freebsd-nq/sys/mips/nlm/hal/fmn.c
Jayachandran C. d4aba0f611 Fix credit configuration on Broadcom XLP CMS
The CMS output queue credit configuration register is 64 bit, so use
a 64 bit variable while updating it.
Obtained from:	Venkatesh J V <venkatesh.vivekanandan@broadcom.com>
2013-01-24 15:23:01 +00:00

355 lines
11 KiB
C

/*-
* Copyright 2003-2011 Netlogic Microsystems (Netlogic). All rights
* reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are
* met:
*
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. 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.
*
* THIS SOFTWARE IS PROVIDED BY Netlogic Microsystems ``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 NETLOGIC OR CONTRIBUTORS 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.
*
* NETLOGIC_BSD */
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/types.h>
#include <sys/systm.h>
#include <machine/cpufunc.h>
#include <mips/nlm/hal/mips-extns.h>
#include <mips/nlm/hal/haldefs.h>
#include <mips/nlm/hal/iomap.h>
#include <mips/nlm/hal/fmn.h>
/* XLP can take upto 16K of FMN messages per hardware queue, as spill.
* But, configuring all 16K causes the total spill memory to required
* to blow upto 192MB for single chip configuration, and 768MB in four
* chip configuration. Hence for now, we will setup the per queue spill
* as 1K FMN messages. With this, the total spill memory needed for 1024
* hardware queues (with 12bytes per single entry FMN message) becomes
* (1*1024)*12*1024queues = 12MB. For the four chip config, the memory
* needed = 12 * 4 = 48MB.
*/
uint64_t nlm_cms_spill_total_messages = 1 * 1024;
/* On a XLP832, we have the following FMN stations:
* CPU stations: 8
* PCIE0 stations: 1
* PCIE1 stations: 1
* PCIE2 stations: 1
* PCIE3 stations: 1
* GDX stations: 1
* CRYPTO stations: 1
* RSA stations: 1
* CMP stations: 1
* POE stations: 1
* NAE stations: 1
* ==================
* Total : 18 stations per chip
*
* For all 4 nodes, there are 18*4 = 72 FMN stations
*/
uint32_t nlm_cms_total_stations = 18 * 4 /*xlp_num_nodes*/;
/**
* Takes inputs as node, queue_size and maximum number of queues.
* Calculates the base, start & end and returns the same for a
* defined qid.
*
* The output queues are maintained in the internal output buffer
* which is a on-chip SRAM structure. For the actial hardware
* internal implementation, It is a structure which consists
* of eight banks of 4096-entry x message-width SRAMs. The SRAM
* implementation is designed to run at 1GHz with a 1-cycle read/write
* access. A read/write transaction can be initiated for each bank
* every cycle for a total of eight accesses per cycle. Successive
* entries of the same output queue are placed in successive banks.
* This is done to spread different read & write accesses to same/different
* output queue over as many different banks as possible so that they
* can be scheduled concurrently. Spreading the accesses to as many banks
* as possible to maximize the concurrency internally is important for
* achieving the desired peak throughput. This is done by h/w implementation
* itself.
*
* Output queues are allocated from this internal output buffer by
* software. The total capacity of the output buffer is 32K-entry.
* Each output queue can be sized from 32-entry to 1024-entry in
* increments of 32-entry. This is done by specifying a Start & a
* End pointer: pointers to the first & last 32-entry chunks allocated
* to the output queue.
*
* To optimize the storage required for 1024 OQ pointers, the upper 5-bits
* are shared by the Start & the End pointer. The side-effect of this
* optimization is that an OQ can't cross a 1024-entry boundary. Also, the
* lower 5-bits don't need to be specified in the Start & the End pointer
* as the allocation is in increments of 32-entries.
*
* Queue occupancy is tracked by a Head & a Tail pointer. Tail pointer
* indicates the location to which next entry will be written & Head
* pointer indicates the location from which next entry will be read. When
* these pointers reach the top of the allocated space (indicated by the
* End pointer), they are reset to the bottom of the allocated space
* (indicated by the Start pointer).
*
* Output queue pointer information:
* ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
*
* 14 10 9 5 4 0
* ------------------
* | base ptr |
* ------------------
* ----------------
* | start ptr |
* ----------------
* ----------------
* | end ptr |
* ----------------
* ------------------------------------
* | head ptr |
* ------------------------------------
* ------------------------------------
* | tail ptr |
* ------------------------------------
* Note:
* A total of 1024 segments can sit on one software-visible "bank"
* of internal SRAM. Each segment contains 32 entries. Also note
* that sw-visible "banks" are not the same as the actual internal
* 8-bank implementation of hardware. It is an optimization of
* internal access.
*
*/
void nlm_cms_setup_credits(uint64_t base, int destid, int srcid, int credit)
{
uint64_t val;
val = (((uint64_t)credit << 24) | (destid << 12) | (srcid << 0));
nlm_write_cms_reg(base, CMS_OUTPUTQ_CREDIT_CFG, val);
}
/*
* base - CMS module base address for this node.
* qid - is the output queue id otherwise called as vc id
* spill_base - is the 40-bit physical address of spill memory. Must be
4KB aligned.
* nsegs - No of segments where a "1" indicates 4KB. Spill size must be
* a multiple of 4KB.
*/
int nlm_cms_alloc_spill_q(uint64_t base, int qid, uint64_t spill_base,
int nsegs)
{
uint64_t queue_config;
uint32_t spill_start;
if (nsegs > CMS_MAX_SPILL_SEGMENTS_PER_QUEUE) {
return 1;
}
queue_config = nlm_read_cms_reg(base,(CMS_OUTPUTQ_CONFIG(qid)));
spill_start = ((spill_base >> 12) & 0x3F);
/* Spill configuration */
queue_config = (((uint64_t)CMS_SPILL_ENA << 62) |
(((spill_base >> 18) & 0x3FFFFF) << 27) |
(spill_start + nsegs - 1) << 21 |
(spill_start << 15));
nlm_write_cms_reg(base,(CMS_OUTPUTQ_CONFIG(qid)),queue_config);
return 0;
}
uint64_t nlm_cms_get_onchip_queue (uint64_t base, int qid)
{
return nlm_read_cms_reg(base, CMS_OUTPUTQ_CONFIG(qid));
}
void nlm_cms_set_onchip_queue (uint64_t base, int qid, uint64_t val)
{
uint64_t rdval;
rdval = nlm_read_cms_reg(base, CMS_OUTPUTQ_CONFIG(qid));
rdval |= val;
nlm_write_cms_reg(base, CMS_OUTPUTQ_CONFIG(qid), rdval);
}
void nlm_cms_per_queue_level_intr(uint64_t base, int qid, int sub_type,
int intr_val)
{
uint64_t val;
val = nlm_read_cms_reg(base, CMS_OUTPUTQ_CONFIG(qid));
val &= ~((0x7ULL << 56) | (0x3ULL << 54));
val |= (((uint64_t)sub_type<<54) |
((uint64_t)intr_val<<56));
nlm_write_cms_reg(base, CMS_OUTPUTQ_CONFIG(qid), val);
}
void nlm_cms_per_queue_timer_intr(uint64_t base, int qid, int sub_type,
int intr_val)
{
uint64_t val;
val = nlm_read_cms_reg(base, CMS_OUTPUTQ_CONFIG(qid));
val &= ~((0x7ULL << 51) | (0x3ULL << 49));
val |= (((uint64_t)sub_type<<49) |
((uint64_t)intr_val<<51));
nlm_write_cms_reg(base, CMS_OUTPUTQ_CONFIG(qid), val);
}
/* returns 1 if interrupt has been generated for this output queue */
int nlm_cms_outputq_intr_check(uint64_t base, int qid)
{
uint64_t val;
val = nlm_read_cms_reg(base, CMS_OUTPUTQ_CONFIG(qid));
return ((val >> 59) & 0x1);
}
void nlm_cms_outputq_clr_intr(uint64_t base, int qid)
{
uint64_t val;
val = nlm_read_cms_reg(base, CMS_OUTPUTQ_CONFIG(qid));
val |= (1ULL<<59);
nlm_write_cms_reg(base, CMS_OUTPUTQ_CONFIG(qid), val);
}
void nlm_cms_illegal_dst_error_intr(uint64_t base, int en)
{
uint64_t val;
val = nlm_read_cms_reg(base, CMS_MSG_CONFIG);
val |= (en<<8);
nlm_write_cms_reg(base, CMS_MSG_CONFIG, val);
}
void nlm_cms_timeout_error_intr(uint64_t base, int en)
{
uint64_t val;
val = nlm_read_cms_reg(base, CMS_MSG_CONFIG);
val |= (en<<7);
nlm_write_cms_reg(base, CMS_MSG_CONFIG, val);
}
void nlm_cms_biu_error_resp_intr(uint64_t base, int en)
{
uint64_t val;
val = nlm_read_cms_reg(base, CMS_MSG_CONFIG);
val |= (en<<6);
nlm_write_cms_reg(base, CMS_MSG_CONFIG, val);
}
void nlm_cms_spill_uncorrectable_ecc_error_intr(uint64_t base, int en)
{
uint64_t val;
val = nlm_read_cms_reg(base, CMS_MSG_CONFIG);
val |= (en<<5) | (en<<3);
nlm_write_cms_reg(base, CMS_MSG_CONFIG, val);
}
void nlm_cms_spill_correctable_ecc_error_intr(uint64_t base, int en)
{
uint64_t val;
val = nlm_read_cms_reg(base, CMS_MSG_CONFIG);
val |= (en<<4) | (en<<2);
nlm_write_cms_reg(base, CMS_MSG_CONFIG, val);
}
void nlm_cms_outputq_uncorrectable_ecc_error_intr(uint64_t base, int en)
{
uint64_t val;
val = nlm_read_cms_reg(base, CMS_MSG_CONFIG);
val |= (en<<1);
nlm_write_cms_reg(base, CMS_MSG_CONFIG, val);
}
void nlm_cms_outputq_correctable_ecc_error_intr(uint64_t base, int en)
{
uint64_t val;
val = nlm_read_cms_reg(base, CMS_MSG_CONFIG);
val |= (en<<0);
nlm_write_cms_reg(base, CMS_MSG_CONFIG, val);
}
uint64_t nlm_cms_network_error_status(uint64_t base)
{
return nlm_read_cms_reg(base, CMS_MSG_ERR);
}
int nlm_cms_get_net_error_code(uint64_t err)
{
return ((err >> 12) & 0xf);
}
int nlm_cms_get_net_error_syndrome(uint64_t err)
{
return ((err >> 32) & 0x1ff);
}
int nlm_cms_get_net_error_ramindex(uint64_t err)
{
return ((err >> 44) & 0x7fff);
}
int nlm_cms_get_net_error_outputq(uint64_t err)
{
return ((err >> 16) & 0xfff);
}
/*========================= FMN Tracing related APIs ================*/
void nlm_cms_trace_setup(uint64_t base, int en, uint64_t trace_base,
uint64_t trace_limit, int match_dstid_en,
int dst_id, int match_srcid_en, int src_id,
int wrap)
{
uint64_t val;
nlm_write_cms_reg(base, CMS_TRACE_BASE_ADDR, trace_base);
nlm_write_cms_reg(base, CMS_TRACE_LIMIT_ADDR, trace_limit);
val = nlm_read_cms_reg(base, CMS_TRACE_CONFIG);
val |= (((uint64_t)match_dstid_en << 39) |
((dst_id & 0xfff) << 24) |
(match_srcid_en << 23) |
((src_id & 0xfff) << 8) |
(wrap << 1) |
(en << 0));
nlm_write_cms_reg(base, CMS_MSG_CONFIG, val);
}
void nlm_cms_endian_byte_swap (uint64_t base, int en)
{
nlm_write_cms_reg(base, CMS_MSG_ENDIAN_SWAP, en);
}