From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from dpdk.org (dpdk.org [92.243.14.124]) by dpdk.space (Postfix) with ESMTP id 16578A0471 for ; Sat, 22 Jun 2019 15:27:32 +0200 (CEST) Received: from [92.243.14.124] (localhost [127.0.0.1]) by dpdk.org (Postfix) with ESMTP id DF9681CFC2; Sat, 22 Jun 2019 15:25:52 +0200 (CEST) Received: from mx0b-0016f401.pphosted.com (mx0b-0016f401.pphosted.com [67.231.156.173]) by dpdk.org (Postfix) with ESMTP id 585B91C59E for ; Sat, 22 Jun 2019 15:25:28 +0200 (CEST) Received: from pps.filterd (m0045851.ppops.net [127.0.0.1]) by mx0b-0016f401.pphosted.com (8.16.0.27/8.16.0.27) with SMTP id x5MDP4Te030637 for ; Sat, 22 Jun 2019 06:25:27 -0700 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=marvell.com; h=from : to : cc : subject : date : message-id : in-reply-to : references : mime-version : content-transfer-encoding : content-type; s=pfpt0818; bh=7IQkoq1UwXw3J83NrWFZ/QWyvC0Kl5bkEKEfhbF5Ass=; b=nk6E46TJXErQric0FXgr9lFGv+52viieSiuP8hKCZEZw5AxnPh7e1HPq4TfGJxly/4Bo /5Y62rBfhouys45lsZfheoEN53iPFRWW3BVXktOMKmA+FA/JSAttfZ600A9XQL3ZAGq0 6fi/A+WCYcTcKKDTItPItJDNpWLbtM60gscWSxMsN4+EBHE89WIVqWTH/hyA22q/ze1t FzURjcnvnUtancTLtqc1qsb+G7kD5JQU3hILAgeG5PkHfaWiOSPgRXbVjWh7NZJidT0i QuC9oTnQrUVinxZDZV7g1nu+WK7bmDv+cVyeSO/E5WUH8Ceee5Z2e3t50K0nXcKuwKqI QA== Received: from sc-exch03.marvell.com ([199.233.58.183]) by mx0b-0016f401.pphosted.com with ESMTP id 2t9kuj8654-1 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-SHA384 bits=256 verify=NOT) for ; Sat, 22 Jun 2019 06:25:27 -0700 Received: from SC-EXCH01.marvell.com (10.93.176.81) by SC-EXCH03.marvell.com (10.93.176.83) with Microsoft SMTP Server (TLS) id 15.0.1367.3; Sat, 22 Jun 2019 06:25:26 -0700 Received: from maili.marvell.com (10.93.176.43) by SC-EXCH01.marvell.com (10.93.176.81) with Microsoft SMTP Server id 15.0.1367.3 via Frontend Transport; Sat, 22 Jun 2019 06:25:26 -0700 Received: from jerin-lab.marvell.com (jerin-lab.marvell.com [10.28.34.14]) by maili.marvell.com (Postfix) with ESMTP id 685683F7040; Sat, 22 Jun 2019 06:25:24 -0700 (PDT) From: To: Jerin Jacob , Nithin Dabilpuram , Vamsi Attunuru CC: , Harman Kalra Date: Sat, 22 Jun 2019 18:54:09 +0530 Message-ID: <20190622132417.32694-20-jerinj@marvell.com> X-Mailer: git-send-email 2.21.0 In-Reply-To: <20190622132417.32694-1-jerinj@marvell.com> References: <20190617155537.36144-1-jerinj@marvell.com> <20190622132417.32694-1-jerinj@marvell.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Content-Type: text/plain X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10434:, , definitions=2019-06-22_09:, , signatures=0 Subject: [dpdk-dev] [PATCH v4 19/27] mempool/octeontx2: add NPA IRQ handler X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" From: Jerin Jacob Register and implement NPA IRQ handler for RAS and all type of error interrupts to get the fatal errors from HW. Signed-off-by: Jerin Jacob Signed-off-by: Harman Kalra --- drivers/mempool/octeontx2/Makefile | 3 +- drivers/mempool/octeontx2/meson.build | 1 + drivers/mempool/octeontx2/otx2_mempool.c | 6 + drivers/mempool/octeontx2/otx2_mempool.h | 4 + drivers/mempool/octeontx2/otx2_mempool_irq.c | 302 +++++++++++++++++++ 5 files changed, 315 insertions(+), 1 deletion(-) create mode 100644 drivers/mempool/octeontx2/otx2_mempool_irq.c diff --git a/drivers/mempool/octeontx2/Makefile b/drivers/mempool/octeontx2/Makefile index 6fbb6e291..86950b270 100644 --- a/drivers/mempool/octeontx2/Makefile +++ b/drivers/mempool/octeontx2/Makefile @@ -28,7 +28,8 @@ LIBABIVER := 1 # all source are stored in SRCS-y # SRCS-$(CONFIG_RTE_LIBRTE_OCTEONTX2_MEMPOOL) += \ - otx2_mempool.c + otx2_mempool.c \ + otx2_mempool_irq.c LDLIBS += -lrte_eal -lrte_mempool -lrte_mbuf LDLIBS += -lrte_common_octeontx2 -lrte_kvargs -lrte_bus_pci diff --git a/drivers/mempool/octeontx2/meson.build b/drivers/mempool/octeontx2/meson.build index ec3c59eef..3f93b509d 100644 --- a/drivers/mempool/octeontx2/meson.build +++ b/drivers/mempool/octeontx2/meson.build @@ -3,6 +3,7 @@ # sources = files('otx2_mempool.c', + 'otx2_mempool_irq.c', ) extra_flags = [] diff --git a/drivers/mempool/octeontx2/otx2_mempool.c b/drivers/mempool/octeontx2/otx2_mempool.c index fa74b7532..1bcb86cf4 100644 --- a/drivers/mempool/octeontx2/otx2_mempool.c +++ b/drivers/mempool/octeontx2/otx2_mempool.c @@ -195,6 +195,7 @@ otx2_npa_lf_fini(void) return -ENOMEM; if (rte_atomic16_add_return(&idev->npa_refcnt, -1) == 0) { + otx2_npa_unregister_irqs(idev->npa_lf); rc |= npa_lf_fini(idev->npa_lf); rc |= npa_lf_detach(idev->npa_lf->mbox); otx2_npa_set_defaults(idev); @@ -251,6 +252,9 @@ otx2_npa_lf_init(struct rte_pci_device *pci_dev, void *otx2_dev) idev->npa_pf_func = dev->pf_func; idev->npa_lf = lf; rte_smp_wmb(); + rc = otx2_npa_register_irqs(lf); + if (rc) + goto npa_fini; rte_mbuf_set_platform_mempool_ops("octeontx2_npa"); otx2_npa_dbg("npa_lf=%p pools=%d sz=%d pf_func=0x%x msix=0x%x", @@ -259,6 +263,8 @@ otx2_npa_lf_init(struct rte_pci_device *pci_dev, void *otx2_dev) return 0; +npa_fini: + npa_lf_fini(idev->npa_lf); npa_detach: npa_lf_detach(dev->mbox); fail: diff --git a/drivers/mempool/octeontx2/otx2_mempool.h b/drivers/mempool/octeontx2/otx2_mempool.h index 871b45870..41542cf89 100644 --- a/drivers/mempool/octeontx2/otx2_mempool.h +++ b/drivers/mempool/octeontx2/otx2_mempool.h @@ -198,4 +198,8 @@ npa_lf_aura_op_range_set(uint64_t aura_handle, uint64_t start_iova, int otx2_npa_lf_init(struct rte_pci_device *pci_dev, void *otx2_dev); int otx2_npa_lf_fini(void); +/* IRQ */ +int otx2_npa_register_irqs(struct otx2_npa_lf *lf); +void otx2_npa_unregister_irqs(struct otx2_npa_lf *lf); + #endif /* __OTX2_MEMPOOL_H__ */ diff --git a/drivers/mempool/octeontx2/otx2_mempool_irq.c b/drivers/mempool/octeontx2/otx2_mempool_irq.c new file mode 100644 index 000000000..c026e1eea --- /dev/null +++ b/drivers/mempool/octeontx2/otx2_mempool_irq.c @@ -0,0 +1,302 @@ +/* SPDX-License-Identifier: BSD-3-Clause + * Copyright(C) 2019 Marvell International Ltd. + */ + +#include + +#include +#include + +#include "otx2_common.h" +#include "otx2_irq.h" +#include "otx2_mempool.h" + +static void +npa_lf_err_irq(void *param) +{ + struct otx2_npa_lf *lf = (struct otx2_npa_lf *)param; + uint64_t intr; + + intr = otx2_read64(lf->base + NPA_LF_ERR_INT); + if (intr == 0) + return; + + otx2_err("Err_intr=0x%" PRIx64 "", intr); + + /* Clear interrupt */ + otx2_write64(intr, lf->base + NPA_LF_ERR_INT); +} + +static int +npa_lf_register_err_irq(struct otx2_npa_lf *lf) +{ + struct rte_intr_handle *handle = lf->intr_handle; + int rc, vec; + + vec = lf->npa_msixoff + NPA_LF_INT_VEC_ERR_INT; + + /* Clear err interrupt */ + otx2_write64(~0ull, lf->base + NPA_LF_ERR_INT_ENA_W1C); + /* Register err interrupt vector */ + rc = otx2_register_irq(handle, npa_lf_err_irq, lf, vec); + + /* Enable hw interrupt */ + otx2_write64(~0ull, lf->base + NPA_LF_ERR_INT_ENA_W1S); + + return rc; +} + +static void +npa_lf_unregister_err_irq(struct otx2_npa_lf *lf) +{ + struct rte_intr_handle *handle = lf->intr_handle; + int vec; + + vec = lf->npa_msixoff + NPA_LF_INT_VEC_ERR_INT; + + /* Clear err interrupt */ + otx2_write64(~0ull, lf->base + NPA_LF_ERR_INT_ENA_W1C); + otx2_unregister_irq(handle, npa_lf_err_irq, lf, vec); +} + +static void +npa_lf_ras_irq(void *param) +{ + struct otx2_npa_lf *lf = (struct otx2_npa_lf *)param; + uint64_t intr; + + intr = otx2_read64(lf->base + NPA_LF_RAS); + if (intr == 0) + return; + + otx2_err("Ras_intr=0x%" PRIx64 "", intr); + + /* Clear interrupt */ + otx2_write64(intr, lf->base + NPA_LF_RAS); +} + +static int +npa_lf_register_ras_irq(struct otx2_npa_lf *lf) +{ + struct rte_intr_handle *handle = lf->intr_handle; + int rc, vec; + + vec = lf->npa_msixoff + NPA_LF_INT_VEC_POISON; + + /* Clear err interrupt */ + otx2_write64(~0ull, lf->base + NPA_LF_RAS_ENA_W1C); + /* Set used interrupt vectors */ + rc = otx2_register_irq(handle, npa_lf_ras_irq, lf, vec); + /* Enable hw interrupt */ + otx2_write64(~0ull, lf->base + NPA_LF_RAS_ENA_W1S); + + return rc; +} + +static void +npa_lf_unregister_ras_irq(struct otx2_npa_lf *lf) +{ + int vec; + struct rte_intr_handle *handle = lf->intr_handle; + + vec = lf->npa_msixoff + NPA_LF_INT_VEC_POISON; + + /* Clear err interrupt */ + otx2_write64(~0ull, lf->base + NPA_LF_RAS_ENA_W1C); + otx2_unregister_irq(handle, npa_lf_ras_irq, lf, vec); +} + +static inline uint8_t +npa_lf_q_irq_get_and_clear(struct otx2_npa_lf *lf, uint32_t q, + uint32_t off, uint64_t mask) +{ + uint64_t reg, wdata; + uint8_t qint; + + wdata = (uint64_t)q << 44; + reg = otx2_atomic64_add_nosync(wdata, (int64_t *)(lf->base + off)); + + if (reg & BIT_ULL(42) /* OP_ERR */) { + otx2_err("Failed execute irq get off=0x%x", off); + return 0; + } + + qint = reg & 0xff; + wdata &= mask; + otx2_write64(wdata, lf->base + off); + + return qint; +} + +static inline uint8_t +npa_lf_pool_irq_get_and_clear(struct otx2_npa_lf *lf, uint32_t p) +{ + return npa_lf_q_irq_get_and_clear(lf, p, NPA_LF_POOL_OP_INT, ~0xff00); +} + +static inline uint8_t +npa_lf_aura_irq_get_and_clear(struct otx2_npa_lf *lf, uint32_t a) +{ + return npa_lf_q_irq_get_and_clear(lf, a, NPA_LF_AURA_OP_INT, ~0xff00); +} + +static void +npa_lf_q_irq(void *param) +{ + struct otx2_npa_qint *qint = (struct otx2_npa_qint *)param; + struct otx2_npa_lf *lf = qint->lf; + uint8_t irq, qintx = qint->qintx; + uint32_t q, pool, aura; + uint64_t intr; + + intr = otx2_read64(lf->base + NPA_LF_QINTX_INT(qintx)); + if (intr == 0) + return; + + otx2_err("queue_intr=0x%" PRIx64 " qintx=%d", intr, qintx); + + /* Handle pool queue interrupts */ + for (q = 0; q < lf->nr_pools; q++) { + /* Skip disabled POOL */ + if (rte_bitmap_get(lf->npa_bmp, q)) + continue; + + pool = q % lf->qints; + irq = npa_lf_pool_irq_get_and_clear(lf, pool); + + if (irq & BIT_ULL(NPA_POOL_ERR_INT_OVFLS)) + otx2_err("Pool=%d NPA_POOL_ERR_INT_OVFLS", pool); + + if (irq & BIT_ULL(NPA_POOL_ERR_INT_RANGE)) + otx2_err("Pool=%d NPA_POOL_ERR_INT_RANGE", pool); + + if (irq & BIT_ULL(NPA_POOL_ERR_INT_PERR)) + otx2_err("Pool=%d NPA_POOL_ERR_INT_PERR", pool); + } + + /* Handle aura queue interrupts */ + for (q = 0; q < lf->nr_pools; q++) { + + /* Skip disabled AURA */ + if (rte_bitmap_get(lf->npa_bmp, q)) + continue; + + aura = q % lf->qints; + irq = npa_lf_aura_irq_get_and_clear(lf, aura); + + if (irq & BIT_ULL(NPA_AURA_ERR_INT_AURA_ADD_OVER)) + otx2_err("Aura=%d NPA_AURA_ERR_INT_ADD_OVER", aura); + + if (irq & BIT_ULL(NPA_AURA_ERR_INT_AURA_ADD_UNDER)) + otx2_err("Aura=%d NPA_AURA_ERR_INT_ADD_UNDER", aura); + + if (irq & BIT_ULL(NPA_AURA_ERR_INT_AURA_FREE_UNDER)) + otx2_err("Aura=%d NPA_AURA_ERR_INT_FREE_UNDER", aura); + + if (irq & BIT_ULL(NPA_AURA_ERR_INT_POOL_DIS)) + otx2_err("Aura=%d NPA_AURA_ERR_POOL_DIS", aura); + } + + /* Clear interrupt */ + otx2_write64(intr, lf->base + NPA_LF_QINTX_INT(qintx)); +} + +static int +npa_lf_register_queue_irqs(struct otx2_npa_lf *lf) +{ + struct rte_intr_handle *handle = lf->intr_handle; + int vec, q, qs, rc = 0; + + /* Figure out max qintx required */ + qs = RTE_MIN(lf->qints, lf->nr_pools); + + for (q = 0; q < qs; q++) { + vec = lf->npa_msixoff + NPA_LF_INT_VEC_QINT_START + q; + + /* Clear QINT CNT */ + otx2_write64(0, lf->base + NPA_LF_QINTX_CNT(q)); + + /* Clear interrupt */ + otx2_write64(~0ull, lf->base + NPA_LF_QINTX_ENA_W1C(q)); + + struct otx2_npa_qint *qintmem = lf->npa_qint_mem; + qintmem += q; + + qintmem->lf = lf; + qintmem->qintx = q; + + /* Sync qints_mem update */ + rte_smp_wmb(); + + /* Register queue irq vector */ + rc = otx2_register_irq(handle, npa_lf_q_irq, qintmem, vec); + if (rc) + break; + + otx2_write64(0, lf->base + NPA_LF_QINTX_CNT(q)); + otx2_write64(0, lf->base + NPA_LF_QINTX_INT(q)); + /* Enable QINT interrupt */ + otx2_write64(~0ull, lf->base + NPA_LF_QINTX_ENA_W1S(q)); + } + + return rc; +} + +static void +npa_lf_unregister_queue_irqs(struct otx2_npa_lf *lf) +{ + struct rte_intr_handle *handle = lf->intr_handle; + int vec, q, qs; + + /* Figure out max qintx required */ + qs = RTE_MIN(lf->qints, lf->nr_pools); + + for (q = 0; q < qs; q++) { + vec = lf->npa_msixoff + NPA_LF_INT_VEC_QINT_START + q; + + /* Clear QINT CNT */ + otx2_write64(0, lf->base + NPA_LF_QINTX_CNT(q)); + otx2_write64(0, lf->base + NPA_LF_QINTX_INT(q)); + + /* Clear interrupt */ + otx2_write64(~0ull, lf->base + NPA_LF_QINTX_ENA_W1C(q)); + + struct otx2_npa_qint *qintmem = lf->npa_qint_mem; + qintmem += q; + + /* Unregister queue irq vector */ + otx2_unregister_irq(handle, npa_lf_q_irq, qintmem, vec); + + qintmem->lf = NULL; + qintmem->qintx = 0; + } +} + +int +otx2_npa_register_irqs(struct otx2_npa_lf *lf) +{ + int rc; + + if (lf->npa_msixoff == MSIX_VECTOR_INVALID) { + otx2_err("Invalid NPALF MSIX vector offset vector: 0x%x", + lf->npa_msixoff); + return -EINVAL; + } + + /* Register lf err interrupt */ + rc = npa_lf_register_err_irq(lf); + /* Register RAS interrupt */ + rc |= npa_lf_register_ras_irq(lf); + /* Register queue interrupts */ + rc |= npa_lf_register_queue_irqs(lf); + + return rc; +} + +void +otx2_npa_unregister_irqs(struct otx2_npa_lf *lf) +{ + npa_lf_unregister_err_irq(lf); + npa_lf_unregister_ras_irq(lf); + npa_lf_unregister_queue_irqs(lf); +} -- 2.21.0