linux_dsm_epyc7002/drivers/net/wan/ixp4xx_hss.c

1420 lines
37 KiB
C
Raw Normal View History

/*
* Intel IXP4xx HSS (synchronous serial port) driver for Linux
*
* Copyright (C) 2007-2008 Krzysztof Hałasa <khc@pm.waw.pl>
*
* This program is free software; you can redistribute it and/or modify it
* under the terms of version 2 of the GNU General Public License
* as published by the Free Software Foundation.
*/
#include <linux/bitops.h>
#include <linux/cdev.h>
#include <linux/dma-mapping.h>
#include <linux/dmapool.h>
#include <linux/fs.h>
#include <linux/hdlc.h>
#include <linux/io.h>
#include <linux/kernel.h>
#include <linux/platform_device.h>
#include <linux/poll.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 15:04:11 +07:00
#include <linux/slab.h>
#include <mach/npe.h>
#include <mach/qmgr.h>
#define DEBUG_DESC 0
#define DEBUG_RX 0
#define DEBUG_TX 0
#define DEBUG_PKT_BYTES 0
#define DEBUG_CLOSE 0
#define DRV_NAME "ixp4xx_hss"
#define PKT_EXTRA_FLAGS 0 /* orig 1 */
#define PKT_NUM_PIPES 1 /* 1, 2 or 4 */
#define PKT_PIPE_FIFO_SIZEW 4 /* total 4 dwords per HSS */
#define RX_DESCS 16 /* also length of all RX queues */
#define TX_DESCS 16 /* also length of all TX queues */
#define POOL_ALLOC_SIZE (sizeof(struct desc) * (RX_DESCS + TX_DESCS))
#define RX_SIZE (HDLC_MAX_MRU + 4) /* NPE needs more space */
#define MAX_CLOSE_WAIT 1000 /* microseconds */
#define HSS_COUNT 2
#define FRAME_SIZE 256 /* doesn't matter at this point */
#define FRAME_OFFSET 0
#define MAX_CHANNELS (FRAME_SIZE / 8)
#define NAPI_WEIGHT 16
/* Queue IDs */
#define HSS0_CHL_RXTRIG_QUEUE 12 /* orig size = 32 dwords */
#define HSS0_PKT_RX_QUEUE 13 /* orig size = 32 dwords */
#define HSS0_PKT_TX0_QUEUE 14 /* orig size = 16 dwords */
#define HSS0_PKT_TX1_QUEUE 15
#define HSS0_PKT_TX2_QUEUE 16
#define HSS0_PKT_TX3_QUEUE 17
#define HSS0_PKT_RXFREE0_QUEUE 18 /* orig size = 16 dwords */
#define HSS0_PKT_RXFREE1_QUEUE 19
#define HSS0_PKT_RXFREE2_QUEUE 20
#define HSS0_PKT_RXFREE3_QUEUE 21
#define HSS0_PKT_TXDONE_QUEUE 22 /* orig size = 64 dwords */
#define HSS1_CHL_RXTRIG_QUEUE 10
#define HSS1_PKT_RX_QUEUE 0
#define HSS1_PKT_TX0_QUEUE 5
#define HSS1_PKT_TX1_QUEUE 6
#define HSS1_PKT_TX2_QUEUE 7
#define HSS1_PKT_TX3_QUEUE 8
#define HSS1_PKT_RXFREE0_QUEUE 1
#define HSS1_PKT_RXFREE1_QUEUE 2
#define HSS1_PKT_RXFREE2_QUEUE 3
#define HSS1_PKT_RXFREE3_QUEUE 4
#define HSS1_PKT_TXDONE_QUEUE 9
#define NPE_PKT_MODE_HDLC 0
#define NPE_PKT_MODE_RAW 1
#define NPE_PKT_MODE_56KMODE 2
#define NPE_PKT_MODE_56KENDIAN_MSB 4
/* PKT_PIPE_HDLC_CFG_WRITE flags */
#define PKT_HDLC_IDLE_ONES 0x1 /* default = flags */
#define PKT_HDLC_CRC_32 0x2 /* default = CRC-16 */
#define PKT_HDLC_MSB_ENDIAN 0x4 /* default = LE */
/* hss_config, PCRs */
/* Frame sync sampling, default = active low */
#define PCR_FRM_SYNC_ACTIVE_HIGH 0x40000000
#define PCR_FRM_SYNC_FALLINGEDGE 0x80000000
#define PCR_FRM_SYNC_RISINGEDGE 0xC0000000
/* Frame sync pin: input (default) or output generated off a given clk edge */
#define PCR_FRM_SYNC_OUTPUT_FALLING 0x20000000
#define PCR_FRM_SYNC_OUTPUT_RISING 0x30000000
/* Frame and data clock sampling on edge, default = falling */
#define PCR_FCLK_EDGE_RISING 0x08000000
#define PCR_DCLK_EDGE_RISING 0x04000000
/* Clock direction, default = input */
#define PCR_SYNC_CLK_DIR_OUTPUT 0x02000000
/* Generate/Receive frame pulses, default = enabled */
#define PCR_FRM_PULSE_DISABLED 0x01000000
/* Data rate is full (default) or half the configured clk speed */
#define PCR_HALF_CLK_RATE 0x00200000
/* Invert data between NPE and HSS FIFOs? (default = no) */
#define PCR_DATA_POLARITY_INVERT 0x00100000
/* TX/RX endianness, default = LSB */
#define PCR_MSB_ENDIAN 0x00080000
/* Normal (default) / open drain mode (TX only) */
#define PCR_TX_PINS_OPEN_DRAIN 0x00040000
/* No framing bit transmitted and expected on RX? (default = framing bit) */
#define PCR_SOF_NO_FBIT 0x00020000
/* Drive data pins? */
#define PCR_TX_DATA_ENABLE 0x00010000
/* Voice 56k type: drive the data pins low (default), high, high Z */
#define PCR_TX_V56K_HIGH 0x00002000
#define PCR_TX_V56K_HIGH_IMP 0x00004000
/* Unassigned type: drive the data pins low (default), high, high Z */
#define PCR_TX_UNASS_HIGH 0x00000800
#define PCR_TX_UNASS_HIGH_IMP 0x00001000
/* T1 @ 1.544MHz only: Fbit dictated in FIFO (default) or high Z */
#define PCR_TX_FB_HIGH_IMP 0x00000400
/* 56k data endiannes - which bit unused: high (default) or low */
#define PCR_TX_56KE_BIT_0_UNUSED 0x00000200
/* 56k data transmission type: 32/8 bit data (default) or 56K data */
#define PCR_TX_56KS_56K_DATA 0x00000100
/* hss_config, cCR */
/* Number of packetized clients, default = 1 */
#define CCR_NPE_HFIFO_2_HDLC 0x04000000
#define CCR_NPE_HFIFO_3_OR_4HDLC 0x08000000
/* default = no loopback */
#define CCR_LOOPBACK 0x02000000
/* HSS number, default = 0 (first) */
#define CCR_SECOND_HSS 0x01000000
/* hss_config, clkCR: main:10, num:10, denom:12 */
#define CLK42X_SPEED_EXP ((0x3FF << 22) | ( 2 << 12) | 15) /*65 KHz*/
#define CLK42X_SPEED_512KHZ (( 130 << 22) | ( 2 << 12) | 15)
#define CLK42X_SPEED_1536KHZ (( 43 << 22) | ( 18 << 12) | 47)
#define CLK42X_SPEED_1544KHZ (( 43 << 22) | ( 33 << 12) | 192)
#define CLK42X_SPEED_2048KHZ (( 32 << 22) | ( 34 << 12) | 63)
#define CLK42X_SPEED_4096KHZ (( 16 << 22) | ( 34 << 12) | 127)
#define CLK42X_SPEED_8192KHZ (( 8 << 22) | ( 34 << 12) | 255)
#define CLK46X_SPEED_512KHZ (( 130 << 22) | ( 24 << 12) | 127)
#define CLK46X_SPEED_1536KHZ (( 43 << 22) | (152 << 12) | 383)
#define CLK46X_SPEED_1544KHZ (( 43 << 22) | ( 66 << 12) | 385)
#define CLK46X_SPEED_2048KHZ (( 32 << 22) | (280 << 12) | 511)
#define CLK46X_SPEED_4096KHZ (( 16 << 22) | (280 << 12) | 1023)
#define CLK46X_SPEED_8192KHZ (( 8 << 22) | (280 << 12) | 2047)
IXP42x HSS support for setting internal clock rate HSS usually uses external clocks, so it's not a big deal. Internal clock is used for direct DTE-DTE connections and when the DCE doesn't provide it's own clock. This also depends on the oscillator frequency. Intel seems to have calculated the clock register settings for 33.33 MHz (66.66 MHz timer base). Their settings seem quite suboptimal both in terms of average frequency (60 ppm is unacceptable for G.703 applications, their primary intended usage(?)) and jitter. Many (most?) platforms use a 33.333 MHz oscillator, a 10 ppm difference from Intel's base. Instead of creating static tables, I've created a procedure to program the HSS clock register. The register consists of 3 parts (A, B, C). The average frequency (= bit rate) is: 66.66x MHz / (A + (B + 1) / (C + 1)) The procedure aims at the closest average frequency, possibly at the cost of increased jitter. Nobody would be able to directly drive an unbufferred transmitter with a HSS anyway, and the frequency error is what it really counts. I've verified the above with an oscilloscope on IXP425. It seems IXP46x and possibly IXP43x use a bit different clock generation algorithm - it looks like the avg frequency is: (on IXP465) 66.66x MHz / (A + B / (C + 1)). Also they use much greater precomputed A and B - on IXP425 it would simply result in more jitter, but I don't know how does it work on IXP46x (perhaps 3 least significant bits aren't used?). Anyway it looks that they were aiming for exactly +60 ppm or -60 ppm, while <1 ppm is typically possible (with a synchronized clock, of course). The attached patch makes it possible to set almost any bit rate (my IXP425 533 MHz quits at > 22 Mb/s if a single port is used, and the minimum is ca. 65 Kb/s). This is independent of MVIP (multi-E1/T1 on one HSS) mode. Signed-off-by: Krzysztof Hałasa <khc@pm.waw.pl> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-09-05 10:59:49 +07:00
/*
* HSS_CONFIG_CLOCK_CR register consists of 3 parts:
* A (10 bits), B (10 bits) and C (12 bits).
* IXP42x HSS clock generator operation (verified with an oscilloscope):
* Each clock bit takes 7.5 ns (1 / 133.xx MHz).
* The clock sequence consists of (C - B) states of 0s and 1s, each state is
* A bits wide. It's followed by (B + 1) states of 0s and 1s, each state is
* (A + 1) bits wide.
*
* The resulting average clock frequency (assuming 33.333 MHz oscillator) is:
* freq = 66.666 MHz / (A + (B + 1) / (C + 1))
* minumum freq = 66.666 MHz / (A + 1)
* maximum freq = 66.666 MHz / A
*
* Example: A = 2, B = 2, C = 7, CLOCK_CR register = 2 << 22 | 2 << 12 | 7
* freq = 66.666 MHz / (2 + (2 + 1) / (7 + 1)) = 28.07 MHz (Mb/s).
* The clock sequence is: 1100110011 (5 doubles) 000111000 (3 triples).
* The sequence takes (C - B) * A + (B + 1) * (A + 1) = 5 * 2 + 3 * 3 bits
* = 19 bits (each 7.5 ns long) = 142.5 ns (then the sequence repeats).
* The sequence consists of 4 complete clock periods, thus the average
* frequency (= clock rate) is 4 / 142.5 ns = 28.07 MHz (Mb/s).
* (max specified clock rate for IXP42x HSS is 8.192 Mb/s).
*/
/* hss_config, LUT entries */
#define TDMMAP_UNASSIGNED 0
#define TDMMAP_HDLC 1 /* HDLC - packetized */
#define TDMMAP_VOICE56K 2 /* Voice56K - 7-bit channelized */
#define TDMMAP_VOICE64K 3 /* Voice64K - 8-bit channelized */
/* offsets into HSS config */
#define HSS_CONFIG_TX_PCR 0x00 /* port configuration registers */
#define HSS_CONFIG_RX_PCR 0x04
#define HSS_CONFIG_CORE_CR 0x08 /* loopback control, HSS# */
#define HSS_CONFIG_CLOCK_CR 0x0C /* clock generator control */
#define HSS_CONFIG_TX_FCR 0x10 /* frame configuration registers */
#define HSS_CONFIG_RX_FCR 0x14
#define HSS_CONFIG_TX_LUT 0x18 /* channel look-up tables */
#define HSS_CONFIG_RX_LUT 0x38
/* NPE command codes */
/* writes the ConfigWord value to the location specified by offset */
#define PORT_CONFIG_WRITE 0x40
/* triggers the NPE to load the contents of the configuration table */
#define PORT_CONFIG_LOAD 0x41
/* triggers the NPE to return an HssErrorReadResponse message */
#define PORT_ERROR_READ 0x42
/* triggers the NPE to reset internal status and enable the HssPacketized
operation for the flow specified by pPipe */
#define PKT_PIPE_FLOW_ENABLE 0x50
#define PKT_PIPE_FLOW_DISABLE 0x51
#define PKT_NUM_PIPES_WRITE 0x52
#define PKT_PIPE_FIFO_SIZEW_WRITE 0x53
#define PKT_PIPE_HDLC_CFG_WRITE 0x54
#define PKT_PIPE_IDLE_PATTERN_WRITE 0x55
#define PKT_PIPE_RX_SIZE_WRITE 0x56
#define PKT_PIPE_MODE_WRITE 0x57
/* HDLC packet status values - desc->status */
#define ERR_SHUTDOWN 1 /* stop or shutdown occurrance */
#define ERR_HDLC_ALIGN 2 /* HDLC alignment error */
#define ERR_HDLC_FCS 3 /* HDLC Frame Check Sum error */
#define ERR_RXFREE_Q_EMPTY 4 /* RX-free queue became empty while receiving
this packet (if buf_len < pkt_len) */
#define ERR_HDLC_TOO_LONG 5 /* HDLC frame size too long */
#define ERR_HDLC_ABORT 6 /* abort sequence received */
#define ERR_DISCONNECTING 7 /* disconnect is in progress */
#ifdef __ARMEB__
typedef struct sk_buff buffer_t;
#define free_buffer dev_kfree_skb
#define free_buffer_irq dev_kfree_skb_irq
#else
typedef void buffer_t;
#define free_buffer kfree
#define free_buffer_irq kfree
#endif
struct port {
struct device *dev;
struct npe *npe;
struct net_device *netdev;
struct napi_struct napi;
struct hss_plat_info *plat;
buffer_t *rx_buff_tab[RX_DESCS], *tx_buff_tab[TX_DESCS];
struct desc *desc_tab; /* coherent */
u32 desc_tab_phys;
unsigned int id;
unsigned int clock_type, clock_rate, loopback;
unsigned int initialized, carrier;
u8 hdlc_cfg;
IXP42x HSS support for setting internal clock rate HSS usually uses external clocks, so it's not a big deal. Internal clock is used for direct DTE-DTE connections and when the DCE doesn't provide it's own clock. This also depends on the oscillator frequency. Intel seems to have calculated the clock register settings for 33.33 MHz (66.66 MHz timer base). Their settings seem quite suboptimal both in terms of average frequency (60 ppm is unacceptable for G.703 applications, their primary intended usage(?)) and jitter. Many (most?) platforms use a 33.333 MHz oscillator, a 10 ppm difference from Intel's base. Instead of creating static tables, I've created a procedure to program the HSS clock register. The register consists of 3 parts (A, B, C). The average frequency (= bit rate) is: 66.66x MHz / (A + (B + 1) / (C + 1)) The procedure aims at the closest average frequency, possibly at the cost of increased jitter. Nobody would be able to directly drive an unbufferred transmitter with a HSS anyway, and the frequency error is what it really counts. I've verified the above with an oscilloscope on IXP425. It seems IXP46x and possibly IXP43x use a bit different clock generation algorithm - it looks like the avg frequency is: (on IXP465) 66.66x MHz / (A + B / (C + 1)). Also they use much greater precomputed A and B - on IXP425 it would simply result in more jitter, but I don't know how does it work on IXP46x (perhaps 3 least significant bits aren't used?). Anyway it looks that they were aiming for exactly +60 ppm or -60 ppm, while <1 ppm is typically possible (with a synchronized clock, of course). The attached patch makes it possible to set almost any bit rate (my IXP425 533 MHz quits at > 22 Mb/s if a single port is used, and the minimum is ca. 65 Kb/s). This is independent of MVIP (multi-E1/T1 on one HSS) mode. Signed-off-by: Krzysztof Hałasa <khc@pm.waw.pl> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-09-05 10:59:49 +07:00
u32 clock_reg;
};
/* NPE message structure */
struct msg {
#ifdef __ARMEB__
u8 cmd, unused, hss_port, index;
union {
struct { u8 data8a, data8b, data8c, data8d; };
struct { u16 data16a, data16b; };
struct { u32 data32; };
};
#else
u8 index, hss_port, unused, cmd;
union {
struct { u8 data8d, data8c, data8b, data8a; };
struct { u16 data16b, data16a; };
struct { u32 data32; };
};
#endif
};
/* HDLC packet descriptor */
struct desc {
u32 next; /* pointer to next buffer, unused */
#ifdef __ARMEB__
u16 buf_len; /* buffer length */
u16 pkt_len; /* packet length */
u32 data; /* pointer to data buffer in RAM */
u8 status;
u8 error_count;
u16 __reserved;
#else
u16 pkt_len; /* packet length */
u16 buf_len; /* buffer length */
u32 data; /* pointer to data buffer in RAM */
u16 __reserved;
u8 error_count;
u8 status;
#endif
u32 __reserved1[4];
};
#define rx_desc_phys(port, n) ((port)->desc_tab_phys + \
(n) * sizeof(struct desc))
#define rx_desc_ptr(port, n) (&(port)->desc_tab[n])
#define tx_desc_phys(port, n) ((port)->desc_tab_phys + \
((n) + RX_DESCS) * sizeof(struct desc))
#define tx_desc_ptr(port, n) (&(port)->desc_tab[(n) + RX_DESCS])
/*****************************************************************************
* global variables
****************************************************************************/
static int ports_open;
static struct dma_pool *dma_pool;
static spinlock_t npe_lock;
static const struct {
int tx, txdone, rx, rxfree;
}queue_ids[2] = {{HSS0_PKT_TX0_QUEUE, HSS0_PKT_TXDONE_QUEUE, HSS0_PKT_RX_QUEUE,
HSS0_PKT_RXFREE0_QUEUE},
{HSS1_PKT_TX0_QUEUE, HSS1_PKT_TXDONE_QUEUE, HSS1_PKT_RX_QUEUE,
HSS1_PKT_RXFREE0_QUEUE},
};
/*****************************************************************************
* utility functions
****************************************************************************/
static inline struct port* dev_to_port(struct net_device *dev)
{
return dev_to_hdlc(dev)->priv;
}
#ifndef __ARMEB__
static inline void memcpy_swab32(u32 *dest, u32 *src, int cnt)
{
int i;
for (i = 0; i < cnt; i++)
dest[i] = swab32(src[i]);
}
#endif
/*****************************************************************************
* HSS access
****************************************************************************/
static void hss_npe_send(struct port *port, struct msg *msg, const char* what)
{
u32 *val = (u32*)msg;
if (npe_send_message(port->npe, msg, what)) {
printk(KERN_CRIT "HSS-%i: unable to send command [%08X:%08X]"
" to %s\n", port->id, val[0], val[1],
npe_name(port->npe));
BUG();
}
}
static void hss_config_set_lut(struct port *port)
{
struct msg msg;
int ch;
memset(&msg, 0, sizeof(msg));
msg.cmd = PORT_CONFIG_WRITE;
msg.hss_port = port->id;
for (ch = 0; ch < MAX_CHANNELS; ch++) {
msg.data32 >>= 2;
msg.data32 |= TDMMAP_HDLC << 30;
if (ch % 16 == 15) {
msg.index = HSS_CONFIG_TX_LUT + ((ch / 4) & ~3);
hss_npe_send(port, &msg, "HSS_SET_TX_LUT");
msg.index += HSS_CONFIG_RX_LUT - HSS_CONFIG_TX_LUT;
hss_npe_send(port, &msg, "HSS_SET_RX_LUT");
}
}
}
static void hss_config(struct port *port)
{
struct msg msg;
memset(&msg, 0, sizeof(msg));
msg.cmd = PORT_CONFIG_WRITE;
msg.hss_port = port->id;
msg.index = HSS_CONFIG_TX_PCR;
msg.data32 = PCR_FRM_SYNC_OUTPUT_RISING | PCR_MSB_ENDIAN |
PCR_TX_DATA_ENABLE | PCR_SOF_NO_FBIT;
if (port->clock_type == CLOCK_INT)
msg.data32 |= PCR_SYNC_CLK_DIR_OUTPUT;
hss_npe_send(port, &msg, "HSS_SET_TX_PCR");
msg.index = HSS_CONFIG_RX_PCR;
msg.data32 ^= PCR_TX_DATA_ENABLE | PCR_DCLK_EDGE_RISING;
hss_npe_send(port, &msg, "HSS_SET_RX_PCR");
memset(&msg, 0, sizeof(msg));
msg.cmd = PORT_CONFIG_WRITE;
msg.hss_port = port->id;
msg.index = HSS_CONFIG_CORE_CR;
msg.data32 = (port->loopback ? CCR_LOOPBACK : 0) |
(port->id ? CCR_SECOND_HSS : 0);
hss_npe_send(port, &msg, "HSS_SET_CORE_CR");
memset(&msg, 0, sizeof(msg));
msg.cmd = PORT_CONFIG_WRITE;
msg.hss_port = port->id;
msg.index = HSS_CONFIG_CLOCK_CR;
IXP42x HSS support for setting internal clock rate HSS usually uses external clocks, so it's not a big deal. Internal clock is used for direct DTE-DTE connections and when the DCE doesn't provide it's own clock. This also depends on the oscillator frequency. Intel seems to have calculated the clock register settings for 33.33 MHz (66.66 MHz timer base). Their settings seem quite suboptimal both in terms of average frequency (60 ppm is unacceptable for G.703 applications, their primary intended usage(?)) and jitter. Many (most?) platforms use a 33.333 MHz oscillator, a 10 ppm difference from Intel's base. Instead of creating static tables, I've created a procedure to program the HSS clock register. The register consists of 3 parts (A, B, C). The average frequency (= bit rate) is: 66.66x MHz / (A + (B + 1) / (C + 1)) The procedure aims at the closest average frequency, possibly at the cost of increased jitter. Nobody would be able to directly drive an unbufferred transmitter with a HSS anyway, and the frequency error is what it really counts. I've verified the above with an oscilloscope on IXP425. It seems IXP46x and possibly IXP43x use a bit different clock generation algorithm - it looks like the avg frequency is: (on IXP465) 66.66x MHz / (A + B / (C + 1)). Also they use much greater precomputed A and B - on IXP425 it would simply result in more jitter, but I don't know how does it work on IXP46x (perhaps 3 least significant bits aren't used?). Anyway it looks that they were aiming for exactly +60 ppm or -60 ppm, while <1 ppm is typically possible (with a synchronized clock, of course). The attached patch makes it possible to set almost any bit rate (my IXP425 533 MHz quits at > 22 Mb/s if a single port is used, and the minimum is ca. 65 Kb/s). This is independent of MVIP (multi-E1/T1 on one HSS) mode. Signed-off-by: Krzysztof Hałasa <khc@pm.waw.pl> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-09-05 10:59:49 +07:00
msg.data32 = port->clock_reg;
hss_npe_send(port, &msg, "HSS_SET_CLOCK_CR");
memset(&msg, 0, sizeof(msg));
msg.cmd = PORT_CONFIG_WRITE;
msg.hss_port = port->id;
msg.index = HSS_CONFIG_TX_FCR;
msg.data16a = FRAME_OFFSET;
msg.data16b = FRAME_SIZE - 1;
hss_npe_send(port, &msg, "HSS_SET_TX_FCR");
memset(&msg, 0, sizeof(msg));
msg.cmd = PORT_CONFIG_WRITE;
msg.hss_port = port->id;
msg.index = HSS_CONFIG_RX_FCR;
msg.data16a = FRAME_OFFSET;
msg.data16b = FRAME_SIZE - 1;
hss_npe_send(port, &msg, "HSS_SET_RX_FCR");
hss_config_set_lut(port);
memset(&msg, 0, sizeof(msg));
msg.cmd = PORT_CONFIG_LOAD;
msg.hss_port = port->id;
hss_npe_send(port, &msg, "HSS_LOAD_CONFIG");
if (npe_recv_message(port->npe, &msg, "HSS_LOAD_CONFIG") ||
/* HSS_LOAD_CONFIG for port #1 returns port_id = #4 */
msg.cmd != PORT_CONFIG_LOAD || msg.data32) {
printk(KERN_CRIT "HSS-%i: HSS_LOAD_CONFIG failed\n",
port->id);
BUG();
}
/* HDLC may stop working without this - check FIXME */
npe_recv_message(port->npe, &msg, "FLUSH_IT");
}
static void hss_set_hdlc_cfg(struct port *port)
{
struct msg msg;
memset(&msg, 0, sizeof(msg));
msg.cmd = PKT_PIPE_HDLC_CFG_WRITE;
msg.hss_port = port->id;
msg.data8a = port->hdlc_cfg; /* rx_cfg */
msg.data8b = port->hdlc_cfg | (PKT_EXTRA_FLAGS << 3); /* tx_cfg */
hss_npe_send(port, &msg, "HSS_SET_HDLC_CFG");
}
static u32 hss_get_status(struct port *port)
{
struct msg msg;
memset(&msg, 0, sizeof(msg));
msg.cmd = PORT_ERROR_READ;
msg.hss_port = port->id;
hss_npe_send(port, &msg, "PORT_ERROR_READ");
if (npe_recv_message(port->npe, &msg, "PORT_ERROR_READ")) {
printk(KERN_CRIT "HSS-%i: unable to read HSS status\n",
port->id);
BUG();
}
return msg.data32;
}
static void hss_start_hdlc(struct port *port)
{
struct msg msg;
memset(&msg, 0, sizeof(msg));
msg.cmd = PKT_PIPE_FLOW_ENABLE;
msg.hss_port = port->id;
msg.data32 = 0;
hss_npe_send(port, &msg, "HSS_ENABLE_PKT_PIPE");
}
static void hss_stop_hdlc(struct port *port)
{
struct msg msg;
memset(&msg, 0, sizeof(msg));
msg.cmd = PKT_PIPE_FLOW_DISABLE;
msg.hss_port = port->id;
hss_npe_send(port, &msg, "HSS_DISABLE_PKT_PIPE");
hss_get_status(port); /* make sure it's halted */
}
static int hss_load_firmware(struct port *port)
{
struct msg msg;
int err;
if (port->initialized)
return 0;
if (!npe_running(port->npe) &&
(err = npe_load_firmware(port->npe, npe_name(port->npe),
port->dev)))
return err;
/* HDLC mode configuration */
memset(&msg, 0, sizeof(msg));
msg.cmd = PKT_NUM_PIPES_WRITE;
msg.hss_port = port->id;
msg.data8a = PKT_NUM_PIPES;
hss_npe_send(port, &msg, "HSS_SET_PKT_PIPES");
msg.cmd = PKT_PIPE_FIFO_SIZEW_WRITE;
msg.data8a = PKT_PIPE_FIFO_SIZEW;
hss_npe_send(port, &msg, "HSS_SET_PKT_FIFO");
msg.cmd = PKT_PIPE_MODE_WRITE;
msg.data8a = NPE_PKT_MODE_HDLC;
/* msg.data8b = inv_mask */
/* msg.data8c = or_mask */
hss_npe_send(port, &msg, "HSS_SET_PKT_MODE");
msg.cmd = PKT_PIPE_RX_SIZE_WRITE;
msg.data16a = HDLC_MAX_MRU; /* including CRC */
hss_npe_send(port, &msg, "HSS_SET_PKT_RX_SIZE");
msg.cmd = PKT_PIPE_IDLE_PATTERN_WRITE;
msg.data32 = 0x7F7F7F7F; /* ??? FIXME */
hss_npe_send(port, &msg, "HSS_SET_PKT_IDLE");
port->initialized = 1;
return 0;
}
/*****************************************************************************
* packetized (HDLC) operation
****************************************************************************/
static inline void debug_pkt(struct net_device *dev, const char *func,
u8 *data, int len)
{
#if DEBUG_PKT_BYTES
int i;
printk(KERN_DEBUG "%s: %s(%i)", dev->name, func, len);
for (i = 0; i < len; i++) {
if (i >= DEBUG_PKT_BYTES)
break;
printk("%s%02X", !(i % 4) ? " " : "", data[i]);
}
printk("\n");
#endif
}
static inline void debug_desc(u32 phys, struct desc *desc)
{
#if DEBUG_DESC
printk(KERN_DEBUG "%X: %X %3X %3X %08X %X %X\n",
phys, desc->next, desc->buf_len, desc->pkt_len,
desc->data, desc->status, desc->error_count);
#endif
}
static inline int queue_get_desc(unsigned int queue, struct port *port,
int is_tx)
{
u32 phys, tab_phys, n_desc;
struct desc *tab;
if (!(phys = qmgr_get_entry(queue)))
return -1;
BUG_ON(phys & 0x1F);
tab_phys = is_tx ? tx_desc_phys(port, 0) : rx_desc_phys(port, 0);
tab = is_tx ? tx_desc_ptr(port, 0) : rx_desc_ptr(port, 0);
n_desc = (phys - tab_phys) / sizeof(struct desc);
BUG_ON(n_desc >= (is_tx ? TX_DESCS : RX_DESCS));
debug_desc(phys, &tab[n_desc]);
BUG_ON(tab[n_desc].next);
return n_desc;
}
static inline void queue_put_desc(unsigned int queue, u32 phys,
struct desc *desc)
{
debug_desc(phys, desc);
BUG_ON(phys & 0x1F);
qmgr_put_entry(queue, phys);
/* Don't check for queue overflow here, we've allocated sufficient
length and queues >= 32 don't support this check anyway. */
}
static inline void dma_unmap_tx(struct port *port, struct desc *desc)
{
#ifdef __ARMEB__
dma_unmap_single(&port->netdev->dev, desc->data,
desc->buf_len, DMA_TO_DEVICE);
#else
dma_unmap_single(&port->netdev->dev, desc->data & ~3,
ALIGN((desc->data & 3) + desc->buf_len, 4),
DMA_TO_DEVICE);
#endif
}
static void hss_hdlc_set_carrier(void *pdev, int carrier)
{
struct net_device *netdev = pdev;
struct port *port = dev_to_port(netdev);
unsigned long flags;
spin_lock_irqsave(&npe_lock, flags);
port->carrier = carrier;
if (!port->loopback) {
if (carrier)
netif_carrier_on(netdev);
else
netif_carrier_off(netdev);
}
spin_unlock_irqrestore(&npe_lock, flags);
}
static void hss_hdlc_rx_irq(void *pdev)
{
struct net_device *dev = pdev;
struct port *port = dev_to_port(dev);
#if DEBUG_RX
printk(KERN_DEBUG "%s: hss_hdlc_rx_irq\n", dev->name);
#endif
qmgr_disable_irq(queue_ids[port->id].rx);
napi_schedule(&port->napi);
}
static int hss_hdlc_poll(struct napi_struct *napi, int budget)
{
struct port *port = container_of(napi, struct port, napi);
struct net_device *dev = port->netdev;
unsigned int rxq = queue_ids[port->id].rx;
unsigned int rxfreeq = queue_ids[port->id].rxfree;
int received = 0;
#if DEBUG_RX
printk(KERN_DEBUG "%s: hss_hdlc_poll\n", dev->name);
#endif
while (received < budget) {
struct sk_buff *skb;
struct desc *desc;
int n;
#ifdef __ARMEB__
struct sk_buff *temp;
u32 phys;
#endif
if ((n = queue_get_desc(rxq, port, 0)) < 0) {
#if DEBUG_RX
printk(KERN_DEBUG "%s: hss_hdlc_poll"
" napi_complete\n", dev->name);
#endif
napi_complete(napi);
qmgr_enable_irq(rxq);
if (!qmgr_stat_empty(rxq) &&
napi_reschedule(napi)) {
#if DEBUG_RX
printk(KERN_DEBUG "%s: hss_hdlc_poll"
" napi_reschedule succeeded\n",
dev->name);
#endif
qmgr_disable_irq(rxq);
continue;
}
#if DEBUG_RX
printk(KERN_DEBUG "%s: hss_hdlc_poll all done\n",
dev->name);
#endif
return received; /* all work done */
}
desc = rx_desc_ptr(port, n);
#if 0 /* FIXME - error_count counts modulo 256, perhaps we should use it */
if (desc->error_count)
printk(KERN_DEBUG "%s: hss_hdlc_poll status 0x%02X"
" errors %u\n", dev->name, desc->status,
desc->error_count);
#endif
skb = NULL;
switch (desc->status) {
case 0:
#ifdef __ARMEB__
if ((skb = netdev_alloc_skb(dev, RX_SIZE)) != NULL) {
phys = dma_map_single(&dev->dev, skb->data,
RX_SIZE,
DMA_FROM_DEVICE);
if (dma_mapping_error(&dev->dev, phys)) {
dev_kfree_skb(skb);
skb = NULL;
}
}
#else
skb = netdev_alloc_skb(dev, desc->pkt_len);
#endif
if (!skb)
dev->stats.rx_dropped++;
break;
case ERR_HDLC_ALIGN:
case ERR_HDLC_ABORT:
dev->stats.rx_frame_errors++;
dev->stats.rx_errors++;
break;
case ERR_HDLC_FCS:
dev->stats.rx_crc_errors++;
dev->stats.rx_errors++;
break;
case ERR_HDLC_TOO_LONG:
dev->stats.rx_length_errors++;
dev->stats.rx_errors++;
break;
default: /* FIXME - remove printk */
printk(KERN_ERR "%s: hss_hdlc_poll: status 0x%02X"
" errors %u\n", dev->name, desc->status,
desc->error_count);
dev->stats.rx_errors++;
}
if (!skb) {
/* put the desc back on RX-ready queue */
desc->buf_len = RX_SIZE;
desc->pkt_len = desc->status = 0;
queue_put_desc(rxfreeq, rx_desc_phys(port, n), desc);
continue;
}
/* process received frame */
#ifdef __ARMEB__
temp = skb;
skb = port->rx_buff_tab[n];
dma_unmap_single(&dev->dev, desc->data,
RX_SIZE, DMA_FROM_DEVICE);
#else
dma_sync_single_for_cpu(&dev->dev, desc->data,
RX_SIZE, DMA_FROM_DEVICE);
memcpy_swab32((u32 *)skb->data, (u32 *)port->rx_buff_tab[n],
ALIGN(desc->pkt_len, 4) / 4);
#endif
skb_put(skb, desc->pkt_len);
debug_pkt(dev, "hss_hdlc_poll", skb->data, skb->len);
skb->protocol = hdlc_type_trans(skb, dev);
dev->stats.rx_packets++;
dev->stats.rx_bytes += skb->len;
netif_receive_skb(skb);
/* put the new buffer on RX-free queue */
#ifdef __ARMEB__
port->rx_buff_tab[n] = temp;
desc->data = phys;
#endif
desc->buf_len = RX_SIZE;
desc->pkt_len = 0;
queue_put_desc(rxfreeq, rx_desc_phys(port, n), desc);
received++;
}
#if DEBUG_RX
printk(KERN_DEBUG "hss_hdlc_poll: end, not all work done\n");
#endif
return received; /* not all work done */
}
static void hss_hdlc_txdone_irq(void *pdev)
{
struct net_device *dev = pdev;
struct port *port = dev_to_port(dev);
int n_desc;
#if DEBUG_TX
printk(KERN_DEBUG DRV_NAME ": hss_hdlc_txdone_irq\n");
#endif
while ((n_desc = queue_get_desc(queue_ids[port->id].txdone,
port, 1)) >= 0) {
struct desc *desc;
int start;
desc = tx_desc_ptr(port, n_desc);
dev->stats.tx_packets++;
dev->stats.tx_bytes += desc->pkt_len;
dma_unmap_tx(port, desc);
#if DEBUG_TX
printk(KERN_DEBUG "%s: hss_hdlc_txdone_irq free %p\n",
dev->name, port->tx_buff_tab[n_desc]);
#endif
free_buffer_irq(port->tx_buff_tab[n_desc]);
port->tx_buff_tab[n_desc] = NULL;
start = qmgr_stat_below_low_watermark(port->plat->txreadyq);
queue_put_desc(port->plat->txreadyq,
tx_desc_phys(port, n_desc), desc);
if (start) { /* TX-ready queue was empty */
#if DEBUG_TX
printk(KERN_DEBUG "%s: hss_hdlc_txdone_irq xmit"
" ready\n", dev->name);
#endif
netif_wake_queue(dev);
}
}
}
static int hss_hdlc_xmit(struct sk_buff *skb, struct net_device *dev)
{
struct port *port = dev_to_port(dev);
unsigned int txreadyq = port->plat->txreadyq;
int len, offset, bytes, n;
void *mem;
u32 phys;
struct desc *desc;
#if DEBUG_TX
printk(KERN_DEBUG "%s: hss_hdlc_xmit\n", dev->name);
#endif
if (unlikely(skb->len > HDLC_MAX_MRU)) {
dev_kfree_skb(skb);
dev->stats.tx_errors++;
return NETDEV_TX_OK;
}
debug_pkt(dev, "hss_hdlc_xmit", skb->data, skb->len);
len = skb->len;
#ifdef __ARMEB__
offset = 0; /* no need to keep alignment */
bytes = len;
mem = skb->data;
#else
offset = (int)skb->data & 3; /* keep 32-bit alignment */
bytes = ALIGN(offset + len, 4);
if (!(mem = kmalloc(bytes, GFP_ATOMIC))) {
dev_kfree_skb(skb);
dev->stats.tx_dropped++;
return NETDEV_TX_OK;
}
memcpy_swab32(mem, (u32 *)((int)skb->data & ~3), bytes / 4);
dev_kfree_skb(skb);
#endif
phys = dma_map_single(&dev->dev, mem, bytes, DMA_TO_DEVICE);
if (dma_mapping_error(&dev->dev, phys)) {
#ifdef __ARMEB__
dev_kfree_skb(skb);
#else
kfree(mem);
#endif
dev->stats.tx_dropped++;
return NETDEV_TX_OK;
}
n = queue_get_desc(txreadyq, port, 1);
BUG_ON(n < 0);
desc = tx_desc_ptr(port, n);
#ifdef __ARMEB__
port->tx_buff_tab[n] = skb;
#else
port->tx_buff_tab[n] = mem;
#endif
desc->data = phys + offset;
desc->buf_len = desc->pkt_len = len;
wmb();
queue_put_desc(queue_ids[port->id].tx, tx_desc_phys(port, n), desc);
if (qmgr_stat_below_low_watermark(txreadyq)) { /* empty */
#if DEBUG_TX
printk(KERN_DEBUG "%s: hss_hdlc_xmit queue full\n", dev->name);
#endif
netif_stop_queue(dev);
/* we could miss TX ready interrupt */
if (!qmgr_stat_below_low_watermark(txreadyq)) {
#if DEBUG_TX
printk(KERN_DEBUG "%s: hss_hdlc_xmit ready again\n",
dev->name);
#endif
netif_wake_queue(dev);
}
}
#if DEBUG_TX
printk(KERN_DEBUG "%s: hss_hdlc_xmit end\n", dev->name);
#endif
return NETDEV_TX_OK;
}
static int request_hdlc_queues(struct port *port)
{
int err;
err = qmgr_request_queue(queue_ids[port->id].rxfree, RX_DESCS, 0, 0,
"%s:RX-free", port->netdev->name);
if (err)
return err;
err = qmgr_request_queue(queue_ids[port->id].rx, RX_DESCS, 0, 0,
"%s:RX", port->netdev->name);
if (err)
goto rel_rxfree;
err = qmgr_request_queue(queue_ids[port->id].tx, TX_DESCS, 0, 0,
"%s:TX", port->netdev->name);
if (err)
goto rel_rx;
err = qmgr_request_queue(port->plat->txreadyq, TX_DESCS, 0, 0,
"%s:TX-ready", port->netdev->name);
if (err)
goto rel_tx;
err = qmgr_request_queue(queue_ids[port->id].txdone, TX_DESCS, 0, 0,
"%s:TX-done", port->netdev->name);
if (err)
goto rel_txready;
return 0;
rel_txready:
qmgr_release_queue(port->plat->txreadyq);
rel_tx:
qmgr_release_queue(queue_ids[port->id].tx);
rel_rx:
qmgr_release_queue(queue_ids[port->id].rx);
rel_rxfree:
qmgr_release_queue(queue_ids[port->id].rxfree);
printk(KERN_DEBUG "%s: unable to request hardware queues\n",
port->netdev->name);
return err;
}
static void release_hdlc_queues(struct port *port)
{
qmgr_release_queue(queue_ids[port->id].rxfree);
qmgr_release_queue(queue_ids[port->id].rx);
qmgr_release_queue(queue_ids[port->id].txdone);
qmgr_release_queue(queue_ids[port->id].tx);
qmgr_release_queue(port->plat->txreadyq);
}
static int init_hdlc_queues(struct port *port)
{
int i;
if (!ports_open)
if (!(dma_pool = dma_pool_create(DRV_NAME, NULL,
POOL_ALLOC_SIZE, 32, 0)))
return -ENOMEM;
if (!(port->desc_tab = dma_pool_alloc(dma_pool, GFP_KERNEL,
&port->desc_tab_phys)))
return -ENOMEM;
memset(port->desc_tab, 0, POOL_ALLOC_SIZE);
memset(port->rx_buff_tab, 0, sizeof(port->rx_buff_tab)); /* tables */
memset(port->tx_buff_tab, 0, sizeof(port->tx_buff_tab));
/* Setup RX buffers */
for (i = 0; i < RX_DESCS; i++) {
struct desc *desc = rx_desc_ptr(port, i);
buffer_t *buff;
void *data;
#ifdef __ARMEB__
if (!(buff = netdev_alloc_skb(port->netdev, RX_SIZE)))
return -ENOMEM;
data = buff->data;
#else
if (!(buff = kmalloc(RX_SIZE, GFP_KERNEL)))
return -ENOMEM;
data = buff;
#endif
desc->buf_len = RX_SIZE;
desc->data = dma_map_single(&port->netdev->dev, data,
RX_SIZE, DMA_FROM_DEVICE);
if (dma_mapping_error(&port->netdev->dev, desc->data)) {
free_buffer(buff);
return -EIO;
}
port->rx_buff_tab[i] = buff;
}
return 0;
}
static void destroy_hdlc_queues(struct port *port)
{
int i;
if (port->desc_tab) {
for (i = 0; i < RX_DESCS; i++) {
struct desc *desc = rx_desc_ptr(port, i);
buffer_t *buff = port->rx_buff_tab[i];
if (buff) {
dma_unmap_single(&port->netdev->dev,
desc->data, RX_SIZE,
DMA_FROM_DEVICE);
free_buffer(buff);
}
}
for (i = 0; i < TX_DESCS; i++) {
struct desc *desc = tx_desc_ptr(port, i);
buffer_t *buff = port->tx_buff_tab[i];
if (buff) {
dma_unmap_tx(port, desc);
free_buffer(buff);
}
}
dma_pool_free(dma_pool, port->desc_tab, port->desc_tab_phys);
port->desc_tab = NULL;
}
if (!ports_open && dma_pool) {
dma_pool_destroy(dma_pool);
dma_pool = NULL;
}
}
static int hss_hdlc_open(struct net_device *dev)
{
struct port *port = dev_to_port(dev);
unsigned long flags;
int i, err = 0;
if ((err = hdlc_open(dev)))
return err;
if ((err = hss_load_firmware(port)))
goto err_hdlc_close;
if ((err = request_hdlc_queues(port)))
goto err_hdlc_close;
if ((err = init_hdlc_queues(port)))
goto err_destroy_queues;
spin_lock_irqsave(&npe_lock, flags);
if (port->plat->open)
if ((err = port->plat->open(port->id, dev,
hss_hdlc_set_carrier)))
goto err_unlock;
spin_unlock_irqrestore(&npe_lock, flags);
/* Populate queues with buffers, no failure after this point */
for (i = 0; i < TX_DESCS; i++)
queue_put_desc(port->plat->txreadyq,
tx_desc_phys(port, i), tx_desc_ptr(port, i));
for (i = 0; i < RX_DESCS; i++)
queue_put_desc(queue_ids[port->id].rxfree,
rx_desc_phys(port, i), rx_desc_ptr(port, i));
napi_enable(&port->napi);
netif_start_queue(dev);
qmgr_set_irq(queue_ids[port->id].rx, QUEUE_IRQ_SRC_NOT_EMPTY,
hss_hdlc_rx_irq, dev);
qmgr_set_irq(queue_ids[port->id].txdone, QUEUE_IRQ_SRC_NOT_EMPTY,
hss_hdlc_txdone_irq, dev);
qmgr_enable_irq(queue_ids[port->id].txdone);
ports_open++;
hss_set_hdlc_cfg(port);
hss_config(port);
hss_start_hdlc(port);
/* we may already have RX data, enables IRQ */
napi_schedule(&port->napi);
return 0;
err_unlock:
spin_unlock_irqrestore(&npe_lock, flags);
err_destroy_queues:
destroy_hdlc_queues(port);
release_hdlc_queues(port);
err_hdlc_close:
hdlc_close(dev);
return err;
}
static int hss_hdlc_close(struct net_device *dev)
{
struct port *port = dev_to_port(dev);
unsigned long flags;
int i, buffs = RX_DESCS; /* allocated RX buffers */
spin_lock_irqsave(&npe_lock, flags);
ports_open--;
qmgr_disable_irq(queue_ids[port->id].rx);
netif_stop_queue(dev);
napi_disable(&port->napi);
hss_stop_hdlc(port);
while (queue_get_desc(queue_ids[port->id].rxfree, port, 0) >= 0)
buffs--;
while (queue_get_desc(queue_ids[port->id].rx, port, 0) >= 0)
buffs--;
if (buffs)
printk(KERN_CRIT "%s: unable to drain RX queue, %i buffer(s)"
" left in NPE\n", dev->name, buffs);
buffs = TX_DESCS;
while (queue_get_desc(queue_ids[port->id].tx, port, 1) >= 0)
buffs--; /* cancel TX */
i = 0;
do {
while (queue_get_desc(port->plat->txreadyq, port, 1) >= 0)
buffs--;
if (!buffs)
break;
} while (++i < MAX_CLOSE_WAIT);
if (buffs)
printk(KERN_CRIT "%s: unable to drain TX queue, %i buffer(s) "
"left in NPE\n", dev->name, buffs);
#if DEBUG_CLOSE
if (!buffs)
printk(KERN_DEBUG "Draining TX queues took %i cycles\n", i);
#endif
qmgr_disable_irq(queue_ids[port->id].txdone);
if (port->plat->close)
port->plat->close(port->id, dev);
spin_unlock_irqrestore(&npe_lock, flags);
destroy_hdlc_queues(port);
release_hdlc_queues(port);
hdlc_close(dev);
return 0;
}
static int hss_hdlc_attach(struct net_device *dev, unsigned short encoding,
unsigned short parity)
{
struct port *port = dev_to_port(dev);
if (encoding != ENCODING_NRZ)
return -EINVAL;
switch(parity) {
case PARITY_CRC16_PR1_CCITT:
port->hdlc_cfg = 0;
return 0;
case PARITY_CRC32_PR1_CCITT:
port->hdlc_cfg = PKT_HDLC_CRC_32;
return 0;
default:
return -EINVAL;
}
}
IXP42x HSS support for setting internal clock rate HSS usually uses external clocks, so it's not a big deal. Internal clock is used for direct DTE-DTE connections and when the DCE doesn't provide it's own clock. This also depends on the oscillator frequency. Intel seems to have calculated the clock register settings for 33.33 MHz (66.66 MHz timer base). Their settings seem quite suboptimal both in terms of average frequency (60 ppm is unacceptable for G.703 applications, their primary intended usage(?)) and jitter. Many (most?) platforms use a 33.333 MHz oscillator, a 10 ppm difference from Intel's base. Instead of creating static tables, I've created a procedure to program the HSS clock register. The register consists of 3 parts (A, B, C). The average frequency (= bit rate) is: 66.66x MHz / (A + (B + 1) / (C + 1)) The procedure aims at the closest average frequency, possibly at the cost of increased jitter. Nobody would be able to directly drive an unbufferred transmitter with a HSS anyway, and the frequency error is what it really counts. I've verified the above with an oscilloscope on IXP425. It seems IXP46x and possibly IXP43x use a bit different clock generation algorithm - it looks like the avg frequency is: (on IXP465) 66.66x MHz / (A + B / (C + 1)). Also they use much greater precomputed A and B - on IXP425 it would simply result in more jitter, but I don't know how does it work on IXP46x (perhaps 3 least significant bits aren't used?). Anyway it looks that they were aiming for exactly +60 ppm or -60 ppm, while <1 ppm is typically possible (with a synchronized clock, of course). The attached patch makes it possible to set almost any bit rate (my IXP425 533 MHz quits at > 22 Mb/s if a single port is used, and the minimum is ca. 65 Kb/s). This is independent of MVIP (multi-E1/T1 on one HSS) mode. Signed-off-by: Krzysztof Hałasa <khc@pm.waw.pl> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-09-05 10:59:49 +07:00
static u32 check_clock(u32 rate, u32 a, u32 b, u32 c,
u32 *best, u32 *best_diff, u32 *reg)
{
/* a is 10-bit, b is 10-bit, c is 12-bit */
u64 new_rate;
u32 new_diff;
new_rate = ixp4xx_timer_freq * (u64)(c + 1);
do_div(new_rate, a * (c + 1) + b + 1);
new_diff = abs((u32)new_rate - rate);
if (new_diff < *best_diff) {
*best = new_rate;
*best_diff = new_diff;
*reg = (a << 22) | (b << 12) | c;
}
return new_diff;
}
static void find_best_clock(u32 rate, u32 *best, u32 *reg)
{
u32 a, b, diff = 0xFFFFFFFF;
a = ixp4xx_timer_freq / rate;
if (a > 0x3FF) { /* 10-bit value - we can go as slow as ca. 65 kb/s */
check_clock(rate, 0x3FF, 1, 1, best, &diff, reg);
return;
}
if (a == 0) { /* > 66.666 MHz */
a = 1; /* minimum divider is 1 (a = 0, b = 1, c = 1) */
rate = ixp4xx_timer_freq;
}
if (rate * a == ixp4xx_timer_freq) { /* don't divide by 0 later */
check_clock(rate, a - 1, 1, 1, best, &diff, reg);
return;
}
for (b = 0; b < 0x400; b++) {
u64 c = (b + 1) * (u64)rate;
do_div(c, ixp4xx_timer_freq - rate * a);
c--;
if (c >= 0xFFF) { /* 12-bit - no need to check more 'b's */
if (b == 0 && /* also try a bit higher rate */
!check_clock(rate, a - 1, 1, 1, best, &diff, reg))
return;
check_clock(rate, a, b, 0xFFF, best, &diff, reg);
return;
}
if (!check_clock(rate, a, b, c, best, &diff, reg))
return;
if (!check_clock(rate, a, b, c + 1, best, &diff, reg))
return;
}
}
static int hss_hdlc_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
{
const size_t size = sizeof(sync_serial_settings);
sync_serial_settings new_line;
sync_serial_settings __user *line = ifr->ifr_settings.ifs_ifsu.sync;
struct port *port = dev_to_port(dev);
unsigned long flags;
int clk;
if (cmd != SIOCWANDEV)
return hdlc_ioctl(dev, ifr, cmd);
switch(ifr->ifr_settings.type) {
case IF_GET_IFACE:
ifr->ifr_settings.type = IF_IFACE_V35;
if (ifr->ifr_settings.size < size) {
ifr->ifr_settings.size = size; /* data size wanted */
return -ENOBUFS;
}
memset(&new_line, 0, sizeof(new_line));
new_line.clock_type = port->clock_type;
IXP42x HSS support for setting internal clock rate HSS usually uses external clocks, so it's not a big deal. Internal clock is used for direct DTE-DTE connections and when the DCE doesn't provide it's own clock. This also depends on the oscillator frequency. Intel seems to have calculated the clock register settings for 33.33 MHz (66.66 MHz timer base). Their settings seem quite suboptimal both in terms of average frequency (60 ppm is unacceptable for G.703 applications, their primary intended usage(?)) and jitter. Many (most?) platforms use a 33.333 MHz oscillator, a 10 ppm difference from Intel's base. Instead of creating static tables, I've created a procedure to program the HSS clock register. The register consists of 3 parts (A, B, C). The average frequency (= bit rate) is: 66.66x MHz / (A + (B + 1) / (C + 1)) The procedure aims at the closest average frequency, possibly at the cost of increased jitter. Nobody would be able to directly drive an unbufferred transmitter with a HSS anyway, and the frequency error is what it really counts. I've verified the above with an oscilloscope on IXP425. It seems IXP46x and possibly IXP43x use a bit different clock generation algorithm - it looks like the avg frequency is: (on IXP465) 66.66x MHz / (A + B / (C + 1)). Also they use much greater precomputed A and B - on IXP425 it would simply result in more jitter, but I don't know how does it work on IXP46x (perhaps 3 least significant bits aren't used?). Anyway it looks that they were aiming for exactly +60 ppm or -60 ppm, while <1 ppm is typically possible (with a synchronized clock, of course). The attached patch makes it possible to set almost any bit rate (my IXP425 533 MHz quits at > 22 Mb/s if a single port is used, and the minimum is ca. 65 Kb/s). This is independent of MVIP (multi-E1/T1 on one HSS) mode. Signed-off-by: Krzysztof Hałasa <khc@pm.waw.pl> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-09-05 10:59:49 +07:00
new_line.clock_rate = port->clock_rate;
new_line.loopback = port->loopback;
if (copy_to_user(line, &new_line, size))
return -EFAULT;
return 0;
case IF_IFACE_SYNC_SERIAL:
case IF_IFACE_V35:
if(!capable(CAP_NET_ADMIN))
return -EPERM;
if (copy_from_user(&new_line, line, size))
return -EFAULT;
clk = new_line.clock_type;
if (port->plat->set_clock)
clk = port->plat->set_clock(port->id, clk);
if (clk != CLOCK_EXT && clk != CLOCK_INT)
return -EINVAL; /* No such clock setting */
if (new_line.loopback != 0 && new_line.loopback != 1)
return -EINVAL;
port->clock_type = clk; /* Update settings */
IXP42x HSS support for setting internal clock rate HSS usually uses external clocks, so it's not a big deal. Internal clock is used for direct DTE-DTE connections and when the DCE doesn't provide it's own clock. This also depends on the oscillator frequency. Intel seems to have calculated the clock register settings for 33.33 MHz (66.66 MHz timer base). Their settings seem quite suboptimal both in terms of average frequency (60 ppm is unacceptable for G.703 applications, their primary intended usage(?)) and jitter. Many (most?) platforms use a 33.333 MHz oscillator, a 10 ppm difference from Intel's base. Instead of creating static tables, I've created a procedure to program the HSS clock register. The register consists of 3 parts (A, B, C). The average frequency (= bit rate) is: 66.66x MHz / (A + (B + 1) / (C + 1)) The procedure aims at the closest average frequency, possibly at the cost of increased jitter. Nobody would be able to directly drive an unbufferred transmitter with a HSS anyway, and the frequency error is what it really counts. I've verified the above with an oscilloscope on IXP425. It seems IXP46x and possibly IXP43x use a bit different clock generation algorithm - it looks like the avg frequency is: (on IXP465) 66.66x MHz / (A + B / (C + 1)). Also they use much greater precomputed A and B - on IXP425 it would simply result in more jitter, but I don't know how does it work on IXP46x (perhaps 3 least significant bits aren't used?). Anyway it looks that they were aiming for exactly +60 ppm or -60 ppm, while <1 ppm is typically possible (with a synchronized clock, of course). The attached patch makes it possible to set almost any bit rate (my IXP425 533 MHz quits at > 22 Mb/s if a single port is used, and the minimum is ca. 65 Kb/s). This is independent of MVIP (multi-E1/T1 on one HSS) mode. Signed-off-by: Krzysztof Hałasa <khc@pm.waw.pl> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-09-05 10:59:49 +07:00
if (clk == CLOCK_INT)
find_best_clock(new_line.clock_rate, &port->clock_rate,
&port->clock_reg);
else {
port->clock_rate = 0;
port->clock_reg = CLK42X_SPEED_2048KHZ;
}
port->loopback = new_line.loopback;
spin_lock_irqsave(&npe_lock, flags);
if (dev->flags & IFF_UP)
hss_config(port);
if (port->loopback || port->carrier)
netif_carrier_on(port->netdev);
else
netif_carrier_off(port->netdev);
spin_unlock_irqrestore(&npe_lock, flags);
return 0;
default:
return hdlc_ioctl(dev, ifr, cmd);
}
}
/*****************************************************************************
* initialization
****************************************************************************/
static const struct net_device_ops hss_hdlc_ops = {
.ndo_open = hss_hdlc_open,
.ndo_stop = hss_hdlc_close,
.ndo_change_mtu = hdlc_change_mtu,
.ndo_start_xmit = hdlc_start_xmit,
.ndo_do_ioctl = hss_hdlc_ioctl,
};
static int __devinit hss_init_one(struct platform_device *pdev)
{
struct port *port;
struct net_device *dev;
hdlc_device *hdlc;
int err;
if ((port = kzalloc(sizeof(*port), GFP_KERNEL)) == NULL)
return -ENOMEM;
if ((port->npe = npe_request(0)) == NULL) {
err = -ENODEV;
goto err_free;
}
if ((port->netdev = dev = alloc_hdlcdev(port)) == NULL) {
err = -ENOMEM;
goto err_plat;
}
SET_NETDEV_DEV(dev, &pdev->dev);
hdlc = dev_to_hdlc(dev);
hdlc->attach = hss_hdlc_attach;
hdlc->xmit = hss_hdlc_xmit;
dev->netdev_ops = &hss_hdlc_ops;
dev->tx_queue_len = 100;
port->clock_type = CLOCK_EXT;
IXP42x HSS support for setting internal clock rate HSS usually uses external clocks, so it's not a big deal. Internal clock is used for direct DTE-DTE connections and when the DCE doesn't provide it's own clock. This also depends on the oscillator frequency. Intel seems to have calculated the clock register settings for 33.33 MHz (66.66 MHz timer base). Their settings seem quite suboptimal both in terms of average frequency (60 ppm is unacceptable for G.703 applications, their primary intended usage(?)) and jitter. Many (most?) platforms use a 33.333 MHz oscillator, a 10 ppm difference from Intel's base. Instead of creating static tables, I've created a procedure to program the HSS clock register. The register consists of 3 parts (A, B, C). The average frequency (= bit rate) is: 66.66x MHz / (A + (B + 1) / (C + 1)) The procedure aims at the closest average frequency, possibly at the cost of increased jitter. Nobody would be able to directly drive an unbufferred transmitter with a HSS anyway, and the frequency error is what it really counts. I've verified the above with an oscilloscope on IXP425. It seems IXP46x and possibly IXP43x use a bit different clock generation algorithm - it looks like the avg frequency is: (on IXP465) 66.66x MHz / (A + B / (C + 1)). Also they use much greater precomputed A and B - on IXP425 it would simply result in more jitter, but I don't know how does it work on IXP46x (perhaps 3 least significant bits aren't used?). Anyway it looks that they were aiming for exactly +60 ppm or -60 ppm, while <1 ppm is typically possible (with a synchronized clock, of course). The attached patch makes it possible to set almost any bit rate (my IXP425 533 MHz quits at > 22 Mb/s if a single port is used, and the minimum is ca. 65 Kb/s). This is independent of MVIP (multi-E1/T1 on one HSS) mode. Signed-off-by: Krzysztof Hałasa <khc@pm.waw.pl> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-09-05 10:59:49 +07:00
port->clock_rate = 0;
port->clock_reg = CLK42X_SPEED_2048KHZ;
port->id = pdev->id;
port->dev = &pdev->dev;
port->plat = pdev->dev.platform_data;
netif_napi_add(dev, &port->napi, hss_hdlc_poll, NAPI_WEIGHT);
if ((err = register_hdlc_device(dev)))
goto err_free_netdev;
platform_set_drvdata(pdev, port);
printk(KERN_INFO "%s: HSS-%i\n", dev->name, port->id);
return 0;
err_free_netdev:
free_netdev(dev);
err_plat:
npe_release(port->npe);
err_free:
kfree(port);
return err;
}
static int __devexit hss_remove_one(struct platform_device *pdev)
{
struct port *port = platform_get_drvdata(pdev);
unregister_hdlc_device(port->netdev);
free_netdev(port->netdev);
npe_release(port->npe);
platform_set_drvdata(pdev, NULL);
kfree(port);
return 0;
}
static struct platform_driver ixp4xx_hss_driver = {
.driver.name = DRV_NAME,
.probe = hss_init_one,
.remove = hss_remove_one,
};
static int __init hss_init_module(void)
{
if ((ixp4xx_read_feature_bits() &
(IXP4XX_FEATURE_HDLC | IXP4XX_FEATURE_HSS)) !=
(IXP4XX_FEATURE_HDLC | IXP4XX_FEATURE_HSS))
return -ENODEV;
spin_lock_init(&npe_lock);
return platform_driver_register(&ixp4xx_hss_driver);
}
static void __exit hss_cleanup_module(void)
{
platform_driver_unregister(&ixp4xx_hss_driver);
}
MODULE_AUTHOR("Krzysztof Halasa");
MODULE_DESCRIPTION("Intel IXP4xx HSS driver");
MODULE_LICENSE("GPL v2");
MODULE_ALIAS("platform:ixp4xx_hss");
module_init(hss_init_module);
module_exit(hss_cleanup_module);