mips: octeon: Add cvmx-helper-util.c
Import cvmx-helper-util.c from 2013 U-Boot. It will be used by the later
added drivers to support PCIe and networking on the MIPS Octeon II / III
platforms.
Signed-off-by: Aaron Williams <awilliams@marvell.com>
Signed-off-by: Stefan Roese <sr@denx.de>
diff --git a/arch/mips/mach-octeon/cvmx-helper-util.c b/arch/mips/mach-octeon/cvmx-helper-util.c
new file mode 100644
index 0000000..4625b45
--- /dev/null
+++ b/arch/mips/mach-octeon/cvmx-helper-util.c
@@ -0,0 +1,1225 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2020 Marvell International Ltd.
+ *
+ * Small helper utilities.
+ */
+
+#include <log.h>
+#include <time.h>
+#include <linux/delay.h>
+
+#include <mach/cvmx-regs.h>
+#include <mach/cvmx-csr-enums.h>
+#include <mach/octeon-model.h>
+#include <mach/octeon-feature.h>
+#include <mach/cvmx-gmxx-defs.h>
+#include <mach/cvmx-ipd-defs.h>
+#include <mach/cvmx-pko-defs.h>
+#include <mach/cvmx-ipd.h>
+#include <mach/cvmx-hwpko.h>
+#include <mach/cvmx-pki.h>
+#include <mach/cvmx-pip.h>
+#include <mach/cvmx-helper.h>
+#include <mach/cvmx-helper-util.h>
+#include <mach/cvmx-helper-pki.h>
+
+/**
+ * @INTERNAL
+ * These are the interface types needed to convert interface numbers to ipd
+ * ports.
+ *
+ * @param GMII
+ * This type is used for sgmii, rgmii, xaui and rxaui interfaces.
+ * @param ILK
+ * This type is used for ilk interfaces.
+ * @param SRIO
+ * This type is used for serial-RapidIo interfaces.
+ * @param NPI
+ * This type is used for npi interfaces.
+ * @param LB
+ * This type is used for loopback interfaces.
+ * @param INVALID_IF_TYPE
+ * This type indicates the interface hasn't been configured.
+ */
+enum port_map_if_type { INVALID_IF_TYPE = 0, GMII, ILK, SRIO, NPI, LB };
+
+/**
+ * @INTERNAL
+ * This structure is used to map interface numbers to ipd ports.
+ *
+ * @param type
+ * Interface type
+ * @param first_ipd_port
+ * First IPD port number assigned to this interface.
+ * @param last_ipd_port
+ * Last IPD port number assigned to this interface.
+ * @param ipd_port_adj
+ * Different octeon chips require different ipd ports for the
+ * same interface port/mode configuration. This value is used
+ * to account for that difference.
+ */
+struct ipd_port_map {
+ enum port_map_if_type type;
+ int first_ipd_port;
+ int last_ipd_port;
+ int ipd_port_adj;
+};
+
+/**
+ * @INTERNAL
+ * Interface number to ipd port map for the octeon 68xx.
+ */
+static const struct ipd_port_map ipd_port_map_68xx[CVMX_HELPER_MAX_IFACE] = {
+ { GMII, 0x800, 0x8ff, 0x40 }, /* Interface 0 */
+ { GMII, 0x900, 0x9ff, 0x40 }, /* Interface 1 */
+ { GMII, 0xa00, 0xaff, 0x40 }, /* Interface 2 */
+ { GMII, 0xb00, 0xbff, 0x40 }, /* Interface 3 */
+ { GMII, 0xc00, 0xcff, 0x40 }, /* Interface 4 */
+ { ILK, 0x400, 0x4ff, 0x00 }, /* Interface 5 */
+ { ILK, 0x500, 0x5ff, 0x00 }, /* Interface 6 */
+ { NPI, 0x100, 0x120, 0x00 }, /* Interface 7 */
+ { LB, 0x000, 0x008, 0x00 }, /* Interface 8 */
+};
+
+/**
+ * @INTERNAL
+ * Interface number to ipd port map for the octeon 78xx.
+ *
+ * This mapping corresponds to WQE(CHAN) enumeration in
+ * HRM Sections 11.15, PKI_CHAN_E, Section 11.6
+ *
+ */
+static const struct ipd_port_map ipd_port_map_78xx[CVMX_HELPER_MAX_IFACE] = {
+ { GMII, 0x800, 0x83f, 0x00 }, /* Interface 0 - BGX0 */
+ { GMII, 0x900, 0x93f, 0x00 }, /* Interface 1 -BGX1 */
+ { GMII, 0xa00, 0xa3f, 0x00 }, /* Interface 2 -BGX2 */
+ { GMII, 0xb00, 0xb3f, 0x00 }, /* Interface 3 - BGX3 */
+ { GMII, 0xc00, 0xc3f, 0x00 }, /* Interface 4 - BGX4 */
+ { GMII, 0xd00, 0xd3f, 0x00 }, /* Interface 5 - BGX5 */
+ { ILK, 0x400, 0x4ff, 0x00 }, /* Interface 6 - ILK0 */
+ { ILK, 0x500, 0x5ff, 0x00 }, /* Interface 7 - ILK1 */
+ { NPI, 0x100, 0x13f, 0x00 }, /* Interface 8 - DPI */
+ { LB, 0x000, 0x03f, 0x00 }, /* Interface 9 - LOOPBACK */
+};
+
+/**
+ * @INTERNAL
+ * Interface number to ipd port map for the octeon 73xx.
+ */
+static const struct ipd_port_map ipd_port_map_73xx[CVMX_HELPER_MAX_IFACE] = {
+ { GMII, 0x800, 0x83f, 0x00 }, /* Interface 0 - BGX(0,0-3) */
+ { GMII, 0x900, 0x93f, 0x00 }, /* Interface 1 -BGX(1,0-3) */
+ { GMII, 0xa00, 0xa3f, 0x00 }, /* Interface 2 -BGX(2,0-3) */
+ { NPI, 0x100, 0x17f, 0x00 }, /* Interface 3 - DPI */
+ { LB, 0x000, 0x03f, 0x00 }, /* Interface 4 - LOOPBACK */
+};
+
+/**
+ * @INTERNAL
+ * Interface number to ipd port map for the octeon 75xx.
+ */
+static const struct ipd_port_map ipd_port_map_75xx[CVMX_HELPER_MAX_IFACE] = {
+ { GMII, 0x800, 0x83f, 0x00 }, /* Interface 0 - BGX0 */
+ { SRIO, 0x240, 0x241, 0x00 }, /* Interface 1 - SRIO 0 */
+ { SRIO, 0x242, 0x243, 0x00 }, /* Interface 2 - SRIO 1 */
+ { NPI, 0x100, 0x13f, 0x00 }, /* Interface 3 - DPI */
+ { LB, 0x000, 0x03f, 0x00 }, /* Interface 4 - LOOPBACK */
+};
+
+/**
+ * Convert a interface mode into a human readable string
+ *
+ * @param mode Mode to convert
+ *
+ * @return String
+ */
+const char *cvmx_helper_interface_mode_to_string(cvmx_helper_interface_mode_t mode)
+{
+ switch (mode) {
+ case CVMX_HELPER_INTERFACE_MODE_DISABLED:
+ return "DISABLED";
+ case CVMX_HELPER_INTERFACE_MODE_RGMII:
+ return "RGMII";
+ case CVMX_HELPER_INTERFACE_MODE_GMII:
+ return "GMII";
+ case CVMX_HELPER_INTERFACE_MODE_SPI:
+ return "SPI";
+ case CVMX_HELPER_INTERFACE_MODE_PCIE:
+ return "PCIE";
+ case CVMX_HELPER_INTERFACE_MODE_XAUI:
+ return "XAUI";
+ case CVMX_HELPER_INTERFACE_MODE_RXAUI:
+ return "RXAUI";
+ case CVMX_HELPER_INTERFACE_MODE_SGMII:
+ return "SGMII";
+ case CVMX_HELPER_INTERFACE_MODE_QSGMII:
+ return "QSGMII";
+ case CVMX_HELPER_INTERFACE_MODE_PICMG:
+ return "PICMG";
+ case CVMX_HELPER_INTERFACE_MODE_NPI:
+ return "NPI";
+ case CVMX_HELPER_INTERFACE_MODE_LOOP:
+ return "LOOP";
+ case CVMX_HELPER_INTERFACE_MODE_SRIO:
+ return "SRIO";
+ case CVMX_HELPER_INTERFACE_MODE_ILK:
+ return "ILK";
+ case CVMX_HELPER_INTERFACE_MODE_AGL:
+ return "AGL";
+ case CVMX_HELPER_INTERFACE_MODE_XLAUI:
+ return "XLAUI";
+ case CVMX_HELPER_INTERFACE_MODE_XFI:
+ return "XFI";
+ case CVMX_HELPER_INTERFACE_MODE_40G_KR4:
+ return "40G_KR4";
+ case CVMX_HELPER_INTERFACE_MODE_10G_KR:
+ return "10G_KR";
+ case CVMX_HELPER_INTERFACE_MODE_MIXED:
+ return "MIXED";
+ }
+ return "UNKNOWN";
+}
+
+/**
+ * Debug routine to dump the packet structure to the console
+ *
+ * @param work Work queue entry containing the packet to dump
+ * @return
+ */
+int cvmx_helper_dump_packet(cvmx_wqe_t *work)
+{
+ u64 count;
+ u64 remaining_bytes;
+ union cvmx_buf_ptr buffer_ptr;
+ cvmx_buf_ptr_pki_t bptr;
+ cvmx_wqe_78xx_t *wqe = (void *)work;
+ u64 start_of_buffer;
+ u8 *data_address;
+ u8 *end_of_data;
+
+ if (octeon_has_feature(OCTEON_FEATURE_CN78XX_WQE)) {
+ cvmx_pki_dump_wqe(wqe);
+ cvmx_wqe_pki_errata_20776(work);
+ } else {
+ debug("WORD0 = %lx\n", (unsigned long)work->word0.u64);
+ debug("WORD1 = %lx\n", (unsigned long)work->word1.u64);
+ debug("WORD2 = %lx\n", (unsigned long)work->word2.u64);
+ debug("Packet Length: %u\n", cvmx_wqe_get_len(work));
+ debug(" Input Port: %u\n", cvmx_wqe_get_port(work));
+ debug(" QoS: %u\n", cvmx_wqe_get_qos(work));
+ debug(" Buffers: %u\n", cvmx_wqe_get_bufs(work));
+ }
+
+ if (cvmx_wqe_get_bufs(work) == 0) {
+ int wqe_pool;
+
+ if (octeon_has_feature(OCTEON_FEATURE_CN78XX_WQE)) {
+ debug("%s: ERROR: Unexpected bufs==0 in WQE\n", __func__);
+ return -1;
+ }
+ wqe_pool = (int)cvmx_fpa_get_wqe_pool();
+ buffer_ptr.u64 = 0;
+ buffer_ptr.s.pool = wqe_pool;
+
+ buffer_ptr.s.size = 128;
+ buffer_ptr.s.addr = cvmx_ptr_to_phys(work->packet_data);
+ if (cvmx_likely(!work->word2.s.not_IP)) {
+ union cvmx_pip_ip_offset pip_ip_offset;
+
+ pip_ip_offset.u64 = csr_rd(CVMX_PIP_IP_OFFSET);
+ buffer_ptr.s.addr +=
+ (pip_ip_offset.s.offset << 3) - work->word2.s.ip_offset;
+ buffer_ptr.s.addr += (work->word2.s.is_v6 ^ 1) << 2;
+ } else {
+ /*
+ * WARNING: This code assume that the packet
+ * is not RAW. If it was, we would use
+ * PIP_GBL_CFG[RAW_SHF] instead of
+ * PIP_GBL_CFG[NIP_SHF].
+ */
+ union cvmx_pip_gbl_cfg pip_gbl_cfg;
+
+ pip_gbl_cfg.u64 = csr_rd(CVMX_PIP_GBL_CFG);
+ buffer_ptr.s.addr += pip_gbl_cfg.s.nip_shf;
+ }
+ } else {
+ buffer_ptr = work->packet_ptr;
+ }
+
+ remaining_bytes = cvmx_wqe_get_len(work);
+
+ while (remaining_bytes) {
+ /* native cn78xx buffer format, unless legacy-translated */
+ if (octeon_has_feature(OCTEON_FEATURE_CN78XX_WQE) && !wqe->pki_wqe_translated) {
+ bptr.u64 = buffer_ptr.u64;
+ /* XXX- assumes cache-line aligned buffer */
+ start_of_buffer = (bptr.addr >> 7) << 7;
+ debug(" Buffer Start:%llx\n", (unsigned long long)start_of_buffer);
+ debug(" Buffer Data: %llx\n", (unsigned long long)bptr.addr);
+ debug(" Buffer Size: %u\n", bptr.size);
+ data_address = (uint8_t *)cvmx_phys_to_ptr(bptr.addr);
+ end_of_data = data_address + bptr.size;
+ } else {
+ start_of_buffer = ((buffer_ptr.s.addr >> 7) - buffer_ptr.s.back) << 7;
+ debug(" Buffer Start:%llx\n", (unsigned long long)start_of_buffer);
+ debug(" Buffer I : %u\n", buffer_ptr.s.i);
+ debug(" Buffer Back: %u\n", buffer_ptr.s.back);
+ debug(" Buffer Pool: %u\n", buffer_ptr.s.pool);
+ debug(" Buffer Data: %llx\n", (unsigned long long)buffer_ptr.s.addr);
+ debug(" Buffer Size: %u\n", buffer_ptr.s.size);
+ data_address = (uint8_t *)cvmx_phys_to_ptr(buffer_ptr.s.addr);
+ end_of_data = data_address + buffer_ptr.s.size;
+ }
+
+ debug("\t\t");
+ count = 0;
+ while (data_address < end_of_data) {
+ if (remaining_bytes == 0)
+ break;
+
+ remaining_bytes--;
+ debug("%02x", (unsigned int)*data_address);
+ data_address++;
+ if (remaining_bytes && count == 7) {
+ debug("\n\t\t");
+ count = 0;
+ } else {
+ count++;
+ }
+ }
+ debug("\n");
+
+ if (remaining_bytes) {
+ if (octeon_has_feature(OCTEON_FEATURE_CN78XX_WQE) &&
+ !wqe->pki_wqe_translated)
+ buffer_ptr.u64 = *(uint64_t *)cvmx_phys_to_ptr(bptr.addr - 8);
+ else
+ buffer_ptr.u64 =
+ *(uint64_t *)cvmx_phys_to_ptr(buffer_ptr.s.addr - 8);
+ }
+ }
+ return 0;
+}
+
+/**
+ * @INTERNAL
+ *
+ * Extract NO_WPTR mode from PIP/IPD register
+ */
+static int __cvmx_ipd_mode_no_wptr(void)
+{
+ if (octeon_has_feature(OCTEON_FEATURE_NO_WPTR)) {
+ cvmx_ipd_ctl_status_t ipd_ctl_status;
+
+ ipd_ctl_status.u64 = csr_rd(CVMX_IPD_CTL_STATUS);
+ return ipd_ctl_status.s.no_wptr;
+ }
+ return 0;
+}
+
+static cvmx_buf_ptr_t __cvmx_packet_short_ptr[4];
+static int8_t __cvmx_wqe_pool = -1;
+
+/**
+ * @INTERNAL
+ * Prepare packet pointer templace for dynamic short
+ * packets.
+ */
+static void cvmx_packet_short_ptr_calculate(void)
+{
+ unsigned int i, off;
+ union cvmx_pip_gbl_cfg pip_gbl_cfg;
+ union cvmx_pip_ip_offset pip_ip_offset;
+
+ /* Fill in the common values for all cases */
+ for (i = 0; i < 4; i++) {
+ if (__cvmx_ipd_mode_no_wptr())
+ /* packet pool, set to 0 in hardware */
+ __cvmx_wqe_pool = 0;
+ else
+ /* WQE pool as configured */
+ __cvmx_wqe_pool = csr_rd(CVMX_IPD_WQE_FPA_QUEUE) & 7;
+
+ __cvmx_packet_short_ptr[i].s.pool = __cvmx_wqe_pool;
+ __cvmx_packet_short_ptr[i].s.size = cvmx_fpa_get_block_size(__cvmx_wqe_pool);
+ __cvmx_packet_short_ptr[i].s.size -= 32;
+ __cvmx_packet_short_ptr[i].s.addr = 32;
+ }
+
+ pip_gbl_cfg.u64 = csr_rd(CVMX_PIP_GBL_CFG);
+ pip_ip_offset.u64 = csr_rd(CVMX_PIP_IP_OFFSET);
+
+ /* RAW_FULL: index = 0 */
+ i = 0;
+ off = pip_gbl_cfg.s.raw_shf;
+ __cvmx_packet_short_ptr[i].s.addr += off;
+ __cvmx_packet_short_ptr[i].s.size -= off;
+ __cvmx_packet_short_ptr[i].s.back += off >> 7;
+
+ /* NON-IP: index = 1 */
+ i = 1;
+ off = pip_gbl_cfg.s.nip_shf;
+ __cvmx_packet_short_ptr[i].s.addr += off;
+ __cvmx_packet_short_ptr[i].s.size -= off;
+ __cvmx_packet_short_ptr[i].s.back += off >> 7;
+
+ /* IPv4: index = 2 */
+ i = 2;
+ off = (pip_ip_offset.s.offset << 3) + 4;
+ __cvmx_packet_short_ptr[i].s.addr += off;
+ __cvmx_packet_short_ptr[i].s.size -= off;
+ __cvmx_packet_short_ptr[i].s.back += off >> 7;
+
+ /* IPv6: index = 3 */
+ i = 3;
+ off = (pip_ip_offset.s.offset << 3) + 0;
+ __cvmx_packet_short_ptr[i].s.addr += off;
+ __cvmx_packet_short_ptr[i].s.size -= off;
+ __cvmx_packet_short_ptr[i].s.back += off >> 7;
+
+ /* For IPv4/IPv6: subtract work->word2.s.ip_offset
+ * to addr, if it is smaller than IP_OFFSET[OFFSET]*8
+ * which is stored in __cvmx_packet_short_ptr[3].s.addr
+ */
+}
+
+/**
+ * Extract packet data buffer pointer from work queue entry.
+ *
+ * Returns the legacy (Octeon1/Octeon2) buffer pointer structure
+ * for the linked buffer list.
+ * On CN78XX, the native buffer pointer structure is converted into
+ * the legacy format.
+ * The legacy buf_ptr is then stored in the WQE, and word0 reserved
+ * field is set to indicate that the buffer pointers were translated.
+ * If the packet data is only found inside the work queue entry,
+ * a standard buffer pointer structure is created for it.
+ */
+cvmx_buf_ptr_t cvmx_wqe_get_packet_ptr(cvmx_wqe_t *work)
+{
+ if (octeon_has_feature(OCTEON_FEATURE_CN78XX_WQE)) {
+ cvmx_wqe_78xx_t *wqe = (void *)work;
+ cvmx_buf_ptr_t optr, lptr;
+ cvmx_buf_ptr_pki_t nptr;
+ unsigned int pool, bufs;
+ int node = cvmx_get_node_num();
+
+ /* In case of repeated calls of this function */
+ if (wqe->pki_wqe_translated || wqe->word2.software) {
+ optr.u64 = wqe->packet_ptr.u64;
+ return optr;
+ }
+
+ bufs = wqe->word0.bufs;
+ pool = wqe->word0.aura;
+ nptr.u64 = wqe->packet_ptr.u64;
+
+ optr.u64 = 0;
+ optr.s.pool = pool;
+ optr.s.addr = nptr.addr;
+ if (bufs == 1) {
+ optr.s.size = pki_dflt_pool[node].buffer_size -
+ pki_dflt_style[node].parm_cfg.first_skip - 8 -
+ wqe->word0.apad;
+ } else {
+ optr.s.size = nptr.size;
+ }
+
+ /* Calculate the "back" offset */
+ if (!nptr.packet_outside_wqe) {
+ optr.s.back = (nptr.addr -
+ cvmx_ptr_to_phys(wqe)) >> 7;
+ } else {
+ optr.s.back =
+ (pki_dflt_style[node].parm_cfg.first_skip +
+ 8 + wqe->word0.apad) >> 7;
+ }
+ lptr = optr;
+
+ /* Follow pointer and convert all linked pointers */
+ while (bufs > 1) {
+ void *vptr;
+
+ vptr = cvmx_phys_to_ptr(lptr.s.addr);
+
+ memcpy(&nptr, vptr - 8, 8);
+ /*
+ * Errata (PKI-20776) PKI_BUFLINK_S's are endian-swapped
+ * CN78XX pass 1.x has a bug where the packet pointer
+ * in each segment is written in the opposite
+ * endianness of the configured mode. Fix these here
+ */
+ if (OCTEON_IS_MODEL(OCTEON_CN78XX_PASS1_X))
+ nptr.u64 = __builtin_bswap64(nptr.u64);
+ lptr.u64 = 0;
+ lptr.s.pool = pool;
+ lptr.s.addr = nptr.addr;
+ lptr.s.size = nptr.size;
+ lptr.s.back = (pki_dflt_style[0].parm_cfg.later_skip + 8) >>
+ 7; /* TBD: not guaranteed !! */
+
+ memcpy(vptr - 8, &lptr, 8);
+ bufs--;
+ }
+ /* Store translated bufptr in WQE, and set indicator */
+ wqe->pki_wqe_translated = 1;
+ wqe->packet_ptr.u64 = optr.u64;
+ return optr;
+
+ } else {
+ unsigned int i;
+ unsigned int off = 0;
+ cvmx_buf_ptr_t bptr;
+
+ if (cvmx_likely(work->word2.s.bufs > 0))
+ return work->packet_ptr;
+
+ if (cvmx_unlikely(work->word2.s.software))
+ return work->packet_ptr;
+
+ /* first packet, precalculate packet_ptr templaces */
+ if (cvmx_unlikely(__cvmx_packet_short_ptr[0].u64 == 0))
+ cvmx_packet_short_ptr_calculate();
+
+ /* calculate templace index */
+ i = work->word2.s_cn38xx.not_IP | work->word2.s_cn38xx.rcv_error;
+ i = 2 ^ (i << 1);
+
+ /* IPv4/IPv6: Adjust IP offset */
+ if (cvmx_likely(i & 2)) {
+ i |= work->word2.s.is_v6;
+ off = work->word2.s.ip_offset;
+ } else {
+ /* RAWFULL/RAWSCHED should be handled here */
+ i = 1; /* not-IP */
+ off = 0;
+ }
+
+ /* Get the right templace */
+ bptr = __cvmx_packet_short_ptr[i];
+ bptr.s.addr -= off;
+ bptr.s.back = bptr.s.addr >> 7;
+
+ /* Add actual WQE paddr to the templace offset */
+ bptr.s.addr += cvmx_ptr_to_phys(work);
+
+ /* Adjust word2.bufs so that _free_data() handles it
+ * in the same way as PKO
+ */
+ work->word2.s.bufs = 1;
+
+ /* Store the new buffer pointer back into WQE */
+ work->packet_ptr = bptr;
+
+ /* Returned the synthetic buffer_pointer */
+ return bptr;
+ }
+}
+
+void cvmx_wqe_free(cvmx_wqe_t *work)
+{
+ unsigned int bufs, ncl = 1;
+ u64 paddr, paddr1;
+
+ if (octeon_has_feature(OCTEON_FEATURE_CN78XX_WQE)) {
+ cvmx_wqe_78xx_t *wqe = (void *)work;
+ cvmx_fpa3_gaura_t aura;
+ cvmx_buf_ptr_pki_t bptr;
+
+ bufs = wqe->word0.bufs;
+
+ if (!wqe->pki_wqe_translated && bufs != 0) {
+ /* Handle cn78xx native untralsated WQE */
+
+ bptr = wqe->packet_ptr;
+
+ /* Do nothing - first packet buffer shares WQE buffer */
+ if (!bptr.packet_outside_wqe)
+ return;
+ } else if (cvmx_likely(bufs != 0)) {
+ /* Handle translated 78XX WQE */
+ paddr = (work->packet_ptr.s.addr & (~0x7full)) -
+ (work->packet_ptr.s.back << 7);
+ paddr1 = cvmx_ptr_to_phys(work);
+
+ /* do not free WQE if contains first data buffer */
+ if (paddr == paddr1)
+ return;
+ }
+
+ /* WQE is separate from packet buffer, free it */
+ aura = __cvmx_fpa3_gaura(wqe->word0.aura >> 10, wqe->word0.aura & 0x3ff);
+
+ cvmx_fpa3_free(work, aura, ncl);
+ } else {
+ /* handle legacy WQE */
+ bufs = work->word2.s_cn38xx.bufs;
+
+ if (cvmx_likely(bufs != 0)) {
+ /* Check if the first data buffer is inside WQE */
+ paddr = (work->packet_ptr.s.addr & (~0x7full)) -
+ (work->packet_ptr.s.back << 7);
+ paddr1 = cvmx_ptr_to_phys(work);
+
+ /* do not free WQE if contains first data buffer */
+ if (paddr == paddr1)
+ return;
+ }
+
+ /* precalculate packet_ptr, WQE pool number */
+ if (cvmx_unlikely(__cvmx_wqe_pool < 0))
+ cvmx_packet_short_ptr_calculate();
+ cvmx_fpa1_free(work, __cvmx_wqe_pool, ncl);
+ }
+}
+
+/**
+ * Free the packet buffers contained in a work queue entry.
+ * The work queue entry is also freed if it contains packet data.
+ * If however the packet starts outside the WQE, the WQE will
+ * not be freed. The application should call cvmx_wqe_free()
+ * to free the WQE buffer that contains no packet data.
+ *
+ * @param work Work queue entry with packet to free
+ */
+void cvmx_helper_free_packet_data(cvmx_wqe_t *work)
+{
+ u64 number_buffers;
+ u64 start_of_buffer;
+ u64 next_buffer_ptr;
+ cvmx_fpa3_gaura_t aura;
+ unsigned int ncl;
+ cvmx_buf_ptr_t buffer_ptr;
+ cvmx_buf_ptr_pki_t bptr;
+ cvmx_wqe_78xx_t *wqe = (void *)work;
+ int o3_pki_wqe = 0;
+
+ number_buffers = cvmx_wqe_get_bufs(work);
+
+ buffer_ptr.u64 = work->packet_ptr.u64;
+
+ /* Zero-out WQE WORD3 so that the WQE is freed by cvmx_wqe_free() */
+ work->packet_ptr.u64 = 0;
+
+ if (number_buffers == 0)
+ return;
+
+ /* Interpret PKI-style bufptr unless it has been translated */
+ if (octeon_has_feature(OCTEON_FEATURE_CN78XX_WQE) &&
+ !wqe->pki_wqe_translated) {
+ o3_pki_wqe = 1;
+ cvmx_wqe_pki_errata_20776(work);
+ aura = __cvmx_fpa3_gaura(wqe->word0.aura >> 10,
+ wqe->word0.aura & 0x3ff);
+ } else {
+ start_of_buffer = ((buffer_ptr.s.addr >> 7) -
+ buffer_ptr.s.back) << 7;
+ next_buffer_ptr =
+ *(uint64_t *)cvmx_phys_to_ptr(buffer_ptr.s.addr - 8);
+ /*
+ * Since the number of buffers is not zero, we know this is not
+ * a dynamic short packet. We need to check if it is a packet
+ * received with IPD_CTL_STATUS[NO_WPTR]. If this is true,
+ * we need to free all buffers except for the first one.
+ * The caller doesn't expect their WQE pointer to be freed
+ */
+ if (cvmx_ptr_to_phys(work) == start_of_buffer) {
+ buffer_ptr.u64 = next_buffer_ptr;
+ number_buffers--;
+ }
+ }
+ while (number_buffers--) {
+ if (o3_pki_wqe) {
+ bptr.u64 = buffer_ptr.u64;
+
+ ncl = (bptr.size + CVMX_CACHE_LINE_SIZE - 1) /
+ CVMX_CACHE_LINE_SIZE;
+
+ /* XXX- assumes the buffer is cache-line aligned */
+ start_of_buffer = (bptr.addr >> 7) << 7;
+
+ /*
+ * Read pointer to next buffer before we free the
+ * current buffer.
+ */
+ next_buffer_ptr = *(uint64_t *)cvmx_phys_to_ptr(bptr.addr - 8);
+ /* FPA AURA comes from WQE, includes node */
+ cvmx_fpa3_free(cvmx_phys_to_ptr(start_of_buffer),
+ aura, ncl);
+ } else {
+ ncl = (buffer_ptr.s.size + CVMX_CACHE_LINE_SIZE - 1) /
+ CVMX_CACHE_LINE_SIZE +
+ buffer_ptr.s.back;
+ /*
+ * Calculate buffer start using "back" offset,
+ * Remember the back pointer is in cache lines,
+ * not 64bit words
+ */
+ start_of_buffer = ((buffer_ptr.s.addr >> 7) -
+ buffer_ptr.s.back) << 7;
+ /*
+ * Read pointer to next buffer before we free
+ * the current buffer.
+ */
+ next_buffer_ptr =
+ *(uint64_t *)cvmx_phys_to_ptr(buffer_ptr.s.addr - 8);
+ /* FPA pool comes from buf_ptr itself */
+ if (octeon_has_feature(OCTEON_FEATURE_CN78XX_WQE)) {
+ aura = cvmx_fpa1_pool_to_fpa3_aura(buffer_ptr.s.pool);
+ cvmx_fpa3_free(cvmx_phys_to_ptr(start_of_buffer),
+ aura, ncl);
+ } else {
+ cvmx_fpa1_free(cvmx_phys_to_ptr(start_of_buffer),
+ buffer_ptr.s.pool, ncl);
+ }
+ }
+ buffer_ptr.u64 = next_buffer_ptr;
+ }
+}
+
+void cvmx_helper_setup_legacy_red(int pass_thresh, int drop_thresh)
+{
+ unsigned int node = cvmx_get_node_num();
+ int aura, bpid;
+ int buf_cnt;
+ bool ena_red = 0, ena_drop = 0, ena_bp = 0;
+
+#define FPA_RED_AVG_DLY 1
+#define FPA_RED_LVL_DLY 3
+#define FPA_QOS_AVRG 0
+ /* Trying to make it backward compatible with older chips */
+
+ /* Setting up avg_dly and prb_dly, enable bits */
+ if (octeon_has_feature(OCTEON_FEATURE_FPA3)) {
+ cvmx_fpa3_config_red_params(node, FPA_QOS_AVRG,
+ FPA_RED_LVL_DLY, FPA_RED_AVG_DLY);
+ }
+
+ /* Disable backpressure on queued buffers which is aura in 78xx*/
+ /*
+ * Assumption is that all packets from all interface and ports goes
+ * in same poolx/aurax for backward compatibility
+ */
+ aura = cvmx_fpa_get_packet_pool();
+ buf_cnt = cvmx_fpa_get_packet_pool_buffer_count();
+ pass_thresh = buf_cnt - pass_thresh;
+ drop_thresh = buf_cnt - drop_thresh;
+ /* Map aura to bpid 0*/
+ bpid = 0;
+ cvmx_pki_write_aura_bpid(node, aura, bpid);
+ /* Don't enable back pressure */
+ ena_bp = 0;
+ /* enable RED */
+ ena_red = 1;
+ /*
+ * This will enable RED on all interfaces since
+ * they all have packet buffer coming from same aura
+ */
+ cvmx_helper_setup_aura_qos(node, aura, ena_red, ena_drop, pass_thresh,
+ drop_thresh, ena_bp, 0);
+}
+
+/**
+ * Setup Random Early Drop to automatically begin dropping packets.
+ *
+ * @param pass_thresh
+ * Packets will begin slowly dropping when there are less than
+ * this many packet buffers free in FPA 0.
+ * @param drop_thresh
+ * All incoming packets will be dropped when there are less
+ * than this many free packet buffers in FPA 0.
+ * @return Zero on success. Negative on failure
+ */
+int cvmx_helper_setup_red(int pass_thresh, int drop_thresh)
+{
+ if (octeon_has_feature(OCTEON_FEATURE_PKI))
+ cvmx_helper_setup_legacy_red(pass_thresh, drop_thresh);
+ else
+ cvmx_ipd_setup_red(pass_thresh, drop_thresh);
+ return 0;
+}
+
+/**
+ * @INTERNAL
+ * Setup the common GMX settings that determine the number of
+ * ports. These setting apply to almost all configurations of all
+ * chips.
+ *
+ * @param xiface Interface to configure
+ * @param num_ports Number of ports on the interface
+ *
+ * @return Zero on success, negative on failure
+ */
+int __cvmx_helper_setup_gmx(int xiface, int num_ports)
+{
+ union cvmx_gmxx_tx_prts gmx_tx_prts;
+ union cvmx_gmxx_rx_prts gmx_rx_prts;
+ union cvmx_pko_reg_gmx_port_mode pko_mode;
+ union cvmx_gmxx_txx_thresh gmx_tx_thresh;
+ struct cvmx_xiface xi = cvmx_helper_xiface_to_node_interface(xiface);
+ int index;
+
+ /*
+ * The common BGX settings are already done in the appropriate
+ * enable functions, nothing to do here.
+ */
+ if (octeon_has_feature(OCTEON_FEATURE_BGX))
+ return 0;
+
+ /* Tell GMX the number of TX ports on this interface */
+ gmx_tx_prts.u64 = csr_rd(CVMX_GMXX_TX_PRTS(xi.interface));
+ gmx_tx_prts.s.prts = num_ports;
+ csr_wr(CVMX_GMXX_TX_PRTS(xi.interface), gmx_tx_prts.u64);
+
+ /*
+ * Tell GMX the number of RX ports on this interface. This only applies
+ * to *GMII and XAUI ports.
+ */
+ switch (cvmx_helper_interface_get_mode(xiface)) {
+ case CVMX_HELPER_INTERFACE_MODE_RGMII:
+ case CVMX_HELPER_INTERFACE_MODE_SGMII:
+ case CVMX_HELPER_INTERFACE_MODE_QSGMII:
+ case CVMX_HELPER_INTERFACE_MODE_GMII:
+ case CVMX_HELPER_INTERFACE_MODE_XAUI:
+ case CVMX_HELPER_INTERFACE_MODE_RXAUI:
+ if (num_ports > 4) {
+ debug("%s: Illegal num_ports\n", __func__);
+ return -1;
+ }
+
+ gmx_rx_prts.u64 = csr_rd(CVMX_GMXX_RX_PRTS(xi.interface));
+ gmx_rx_prts.s.prts = num_ports;
+ csr_wr(CVMX_GMXX_RX_PRTS(xi.interface), gmx_rx_prts.u64);
+ break;
+
+ default:
+ break;
+ }
+
+ /*
+ * Skip setting CVMX_PKO_REG_GMX_PORT_MODE on 30XX, 31XX, 50XX,
+ * and 68XX.
+ */
+ if (!OCTEON_IS_MODEL(OCTEON_CN68XX)) {
+ /* Tell PKO the number of ports on this interface */
+ pko_mode.u64 = csr_rd(CVMX_PKO_REG_GMX_PORT_MODE);
+ if (xi.interface == 0) {
+ if (num_ports == 1)
+ pko_mode.s.mode0 = 4;
+ else if (num_ports == 2)
+ pko_mode.s.mode0 = 3;
+ else if (num_ports <= 4)
+ pko_mode.s.mode0 = 2;
+ else if (num_ports <= 8)
+ pko_mode.s.mode0 = 1;
+ else
+ pko_mode.s.mode0 = 0;
+ } else {
+ if (num_ports == 1)
+ pko_mode.s.mode1 = 4;
+ else if (num_ports == 2)
+ pko_mode.s.mode1 = 3;
+ else if (num_ports <= 4)
+ pko_mode.s.mode1 = 2;
+ else if (num_ports <= 8)
+ pko_mode.s.mode1 = 1;
+ else
+ pko_mode.s.mode1 = 0;
+ }
+ csr_wr(CVMX_PKO_REG_GMX_PORT_MODE, pko_mode.u64);
+ }
+
+ /*
+ * Set GMX to buffer as much data as possible before starting
+ * transmit. This reduces the chances that we have a TX under run
+ * due to memory contention. Any packet that fits entirely in the
+ * GMX FIFO can never have an under run regardless of memory load.
+ */
+ gmx_tx_thresh.u64 = csr_rd(CVMX_GMXX_TXX_THRESH(0, xi.interface));
+ /* ccn - common cnt numberator */
+ int ccn = 0x100;
+
+ /* Choose the max value for the number of ports */
+ if (num_ports <= 1)
+ gmx_tx_thresh.s.cnt = ccn / 1;
+ else if (num_ports == 2)
+ gmx_tx_thresh.s.cnt = ccn / 2;
+ else
+ gmx_tx_thresh.s.cnt = ccn / 4;
+
+ /*
+ * SPI and XAUI can have lots of ports but the GMX hardware
+ * only ever has a max of 4
+ */
+ if (num_ports > 4)
+ num_ports = 4;
+ for (index = 0; index < num_ports; index++)
+ csr_wr(CVMX_GMXX_TXX_THRESH(index, xi.interface), gmx_tx_thresh.u64);
+
+ /*
+ * For o68, we need to setup the pipes
+ */
+ if (OCTEON_IS_MODEL(OCTEON_CN68XX) && xi.interface < CVMX_HELPER_MAX_GMX) {
+ union cvmx_gmxx_txx_pipe config;
+
+ for (index = 0; index < num_ports; index++) {
+ config.u64 = 0;
+
+ if (__cvmx_helper_cfg_pko_port_base(xiface, index) >= 0) {
+ config.u64 = csr_rd(CVMX_GMXX_TXX_PIPE(index,
+ xi.interface));
+ config.s.nump = __cvmx_helper_cfg_pko_port_num(xiface,
+ index);
+ config.s.base = __cvmx_helper_cfg_pko_port_base(xiface,
+ index);
+ csr_wr(CVMX_GMXX_TXX_PIPE(index, xi.interface),
+ config.u64);
+ }
+ }
+ }
+
+ return 0;
+}
+
+int cvmx_helper_get_pko_port(int interface, int port)
+{
+ return cvmx_pko_get_base_pko_port(interface, port);
+}
+
+int cvmx_helper_get_ipd_port(int xiface, int index)
+{
+ struct cvmx_xiface xi = cvmx_helper_xiface_to_node_interface(xiface);
+
+ if (octeon_has_feature(OCTEON_FEATURE_PKND)) {
+ const struct ipd_port_map *port_map;
+ int ipd_port;
+
+ if (OCTEON_IS_MODEL(OCTEON_CN68XX)) {
+ port_map = ipd_port_map_68xx;
+ ipd_port = 0;
+ } else if (OCTEON_IS_MODEL(OCTEON_CN78XX)) {
+ port_map = ipd_port_map_78xx;
+ ipd_port = cvmx_helper_node_to_ipd_port(xi.node, 0);
+ } else if (OCTEON_IS_MODEL(OCTEON_CN73XX)) {
+ port_map = ipd_port_map_73xx;
+ ipd_port = 0;
+ } else if (OCTEON_IS_MODEL(OCTEON_CNF75XX)) {
+ port_map = ipd_port_map_75xx;
+ ipd_port = 0;
+ } else {
+ return -1;
+ }
+
+ ipd_port += port_map[xi.interface].first_ipd_port;
+ if (port_map[xi.interface].type == GMII) {
+ cvmx_helper_interface_mode_t mode;
+
+ mode = cvmx_helper_interface_get_mode(xiface);
+ if (mode == CVMX_HELPER_INTERFACE_MODE_XAUI ||
+ (mode == CVMX_HELPER_INTERFACE_MODE_RXAUI &&
+ OCTEON_IS_MODEL(OCTEON_CN68XX))) {
+ ipd_port += port_map[xi.interface].ipd_port_adj;
+ return ipd_port;
+ } else {
+ return ipd_port + (index * 16);
+ }
+ } else if (port_map[xi.interface].type == ILK) {
+ return ipd_port + index;
+ } else if (port_map[xi.interface].type == NPI) {
+ return ipd_port + index;
+ } else if (port_map[xi.interface].type == SRIO) {
+ return ipd_port + index;
+ } else if (port_map[xi.interface].type == LB) {
+ return ipd_port + index;
+ }
+
+ debug("ERROR: %s: interface %u:%u bad mode\n",
+ __func__, xi.node, xi.interface);
+ return -1;
+ } else if (cvmx_helper_interface_get_mode(xiface) ==
+ CVMX_HELPER_INTERFACE_MODE_AGL) {
+ return 24;
+ }
+
+ switch (xi.interface) {
+ case 0:
+ return index;
+ case 1:
+ return index + 16;
+ case 2:
+ return index + 32;
+ case 3:
+ return index + 36;
+ case 4:
+ return index + 40;
+ case 5:
+ return index + 42;
+ case 6:
+ return index + 44;
+ case 7:
+ return index + 46;
+ }
+ return -1;
+}
+
+int cvmx_helper_get_pknd(int xiface, int index)
+{
+ if (octeon_has_feature(OCTEON_FEATURE_PKND))
+ return __cvmx_helper_cfg_pknd(xiface, index);
+
+ return CVMX_INVALID_PKND;
+}
+
+int cvmx_helper_get_bpid(int interface, int port)
+{
+ if (octeon_has_feature(OCTEON_FEATURE_PKND))
+ return __cvmx_helper_cfg_bpid(interface, port);
+
+ return CVMX_INVALID_BPID;
+}
+
+/**
+ * Display interface statistics.
+ *
+ * @param port IPD/PKO port number
+ *
+ * @return none
+ */
+void cvmx_helper_show_stats(int port)
+{
+ cvmx_pip_port_status_t status;
+ cvmx_pko_port_status_t pko_status;
+
+ /* ILK stats */
+ if (octeon_has_feature(OCTEON_FEATURE_ILK))
+ __cvmx_helper_ilk_show_stats();
+
+ /* PIP stats */
+ cvmx_pip_get_port_stats(port, 0, &status);
+ debug("port %d: the number of packets - ipd: %d\n", port,
+ (int)status.packets);
+
+ /* PKO stats */
+ cvmx_pko_get_port_status(port, 0, &pko_status);
+ debug("port %d: the number of packets - pko: %d\n", port,
+ (int)pko_status.packets);
+
+ /* TODO: other stats */
+}
+
+/**
+ * Returns the interface number for an IPD/PKO port number.
+ *
+ * @param ipd_port IPD/PKO port number
+ *
+ * @return Interface number
+ */
+int cvmx_helper_get_interface_num(int ipd_port)
+{
+ if (OCTEON_IS_MODEL(OCTEON_CN68XX)) {
+ const struct ipd_port_map *port_map;
+ int i;
+ struct cvmx_xport xp = cvmx_helper_ipd_port_to_xport(ipd_port);
+
+ port_map = ipd_port_map_68xx;
+ for (i = 0; i < CVMX_HELPER_MAX_IFACE; i++) {
+ if (xp.port >= port_map[i].first_ipd_port &&
+ xp.port <= port_map[i].last_ipd_port)
+ return i;
+ }
+ return -1;
+ } else if (OCTEON_IS_MODEL(OCTEON_CN78XX)) {
+ const struct ipd_port_map *port_map;
+ int i;
+ struct cvmx_xport xp = cvmx_helper_ipd_port_to_xport(ipd_port);
+
+ port_map = ipd_port_map_78xx;
+ for (i = 0; i < CVMX_HELPER_MAX_IFACE; i++) {
+ if (xp.port >= port_map[i].first_ipd_port &&
+ xp.port <= port_map[i].last_ipd_port)
+ return cvmx_helper_node_interface_to_xiface(xp.node, i);
+ }
+ return -1;
+ } else if (OCTEON_IS_MODEL(OCTEON_CN73XX)) {
+ const struct ipd_port_map *port_map;
+ int i;
+ struct cvmx_xport xp = cvmx_helper_ipd_port_to_xport(ipd_port);
+
+ port_map = ipd_port_map_73xx;
+ for (i = 0; i < CVMX_HELPER_MAX_IFACE; i++) {
+ if (xp.port >= port_map[i].first_ipd_port &&
+ xp.port <= port_map[i].last_ipd_port)
+ return i;
+ }
+ return -1;
+ } else if (OCTEON_IS_MODEL(OCTEON_CNF75XX)) {
+ const struct ipd_port_map *port_map;
+ int i;
+ struct cvmx_xport xp = cvmx_helper_ipd_port_to_xport(ipd_port);
+
+ port_map = ipd_port_map_75xx;
+ for (i = 0; i < CVMX_HELPER_MAX_IFACE; i++) {
+ if (xp.port >= port_map[i].first_ipd_port &&
+ xp.port <= port_map[i].last_ipd_port)
+ return i;
+ }
+ return -1;
+ } else if (OCTEON_IS_MODEL(OCTEON_CN70XX) && ipd_port == 24) {
+ return 4;
+ }
+
+ if (ipd_port < 16)
+ return 0;
+ else if (ipd_port < 32)
+ return 1;
+ else if (ipd_port < 36)
+ return 2;
+ else if (ipd_port < 40)
+ return 3;
+ else if (ipd_port < 42)
+ return 4;
+ else if (ipd_port < 44)
+ return 5;
+ else if (ipd_port < 46)
+ return 6;
+ else if (ipd_port < 48)
+ return 7;
+
+ debug("%s: Illegal IPD port number %d\n", __func__, ipd_port);
+ return -1;
+}
+
+/**
+ * Returns the interface index number for an IPD/PKO port
+ * number.
+ *
+ * @param ipd_port IPD/PKO port number
+ *
+ * @return Interface index number
+ */
+int cvmx_helper_get_interface_index_num(int ipd_port)
+{
+ if (octeon_has_feature(OCTEON_FEATURE_PKND)) {
+ const struct ipd_port_map *port_map;
+ int port;
+ enum port_map_if_type type = INVALID_IF_TYPE;
+ int i;
+ int num_interfaces;
+
+ if (OCTEON_IS_MODEL(OCTEON_CN68XX)) {
+ port_map = ipd_port_map_68xx;
+ } else if (OCTEON_IS_MODEL(OCTEON_CN78XX)) {
+ struct cvmx_xport xp = cvmx_helper_ipd_port_to_xport(ipd_port);
+
+ port_map = ipd_port_map_78xx;
+ ipd_port = xp.port;
+ } else if (OCTEON_IS_MODEL(OCTEON_CN73XX)) {
+ struct cvmx_xport xp = cvmx_helper_ipd_port_to_xport(ipd_port);
+
+ port_map = ipd_port_map_73xx;
+ ipd_port = xp.port;
+ } else if (OCTEON_IS_MODEL(OCTEON_CNF75XX)) {
+ struct cvmx_xport xp = cvmx_helper_ipd_port_to_xport(ipd_port);
+
+ port_map = ipd_port_map_75xx;
+ ipd_port = xp.port;
+ } else {
+ return -1;
+ }
+
+ num_interfaces = cvmx_helper_get_number_of_interfaces();
+
+ /* Get the interface type of the ipd port */
+ for (i = 0; i < num_interfaces; i++) {
+ if (ipd_port >= port_map[i].first_ipd_port &&
+ ipd_port <= port_map[i].last_ipd_port) {
+ type = port_map[i].type;
+ break;
+ }
+ }
+
+ /* Convert the ipd port to the interface port */
+ switch (type) {
+ /* Ethernet interfaces have a channel in lower 4 bits
+ * that is does not discriminate traffic, and is ignored.
+ */
+ case GMII:
+ port = ipd_port - port_map[i].first_ipd_port;
+
+ /* CN68XX adds 0x40 to IPD_PORT when in XAUI/RXAUI
+ * mode of operation, adjust for that case
+ */
+ if (port >= port_map[i].ipd_port_adj)
+ port -= port_map[i].ipd_port_adj;
+
+ port >>= 4;
+ return port;
+
+ /*
+ * These interfaces do not have physical ports,
+ * but have logical channels instead that separate
+ * traffic into logical streams
+ */
+ case ILK:
+ case SRIO:
+ case NPI:
+ case LB:
+ port = ipd_port - port_map[i].first_ipd_port;
+ return port;
+
+ default:
+ printf("ERROR: %s: Illegal IPD port number %#x\n",
+ __func__, ipd_port);
+ return -1;
+ }
+ }
+ if (OCTEON_IS_MODEL(OCTEON_CN70XX))
+ return ipd_port & 3;
+ if (ipd_port < 32)
+ return ipd_port & 15;
+ else if (ipd_port < 40)
+ return ipd_port & 3;
+ else if (ipd_port < 48)
+ return ipd_port & 1;
+
+ debug("%s: Illegal IPD port number\n", __func__);
+
+ return -1;
+}
+
+/**
+ * Prints out a buffer with the address, hex bytes, and ASCII
+ *
+ * @param addr Start address to print on the left
+ * @param[in] buffer array of bytes to print
+ * @param count Number of bytes to print
+ */
+void cvmx_print_buffer_u8(unsigned int addr, const uint8_t *buffer,
+ size_t count)
+{
+ uint i;
+
+ while (count) {
+ unsigned int linelen = count < 16 ? count : 16;
+
+ debug("%08x:", addr);
+
+ for (i = 0; i < linelen; i++)
+ debug(" %0*x", 2, buffer[i]);
+
+ while (i++ < 17)
+ debug(" ");
+
+ for (i = 0; i < linelen; i++) {
+ if (buffer[i] >= 0x20 && buffer[i] < 0x7f)
+ debug("%c", buffer[i]);
+ else
+ debug(".");
+ }
+ debug("\n");
+ addr += linelen;
+ buffer += linelen;
+ count -= linelen;
+ }
+}