_read64(rvu, blkaddr, NDC_AF_CAMS_RD_INTERVAL);
+	rvu_write64(rvu, blkaddr, NDC_AF_CAMS_RD_INTERVAL, reg & GENMASK_ULL(62, 0));
+
+	/* Poll until the BUSY bits(47:32) are set to '0' */
+	err = rvu_poll_reg(rvu, blkaddr, NDC_AF_CAMS_RD_INTERVAL, GENMASK_ULL(47, 32), true);
+	if (err) {
+		dev_err(rvu->dev, "Timed out while polling for NDC CAM busy bits.\n");
+		return err;
+	}
+
+	ndc_af_const = rvu_read64(rvu, blkaddr, NDC_AF_CONST);
+	max_bank = FIELD_GET(NDC_AF_BANK_MASK, ndc_af_const);
+	max_line = FIELD_GET(NDC_AF_BANK_LINE_MASK, ndc_af_const);
+	for (bank = 0; bank < max_bank; bank++) {
+		for (line = 0; line < max_line; line++) {
+			/* Check if 'cache line valid bit(63)' is not set
+			 * but 'cache line lock bit(60)' is set and on
+			 * success, reset the lock bit(60).
+			 */
+			reg = rvu_read64(rvu, blkaddr,
+					 NDC_AF_BANKX_LINEX_METADATA(bank, line));
+			if (!(reg & BIT_ULL(63)) && (reg & BIT_ULL(60))) {
+				rvu_write64(rvu, blkaddr,
+					    NDC_AF_BANKX_LINEX_METADATA(bank, line),
+					    reg & ~BIT_ULL(60));
+			}
+		}
+	}
+
+	return 0;
+}