diff options
Diffstat (limited to 'drivers/soc/fsl/qbman/qman.c')
| -rw-r--r-- | drivers/soc/fsl/qbman/qman.c | 3053 |
1 files changed, 3053 insertions, 0 deletions
diff --git a/drivers/soc/fsl/qbman/qman.c b/drivers/soc/fsl/qbman/qman.c new file mode 100644 index 000000000..739e4eee6 --- /dev/null +++ b/drivers/soc/fsl/qbman/qman.c @@ -0,0 +1,3053 @@ +/* Copyright 2008 - 2016 Freescale Semiconductor, Inc. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions are met: + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * * Neither the name of Freescale Semiconductor nor the + * names of its contributors may be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * ALTERNATIVELY, this software may be distributed under the terms of the + * GNU General Public License ("GPL") as published by the Free Software + * Foundation, either version 2 of that License or (at your option) any + * later version. + * + * THIS SOFTWARE IS PROVIDED BY Freescale Semiconductor ``AS IS'' AND ANY + * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED + * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE + * DISCLAIMED. IN NO EVENT SHALL Freescale Semiconductor BE LIABLE FOR ANY + * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES + * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; + * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND + * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS + * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "qman_priv.h" + +#define DQRR_MAXFILL 15 +#define EQCR_ITHRESH 4 /* if EQCR congests, interrupt threshold */ +#define IRQNAME "QMan portal %d" +#define MAX_IRQNAME 16 /* big enough for "QMan portal %d" */ +#define QMAN_POLL_LIMIT 32 +#define QMAN_PIRQ_DQRR_ITHRESH 12 +#define QMAN_DQRR_IT_MAX 15 +#define QMAN_ITP_MAX 0xFFF +#define QMAN_PIRQ_MR_ITHRESH 4 +#define QMAN_PIRQ_IPERIOD 100 + +/* Portal register assists */ + +#if defined(CONFIG_ARM) || defined(CONFIG_ARM64) +/* Cache-inhibited register offsets */ +#define QM_REG_EQCR_PI_CINH 0x3000 +#define QM_REG_EQCR_CI_CINH 0x3040 +#define QM_REG_EQCR_ITR 0x3080 +#define QM_REG_DQRR_PI_CINH 0x3100 +#define QM_REG_DQRR_CI_CINH 0x3140 +#define QM_REG_DQRR_ITR 0x3180 +#define QM_REG_DQRR_DCAP 0x31C0 +#define QM_REG_DQRR_SDQCR 0x3200 +#define QM_REG_DQRR_VDQCR 0x3240 +#define QM_REG_DQRR_PDQCR 0x3280 +#define QM_REG_MR_PI_CINH 0x3300 +#define QM_REG_MR_CI_CINH 0x3340 +#define QM_REG_MR_ITR 0x3380 +#define QM_REG_CFG 0x3500 +#define QM_REG_ISR 0x3600 +#define QM_REG_IER 0x3640 +#define QM_REG_ISDR 0x3680 +#define QM_REG_IIR 0x36C0 +#define QM_REG_ITPR 0x3740 + +/* Cache-enabled register offsets */ +#define QM_CL_EQCR 0x0000 +#define QM_CL_DQRR 0x1000 +#define QM_CL_MR 0x2000 +#define QM_CL_EQCR_PI_CENA 0x3000 +#define QM_CL_EQCR_CI_CENA 0x3040 +#define QM_CL_DQRR_PI_CENA 0x3100 +#define QM_CL_DQRR_CI_CENA 0x3140 +#define QM_CL_MR_PI_CENA 0x3300 +#define QM_CL_MR_CI_CENA 0x3340 +#define QM_CL_CR 0x3800 +#define QM_CL_RR0 0x3900 +#define QM_CL_RR1 0x3940 + +#else +/* Cache-inhibited register offsets */ +#define QM_REG_EQCR_PI_CINH 0x0000 +#define QM_REG_EQCR_CI_CINH 0x0004 +#define QM_REG_EQCR_ITR 0x0008 +#define QM_REG_DQRR_PI_CINH 0x0040 +#define QM_REG_DQRR_CI_CINH 0x0044 +#define QM_REG_DQRR_ITR 0x0048 +#define QM_REG_DQRR_DCAP 0x0050 +#define QM_REG_DQRR_SDQCR 0x0054 +#define QM_REG_DQRR_VDQCR 0x0058 +#define QM_REG_DQRR_PDQCR 0x005c +#define QM_REG_MR_PI_CINH 0x0080 +#define QM_REG_MR_CI_CINH 0x0084 +#define QM_REG_MR_ITR 0x0088 +#define QM_REG_CFG 0x0100 +#define QM_REG_ISR 0x0e00 +#define QM_REG_IER 0x0e04 +#define QM_REG_ISDR 0x0e08 +#define QM_REG_IIR 0x0e0c +#define QM_REG_ITPR 0x0e14 + +/* Cache-enabled register offsets */ +#define QM_CL_EQCR 0x0000 +#define QM_CL_DQRR 0x1000 +#define QM_CL_MR 0x2000 +#define QM_CL_EQCR_PI_CENA 0x3000 +#define QM_CL_EQCR_CI_CENA 0x3100 +#define QM_CL_DQRR_PI_CENA 0x3200 +#define QM_CL_DQRR_CI_CENA 0x3300 +#define QM_CL_MR_PI_CENA 0x3400 +#define QM_CL_MR_CI_CENA 0x3500 +#define QM_CL_CR 0x3800 +#define QM_CL_RR0 0x3900 +#define QM_CL_RR1 0x3940 +#endif + +/* + * BTW, the drivers (and h/w programming model) already obtain the required + * synchronisation for portal accesses and data-dependencies. Use of barrier()s + * or other order-preserving primitives simply degrade performance. Hence the + * use of the __raw_*() interfaces, which simply ensure that the compiler treats + * the portal registers as volatile + */ + +/* Cache-enabled ring access */ +#define qm_cl(base, idx) ((void *)base + ((idx) << 6)) + +/* + * Portal modes. + * Enum types; + * pmode == production mode + * cmode == consumption mode, + * dmode == h/w dequeue mode. + * Enum values use 3 letter codes. First letter matches the portal mode, + * remaining two letters indicate; + * ci == cache-inhibited portal register + * ce == cache-enabled portal register + * vb == in-band valid-bit (cache-enabled) + * dc == DCA (Discrete Consumption Acknowledgment), DQRR-only + * As for "enum qm_dqrr_dmode", it should be self-explanatory. + */ +enum qm_eqcr_pmode { /* matches QCSP_CFG::EPM */ + qm_eqcr_pci = 0, /* PI index, cache-inhibited */ + qm_eqcr_pce = 1, /* PI index, cache-enabled */ + qm_eqcr_pvb = 2 /* valid-bit */ +}; +enum qm_dqrr_dmode { /* matches QCSP_CFG::DP */ + qm_dqrr_dpush = 0, /* SDQCR + VDQCR */ + qm_dqrr_dpull = 1 /* PDQCR */ +}; +enum qm_dqrr_pmode { /* s/w-only */ + qm_dqrr_pci, /* reads DQRR_PI_CINH */ + qm_dqrr_pce, /* reads DQRR_PI_CENA */ + qm_dqrr_pvb /* reads valid-bit */ +}; +enum qm_dqrr_cmode { /* matches QCSP_CFG::DCM */ + qm_dqrr_cci = 0, /* CI index, cache-inhibited */ + qm_dqrr_cce = 1, /* CI index, cache-enabled */ + qm_dqrr_cdc = 2 /* Discrete Consumption Acknowledgment */ +}; +enum qm_mr_pmode { /* s/w-only */ + qm_mr_pci, /* reads MR_PI_CINH */ + qm_mr_pce, /* reads MR_PI_CENA */ + qm_mr_pvb /* reads valid-bit */ +}; +enum qm_mr_cmode { /* matches QCSP_CFG::MM */ + qm_mr_cci = 0, /* CI index, cache-inhibited */ + qm_mr_cce = 1 /* CI index, cache-enabled */ +}; + +/* --- Portal structures --- */ + +#define QM_EQCR_SIZE 8 +#define QM_DQRR_SIZE 16 +#define QM_MR_SIZE 8 + +/* "Enqueue Command" */ +struct qm_eqcr_entry { + u8 _ncw_verb; /* writes to this are non-coherent */ + u8 dca; + __be16 seqnum; + u8 __reserved[4]; + __be32 fqid; /* 24-bit */ + __be32 tag; + struct qm_fd fd; + u8 __reserved3[32]; +} __packed __aligned(8); +#define QM_EQCR_VERB_VBIT 0x80 +#define QM_EQCR_VERB_CMD_MASK 0x61 /* but only one value; */ +#define QM_EQCR_VERB_CMD_ENQUEUE 0x01 +#define QM_EQCR_SEQNUM_NESN 0x8000 /* Advance NESN */ +#define QM_EQCR_SEQNUM_NLIS 0x4000 /* More fragments to come */ +#define QM_EQCR_SEQNUM_SEQMASK 0x3fff /* sequence number goes here */ + +struct qm_eqcr { + struct qm_eqcr_entry *ring, *cursor; + u8 ci, available, ithresh, vbit; +#ifdef CONFIG_FSL_DPAA_CHECKING + u32 busy; + enum qm_eqcr_pmode pmode; +#endif +}; + +struct qm_dqrr { + const struct qm_dqrr_entry *ring, *cursor; + u8 pi, ci, fill, ithresh, vbit; +#ifdef CONFIG_FSL_DPAA_CHECKING + enum qm_dqrr_dmode dmode; + enum qm_dqrr_pmode pmode; + enum qm_dqrr_cmode cmode; +#endif +}; + +struct qm_mr { + union qm_mr_entry *ring, *cursor; + u8 pi, ci, fill, ithresh, vbit; +#ifdef CONFIG_FSL_DPAA_CHECKING + enum qm_mr_pmode pmode; + enum qm_mr_cmode cmode; +#endif +}; + +/* MC (Management Command) command */ +/* "FQ" command layout */ +struct qm_mcc_fq { + u8 _ncw_verb; + u8 __reserved1[3]; + __be32 fqid; /* 24-bit */ + u8 __reserved2[56]; +} __packed; + +/* "CGR" command layout */ +struct qm_mcc_cgr { + u8 _ncw_verb; + u8 __reserved1[30]; + u8 cgid; + u8 __reserved2[32]; +}; + +#define QM_MCC_VERB_VBIT 0x80 +#define QM_MCC_VERB_MASK 0x7f /* where the verb contains; */ +#define QM_MCC_VERB_INITFQ_PARKED 0x40 +#define QM_MCC_VERB_INITFQ_SCHED 0x41 +#define QM_MCC_VERB_QUERYFQ 0x44 +#define QM_MCC_VERB_QUERYFQ_NP 0x45 /* "non-programmable" fields */ +#define QM_MCC_VERB_QUERYWQ 0x46 +#define QM_MCC_VERB_QUERYWQ_DEDICATED 0x47 +#define QM_MCC_VERB_ALTER_SCHED 0x48 /* Schedule FQ */ +#define QM_MCC_VERB_ALTER_FE 0x49 /* Force Eligible FQ */ +#define QM_MCC_VERB_ALTER_RETIRE 0x4a /* Retire FQ */ +#define QM_MCC_VERB_ALTER_OOS 0x4b /* Take FQ out of service */ +#define QM_MCC_VERB_ALTER_FQXON 0x4d /* FQ XON */ +#define QM_MCC_VERB_ALTER_FQXOFF 0x4e /* FQ XOFF */ +#define QM_MCC_VERB_INITCGR 0x50 +#define QM_MCC_VERB_MODIFYCGR 0x51 +#define QM_MCC_VERB_CGRTESTWRITE 0x52 +#define QM_MCC_VERB_QUERYCGR 0x58 +#define QM_MCC_VERB_QUERYCONGESTION 0x59 +union qm_mc_command { + struct { + u8 _ncw_verb; /* writes to this are non-coherent */ + u8 __reserved[63]; + }; + struct qm_mcc_initfq initfq; + struct qm_mcc_initcgr initcgr; + struct qm_mcc_fq fq; + struct qm_mcc_cgr cgr; +}; + +/* MC (Management Command) result */ +/* "Query FQ" */ +struct qm_mcr_queryfq { + u8 verb; + u8 result; + u8 __reserved1[8]; + struct qm_fqd fqd; /* the FQD fields are here */ + u8 __reserved2[30]; +} __packed; + +/* "Alter FQ State Commands" */ +struct qm_mcr_alterfq { + u8 verb; + u8 result; + u8 fqs; /* Frame Queue Status */ + u8 __reserved1[61]; +}; +#define QM_MCR_VERB_RRID 0x80 +#define QM_MCR_VERB_MASK QM_MCC_VERB_MASK +#define QM_MCR_VERB_INITFQ_PARKED QM_MCC_VERB_INITFQ_PARKED +#define QM_MCR_VERB_INITFQ_SCHED QM_MCC_VERB_INITFQ_SCHED +#define QM_MCR_VERB_QUERYFQ QM_MCC_VERB_QUERYFQ +#define QM_MCR_VERB_QUERYFQ_NP QM_MCC_VERB_QUERYFQ_NP +#define QM_MCR_VERB_QUERYWQ QM_MCC_VERB_QUERYWQ +#define QM_MCR_VERB_QUERYWQ_DEDICATED QM_MCC_VERB_QUERYWQ_DEDICATED +#define QM_MCR_VERB_ALTER_SCHED QM_MCC_VERB_ALTER_SCHED +#define QM_MCR_VERB_ALTER_FE QM_MCC_VERB_ALTER_FE +#define QM_MCR_VERB_ALTER_RETIRE QM_MCC_VERB_ALTER_RETIRE +#define QM_MCR_VERB_ALTER_OOS QM_MCC_VERB_ALTER_OOS +#define QM_MCR_RESULT_NULL 0x00 +#define QM_MCR_RESULT_OK 0xf0 +#define QM_MCR_RESULT_ERR_FQID 0xf1 +#define QM_MCR_RESULT_ERR_FQSTATE 0xf2 +#define QM_MCR_RESULT_ERR_NOTEMPTY 0xf3 /* OOS fails if FQ is !empty */ +#define QM_MCR_RESULT_ERR_BADCHANNEL 0xf4 +#define QM_MCR_RESULT_PENDING 0xf8 +#define QM_MCR_RESULT_ERR_BADCOMMAND 0xff +#define QM_MCR_FQS_ORLPRESENT 0x02 /* ORL fragments to come */ +#define QM_MCR_FQS_NOTEMPTY 0x01 /* FQ has enqueued frames */ +#define QM_MCR_TIMEOUT 10000 /* us */ +union qm_mc_result { + struct { + u8 verb; + u8 result; + u8 __reserved1[62]; + }; + struct qm_mcr_queryfq queryfq; + struct qm_mcr_alterfq alterfq; + struct qm_mcr_querycgr querycgr; + struct qm_mcr_querycongestion querycongestion; + struct qm_mcr_querywq querywq; + struct qm_mcr_queryfq_np queryfq_np; +}; + +struct qm_mc { + union qm_mc_command *cr; + union qm_mc_result *rr; + u8 rridx, vbit; +#ifdef CONFIG_FSL_DPAA_CHECKING + enum { + /* Can be _mc_start()ed */ + qman_mc_idle, + /* Can be _mc_commit()ed or _mc_abort()ed */ + qman_mc_user, + /* Can only be _mc_retry()ed */ + qman_mc_hw + } state; +#endif +}; + +struct qm_addr { + void *ce; /* cache-enabled */ + __be32 *ce_be; /* same value as above but for direct access */ + void __iomem *ci; /* cache-inhibited */ +}; + +struct qm_portal { + /* + * In the non-CONFIG_FSL_DPAA_CHECKING case, the following stuff up to + * and including 'mc' fits within a cacheline (yay!). The 'config' part + * is setup-only, so isn't a cause for a concern. In other words, don't + * rearrange this structure on a whim, there be dragons ... + */ + struct qm_addr addr; + struct qm_eqcr eqcr; + struct qm_dqrr dqrr; + struct qm_mr mr; + struct qm_mc mc; +} ____cacheline_aligned; + +/* Cache-inhibited register access. */ +static inline u32 qm_in(struct qm_portal *p, u32 offset) +{ + return ioread32be(p->addr.ci + offset); +} + +static inline void qm_out(struct qm_portal *p, u32 offset, u32 val) +{ + iowrite32be(val, p->addr.ci + offset); +} + +/* Cache Enabled Portal Access */ +static inline void qm_cl_invalidate(struct qm_portal *p, u32 offset) +{ + dpaa_invalidate(p->addr.ce + offset); +} + +static inline void qm_cl_touch_ro(struct qm_portal *p, u32 offset) +{ + dpaa_touch_ro(p->addr.ce + offset); +} + +static inline u32 qm_ce_in(struct qm_portal *p, u32 offset) +{ + return be32_to_cpu(*(p->addr.ce_be + (offset/4))); +} + +/* --- EQCR API --- */ + +#define EQCR_SHIFT ilog2(sizeof(struct qm_eqcr_entry)) +#define EQCR_CARRY (uintptr_t)(QM_EQCR_SIZE << EQCR_SHIFT) + +/* Bit-wise logic to wrap a ring pointer by clearing the "carry bit" */ +static struct qm_eqcr_entry *eqcr_carryclear(struct qm_eqcr_entry *p) +{ + uintptr_t addr = (uintptr_t)p; + + addr &= ~EQCR_CARRY; + + return (struct qm_eqcr_entry *)addr; +} + +/* Bit-wise logic to convert a ring pointer to a ring index */ +static int eqcr_ptr2idx(struct qm_eqcr_entry *e) +{ + return ((uintptr_t)e >> EQCR_SHIFT) & (QM_EQCR_SIZE - 1); +} + +/* Increment the 'cursor' ring pointer, taking 'vbit' into account */ +static inline void eqcr_inc(struct qm_eqcr *eqcr) +{ + /* increment to the next EQCR pointer and handle overflow and 'vbit' */ + struct qm_eqcr_entry *partial = eqcr->cursor + 1; + + eqcr->cursor = eqcr_carryclear(partial); + if (partial != eqcr->cursor) + eqcr->vbit ^= QM_EQCR_VERB_VBIT; +} + +static inline int qm_eqcr_init(struct qm_portal *portal, + enum qm_eqcr_pmode pmode, + unsigned int eq_stash_thresh, + int eq_stash_prio) +{ + struct qm_eqcr *eqcr = &portal->eqcr; + u32 cfg; + u8 pi; + + eqcr->ring = portal->addr.ce + QM_CL_EQCR; + eqcr->ci = qm_in(portal, QM_REG_EQCR_CI_CINH) & (QM_EQCR_SIZE - 1); + qm_cl_invalidate(portal, QM_CL_EQCR_CI_CENA); + pi = qm_in(portal, QM_REG_EQCR_PI_CINH) & (QM_EQCR_SIZE - 1); + eqcr->cursor = eqcr->ring + pi; + eqcr->vbit = (qm_in(portal, QM_REG_EQCR_PI_CINH) & QM_EQCR_SIZE) ? + QM_EQCR_VERB_VBIT : 0; + eqcr->available = QM_EQCR_SIZE - 1 - + dpaa_cyc_diff(QM_EQCR_SIZE, eqcr->ci, pi); + eqcr->ithresh = qm_in(portal, QM_REG_EQCR_ITR); +#ifdef CONFIG_FSL_DPAA_CHECKING + eqcr->busy = 0; + eqcr->pmode = pmode; +#endif + cfg = (qm_in(portal, QM_REG_CFG) & 0x00ffffff) | + (eq_stash_thresh << 28) | /* QCSP_CFG: EST */ + (eq_stash_prio << 26) | /* QCSP_CFG: EP */ + ((pmode & 0x3) << 24); /* QCSP_CFG::EPM */ + qm_out(portal, QM_REG_CFG, cfg); + return 0; +} + +static inline void qm_eqcr_finish(struct qm_portal *portal) +{ + struct qm_eqcr *eqcr = &portal->eqcr; + u8 pi = qm_in(portal, QM_REG_EQCR_PI_CINH) & (QM_EQCR_SIZE - 1); + u8 ci = qm_in(portal, QM_REG_EQCR_CI_CINH) & (QM_EQCR_SIZE - 1); + + DPAA_ASSERT(!eqcr->busy); + if (pi != eqcr_ptr2idx(eqcr->cursor)) + pr_crit("losing uncommitted EQCR entries\n"); + if (ci != eqcr->ci) + pr_crit("missing existing EQCR completions\n"); + if (eqcr->ci != eqcr_ptr2idx(eqcr->cursor)) + pr_crit("EQCR destroyed unquiesced\n"); +} + +static inline struct qm_eqcr_entry *qm_eqcr_start_no_stash(struct qm_portal + *portal) +{ + struct qm_eqcr *eqcr = &portal->eqcr; + + DPAA_ASSERT(!eqcr->busy); + if (!eqcr->available) + return NULL; + +#ifdef CONFIG_FSL_DPAA_CHECKING + eqcr->busy = 1; +#endif + dpaa_zero(eqcr->cursor); + return eqcr->cursor; +} + +static inline struct qm_eqcr_entry *qm_eqcr_start_stash(struct qm_portal + *portal) +{ + struct qm_eqcr *eqcr = &portal->eqcr; + u8 diff, old_ci; + + DPAA_ASSERT(!eqcr->busy); + if (!eqcr->available) { + old_ci = eqcr->ci; + eqcr->ci = qm_ce_in(portal, QM_CL_EQCR_CI_CENA) & + (QM_EQCR_SIZE - 1); + diff = dpaa_cyc_diff(QM_EQCR_SIZE, old_ci, eqcr->ci); + eqcr->available += diff; + if (!diff) + return NULL; + } +#ifdef CONFIG_FSL_DPAA_CHECKING + eqcr->busy = 1; +#endif + dpaa_zero(eqcr->cursor); + return eqcr->cursor; +} + +static inline void eqcr_commit_checks(struct qm_eqcr *eqcr) +{ + DPAA_ASSERT(eqcr->busy); + DPAA_ASSERT(!(be32_to_cpu(eqcr->cursor->fqid) & ~QM_FQID_MASK)); + DPAA_ASSERT(eqcr->available >= 1); +} + +static inline void qm_eqcr_pvb_commit(struct qm_portal *portal, u8 myverb) +{ + struct qm_eqcr *eqcr = &portal->eqcr; + struct qm_eqcr_entry *eqcursor; + + eqcr_commit_checks(eqcr); + DPAA_ASSERT(eqcr->pmode == qm_eqcr_pvb); + dma_wmb(); + eqcursor = eqcr->cursor; + eqcursor->_ncw_verb = myverb | eqcr->vbit; + dpaa_flush(eqcursor); + eqcr_inc(eqcr); + eqcr->available--; +#ifdef CONFIG_FSL_DPAA_CHECKING + eqcr->busy = 0; +#endif +} + +static inline void qm_eqcr_cce_prefetch(struct qm_portal *portal) +{ + qm_cl_touch_ro(portal, QM_CL_EQCR_CI_CENA); +} + +static inline u8 qm_eqcr_cce_update(struct qm_portal *portal) +{ + struct qm_eqcr *eqcr = &portal->eqcr; + u8 diff, old_ci = eqcr->ci; + + eqcr->ci = qm_ce_in(portal, QM_CL_EQCR_CI_CENA) & (QM_EQCR_SIZE - 1); + qm_cl_invalidate(portal, QM_CL_EQCR_CI_CENA); + diff = dpaa_cyc_diff(QM_EQCR_SIZE, old_ci, eqcr->ci); + eqcr->available += diff; + return diff; +} + +static inline void qm_eqcr_set_ithresh(struct qm_portal *portal, u8 ithresh) +{ + struct qm_eqcr *eqcr = &portal->eqcr; + + eqcr->ithresh = ithresh; + qm_out(portal, QM_REG_EQCR_ITR, ithresh); +} + +static inline u8 qm_eqcr_get_avail(struct qm_portal *portal) +{ + struct qm_eqcr *eqcr = &portal->eqcr; + + return eqcr->available; +} + +static inline u8 qm_eqcr_get_fill(struct qm_portal *portal) +{ + struct qm_eqcr *eqcr = &portal->eqcr; + + return QM_EQCR_SIZE - 1 - eqcr->available; +} + +/* --- DQRR API --- */ + +#define DQRR_SHIFT ilog2(sizeof(struct qm_dqrr_entry)) +#define DQRR_CARRY (uintptr_t)(QM_DQRR_SIZE << DQRR_SHIFT) + +static const struct qm_dqrr_entry *dqrr_carryclear( + const struct qm_dqrr_entry *p) +{ + uintptr_t addr = (uintptr_t)p; + + addr &= ~DQRR_CARRY; + + return (const struct qm_dqrr_entry *)addr; +} + +static inline int dqrr_ptr2idx(const struct qm_dqrr_entry *e) +{ + return ((uintptr_t)e >> DQRR_SHIFT) & (QM_DQRR_SIZE - 1); +} + +static const struct qm_dqrr_entry *dqrr_inc(const struct qm_dqrr_entry *e) +{ + return dqrr_carryclear(e + 1); +} + +static inline void qm_dqrr_set_maxfill(struct qm_portal *portal, u8 mf) +{ + qm_out(portal, QM_REG_CFG, (qm_in(portal, QM_REG_CFG) & 0xff0fffff) | + ((mf & (QM_DQRR_SIZE - 1)) << 20)); +} + +static inline int qm_dqrr_init(struct qm_portal *portal, + const struct qm_portal_config *config, + enum qm_dqrr_dmode dmode, + enum qm_dqrr_pmode pmode, + enum qm_dqrr_cmode cmode, u8 max_fill) +{ + struct qm_dqrr *dqrr = &portal->dqrr; + u32 cfg; + + /* Make sure the DQRR will be idle when we enable */ + qm_out(portal, QM_REG_DQRR_SDQCR, 0); + qm_out(portal, QM_REG_DQRR_VDQCR, 0); + qm_out(portal, QM_REG_DQRR_PDQCR, 0); + dqrr->ring = portal->addr.ce + QM_CL_DQRR; + dqrr->pi = qm_in(portal, QM_REG_DQRR_PI_CINH) & (QM_DQRR_SIZE - 1); + dqrr->ci = qm_in(portal, QM_REG_DQRR_CI_CINH) & (QM_DQRR_SIZE - 1); + dqrr->cursor = dqrr->ring + dqrr->ci; + dqrr->fill = dpaa_cyc_diff(QM_DQRR_SIZE, dqrr->ci, dqrr->pi); + dqrr->vbit = (qm_in(portal, QM_REG_DQRR_PI_CINH) & QM_DQRR_SIZE) ? + QM_DQRR_VERB_VBIT : 0; + dqrr->ithresh = qm_in(portal, QM_REG_DQRR_ITR); +#ifdef CONFIG_FSL_DPAA_CHECKING + dqrr->dmode = dmode; + dqrr->pmode = pmode; + dqrr->cmode = cmode; +#endif + /* Invalidate every ring entry before beginning */ + for (cfg = 0; cfg < QM_DQRR_SIZE; cfg++) + dpaa_invalidate(qm_cl(dqrr->ring, cfg)); + cfg = (qm_in(portal, QM_REG_CFG) & 0xff000f00) | + ((max_fill & (QM_DQRR_SIZE - 1)) << 20) | /* DQRR_MF */ + ((dmode & 1) << 18) | /* DP */ + ((cmode & 3) << 16) | /* DCM */ + 0xa0 | /* RE+SE */ + (0 ? 0x40 : 0) | /* Ignore RP */ + (0 ? 0x10 : 0); /* Ignore SP */ + qm_out(portal, QM_REG_CFG, cfg); + qm_dqrr_set_maxfill(portal, max_fill); + return 0; +} + +static inline void qm_dqrr_finish(struct qm_portal *portal) +{ +#ifdef CONFIG_FSL_DPAA_CHECKING + struct qm_dqrr *dqrr = &portal->dqrr; + + if (dqrr->cmode != qm_dqrr_cdc && + dqrr->ci != dqrr_ptr2idx(dqrr->cursor)) + pr_crit("Ignoring completed DQRR entries\n"); +#endif +} + +static inline const struct qm_dqrr_entry *qm_dqrr_current( + struct qm_portal *portal) +{ + struct qm_dqrr *dqrr = &portal->dqrr; + + if (!dqrr->fill) + return NULL; + return dqrr->cursor; +} + +static inline u8 qm_dqrr_next(struct qm_portal *portal) +{ + struct qm_dqrr *dqrr = &portal->dqrr; + + DPAA_ASSERT(dqrr->fill); + dqrr->cursor = dqrr_inc(dqrr->cursor); + return --dqrr->fill; +} + +static inline void qm_dqrr_pvb_update(struct qm_portal *portal) +{ + struct qm_dqrr *dqrr = &portal->dqrr; + struct qm_dqrr_entry *res = qm_cl(dqrr->ring, dqrr->pi); + + DPAA_ASSERT(dqrr->pmode == qm_dqrr_pvb); +#ifndef CONFIG_FSL_PAMU + /* + * If PAMU is not available we need to invalidate the cache. + * When PAMU is available the cache is updated by stash + */ + dpaa_invalidate_touch_ro(res); +#endif + if ((res->verb & QM_DQRR_VERB_VBIT) == dqrr->vbit) { + dqrr->pi = (dqrr->pi + 1) & (QM_DQRR_SIZE - 1); + if (!dqrr->pi) + dqrr->vbit ^= QM_DQRR_VERB_VBIT; + dqrr->fill++; + } +} + +static inline void qm_dqrr_cdc_consume_1ptr(struct qm_portal *portal, + const struct qm_dqrr_entry *dq, + int park) +{ + __maybe_unused struct qm_dqrr *dqrr = &portal->dqrr; + int idx = dqrr_ptr2idx(dq); + + DPAA_ASSERT(dqrr->cmode == qm_dqrr_cdc); + DPAA_ASSERT((dqrr->ring + idx) == dq); + DPAA_ASSERT(idx < QM_DQRR_SIZE); + qm_out(portal, QM_REG_DQRR_DCAP, (0 << 8) | /* DQRR_DCAP::S */ + ((park ? 1 : 0) << 6) | /* DQRR_DCAP::PK */ + idx); /* DQRR_DCAP::DCAP_CI */ +} + +static inline void qm_dqrr_cdc_consume_n(struct qm_portal *portal, u32 bitmask) +{ + __maybe_unused struct qm_dqrr *dqrr = &portal->dqrr; + + DPAA_ASSERT(dqrr->cmode == qm_dqrr_cdc); + qm_out(portal, QM_REG_DQRR_DCAP, (1 << 8) | /* DQRR_DCAP::S */ + (bitmask << 16)); /* DQRR_DCAP::DCAP_CI */ +} + +static inline void qm_dqrr_sdqcr_set(struct qm_portal *portal, u32 sdqcr) +{ + qm_out(portal, QM_REG_DQRR_SDQCR, sdqcr); +} + +static inline void qm_dqrr_vdqcr_set(struct qm_portal *portal, u32 vdqcr) +{ + qm_out(portal, QM_REG_DQRR_VDQCR, vdqcr); +} + +static inline int qm_dqrr_set_ithresh(struct qm_portal *portal, u8 ithresh) +{ + + if (ithresh > QMAN_DQRR_IT_MAX) + return -EINVAL; + + qm_out(portal, QM_REG_DQRR_ITR, ithresh); + + return 0; +} + +/* --- MR API --- */ + +#define MR_SHIFT ilog2(sizeof(union qm_mr_entry)) +#define MR_CARRY (uintptr_t)(QM_MR_SIZE << MR_SHIFT) + +static union qm_mr_entry *mr_carryclear(union qm_mr_entry *p) +{ + uintptr_t addr = (uintptr_t)p; + + addr &= ~MR_CARRY; + + return (union qm_mr_entry *)addr; +} + +static inline int mr_ptr2idx(const union qm_mr_entry *e) +{ + return ((uintptr_t)e >> MR_SHIFT) & (QM_MR_SIZE - 1); +} + +static inline union qm_mr_entry *mr_inc(union qm_mr_entry *e) +{ + return mr_carryclear(e + 1); +} + +static inline int qm_mr_init(struct qm_portal *portal, enum qm_mr_pmode pmode, + enum qm_mr_cmode cmode) +{ + struct qm_mr *mr = &portal->mr; + u32 cfg; + + mr->ring = portal->addr.ce + QM_CL_MR; + mr->pi = qm_in(portal, QM_REG_MR_PI_CINH) & (QM_MR_SIZE - 1); + mr->ci = qm_in(portal, QM_REG_MR_CI_CINH) & (QM_MR_SIZE - 1); + mr->cursor = mr->ring + mr->ci; + mr->fill = dpaa_cyc_diff(QM_MR_SIZE, mr->ci, mr->pi); + mr->vbit = (qm_in(portal, QM_REG_MR_PI_CINH) & QM_MR_SIZE) + ? QM_MR_VERB_VBIT : 0; + mr->ithresh = qm_in(portal, QM_REG_MR_ITR); +#ifdef CONFIG_FSL_DPAA_CHECKING + mr->pmode = pmode; + mr->cmode = cmode; +#endif + cfg = (qm_in(portal, QM_REG_CFG) & 0xfffff0ff) | + ((cmode & 1) << 8); /* QCSP_CFG:MM */ + qm_out(portal, QM_REG_CFG, cfg); + return 0; +} + +static inline void qm_mr_finish(struct qm_portal *portal) +{ + struct qm_mr *mr = &portal->mr; + + if (mr->ci != mr_ptr2idx(mr->cursor)) + pr_crit("Ignoring completed MR entries\n"); +} + +static inline const union qm_mr_entry *qm_mr_current(struct qm_portal *portal) +{ + struct qm_mr *mr = &portal->mr; + + if (!mr->fill) + return NULL; + return mr->cursor; +} + +static inline int qm_mr_next(struct qm_portal *portal) +{ + struct qm_mr *mr = &portal->mr; + + DPAA_ASSERT(mr->fill); + mr->cursor = mr_inc(mr->cursor); + return --mr->fill; +} + +static inline void qm_mr_pvb_update(struct qm_portal *portal) +{ + struct qm_mr *mr = &portal->mr; + union qm_mr_entry *res = qm_cl(mr->ring, mr->pi); + + DPAA_ASSERT(mr->pmode == qm_mr_pvb); + + if ((res->verb & QM_MR_VERB_VBIT) == mr->vbit) { + mr->pi = (mr->pi + 1) & (QM_MR_SIZE - 1); + if (!mr->pi) + mr->vbit ^= QM_MR_VERB_VBIT; + mr->fill++; + res = mr_inc(res); + } + dpaa_invalidate_touch_ro(res); +} + +static inline void qm_mr_cci_consume(struct qm_portal *portal, u8 num) +{ + struct qm_mr *mr = &portal->mr; + + DPAA_ASSERT(mr->cmode == qm_mr_cci); + mr->ci = (mr->ci + num) & (QM_MR_SIZE - 1); + qm_out(portal, QM_REG_MR_CI_CINH, mr->ci); +} + +static inline void qm_mr_cci_consume_to_current(struct qm_portal *portal) +{ + struct qm_mr *mr = &portal->mr; + + DPAA_ASSERT(mr->cmode == qm_mr_cci); + mr->ci = mr_ptr2idx(mr->cursor); + qm_out(portal, QM_REG_MR_CI_CINH, mr->ci); +} + +static inline void qm_mr_set_ithresh(struct qm_portal *portal, u8 ithresh) +{ + qm_out(portal, QM_REG_MR_ITR, ithresh); +} + +/* --- Management command API --- */ + +static inline int qm_mc_init(struct qm_portal *portal) +{ + u8 rr0, rr1; + struct qm_mc *mc = &portal->mc; + + mc->cr = portal->addr.ce + QM_CL_CR; + mc->rr = portal->addr.ce + QM_CL_RR0; + /* + * The expected valid bit polarity for the next CR command is 0 + * if RR1 contains a valid response, and is 1 if RR0 contains a + * valid response. If both RR contain all 0, this indicates either + * that no command has been executed since reset (in which case the + * expected valid bit polarity is 1) + */ + rr0 = mc->rr->verb; + rr1 = (mc->rr+1)->verb; + if ((rr0 == 0 && rr1 == 0) || rr0 != 0) + mc->rridx = 1; + else + mc->rridx = 0; + mc->vbit = mc->rridx ? QM_MCC_VERB_VBIT : 0; +#ifdef CONFIG_FSL_DPAA_CHECKING + mc->state = qman_mc_idle; +#endif + return 0; +} + +static inline void qm_mc_finish(struct qm_portal *portal) +{ +#ifdef CONFIG_FSL_DPAA_CHECKING + struct qm_mc *mc = &portal->mc; + + DPAA_ASSERT(mc->state == qman_mc_idle); + if (mc->state != qman_mc_idle) + pr_crit("Losing incomplete MC command\n"); +#endif +} + +static inline union qm_mc_command *qm_mc_start(struct qm_portal *portal) +{ + struct qm_mc *mc = &portal->mc; + + DPAA_ASSERT(mc->state == qman_mc_idle); +#ifdef CONFIG_FSL_DPAA_CHECKING + mc->state = qman_mc_user; +#endif + dpaa_zero(mc->cr); + return mc->cr; +} + +static inline void qm_mc_commit(struct qm_portal *portal, u8 myverb) +{ + struct qm_mc *mc = &portal->mc; + union qm_mc_result *rr = mc->rr + mc->rridx; + + DPAA_ASSERT(mc->state == qman_mc_user); + dma_wmb(); + mc->cr->_ncw_verb = myverb | mc->vbit; + dpaa_flush(mc->cr); + dpaa_invalidate_touch_ro(rr); +#ifdef CONFIG_FSL_DPAA_CHECKING + mc->state = qman_mc_hw; +#endif +} + +static inline union qm_mc_result *qm_mc_result(struct qm_portal *portal) +{ + struct qm_mc *mc = &portal->mc; + union qm_mc_result *rr = mc->rr + mc->rridx; + + DPAA_ASSERT(mc->state == qman_mc_hw); + /* + * The inactive response register's verb byte always returns zero until + * its command is submitted and completed. This includes the valid-bit, + * in case you were wondering... + */ + if (!rr->verb) { + dpaa_invalidate_touch_ro(rr); + return NULL; + } + mc->rridx ^= 1; + mc->vbit ^= QM_MCC_VERB_VBIT; +#ifdef CONFIG_FSL_DPAA_CHECKING + mc->state = qman_mc_idle; +#endif + return rr; +} + +static inline int qm_mc_result_timeout(struct qm_portal *portal, + union qm_mc_result **mcr) +{ + int timeout = QM_MCR_TIMEOUT; + + do { + *mcr = qm_mc_result(portal); + if (*mcr) + break; + udelay(1); + } while (--timeout); + + return timeout; +} + +static inline void fq_set(struct qman_fq *fq, u32 mask) +{ + fq->flags |= mask; +} + +static inline void fq_clear(struct qman_fq *fq, u32 mask) +{ + fq->flags &= ~mask; +} + +static inline int fq_isset(struct qman_fq *fq, u32 mask) +{ + return fq->flags & mask; +} + +static inline int fq_isclear(struct qman_fq *fq, u32 mask) +{ + return !(fq->flags & mask); +} + +struct qman_portal { + struct qm_portal p; + /* PORTAL_BITS_*** - dynamic, strictly internal */ + unsigned long bits; + /* interrupt sources processed by portal_isr(), configurable */ + unsigned long irq_sources; + u32 use_eqcr_ci_stashing; + /* only 1 volatile dequeue at a time */ + struct qman_fq *vdqcr_owned; + u32 sdqcr; + /* probing time config params for cpu-affine portals */ + const struct qm_portal_config *config; + /* 2-element array. cgrs[0] is mask, cgrs[1] is snapshot. */ + struct qman_cgrs *cgrs; + /* linked-list of CSCN handlers. */ + struct list_head cgr_cbs; + /* list lock */ + spinlock_t cgr_lock; + struct work_struct congestion_work; + struct work_struct mr_work; + char irqname[MAX_IRQNAME]; +}; + +static cpumask_t affine_mask; +static DEFINE_SPINLOCK(affine_mask_lock); +static u16 affine_channels[NR_CPUS]; +static DEFINE_PER_CPU(struct qman_portal, qman_affine_portal); +struct qman_portal *affine_portals[NR_CPUS]; + +static inline struct qman_portal *get_affine_portal(void) +{ + return &get_cpu_var(qman_affine_portal); +} + +static inline void put_affine_portal(void) +{ + put_cpu_var(qman_affine_portal); +} + + +static inline struct qman_portal *get_portal_for_channel(u16 channel) +{ + int i; + + for (i = 0; i < num_possible_cpus(); i++) { + if (affine_portals[i] && + affine_portals[i]->config->channel == channel) + return affine_portals[i]; + } + + return NULL; +} + +static struct workqueue_struct *qm_portal_wq; + +int qman_dqrr_set_ithresh(struct qman_portal *portal, u8 ithresh) +{ + int res; + + if (!portal) + return -EINVAL; + + res = qm_dqrr_set_ithresh(&portal->p, ithresh); + if (res) + return res; + + portal->p.dqrr.ithresh = ithresh; + + return 0; +} +EXPORT_SYMBOL(qman_dqrr_set_ithresh); + +void qman_dqrr_get_ithresh(struct qman_portal *portal, u8 *ithresh) +{ + if (portal && ithresh) + *ithresh = qm_in(&portal->p, QM_REG_DQRR_ITR); +} +EXPORT_SYMBOL(qman_dqrr_get_ithresh); + +void qman_portal_get_iperiod(struct qman_portal *portal, u32 *iperiod) +{ + if (portal && iperiod) + *iperiod = qm_in(&portal->p, QM_REG_ITPR); +} +EXPORT_SYMBOL(qman_portal_get_iperiod); + +int qman_portal_set_iperiod(struct qman_portal *portal, u32 iperiod) +{ + if (!portal || iperiod > QMAN_ITP_MAX) + return -EINVAL; + + qm_out(&portal->p, QM_REG_ITPR, iperiod); + + return 0; +} +EXPORT_SYMBOL(qman_portal_set_iperiod); + +int qman_wq_alloc(void) +{ + qm_portal_wq = alloc_workqueue("qman_portal_wq", 0, 1); + if (!qm_portal_wq) + return -ENOMEM; + return 0; +} + + +void qman_enable_irqs(void) +{ + int i; + + for (i = 0; i < num_possible_cpus(); i++) { + if (affine_portals[i]) { + qm_out(&affine_portals[i]->p, QM_REG_ISR, 0xffffffff); + qm_out(&affine_portals[i]->p, QM_REG_IIR, 0); + } + + } +} + +/* + * This is what everything can wait on, even if it migrates to a different cpu + * to the one whose affine portal it is waiting on. + */ +static DECLARE_WAIT_QUEUE_HEAD(affine_queue); + +static struct qman_fq **fq_table; +static u32 num_fqids; + +int qman_alloc_fq_table(u32 _num_fqids) +{ + num_fqids = _num_fqids; + + fq_table = vzalloc(array3_size(sizeof(struct qman_fq *), + num_fqids, 2)); + if (!fq_table) + return -ENOMEM; + + pr_debug("Allocated fq lookup table at %p, entry count %u\n", + fq_table, num_fqids * 2); + return 0; +} + +static struct qman_fq *idx_to_fq(u32 idx) +{ + struct qman_fq *fq; + +#ifdef CONFIG_FSL_DPAA_CHECKING + if (WARN_ON(idx >= num_fqids * 2)) + return NULL; +#endif + fq = fq_table[idx]; + DPAA_ASSERT(!fq || idx == fq->idx); + + return fq; +} + +/* + * Only returns full-service fq objects, not enqueue-only + * references (QMAN_FQ_FLAG_NO_MODIFY). + */ +static struct qman_fq *fqid_to_fq(u32 fqid) +{ + return idx_to_fq(fqid * 2); +} + +static struct qman_fq *tag_to_fq(u32 tag) +{ +#if BITS_PER_LONG == 64 + return idx_to_fq(tag); +#else + return (struct qman_fq *)tag; +#endif +} + +static u32 fq_to_tag(struct qman_fq *fq) +{ +#if BITS_PER_LONG == 64 + return fq->idx; +#else + return (u32)fq; +#endif +} + +static u32 __poll_portal_slow(struct qman_portal *p, u32 is); +static inline unsigned int __poll_portal_fast(struct qman_portal *p, + unsigned int poll_limit, bool sched_napi); +static void qm_congestion_task(struct work_struct *work); +static void qm_mr_process_task(struct work_struct *work); + +static irqreturn_t portal_isr(int irq, void *ptr) +{ + struct qman_portal *p = ptr; + u32 is = qm_in(&p->p, QM_REG_ISR) & p->irq_sources; + u32 clear = 0; + + if (unlikely(!is)) + return IRQ_NONE; + + /* DQRR-handling if it's interrupt-driven */ + if (is & QM_PIRQ_DQRI) { + __poll_portal_fast(p, QMAN_POLL_LIMIT, true); + clear = QM_DQAVAIL_MASK | QM_PIRQ_DQRI; + } + /* Handling of anything else that's interrupt-driven */ + clear |= __poll_portal_slow(p, is) & QM_PIRQ_SLOW; + qm_out(&p->p, QM_REG_ISR, clear); + return IRQ_HANDLED; +} + +static int drain_mr_fqrni(struct qm_portal *p) +{ + const union qm_mr_entry *msg; +loop: + qm_mr_pvb_update(p); + msg = qm_mr_current(p); + if (!msg) { + /* + * if MR was full and h/w had other FQRNI entries to produce, we + * need to allow it time to produce those entries once the + * existing entries are consumed. A worst-case situation + * (fully-loaded system) means h/w sequencers may have to do 3-4 + * other things before servicing the portal's MR pump, each of + * which (if slow) may take ~50 qman cycles (which is ~200 + * processor cycles). So rounding up and then multiplying this + * worst-case estimate by a factor of 10, just to be + * ultra-paranoid, goes as high as 10,000 cycles. NB, we consume + * one entry at a time, so h/w has an opportunity to produce new + * entries well before the ring has been fully consumed, so + * we're being *really* paranoid here. + */ + mdelay(1); + qm_mr_pvb_update(p); + msg = qm_mr_current(p); + if (!msg) + return 0; + } + if ((msg->verb & QM_MR_VERB_TYPE_MASK) != QM_MR_VERB_FQRNI) { + /* We aren't draining anything but FQRNIs */ + pr_err("Found verb 0x%x in MR\n", msg->verb); + return -1; + } + qm_mr_next(p); + qm_mr_cci_consume(p, 1); + goto loop; +} + +static int qman_create_portal(struct qman_portal *portal, + const struct qm_portal_config *c, + const struct qman_cgrs *cgrs) +{ + struct qm_portal *p; + int ret; + u32 isdr; + + p = &portal->p; + +#ifdef CONFIG_FSL_PAMU + /* PAMU is required for stashing */ + portal->use_eqcr_ci_stashing = ((qman_ip_rev >= QMAN_REV30) ? 1 : 0); +#else + portal->use_eqcr_ci_stashing = 0; +#endif + /* + * prep the low-level portal struct with the mapped addresses from the + * config, everything that follows depends on it and "config" is more + * for (de)reference + */ + p->addr.ce = c->addr_virt_ce; + p->addr.ce_be = c->addr_virt_ce; + p->addr.ci = c->addr_virt_ci; + /* + * If CI-stashing is used, the current defaults use a threshold of 3, + * and stash with high-than-DQRR priority. + */ + if (qm_eqcr_init(p, qm_eqcr_pvb, + portal->use_eqcr_ci_stashing ? 3 : 0, 1)) { + dev_err(c->dev, "EQCR initialisation failed\n"); + goto fail_eqcr; + } + if (qm_dqrr_init(p, c, qm_dqrr_dpush, qm_dqrr_pvb, + qm_dqrr_cdc, DQRR_MAXFILL)) { + dev_err(c->dev, "DQRR initialisation failed\n"); + goto fail_dqrr; + } + if (qm_mr_init(p, qm_mr_pvb, qm_mr_cci)) { + dev_err(c->dev, "MR initialisation failed\n"); + goto fail_mr; + } + if (qm_mc_init(p)) { + dev_err(c->dev, "MC initialisation failed\n"); + goto fail_mc; + } + /* static interrupt-gating controls */ + qm_dqrr_set_ithresh(p, QMAN_PIRQ_DQRR_ITHRESH); + qm_mr_set_ithresh(p, QMAN_PIRQ_MR_ITHRESH); + qm_out(p, QM_REG_ITPR, QMAN_PIRQ_IPERIOD); + portal->cgrs = kmalloc_array(2, sizeof(*cgrs), GFP_KERNEL); + if (!portal->cgrs) + goto fail_cgrs; + /* initial snapshot is no-depletion */ + qman_cgrs_init(&portal->cgrs[1]); + if (cgrs) + portal->cgrs[0] = *cgrs; + else + /* if the given mask is NULL, assume all CGRs can be seen */ + qman_cgrs_fill(&portal->cgrs[0]); + INIT_LIST_HEAD(&portal->cgr_cbs); + spin_lock_init(&portal->cgr_lock); + INIT_WORK(&portal->congestion_work, qm_congestion_task); + INIT_WORK(&portal->mr_work, qm_mr_process_task); + portal->bits = 0; + portal->sdqcr = QM_SDQCR_SOURCE_CHANNELS | QM_SDQCR_COUNT_UPTO3 | + QM_SDQCR_DEDICATED_PRECEDENCE | QM_SDQCR_TYPE_PRIO_QOS | + QM_SDQCR_TOKEN_SET(0xab) | QM_SDQCR_CHANNELS_DEDICATED; + isdr = 0xffffffff; + qm_out(p, QM_REG_ISDR, isdr); + portal->irq_sources = 0; + qm_out(p, QM_REG_IER, 0); + snprintf(portal->irqname, MAX_IRQNAME, IRQNAME, c->cpu); + qm_out(p, QM_REG_IIR, 1); + if (request_irq(c->irq, portal_isr, 0, portal->irqname, portal)) { + dev_err(c->dev, "request_irq() failed\n"); + goto fail_irq; + } + + if (dpaa_set_portal_irq_affinity(c->dev, c->irq, c->cpu)) + goto fail_affinity; + + /* Need EQCR to be empty before continuing */ + isdr &= ~QM_PIRQ_EQCI; + qm_out(p, QM_REG_ISDR, isdr); + ret = qm_eqcr_get_fill(p); + if (ret) { + dev_err(c->dev, "EQCR unclean\n"); + goto fail_eqcr_empty; + } + isdr &= ~(QM_PIRQ_DQRI | QM_PIRQ_MRI); + qm_out(p, QM_REG_ISDR, isdr); + if (qm_dqrr_current(p)) { + dev_dbg(c->dev, "DQRR unclean\n"); + qm_dqrr_cdc_consume_n(p, 0xffff); + } + if (qm_mr_current(p) && drain_mr_fqrni(p)) { + /* special handling, drain just in case it's a few FQRNIs */ + const union qm_mr_entry *e = qm_mr_current(p); + + dev_err(c->dev, "MR dirty, VB 0x%x, rc 0x%x, addr 0x%llx\n", + e->verb, e->ern.rc, qm_fd_addr_get64(&e->ern.fd)); + goto fail_dqrr_mr_empty; + } + /* Success */ + portal->config = c; + qm_out(p, QM_REG_ISR, 0xffffffff); + qm_out(p, QM_REG_ISDR, 0); + if (!qman_requires_cleanup()) + qm_out(p, QM_REG_IIR, 0); + /* Write a sane SDQCR */ + qm_dqrr_sdqcr_set(p, portal->sdqcr); + return 0; + +fail_dqrr_mr_empty: +fail_eqcr_empty: +fail_affinity: + free_irq(c->irq, portal); +fail_irq: + kfree(portal->cgrs); +fail_cgrs: + qm_mc_finish(p); +fail_mc: + qm_mr_finish(p); +fail_mr: + qm_dqrr_finish(p); +fail_dqrr: + qm_eqcr_finish(p); +fail_eqcr: + return -EIO; +} + +struct qman_portal *qman_create_affine_portal(const struct qm_portal_config *c, + const struct qman_cgrs *cgrs) +{ + struct qman_portal *portal; + int err; + + portal = &per_cpu(qman_affine_portal, c->cpu); + err = qman_create_portal(portal, c, cgrs); + if (err) + return NULL; + + spin_lock(&affine_mask_lock); + cpumask_set_cpu(c->cpu, &affine_mask); + affine_channels[c->cpu] = c->channel; + affine_portals[c->cpu] = portal; + spin_unlock(&affine_mask_lock); + + return portal; +} + +static void qman_destroy_portal(struct qman_portal *qm) +{ + const struct qm_portal_config *pcfg; + + /* Stop dequeues on the portal */ + qm_dqrr_sdqcr_set(&qm->p, 0); + + /* + * NB we do this to "quiesce" EQCR. If we add enqueue-completions or + * something related to QM_PIRQ_EQCI, this may need fixing. + * Also, due to the prefetching model used for CI updates in the enqueue + * path, this update will only invalidate the CI cacheline *after* + * working on it, so we need to call this twice to ensure a full update + * irrespective of where the enqueue processing was at when the teardown + * began. + */ + qm_eqcr_cce_update(&qm->p); + qm_eqcr_cce_update(&qm->p); + pcfg = qm->config; + + free_irq(pcfg->irq, qm); + + kfree(qm->cgrs); + qm_mc_finish(&qm->p); + qm_mr_finish(&qm->p); + qm_dqrr_finish(&qm->p); + qm_eqcr_finish(&qm->p); + + qm->config = NULL; +} + +const struct qm_portal_config *qman_destroy_affine_portal(void) +{ + struct qman_portal *qm = get_affine_portal(); + const struct qm_portal_config *pcfg; + int cpu; + + pcfg = qm->config; + cpu = pcfg->cpu; + + qman_destroy_portal(qm); + + spin_lock(&affine_mask_lock); + cpumask_clear_cpu(cpu, &affine_mask); + spin_unlock(&affine_mask_lock); + put_affine_portal(); + return pcfg; +} + +/* Inline helper to reduce nesting in __poll_portal_slow() */ +static inline void fq_state_change(struct qman_portal *p, struct qman_fq *fq, + const union qm_mr_entry *msg, u8 verb) +{ + switch (verb) { + case QM_MR_VERB_FQRL: + DPAA_ASSERT(fq_isset(fq, QMAN_FQ_STATE_ORL)); + fq_clear(fq, QMAN_FQ_STATE_ORL); + break; + case QM_MR_VERB_FQRN: + DPAA_ASSERT(fq->state == qman_fq_state_parked || + fq->state == qman_fq_state_sched); + DPAA_ASSERT(fq_isset(fq, QMAN_FQ_STATE_CHANGING)); + fq_clear(fq, QMAN_FQ_STATE_CHANGING); + if (msg->fq.fqs & QM_MR_FQS_NOTEMPTY) + fq_set(fq, QMAN_FQ_STATE_NE); + if (msg->fq.fqs & QM_MR_FQS_ORLPRESENT) + fq_set(fq, QMAN_FQ_STATE_ORL); + fq->state = qman_fq_state_retired; + break; + case QM_MR_VERB_FQPN: + DPAA_ASSERT(fq->state == qman_fq_state_sched); + DPAA_ASSERT(fq_isclear(fq, QMAN_FQ_STATE_CHANGING)); + fq->state = qman_fq_state_parked; + } +} + +static void qm_congestion_task(struct work_struct *work) +{ + struct qman_portal *p = container_of(work, struct qman_portal, + congestion_work); + struct qman_cgrs rr, c; + union qm_mc_result *mcr; + struct qman_cgr *cgr; + + spin_lock(&p->cgr_lock); + qm_mc_start(&p->p); + qm_mc_commit(&p->p, QM_MCC_VERB_QUERYCONGESTION); + if (!qm_mc_result_timeout(&p->p, &mcr)) { + spin_unlock(&p->cgr_lock); + dev_crit(p->config->dev, "QUERYCONGESTION timeout\n"); + qman_p_irqsource_add(p, QM_PIRQ_CSCI); + return; + } + /* mask out the ones I'm not interested in */ + qman_cgrs_and(&rr, (struct qman_cgrs *)&mcr->querycongestion.state, + &p->cgrs[0]); + /* check previous snapshot for delta, enter/exit congestion */ + qman_cgrs_xor(&c, &rr, &p->cgrs[1]); + /* update snapshot */ + qman_cgrs_cp(&p->cgrs[1], &rr); + /* Invoke callback */ + list_for_each_entry(cgr, &p->cgr_cbs, node) + if (cgr->cb && qman_cgrs_get(&c, cgr->cgrid)) + cgr->cb(p, cgr, qman_cgrs_get(&rr, cgr->cgrid)); + spin_unlock(&p->cgr_lock); + qman_p_irqsource_add(p, QM_PIRQ_CSCI); +} + +static void qm_mr_process_task(struct work_struct *work) +{ + struct qman_portal *p = container_of(work, struct qman_portal, + mr_work); + const union qm_mr_entry *msg; + struct qman_fq *fq; + u8 verb, num = 0; + + preempt_disable(); + + while (1) { + qm_mr_pvb_update(&p->p); + msg = qm_mr_current(&p->p); + if (!msg) + break; + + verb = msg->verb & QM_MR_VERB_TYPE_MASK; + /* The message is a software ERN iff the 0x20 bit is clear */ + if (verb & 0x20) { + switch (verb) { + case QM_MR_VERB_FQRNI: + /* nada, we drop FQRNIs on the floor */ + break; + case QM_MR_VERB_FQRN: + case QM_MR_VERB_FQRL: + /* Lookup in the retirement table */ + fq = fqid_to_fq(qm_fqid_get(&msg->fq)); + if (WARN_ON(!fq)) + break; + fq_state_change(p, fq, msg, verb); + if (fq->cb.fqs) + fq->cb.fqs(p, fq, msg); + break; + case QM_MR_VERB_FQPN: + /* Parked */ + fq = tag_to_fq(be32_to_cpu(msg->fq.context_b)); + fq_state_change(p, fq, msg, verb); + if (fq->cb.fqs) + fq->cb.fqs(p, fq, msg); + break; + case QM_MR_VERB_DC_ERN: + /* DCP ERN */ + pr_crit_once("Leaking DCP ERNs!\n"); + break; + default: + pr_crit("Invalid MR verb 0x%02x\n", verb); + } + } else { + /* Its a software ERN */ + fq = tag_to_fq(be32_to_cpu(msg->ern.tag)); + fq->cb.ern(p, fq, msg); + } + num++; + qm_mr_next(&p->p); + } + + qm_mr_cci_consume(&p->p, num); + qman_p_irqsource_add(p, QM_PIRQ_MRI); + preempt_enable(); +} + +static u32 __poll_portal_slow(struct qman_portal *p, u32 is) +{ + if (is & QM_PIRQ_CSCI) { + qman_p_irqsource_remove(p, QM_PIRQ_CSCI); + queue_work_on(smp_processor_id(), qm_portal_wq, + &p->congestion_work); + } + + if (is & QM_PIRQ_EQRI) { + qm_eqcr_cce_update(&p->p); + qm_eqcr_set_ithresh(&p->p, 0); + wake_up(&affine_queue); + } + + if (is & QM_PIRQ_MRI) { + qman_p_irqsource_remove(p, QM_PIRQ_MRI); + queue_work_on(smp_processor_id(), qm_portal_wq, + &p->mr_work); + } + + return is; +} + +/* + * remove some slowish-path stuff from the "fast path" and make sure it isn't + * inlined. + */ +static noinline void clear_vdqcr(struct qman_portal *p, struct qman_fq *fq) +{ + p->vdqcr_owned = NULL; + fq_clear(fq, QMAN_FQ_STATE_VDQCR); + wake_up(&affine_queue); +} + +/* + * The only states that would conflict with other things if they ran at the + * same time on the same cpu are: + * + * (i) setting/clearing vdqcr_owned, and + * (ii) clearing the NE (Not Empty) flag. + * + * Both are safe. Because; + * + * (i) this clearing can only occur after qman_volatile_dequeue() has set the + * vdqcr_owned field (which it does before setting VDQCR), and + * qman_volatile_dequeue() blocks interrupts and preemption while this is + * done so that we can't interfere. + * (ii) the NE flag is only cleared after qman_retire_fq() has set it, and as + * with (i) that API prevents us from interfering until it's safe. + * + * The good thing is that qman_volatile_dequeue() and qman_retire_fq() run far + * less frequently (ie. per-FQ) than __poll_portal_fast() does, so the nett + * advantage comes from this function not having to "lock" anything at all. + * + * Note also that the callbacks are invoked at points which are safe against the + * above potential conflicts, but that this function itself is not re-entrant + * (this is because the function tracks one end of each FIFO in the portal and + * we do *not* want to lock that). So the consequence is that it is safe for + * user callbacks to call into any QMan API. + */ +static inline unsigned int __poll_portal_fast(struct qman_portal *p, + unsigned int poll_limit, bool sched_napi) +{ + const struct qm_dqrr_entry *dq; + struct qman_fq *fq; + enum qman_cb_dqrr_result res; + unsigned int limit = 0; + + do { + qm_dqrr_pvb_update(&p->p); + dq = qm_dqrr_current(&p->p); + if (!dq) + break; + + if (dq->stat & QM_DQRR_STAT_UNSCHEDULED) { + /* + * VDQCR: don't trust context_b as the FQ may have + * been configured for h/w consumption and we're + * draining it post-retirement. + */ + fq = p->vdqcr_owned; + /* + * We only set QMAN_FQ_STATE_NE when retiring, so we + * only need to check for clearing it when doing + * volatile dequeues. It's one less thing to check + * in the critical path (SDQCR). + */ + if (dq->stat & QM_DQRR_STAT_FQ_EMPTY) + fq_clear(fq, QMAN_FQ_STATE_NE); + /* + * This is duplicated from the SDQCR code, but we + * have stuff to do before *and* after this callback, + * and we don't want multiple if()s in the critical + * path (SDQCR). + */ + res = fq->cb.dqrr(p, fq, dq, sched_napi); + if (res == qman_cb_dqrr_stop) + break; + /* Check for VDQCR completion */ + if (dq->stat & QM_DQRR_STAT_DQCR_EXPIRED) + clear_vdqcr(p, fq); + } else { + /* SDQCR: context_b points to the FQ */ + fq = tag_to_fq(be32_to_cpu(dq->context_b)); + /* Now let the callback do its stuff */ + res = fq->cb.dqrr(p, fq, dq, sched_napi); + /* + * The callback can request that we exit without + * consuming this entry nor advancing; + */ + if (res == qman_cb_dqrr_stop) + break; + } + /* Interpret 'dq' from a driver perspective. */ + /* + * Parking isn't possible unless HELDACTIVE was set. NB, + * FORCEELIGIBLE implies HELDACTIVE, so we only need to + * check for HELDACTIVE to cover both. + */ + DPAA_ASSERT((dq->stat & QM_DQRR_STAT_FQ_HELDACTIVE) || + (res != qman_cb_dqrr_park)); + /* just means "skip it, I'll consume it myself later on" */ + if (res != qman_cb_dqrr_defer) + qm_dqrr_cdc_consume_1ptr(&p->p, dq, + res == qman_cb_dqrr_park); + /* Move forward */ + qm_dqrr_next(&p->p); + /* + * Entry processed and consumed, increment our counter. The + * callback can request that we exit after consuming the + * entry, and we also exit if we reach our processing limit, + * so loop back only if neither of these conditions is met. + */ + } while (++limit < poll_limit && res != qman_cb_dqrr_consume_stop); + + return limit; +} + +void qman_p_irqsource_add(struct qman_portal *p, u32 bits) +{ + unsigned long irqflags; + + local_irq_save(irqflags); + p->irq_sources |= bits & QM_PIRQ_VISIBLE; + qm_out(&p->p, QM_REG_IER, p->irq_sources); + local_irq_restore(irqflags); +} +EXPORT_SYMBOL(qman_p_irqsource_add); + +void qman_p_irqsource_remove(struct qman_portal *p, u32 bits) +{ + unsigned long irqflags; + u32 ier; + + /* + * Our interrupt handler only processes+clears status register bits that + * are in p->irq_sources. As we're trimming that mask, if one of them + * were to assert in the status register just before we remove it from + * the enable register, there would be an interrupt-storm when we + * release the IRQ lock. So we wait for the enable register update to + * take effect in h/w (by reading it back) and then clear all other bits + * in the status register. Ie. we clear them from ISR once it's certain + * IER won't allow them to reassert. + */ + local_irq_save(irqflags); + bits &= QM_PIRQ_VISIBLE; + p->irq_sources &= ~bits; + qm_out(&p->p, QM_REG_IER, p->irq_sources); + ier = qm_in(&p->p, QM_REG_IER); + /* + * Using "~ier" (rather than "bits" or "~p->irq_sources") creates a + * data-dependency, ie. to protect against re-ordering. + */ + qm_out(&p->p, QM_REG_ISR, ~ier); + local_irq_restore(irqflags); +} +EXPORT_SYMBOL(qman_p_irqsource_remove); + +const cpumask_t *qman_affine_cpus(void) +{ + return &affine_mask; +} +EXPORT_SYMBOL(qman_affine_cpus); + +u16 qman_affine_channel(int cpu) +{ + if (cpu < 0) { + struct qman_portal *portal = get_affine_portal(); + + cpu = portal->config->cpu; + put_affine_portal(); + } + WARN_ON(!cpumask_test_cpu(cpu, &affine_mask)); + return affine_channels[cpu]; +} +EXPORT_SYMBOL(qman_affine_channel); + +struct qman_portal *qman_get_affine_portal(int cpu) +{ + return affine_portals[cpu]; +} +EXPORT_SYMBOL(qman_get_affine_portal); + +int qman_start_using_portal(struct qman_portal *p, struct device *dev) +{ + return (!device_link_add(dev, p->config->dev, + DL_FLAG_AUTOREMOVE_CONSUMER)) ? -EINVAL : 0; +} +EXPORT_SYMBOL(qman_start_using_portal); + +int qman_p_poll_dqrr(struct qman_portal *p, unsigned int limit) +{ + return __poll_portal_fast(p, limit, false); +} +EXPORT_SYMBOL(qman_p_poll_dqrr); + +void qman_p_static_dequeue_add(struct qman_portal *p, u32 pools) +{ + unsigned long irqflags; + + local_irq_save(irqflags); + pools &= p->config->pools; + p->sdqcr |= pools; + qm_dqrr_sdqcr_set(&p->p, p->sdqcr); + local_irq_restore(irqflags); +} +EXPORT_SYMBOL(qman_p_static_dequeue_add); + +/* Frame queue API */ + +static const char *mcr_result_str(u8 result) +{ + switch (result) { + case QM_MCR_RESULT_NULL: + return "QM_MCR_RESULT_NULL"; + case QM_MCR_RESULT_OK: + return "QM_MCR_RESULT_OK"; + case QM_MCR_RESULT_ERR_FQID: + return "QM_MCR_RESULT_ERR_FQID"; + case QM_MCR_RESULT_ERR_FQSTATE: + return "QM_MCR_RESULT_ERR_FQSTATE"; + case QM_MCR_RESULT_ERR_NOTEMPTY: + return "QM_MCR_RESULT_ERR_NOTEMPTY"; + case QM_MCR_RESULT_PENDING: + return "QM_MCR_RESULT_PENDING"; + case QM_MCR_RESULT_ERR_BADCOMMAND: + return "QM_MCR_RESULT_ERR_BADCOMMAND"; + } + return "<unknown MCR result>"; +} + +int qman_create_fq(u32 fqid, u32 flags, struct qman_fq *fq) +{ + if (flags & QMAN_FQ_FLAG_DYNAMIC_FQID) { + int ret = qman_alloc_fqid(&fqid); + + if (ret) + return ret; + } + fq->fqid = fqid; + fq->flags = flags; + fq->state = qman_fq_state_oos; + fq->cgr_groupid = 0; + + /* A context_b of 0 is allegedly special, so don't use that fqid */ + if (fqid == 0 || fqid >= num_fqids) { + WARN(1, "bad fqid %d\n", fqid); + return -EINVAL; + } + + fq->idx = fqid * 2; + if (flags & QMAN_FQ_FLAG_NO_MODIFY) + fq->idx++; + + WARN_ON(fq_table[fq->idx]); + fq_table[fq->idx] = fq; + + return 0; +} +EXPORT_SYMBOL(qman_create_fq); + +void qman_destroy_fq(struct qman_fq *fq) +{ + /* + * We don't need to lock the FQ as it is a pre-condition that the FQ be + * quiesced. Instead, run some checks. + */ + switch (fq->state) { + case qman_fq_state_parked: + case qman_fq_state_oos: + if (fq_isset(fq, QMAN_FQ_FLAG_DYNAMIC_FQID)) + qman_release_fqid(fq->fqid); + + DPAA_ASSERT(fq_table[fq->idx]); + fq_table[fq->idx] = NULL; + return; + default: + break; + } + DPAA_ASSERT(NULL == "qman_free_fq() on unquiesced FQ!"); +} +EXPORT_SYMBOL(qman_destroy_fq); + +u32 qman_fq_fqid(struct qman_fq *fq) +{ + return fq->fqid; +} +EXPORT_SYMBOL(qman_fq_fqid); + +int qman_init_fq(struct qman_fq *fq, u32 flags, struct qm_mcc_initfq *opts) +{ + union qm_mc_command *mcc; + union qm_mc_result *mcr; + struct qman_portal *p; + u8 res, myverb; + int ret = 0; + + myverb = (flags & QMAN_INITFQ_FLAG_SCHED) + ? QM_MCC_VERB_INITFQ_SCHED : QM_MCC_VERB_INITFQ_PARKED; + + if (fq->state != qman_fq_state_oos && + fq->state != qman_fq_state_parked) + return -EINVAL; +#ifdef CONFIG_FSL_DPAA_CHECKING + if (fq_isset(fq, QMAN_FQ_FLAG_NO_MODIFY)) + return -EINVAL; +#endif + if (opts && (be16_to_cpu(opts->we_mask) & QM_INITFQ_WE_OAC)) { + /* And can't be set at the same time as TDTHRESH */ + if (be16_to_cpu(opts->we_mask) & QM_INITFQ_WE_TDTHRESH) + return -EINVAL; + } + /* Issue an INITFQ_[PARKED|SCHED] management command */ + p = get_affine_portal(); + if (fq_isset(fq, QMAN_FQ_STATE_CHANGING) || + (fq->state != qman_fq_state_oos && + fq->state != qman_fq_state_parked)) { + ret = -EBUSY; + goto out; + } + mcc = qm_mc_start(&p->p); + if (opts) + mcc->initfq = *opts; + qm_fqid_set(&mcc->fq, fq->fqid); + mcc->initfq.count = 0; + /* + * If the FQ does *not* have the TO_DCPORTAL flag, context_b is set as a + * demux pointer. Otherwise, the caller-provided value is allowed to + * stand, don't overwrite it. + */ + if (fq_isclear(fq, QMAN_FQ_FLAG_TO_DCPORTAL)) { + dma_addr_t phys_fq; + + mcc->initfq.we_mask |= cpu_to_be16(QM_INITFQ_WE_CONTEXTB); + mcc->initfq.fqd.context_b = cpu_to_be32(fq_to_tag(fq)); + /* + * and the physical address - NB, if the user wasn't trying to + * set CONTEXTA, clear the stashing settings. + */ + if (!(be16_to_cpu(mcc->initfq.we_mask) & + QM_INITFQ_WE_CONTEXTA)) { + mcc->initfq.we_mask |= + cpu_to_be16(QM_INITFQ_WE_CONTEXTA); + memset(&mcc->initfq.fqd.context_a, 0, + sizeof(mcc->initfq.fqd.context_a)); + } else { + struct qman_portal *p = qman_dma_portal; + + phys_fq = dma_map_single(p->config->dev, fq, + sizeof(*fq), DMA_TO_DEVICE); + if (dma_mapping_error(p->config->dev, phys_fq)) { + dev_err(p->config->dev, "dma_mapping failed\n"); + ret = -EIO; + goto out; + } + + qm_fqd_stashing_set64(&mcc->initfq.fqd, phys_fq); + } + } + if (flags & QMAN_INITFQ_FLAG_LOCAL) { + int wq = 0; + + if (!(be16_to_cpu(mcc->initfq.we_mask) & + QM_INITFQ_WE_DESTWQ)) { + mcc->initfq.we_mask |= + cpu_to_be16(QM_INITFQ_WE_DESTWQ); + wq = 4; + } + qm_fqd_set_destwq(&mcc->initfq.fqd, p->config->channel, wq); + } + qm_mc_commit(&p->p, myverb); + if (!qm_mc_result_timeout(&p->p, &mcr)) { + dev_err(p->config->dev, "MCR timeout\n"); + ret = -ETIMEDOUT; + goto out; + } + + DPAA_ASSERT((mcr->verb & QM_MCR_VERB_MASK) == myverb); + res = mcr->result; + if (res != QM_MCR_RESULT_OK) { + ret = -EIO; + goto out; + } + if (opts) { + if (be16_to_cpu(opts->we_mask) & QM_INITFQ_WE_FQCTRL) { + if (be16_to_cpu(opts->fqd.fq_ctrl) & QM_FQCTRL_CGE) + fq_set(fq, QMAN_FQ_STATE_CGR_EN); + else + fq_clear(fq, QMAN_FQ_STATE_CGR_EN); + } + if (be16_to_cpu(opts->we_mask) & QM_INITFQ_WE_CGID) + fq->cgr_groupid = opts->fqd.cgid; + } + fq->state = (flags & QMAN_INITFQ_FLAG_SCHED) ? + qman_fq_state_sched : qman_fq_state_parked; + +out: + put_affine_portal(); + return ret; +} +EXPORT_SYMBOL(qman_init_fq); + +int qman_schedule_fq(struct qman_fq *fq) +{ + union qm_mc_command *mcc; + union qm_mc_result *mcr; + struct qman_portal *p; + int ret = 0; + + if (fq->state != qman_fq_state_parked) + return -EINVAL; +#ifdef CONFIG_FSL_DPAA_CHECKING + if (fq_isset(fq, QMAN_FQ_FLAG_NO_MODIFY)) + return -EINVAL; +#endif + /* Issue a ALTERFQ_SCHED management command */ + p = get_affine_portal(); + if (fq_isset(fq, QMAN_FQ_STATE_CHANGING) || + fq->state != qman_fq_state_parked) { + ret = -EBUSY; + goto out; + } + mcc = qm_mc_start(&p->p); + qm_fqid_set(&mcc->fq, fq->fqid); + qm_mc_commit(&p->p, QM_MCC_VERB_ALTER_SCHED); + if (!qm_mc_result_timeout(&p->p, &mcr)) { + dev_err(p->config->dev, "ALTER_SCHED timeout\n"); + ret = -ETIMEDOUT; + goto out; + } + + DPAA_ASSERT((mcr->verb & QM_MCR_VERB_MASK) == QM_MCR_VERB_ALTER_SCHED); + if (mcr->result != QM_MCR_RESULT_OK) { + ret = -EIO; + goto out; + } + fq->state = qman_fq_state_sched; +out: + put_affine_portal(); + return ret; +} +EXPORT_SYMBOL(qman_schedule_fq); + +int qman_retire_fq(struct qman_fq *fq, u32 *flags) +{ + union qm_mc_command *mcc; + union qm_mc_result *mcr; + struct qman_portal *p; + int ret; + u8 res; + + if (fq->state != qman_fq_state_parked && + fq->state != qman_fq_state_sched) + return -EINVAL; +#ifdef CONFIG_FSL_DPAA_CHECKING + if (fq_isset(fq, QMAN_FQ_FLAG_NO_MODIFY)) + return -EINVAL; +#endif + p = get_affine_portal(); + if (fq_isset(fq, QMAN_FQ_STATE_CHANGING) || + fq->state == qman_fq_state_retired || + fq->state == qman_fq_state_oos) { + ret = -EBUSY; + goto out; + } + mcc = qm_mc_start(&p->p); + qm_fqid_set(&mcc->fq, fq->fqid); + qm_mc_commit(&p->p, QM_MCC_VERB_ALTER_RETIRE); + if (!qm_mc_result_timeout(&p->p, &mcr)) { + dev_crit(p->config->dev, "ALTER_RETIRE timeout\n"); + ret = -ETIMEDOUT; + goto out; + } + + DPAA_ASSERT((mcr->verb & QM_MCR_VERB_MASK) == QM_MCR_VERB_ALTER_RETIRE); + res = mcr->result; + /* + * "Elegant" would be to treat OK/PENDING the same way; set CHANGING, + * and defer the flags until FQRNI or FQRN (respectively) show up. But + * "Friendly" is to process OK immediately, and not set CHANGING. We do + * friendly, otherwise the caller doesn't necessarily have a fully + * "retired" FQ on return even if the retirement was immediate. However + * this does mean some code duplication between here and + * fq_state_change(). + */ + if (res == QM_MCR_RESULT_OK) { + ret = 0; + /* Process 'fq' right away, we'll ignore FQRNI */ + if (mcr->alterfq.fqs & QM_MCR_FQS_NOTEMPTY) + fq_set(fq, QMAN_FQ_STATE_NE); + if (mcr->alterfq.fqs & QM_MCR_FQS_ORLPRESENT) + fq_set(fq, QMAN_FQ_STATE_ORL); + if (flags) + *flags = fq->flags; + fq->state = qman_fq_state_retired; + if (fq->cb.fqs) { + /* + * Another issue with supporting "immediate" retirement + * is that we're forced to drop FQRNIs, because by the + * time they're seen it may already be "too late" (the + * fq may have been OOS'd and free()'d already). But if + * the upper layer wants a callback whether it's + * immediate or not, we have to fake a "MR" entry to + * look like an FQRNI... + */ + union qm_mr_entry msg; + + msg.verb = QM_MR_VERB_FQRNI; + msg.fq.fqs = mcr->alterfq.fqs; + qm_fqid_set(&msg.fq, fq->fqid); + msg.fq.context_b = cpu_to_be32(fq_to_tag(fq)); + fq->cb.fqs(p, fq, &msg); + } + } else if (res == QM_MCR_RESULT_PENDING) { + ret = 1; + fq_set(fq, QMAN_FQ_STATE_CHANGING); + } else { + ret = -EIO; + } +out: + put_affine_portal(); + return ret; +} +EXPORT_SYMBOL(qman_retire_fq); + +int qman_oos_fq(struct qman_fq *fq) +{ + union qm_mc_command *mcc; + union qm_mc_result *mcr; + struct qman_portal *p; + int ret = 0; + + if (fq->state != qman_fq_state_retired) + return -EINVAL; +#ifdef CONFIG_FSL_DPAA_CHECKING + if (fq_isset(fq, QMAN_FQ_FLAG_NO_MODIFY)) + return -EINVAL; +#endif + p = get_affine_portal(); + if (fq_isset(fq, QMAN_FQ_STATE_BLOCKOOS) || + fq->state != qman_fq_state_retired) { + ret = -EBUSY; + goto out; + } + mcc = qm_mc_start(&p->p); + qm_fqid_set(&mcc->fq, fq->fqid); + qm_mc_commit(&p->p, QM_MCC_VERB_ALTER_OOS); + if (!qm_mc_result_timeout(&p->p, &mcr)) { + ret = -ETIMEDOUT; + goto out; + } + DPAA_ASSERT((mcr->verb & QM_MCR_VERB_MASK) == QM_MCR_VERB_ALTER_OOS); + if (mcr->result != QM_MCR_RESULT_OK) { + ret = -EIO; + goto out; + } + fq->state = qman_fq_state_oos; +out: + put_affine_portal(); + return ret; +} +EXPORT_SYMBOL(qman_oos_fq); + +int qman_query_fq(struct qman_fq *fq, struct qm_fqd *fqd) +{ + union qm_mc_command *mcc; + union qm_mc_result *mcr; + struct qman_portal *p = get_affine_portal(); + int ret = 0; + + mcc = qm_mc_start(&p->p); + qm_fqid_set(&mcc->fq, fq->fqid); + qm_mc_commit(&p->p, QM_MCC_VERB_QUERYFQ); + if (!qm_mc_result_timeout(&p->p, &mcr)) { + ret = -ETIMEDOUT; + goto out; + } + + DPAA_ASSERT((mcr->verb & QM_MCR_VERB_MASK) == QM_MCR_VERB_QUERYFQ); + if (mcr->result == QM_MCR_RESULT_OK) + *fqd = mcr->queryfq.fqd; + else + ret = -EIO; +out: + put_affine_portal(); + return ret; +} + +int qman_query_fq_np(struct qman_fq *fq, struct qm_mcr_queryfq_np *np) +{ + union qm_mc_command *mcc; + union qm_mc_result *mcr; + struct qman_portal *p = get_affine_portal(); + int ret = 0; + + mcc = qm_mc_start(&p->p); + qm_fqid_set(&mcc->fq, fq->fqid); + qm_mc_commit(&p->p, QM_MCC_VERB_QUERYFQ_NP); + if (!qm_mc_result_timeout(&p->p, &mcr)) { + ret = -ETIMEDOUT; + goto out; + } + + DPAA_ASSERT((mcr->verb & QM_MCR_VERB_MASK) == QM_MCR_VERB_QUERYFQ_NP); + if (mcr->result == QM_MCR_RESULT_OK) + *np = mcr->queryfq_np; + else if (mcr->result == QM_MCR_RESULT_ERR_FQID) + ret = -ERANGE; + else + ret = -EIO; +out: + put_affine_portal(); + return ret; +} +EXPORT_SYMBOL(qman_query_fq_np); + +static int qman_query_cgr(struct qman_cgr *cgr, + struct qm_mcr_querycgr *cgrd) +{ + union qm_mc_command *mcc; + union qm_mc_result *mcr; + struct qman_portal *p = get_affine_portal(); + int ret = 0; + + mcc = qm_mc_start(&p->p); + mcc->cgr.cgid = cgr->cgrid; + qm_mc_commit(&p->p, QM_MCC_VERB_QUERYCGR); + if (!qm_mc_result_timeout(&p->p, &mcr)) { + ret = -ETIMEDOUT; + goto out; + } + DPAA_ASSERT((mcr->verb & QM_MCR_VERB_MASK) == QM_MCC_VERB_QUERYCGR); + if (mcr->result == QM_MCR_RESULT_OK) + *cgrd = mcr->querycgr; + else { + dev_err(p->config->dev, "QUERY_CGR failed: %s\n", + mcr_result_str(mcr->result)); + ret = -EIO; + } +out: + put_affine_portal(); + return ret; +} + +int qman_query_cgr_congested(struct qman_cgr *cgr, bool *result) +{ + struct qm_mcr_querycgr query_cgr; + int err; + + err = qman_query_cgr(cgr, &query_cgr); + if (err) + return err; + + *result = !!query_cgr.cgr.cs; + return 0; +} +EXPORT_SYMBOL(qman_query_cgr_congested); + +/* internal function used as a wait_event() expression */ +static int set_p_vdqcr(struct qman_portal *p, struct qman_fq *fq, u32 vdqcr) +{ + unsigned long irqflags; + int ret = -EBUSY; + + local_irq_save(irqflags); + if (p->vdqcr_owned) + goto out; + if (fq_isset(fq, QMAN_FQ_STATE_VDQCR)) + goto out; + + fq_set(fq, QMAN_FQ_STATE_VDQCR); + p->vdqcr_owned = fq; + qm_dqrr_vdqcr_set(&p->p, vdqcr); + ret = 0; +out: + local_irq_restore(irqflags); + return ret; +} + +static int set_vdqcr(struct qman_portal **p, struct qman_fq *fq, u32 vdqcr) +{ + int ret; + + *p = get_affine_portal(); + ret = set_p_vdqcr(*p, fq, vdqcr); + put_affine_portal(); + return ret; +} + +static int wait_vdqcr_start(struct qman_portal **p, struct qman_fq *fq, + u32 vdqcr, u32 flags) +{ + int ret = 0; + + if (flags & QMAN_VOLATILE_FLAG_WAIT_INT) + ret = wait_event_interruptible(affine_queue, + !set_vdqcr(p, fq, vdqcr)); + else + wait_event(affine_queue, !set_vdqcr(p, fq, vdqcr)); + return ret; +} + +int qman_volatile_dequeue(struct qman_fq *fq, u32 flags, u32 vdqcr) +{ + struct qman_portal *p; + int ret; + + if (fq->state != qman_fq_state_parked && + fq->state != qman_fq_state_retired) + return -EINVAL; + if (vdqcr & QM_VDQCR_FQID_MASK) + return -EINVAL; + if (fq_isset(fq, QMAN_FQ_STATE_VDQCR)) + return -EBUSY; + vdqcr = (vdqcr & ~QM_VDQCR_FQID_MASK) | fq->fqid; + if (flags & QMAN_VOLATILE_FLAG_WAIT) + ret = wait_vdqcr_start(&p, fq, vdqcr, flags); + else + ret = set_vdqcr(&p, fq, vdqcr); + if (ret) + return ret; + /* VDQCR is set */ + if (flags & QMAN_VOLATILE_FLAG_FINISH) { + if (flags & QMAN_VOLATILE_FLAG_WAIT_INT) + /* + * NB: don't propagate any error - the caller wouldn't + * know whether the VDQCR was issued or not. A signal + * could arrive after returning anyway, so the caller + * can check signal_pending() if that's an issue. + */ + wait_event_interruptible(affine_queue, + !fq_isset(fq, QMAN_FQ_STATE_VDQCR)); + else + wait_event(affine_queue, + !fq_isset(fq, QMAN_FQ_STATE_VDQCR)); + } + return 0; +} +EXPORT_SYMBOL(qman_volatile_dequeue); + +static void update_eqcr_ci(struct qman_portal *p, u8 avail) +{ + if (avail) + qm_eqcr_cce_prefetch(&p->p); + else + qm_eqcr_cce_update(&p->p); +} + +int qman_enqueue(struct qman_fq *fq, const struct qm_fd *fd) +{ + struct qman_portal *p; + struct qm_eqcr_entry *eq; + unsigned long irqflags; + u8 avail; + + p = get_affine_portal(); + local_irq_save(irqflags); + + if (p->use_eqcr_ci_stashing) { + /* + * The stashing case is easy, only update if we need to in + * order to try and liberate ring entries. + */ + eq = qm_eqcr_start_stash(&p->p); + } else { + /* + * The non-stashing case is harder, need to prefetch ahead of + * time. + */ + avail = qm_eqcr_get_avail(&p->p); + if (avail < 2) + update_eqcr_ci(p, avail); + eq = qm_eqcr_start_no_stash(&p->p); + } + + if (unlikely(!eq)) + goto out; + + qm_fqid_set(eq, fq->fqid); + eq->tag = cpu_to_be32(fq_to_tag(fq)); + eq->fd = *fd; + + qm_eqcr_pvb_commit(&p->p, QM_EQCR_VERB_CMD_ENQUEUE); +out: + local_irq_restore(irqflags); + put_affine_portal(); + return 0; +} +EXPORT_SYMBOL(qman_enqueue); + +static int qm_modify_cgr(struct qman_cgr *cgr, u32 flags, + struct qm_mcc_initcgr *opts) +{ + union qm_mc_command *mcc; + union qm_mc_result *mcr; + struct qman_portal *p = get_affine_portal(); + u8 verb = QM_MCC_VERB_MODIFYCGR; + int ret = 0; + + mcc = qm_mc_start(&p->p); + if (opts) + mcc->initcgr = *opts; + mcc->initcgr.cgid = cgr->cgrid; + if (flags & QMAN_CGR_FLAG_USE_INIT) + verb = QM_MCC_VERB_INITCGR; + qm_mc_commit(&p->p, verb); + if (!qm_mc_result_timeout(&p->p, &mcr)) { + ret = -ETIMEDOUT; + goto out; + } + + DPAA_ASSERT((mcr->verb & QM_MCR_VERB_MASK) == verb); + if (mcr->result != QM_MCR_RESULT_OK) + ret = -EIO; + +out: + put_affine_portal(); + return ret; +} + +#define PORTAL_IDX(n) (n->config->channel - QM_CHANNEL_SWPORTAL0) + +/* congestion state change notification target update control */ +static void qm_cgr_cscn_targ_set(struct __qm_mc_cgr *cgr, int pi, u32 val) +{ + if (qman_ip_rev >= QMAN_REV30) + cgr->cscn_targ_upd_ctrl = cpu_to_be16(pi | + QM_CGR_TARG_UDP_CTRL_WRITE_BIT); + else + cgr->cscn_targ = cpu_to_be32(val | QM_CGR_TARG_PORTAL(pi)); +} + +static void qm_cgr_cscn_targ_clear(struct __qm_mc_cgr *cgr, int pi, u32 val) +{ + if (qman_ip_rev >= QMAN_REV30) + cgr->cscn_targ_upd_ctrl = cpu_to_be16(pi); + else + cgr->cscn_targ = cpu_to_be32(val & ~QM_CGR_TARG_PORTAL(pi)); +} + +static u8 qman_cgr_cpus[CGR_NUM]; + +void qman_init_cgr_all(void) +{ + struct qman_cgr cgr; + int err_cnt = 0; + + for (cgr.cgrid = 0; cgr.cgrid < CGR_NUM; cgr.cgrid++) { + if (qm_modify_cgr(&cgr, QMAN_CGR_FLAG_USE_INIT, NULL)) + err_cnt++; + } + + if (err_cnt) + pr_err("Warning: %d error%s while initialising CGR h/w\n", + err_cnt, (err_cnt > 1) ? "s" : ""); +} + +int qman_create_cgr(struct qman_cgr *cgr, u32 flags, + struct qm_mcc_initcgr *opts) +{ + struct qm_mcr_querycgr cgr_state; + int ret; + struct qman_portal *p; + + /* + * We have to check that the provided CGRID is within the limits of the + * data-structures, for obvious reasons. However we'll let h/w take + * care of determining whether it's within the limits of what exists on + * the SoC. + */ + if (cgr->cgrid >= CGR_NUM) + return -EINVAL; + + preempt_disable(); + p = get_affine_portal(); + qman_cgr_cpus[cgr->cgrid] = smp_processor_id(); + preempt_enable(); + + cgr->chan = p->config->channel; + spin_lock(&p->cgr_lock); + + if (opts) { + struct qm_mcc_initcgr local_opts = *opts; + + ret = qman_query_cgr(cgr, &cgr_state); + if (ret) + goto out; + + qm_cgr_cscn_targ_set(&local_opts.cgr, PORTAL_IDX(p), + be32_to_cpu(cgr_state.cgr.cscn_targ)); + local_opts.we_mask |= cpu_to_be16(QM_CGR_WE_CSCN_TARG); + + /* send init if flags indicate so */ + if (flags & QMAN_CGR_FLAG_USE_INIT) + ret = qm_modify_cgr(cgr, QMAN_CGR_FLAG_USE_INIT, + &local_opts); + else + ret = qm_modify_cgr(cgr, 0, &local_opts); + if (ret) + goto out; + } + + list_add(&cgr->node, &p->cgr_cbs); + + /* Determine if newly added object requires its callback to be called */ + ret = qman_query_cgr(cgr, &cgr_state); + if (ret) { + /* we can't go back, so proceed and return success */ + dev_err(p->config->dev, "CGR HW state partially modified\n"); + ret = 0; + goto out; + } + if (cgr->cb && cgr_state.cgr.cscn_en && + qman_cgrs_get(&p->cgrs[1], cgr->cgrid)) + cgr->cb(p, cgr, 1); +out: + spin_unlock(&p->cgr_lock); + put_affine_portal(); + return ret; +} +EXPORT_SYMBOL(qman_create_cgr); + +static struct qman_portal *qman_cgr_get_affine_portal(struct qman_cgr *cgr) +{ + struct qman_portal *p = get_affine_portal(); + + if (cgr->chan != p->config->channel) { + /* attempt to delete from other portal than creator */ + dev_err(p->config->dev, "CGR not owned by current portal"); + dev_dbg(p->config->dev, " create 0x%x, delete 0x%x\n", + cgr->chan, p->config->channel); + put_affine_portal(); + return NULL; + } + + return p; +} + +int qman_delete_cgr(struct qman_cgr *cgr) +{ + unsigned long irqflags; + struct qm_mcr_querycgr cgr_state; + struct qm_mcc_initcgr local_opts; + int ret = 0; + struct qman_cgr *i; + struct qman_portal *p = qman_cgr_get_affine_portal(cgr); + + if (!p) + return -EINVAL; + + memset(&local_opts, 0, sizeof(struct qm_mcc_initcgr)); + spin_lock_irqsave(&p->cgr_lock, irqflags); + list_del(&cgr->node); + /* + * If there are no other CGR objects for this CGRID in the list, + * update CSCN_TARG accordingly + */ + list_for_each_entry(i, &p->cgr_cbs, node) + if (i->cgrid == cgr->cgrid && i->cb) + goto release_lock; + ret = qman_query_cgr(cgr, &cgr_state); + if (ret) { + /* add back to the list */ + list_add(&cgr->node, &p->cgr_cbs); + goto release_lock; + } + + local_opts.we_mask = cpu_to_be16(QM_CGR_WE_CSCN_TARG); + qm_cgr_cscn_targ_clear(&local_opts.cgr, PORTAL_IDX(p), + be32_to_cpu(cgr_state.cgr.cscn_targ)); + + ret = qm_modify_cgr(cgr, 0, &local_opts); + if (ret) + /* add back to the list */ + list_add(&cgr->node, &p->cgr_cbs); +release_lock: + spin_unlock_irqrestore(&p->cgr_lock, irqflags); + put_affine_portal(); + return ret; +} +EXPORT_SYMBOL(qman_delete_cgr); + +struct cgr_comp { + struct qman_cgr *cgr; + struct completion completion; +}; + +static void qman_delete_cgr_smp_call(void *p) +{ + qman_delete_cgr((struct qman_cgr *)p); +} + +void qman_delete_cgr_safe(struct qman_cgr *cgr) +{ + preempt_disable(); + if (qman_cgr_cpus[cgr->cgrid] != smp_processor_id()) { + smp_call_function_single(qman_cgr_cpus[cgr->cgrid], + qman_delete_cgr_smp_call, cgr, true); + preempt_enable(); + return; + } + + qman_delete_cgr(cgr); + preempt_enable(); +} +EXPORT_SYMBOL(qman_delete_cgr_safe); + +static int qman_update_cgr(struct qman_cgr *cgr, struct qm_mcc_initcgr *opts) +{ + int ret; + unsigned long irqflags; + struct qman_portal *p = qman_cgr_get_affine_portal(cgr); + + if (!p) + return -EINVAL; + + spin_lock_irqsave(&p->cgr_lock, irqflags); + ret = qm_modify_cgr(cgr, 0, opts); + spin_unlock_irqrestore(&p->cgr_lock, irqflags); + put_affine_portal(); + return ret; +} + +struct update_cgr_params { + struct qman_cgr *cgr; + struct qm_mcc_initcgr *opts; + int ret; +}; + +static void qman_update_cgr_smp_call(void *p) +{ + struct update_cgr_params *params = p; + + params->ret = qman_update_cgr(params->cgr, params->opts); +} + +int qman_update_cgr_safe(struct qman_cgr *cgr, struct qm_mcc_initcgr *opts) +{ + struct update_cgr_params params = { + .cgr = cgr, + .opts = opts, + }; + + preempt_disable(); + if (qman_cgr_cpus[cgr->cgrid] != smp_processor_id()) + smp_call_function_single(qman_cgr_cpus[cgr->cgrid], + qman_update_cgr_smp_call, ¶ms, + true); + else + params.ret = qman_update_cgr(cgr, opts); + preempt_enable(); + return params.ret; +} +EXPORT_SYMBOL(qman_update_cgr_safe); + +/* Cleanup FQs */ + +static int _qm_mr_consume_and_match_verb(struct qm_portal *p, int v) +{ + const union qm_mr_entry *msg; + int found = 0; + + qm_mr_pvb_update(p); + msg = qm_mr_current(p); + while (msg) { + if ((msg->verb & QM_MR_VERB_TYPE_MASK) == v) + found = 1; + qm_mr_next(p); + qm_mr_cci_consume_to_current(p); + qm_mr_pvb_update(p); + msg = qm_mr_current(p); + } + return found; +} + +static int _qm_dqrr_consume_and_match(struct qm_portal *p, u32 fqid, int s, + bool wait) +{ + const struct qm_dqrr_entry *dqrr; + int found = 0; + + do { + qm_dqrr_pvb_update(p); + dqrr = qm_dqrr_current(p); + if (!dqrr) + cpu_relax(); + } while (wait && !dqrr); + + while (dqrr) { + if (qm_fqid_get(dqrr) == fqid && (dqrr->stat & s)) + found = 1; + qm_dqrr_cdc_consume_1ptr(p, dqrr, 0); + qm_dqrr_pvb_update(p); + qm_dqrr_next(p); + dqrr = qm_dqrr_current(p); + } + return found; +} + +#define qm_mr_drain(p, V) \ + _qm_mr_consume_and_match_verb(p, QM_MR_VERB_##V) + +#define qm_dqrr_drain(p, f, S) \ + _qm_dqrr_consume_and_match(p, f, QM_DQRR_STAT_##S, false) + +#define qm_dqrr_drain_wait(p, f, S) \ + _qm_dqrr_consume_and_match(p, f, QM_DQRR_STAT_##S, true) + +#define qm_dqrr_drain_nomatch(p) \ + _qm_dqrr_consume_and_match(p, 0, 0, false) + +int qman_shutdown_fq(u32 fqid) +{ + struct qman_portal *p, *channel_portal; + struct device *dev; + union qm_mc_command *mcc; + union qm_mc_result *mcr; + int orl_empty, drain = 0, ret = 0; + u32 channel, res; + u8 state; + + p = get_affine_portal(); + dev = p->config->dev; + /* Determine the state of the FQID */ + mcc = qm_mc_start(&p->p); + qm_fqid_set(&mcc->fq, fqid); + qm_mc_commit(&p->p, QM_MCC_VERB_QUERYFQ_NP); + if (!qm_mc_result_timeout(&p->p, &mcr)) { + dev_err(dev, "QUERYFQ_NP timeout\n"); + ret = -ETIMEDOUT; + goto out; + } + + DPAA_ASSERT((mcr->verb & QM_MCR_VERB_MASK) == QM_MCR_VERB_QUERYFQ_NP); + state = mcr->queryfq_np.state & QM_MCR_NP_STATE_MASK; + if (state == QM_MCR_NP_STATE_OOS) + goto out; /* Already OOS, no need to do anymore checks */ + + /* Query which channel the FQ is using */ + mcc = qm_mc_start(&p->p); + qm_fqid_set(&mcc->fq, fqid); + qm_mc_commit(&p->p, QM_MCC_VERB_QUERYFQ); + if (!qm_mc_result_timeout(&p->p, &mcr)) { + dev_err(dev, "QUERYFQ timeout\n"); + ret = -ETIMEDOUT; + goto out; + } + + DPAA_ASSERT((mcr->verb & QM_MCR_VERB_MASK) == QM_MCR_VERB_QUERYFQ); + /* Need to store these since the MCR gets reused */ + channel = qm_fqd_get_chan(&mcr->queryfq.fqd); + qm_fqd_get_wq(&mcr->queryfq.fqd); + + if (channel < qm_channel_pool1) { + channel_portal = get_portal_for_channel(channel); + if (channel_portal == NULL) { + dev_err(dev, "Can't find portal for dedicated channel 0x%x\n", + channel); + ret = -EIO; + goto out; + } + } else + channel_portal = p; + + switch (state) { + case QM_MCR_NP_STATE_TEN_SCHED: + case QM_MCR_NP_STATE_TRU_SCHED: + case QM_MCR_NP_STATE_ACTIVE: + case QM_MCR_NP_STATE_PARKED: + orl_empty = 0; + mcc = qm_mc_start(&channel_portal->p); + qm_fqid_set(&mcc->fq, fqid); + qm_mc_commit(&channel_portal->p, QM_MCC_VERB_ALTER_RETIRE); + if (!qm_mc_result_timeout(&channel_portal->p, &mcr)) { + dev_err(dev, "ALTER_RETIRE timeout\n"); + ret = -ETIMEDOUT; + goto out; + } + DPAA_ASSERT((mcr->verb & QM_MCR_VERB_MASK) == + QM_MCR_VERB_ALTER_RETIRE); + res = mcr->result; /* Make a copy as we reuse MCR below */ + + if (res == QM_MCR_RESULT_OK) + drain_mr_fqrni(&channel_portal->p); + + if (res == QM_MCR_RESULT_PENDING) { + /* + * Need to wait for the FQRN in the message ring, which + * will only occur once the FQ has been drained. In + * order for the FQ to drain the portal needs to be set + * to dequeue from the channel the FQ is scheduled on + */ + int found_fqrn = 0; + + /* Flag that we need to drain FQ */ + drain = 1; + + if (channel >= qm_channel_pool1 && + channel < qm_channel_pool1 + 15) { + /* Pool channel, enable the bit in the portal */ + } else if (channel < qm_channel_pool1) { + /* Dedicated channel */ + } else { + dev_err(dev, "Can't recover FQ 0x%x, ch: 0x%x", + fqid, channel); + ret = -EBUSY; + goto out; + } + /* Set the sdqcr to drain this channel */ + if (channel < qm_channel_pool1) + qm_dqrr_sdqcr_set(&channel_portal->p, + QM_SDQCR_TYPE_ACTIVE | + QM_SDQCR_CHANNELS_DEDICATED); + else + qm_dqrr_sdqcr_set(&channel_portal->p, + QM_SDQCR_TYPE_ACTIVE | + QM_SDQCR_CHANNELS_POOL_CONV + (channel)); + do { + /* Keep draining DQRR while checking the MR*/ + qm_dqrr_drain_nomatch(&channel_portal->p); + /* Process message ring too */ + found_fqrn = qm_mr_drain(&channel_portal->p, + FQRN); + cpu_relax(); + } while (!found_fqrn); + /* Restore SDQCR */ + qm_dqrr_sdqcr_set(&channel_portal->p, + channel_portal->sdqcr); + + } + if (res != QM_MCR_RESULT_OK && + res != QM_MCR_RESULT_PENDING) { + dev_err(dev, "retire_fq failed: FQ 0x%x, res=0x%x\n", + fqid, res); + ret = -EIO; + goto out; + } + if (!(mcr->alterfq.fqs & QM_MCR_FQS_ORLPRESENT)) { + /* + * ORL had no entries, no need to wait until the + * ERNs come in + */ + orl_empty = 1; + } + /* + * Retirement succeeded, check to see if FQ needs + * to be drained + */ + if (drain || mcr->alterfq.fqs & QM_MCR_FQS_NOTEMPTY) { + /* FQ is Not Empty, drain using volatile DQ commands */ + do { + u32 vdqcr = fqid | QM_VDQCR_NUMFRAMES_SET(3); + + qm_dqrr_vdqcr_set(&p->p, vdqcr); + /* + * Wait for a dequeue and process the dequeues, + * making sure to empty the ring completely + */ + } while (!qm_dqrr_drain_wait(&p->p, fqid, FQ_EMPTY)); + } + + while (!orl_empty) { + /* Wait for the ORL to have been completely drained */ + orl_empty = qm_mr_drain(&p->p, FQRL); + cpu_relax(); + } + mcc = qm_mc_start(&p->p); + qm_fqid_set(&mcc->fq, fqid); + qm_mc_commit(&p->p, QM_MCC_VERB_ALTER_OOS); + if (!qm_mc_result_timeout(&p->p, &mcr)) { + ret = -ETIMEDOUT; + goto out; + } + + DPAA_ASSERT((mcr->verb & QM_MCR_VERB_MASK) == + QM_MCR_VERB_ALTER_OOS); + if (mcr->result != QM_MCR_RESULT_OK) { + dev_err(dev, "OOS after drain fail: FQ 0x%x (0x%x)\n", + fqid, mcr->result); + ret = -EIO; + goto out; + } + break; + + case QM_MCR_NP_STATE_RETIRED: + /* Send OOS Command */ + mcc = qm_mc_start(&p->p); + qm_fqid_set(&mcc->fq, fqid); + qm_mc_commit(&p->p, QM_MCC_VERB_ALTER_OOS); + if (!qm_mc_result_timeout(&p->p, &mcr)) { + ret = -ETIMEDOUT; + goto out; + } + + DPAA_ASSERT((mcr->verb & QM_MCR_VERB_MASK) == + QM_MCR_VERB_ALTER_OOS); + if (mcr->result != QM_MCR_RESULT_OK) { + dev_err(dev, "OOS fail: FQ 0x%x (0x%x)\n", + fqid, mcr->result); + ret = -EIO; + goto out; + } + break; + + case QM_MCR_NP_STATE_OOS: + /* Done */ + break; + + default: + ret = -EIO; + } + +out: + put_affine_portal(); + return ret; +} + +const struct qm_portal_config *qman_get_qm_portal_config( + struct qman_portal *portal) +{ + return portal->config; +} +EXPORT_SYMBOL(qman_get_qm_portal_config); + +struct gen_pool *qm_fqalloc; /* FQID allocator */ +struct gen_pool *qm_qpalloc; /* pool-channel allocator */ +struct gen_pool *qm_cgralloc; /* CGR ID allocator */ + +static int qman_alloc_range(struct gen_pool *p, u32 *result, u32 cnt) +{ + unsigned long addr; + + if (!p) + return -ENODEV; + + addr = gen_pool_alloc(p, cnt); + if (!addr) + return -ENOMEM; + + *result = addr & ~DPAA_GENALLOC_OFF; + + return 0; +} + +int qman_alloc_fqid_range(u32 *result, u32 count) +{ + return qman_alloc_range(qm_fqalloc, result, count); +} +EXPORT_SYMBOL(qman_alloc_fqid_range); + +int qman_alloc_pool_range(u32 *result, u32 count) +{ + return qman_alloc_range(qm_qpalloc, result, count); +} +EXPORT_SYMBOL(qman_alloc_pool_range); + +int qman_alloc_cgrid_range(u32 *result, u32 count) +{ + return qman_alloc_range(qm_cgralloc, result, count); +} +EXPORT_SYMBOL(qman_alloc_cgrid_range); + +int qman_release_fqid(u32 fqid) +{ + int ret = qman_shutdown_fq(fqid); + + if (ret) { + pr_debug("FQID %d leaked\n", fqid); + return ret; + } + + gen_pool_free(qm_fqalloc, fqid | DPAA_GENALLOC_OFF, 1); + return 0; +} +EXPORT_SYMBOL(qman_release_fqid); + +static int qpool_cleanup(u32 qp) +{ + /* + * We query all FQDs starting from + * FQID 1 until we get an "invalid FQID" error, looking for non-OOS FQDs + * whose destination channel is the pool-channel being released. + * When a non-OOS FQD is found we attempt to clean it up + */ + struct qman_fq fq = { + .fqid = QM_FQID_RANGE_START + }; + int err; + + do { + struct qm_mcr_queryfq_np np; + + err = qman_query_fq_np(&fq, &np); + if (err == -ERANGE) + /* FQID range exceeded, found no problems */ + return 0; + else if (WARN_ON(err)) + return err; + + if ((np.state & QM_MCR_NP_STATE_MASK) != QM_MCR_NP_STATE_OOS) { + struct qm_fqd fqd; + + err = qman_query_fq(&fq, &fqd); + if (WARN_ON(err)) + return err; + if (qm_fqd_get_chan(&fqd) == qp) { + /* The channel is the FQ's target, clean it */ + err = qman_shutdown_fq(fq.fqid); + if (err) + /* + * Couldn't shut down the FQ + * so the pool must be leaked + */ + return err; + } + } + /* Move to the next FQID */ + fq.fqid++; + } while (1); +} + +int qman_release_pool(u32 qp) +{ + int ret; + + ret = qpool_cleanup(qp); + if (ret) { + pr_debug("CHID %d leaked\n", qp); + return ret; + } + + gen_pool_free(qm_qpalloc, qp | DPAA_GENALLOC_OFF, 1); + return 0; +} +EXPORT_SYMBOL(qman_release_pool); + +static int cgr_cleanup(u32 cgrid) +{ + /* + * query all FQDs starting from FQID 1 until we get an "invalid FQID" + * error, looking for non-OOS FQDs whose CGR is the CGR being released + */ + struct qman_fq fq = { + .fqid = QM_FQID_RANGE_START + }; + int err; + + do { + struct qm_mcr_queryfq_np np; + + err = qman_query_fq_np(&fq, &np); + if (err == -ERANGE) + /* FQID range exceeded, found no problems */ + return 0; + else if (WARN_ON(err)) + return err; + + if ((np.state & QM_MCR_NP_STATE_MASK) != QM_MCR_NP_STATE_OOS) { + struct qm_fqd fqd; + + err = qman_query_fq(&fq, &fqd); + if (WARN_ON(err)) + return err; + if (be16_to_cpu(fqd.fq_ctrl) & QM_FQCTRL_CGE && + fqd.cgid == cgrid) { + pr_err("CRGID 0x%x is being used by FQID 0x%x, CGR will be leaked\n", + cgrid, fq.fqid); + return -EIO; + } + } + /* Move to the next FQID */ + fq.fqid++; + } while (1); +} + +int qman_release_cgrid(u32 cgrid) +{ + int ret; + + ret = cgr_cleanup(cgrid); + if (ret) { + pr_debug("CGRID %d leaked\n", cgrid); + return ret; + } + + gen_pool_free(qm_cgralloc, cgrid | DPAA_GENALLOC_OFF, 1); + return 0; +} +EXPORT_SYMBOL(qman_release_cgrid); |