upstream u-boot with additional patches for our devices/boards: https://lists.denx.de/pipermail/u-boot/2017-March/282789.html (AXP crashes) ; Gbit ethernet patch for some LIME2 revisions ; with SPI flash support
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
u-boot/arch/arm/cpu/ixp/npe/IxEthAccCommon.c

1050 lines
29 KiB

/**
* @file IxEthAccCommon.c
*
* @author Intel Corporation
* @date 12-Feb-2002
*
* @brief This file contains the implementation common support routines for the component
*
* Design Notes:
*
* @par
* IXP400 SW Release version 2.0
*
* -- Copyright Notice --
*
* @par
* Copyright 2001-2005, Intel Corporation.
* All rights reserved.
*
* @par
* 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.
* 3. Neither the name of the Intel Corporation nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* @par
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS ``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 THE COPYRIGHT OWNER 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.
*
* @par
* -- End of Copyright Notice --
*/
/*
* Component header files
*/
#include "IxOsal.h"
#include "IxEthAcc.h"
#include "IxEthDB.h"
#include "IxNpeMh.h"
#include "IxEthDBPortDefs.h"
#include "IxFeatureCtrl.h"
#include "IxEthAcc_p.h"
#include "IxEthAccQueueAssign_p.h"
#include "IxEthAccDataPlane_p.h"
#include "IxEthAccMii_p.h"
/**
* @addtogroup IxEthAccPri
*@{
*/
extern IxEthAccInfo ixEthAccDataInfo;
/**
*
* @brief Maximum number of RX queues set to be the maximum number
* of traffic calsses.
*
*/
#define IX_ETHACC_MAX_RX_QUEUES \
(IX_ETH_DB_QOS_TRAFFIC_CLASS_7_RX_QUEUE_PROPERTY \
- IX_ETH_DB_QOS_TRAFFIC_CLASS_0_RX_QUEUE_PROPERTY \
+ 1)
/**
*
* @brief Maximum number of 128 entry RX queues
*
*/
#define IX_ETHACC_MAX_LARGE_RX_QUEUES 4
/**
*
* @brief Data structure template for Default RX Queues
*
*/
IX_ETH_ACC_PRIVATE
IxEthAccQregInfo ixEthAccQmgrRxDefaultTemplate =
{
IX_ETH_ACC_RX_FRAME_ETH_Q, /**< Queue ID */
"Eth Rx Q",
ixEthRxFrameQMCallback, /**< Functional callback */
(IxQMgrCallbackId) 0, /**< Callback tag */
IX_QMGR_Q_SIZE128, /**< Allocate Max Size Q */
IX_QMGR_Q_ENTRY_SIZE1, /**< Queue Entry Sizes - all Q entries are single word entries */
TRUE, /**< Enable Q notification at startup */
IX_ETH_ACC_RX_FRAME_ETH_Q_SOURCE,/**< Q Condition to drive callback */
IX_QMGR_Q_WM_LEVEL0, /**< Q Low water mark */
IX_QMGR_Q_WM_LEVEL1, /**< Q High water mark - needed by NPE */
};
/**
*
* @brief Data structure template for Small RX Queues
*
*/
IX_ETH_ACC_PRIVATE
IxEthAccQregInfo ixEthAccQmgrRxSmallTemplate =
{
IX_ETH_ACC_RX_FRAME_ETH_Q, /**< Queue ID */
"Eth Rx Q",
ixEthRxFrameQMCallback, /**< Functional callback */
(IxQMgrCallbackId) 0, /**< Callback tag */
IX_QMGR_Q_SIZE64, /**< Allocate Smaller Q */
IX_QMGR_Q_ENTRY_SIZE1, /**< Queue Entry Sizes - all Q entries are single word entries */
TRUE, /**< Enable Q notification at startup */
IX_ETH_ACC_RX_FRAME_ETH_Q_SOURCE,/**< Q Condition to drive callback */
IX_QMGR_Q_WM_LEVEL0, /**< Q Low water mark */
IX_QMGR_Q_WM_LEVEL1, /**< Q High water mark - needed by NPE */
};
/**
*
* @brief Data structure used to register & initialize the Queues
*
*/
IX_ETH_ACC_PRIVATE
IxEthAccQregInfo ixEthAccQmgrStaticInfo[]=
{
{
IX_ETH_ACC_RX_FREE_BUFF_ENET0_Q,
"Eth Rx Fr Q 1",
ixEthRxFreeQMCallback,
(IxQMgrCallbackId) IX_ETH_PORT_1,
IX_QMGR_Q_SIZE128, /**< Allocate Max Size Q */
IX_QMGR_Q_ENTRY_SIZE1, /**< Queue Entry Sizes - all Q entries are single word entries */
FALSE, /**< Disable Q notification at startup */
IX_ETH_ACC_RX_FREE_BUFF_ENET0_Q_SOURCE, /**< Q Condition to drive callback */
IX_QMGR_Q_WM_LEVEL0, /***< Q Low water mark */
IX_QMGR_Q_WM_LEVEL64, /**< Q High water mark */
},
{
IX_ETH_ACC_RX_FREE_BUFF_ENET1_Q,
"Eth Rx Fr Q 2",
ixEthRxFreeQMCallback,
(IxQMgrCallbackId) IX_ETH_PORT_2,
IX_QMGR_Q_SIZE128, /**< Allocate Max Size Q */
IX_QMGR_Q_ENTRY_SIZE1, /**< Queue Entry Sizes - all Q entries are single word entries */
FALSE, /**< Disable Q notification at startup */
IX_ETH_ACC_RX_FREE_BUFF_ENET1_Q_SOURCE, /**< Q Condition to drive callback */
IX_QMGR_Q_WM_LEVEL0, /**< Q Low water mark */
IX_QMGR_Q_WM_LEVEL64, /**< Q High water mark */
},
#ifdef __ixp46X
{
IX_ETH_ACC_RX_FREE_BUFF_ENET2_Q,
"Eth Rx Fr Q 3",
ixEthRxFreeQMCallback,
(IxQMgrCallbackId) IX_ETH_PORT_3,
IX_QMGR_Q_SIZE128, /**< Allocate Max Size Q */
IX_QMGR_Q_ENTRY_SIZE1, /**< Queue Entry Sizes - all Q entries are single word entries */
FALSE, /**< Disable Q notification at startup */
IX_ETH_ACC_RX_FREE_BUFF_ENET2_Q_SOURCE, /**< Q Condition to drive callback */
IX_QMGR_Q_WM_LEVEL0, /**< Q Low water mark */
IX_QMGR_Q_WM_LEVEL64, /**< Q High water mark */
},
#endif
{
IX_ETH_ACC_TX_FRAME_ENET0_Q,
"Eth Tx Q 1",
ixEthTxFrameQMCallback,
(IxQMgrCallbackId) IX_ETH_PORT_1,
IX_QMGR_Q_SIZE128, /**< Allocate Max Size Q */
IX_QMGR_Q_ENTRY_SIZE1, /**< Queue Entry Sizes - all Q entries are single word entries */
FALSE, /**< Disable Q notification at startup */
IX_ETH_ACC_TX_FRAME_ENET0_Q_SOURCE, /**< Q Condition to drive callback */
IX_QMGR_Q_WM_LEVEL0, /**< Q Low water mark */
IX_QMGR_Q_WM_LEVEL64, /**< Q High water mark */
},
{
IX_ETH_ACC_TX_FRAME_ENET1_Q,
"Eth Tx Q 2",
ixEthTxFrameQMCallback,
(IxQMgrCallbackId) IX_ETH_PORT_2,
IX_QMGR_Q_SIZE128, /**< Allocate Max Size Q */
IX_QMGR_Q_ENTRY_SIZE1, /**< Queue Entry Sizes - all Q entries are single word entries */
FALSE, /**< Disable Q notification at startup */
IX_ETH_ACC_TX_FRAME_ENET1_Q_SOURCE, /**< Q Condition to drive callback */
IX_QMGR_Q_WM_LEVEL0, /**< Q Low water mark */
IX_QMGR_Q_WM_LEVEL64, /**< Q High water mark */
},
#ifdef __ixp46X
{
IX_ETH_ACC_TX_FRAME_ENET2_Q,
"Eth Tx Q 3",
ixEthTxFrameQMCallback,
(IxQMgrCallbackId) IX_ETH_PORT_3,
IX_QMGR_Q_SIZE128, /**< Allocate Max Size Q */
IX_QMGR_Q_ENTRY_SIZE1, /** Queue Entry Sizes - all Q entries are single ord entries */
FALSE, /** Disable Q notification at startup */
IX_ETH_ACC_TX_FRAME_ENET2_Q_SOURCE, /** Q Condition to drive callback */
IX_QMGR_Q_WM_LEVEL0, /* No queues use almost empty */
IX_QMGR_Q_WM_LEVEL64, /** Q High water mark - needed used */
},
#endif
{
IX_ETH_ACC_TX_FRAME_DONE_ETH_Q,
"Eth Tx Done Q",
ixEthTxFrameDoneQMCallback,
(IxQMgrCallbackId) 0,
IX_QMGR_Q_SIZE128, /**< Allocate Max Size Q */
IX_QMGR_Q_ENTRY_SIZE1, /**< Queue Entry Sizes - all Q entries are single word entries */
TRUE, /**< Enable Q notification at startup */
IX_ETH_ACC_TX_FRAME_DONE_ETH_Q_SOURCE, /**< Q Condition to drive callback */
IX_QMGR_Q_WM_LEVEL0, /**< Q Low water mark */
IX_QMGR_Q_WM_LEVEL2, /**< Q High water mark - needed by NPE */
},
{ /* Null Termination entry
*/
(IxQMgrQId)0,
(char *) NULL,
(IxQMgrCallback) NULL,
(IxQMgrCallbackId) 0,
0,
0,
0,
0,
0,
0
}
};
/**
*
* @brief Data structure used to register & initialize the Queues
*
* The structure will be filled at run time depending on the NPE
* image already loaded and the QoS configured in ethDB.
*
*/
IX_ETH_ACC_PRIVATE
IxEthAccQregInfo ixEthAccQmgrRxQueuesInfo[IX_ETHACC_MAX_RX_QUEUES+1]=
{
{ /* PlaceHolder for rx queues
* depending on the QoS configured
*/
(IxQMgrQId)0,
(char *) NULL,
(IxQMgrCallback) NULL,
(IxQMgrCallbackId) 0,
0,
0,
0,
0,
0,
0
},
{ /* PlaceHolder for rx queues
* depending on the QoS configured
*/
(IxQMgrQId)0,
(char *) NULL,
(IxQMgrCallback) NULL,
(IxQMgrCallbackId) 0,
0,
0,
0,
0,
0,
0
},
{ /* PlaceHolder for rx queues
* depending on the QoS configured
*/
(IxQMgrQId)0,
(char *) NULL,
(IxQMgrCallback) NULL,
(IxQMgrCallbackId) 0,
0,
0,
0,
0,
0,
0
},
{ /* PlaceHolder for rx queues
* depending on the QoS configured
*/
(IxQMgrQId)0,
(char *) NULL,
(IxQMgrCallback) NULL,
(IxQMgrCallbackId) 0,
0,
0,
0,
0,
0,
0
},
{ /* PlaceHolder for rx queues
* depending on the QoS configured
*/
(IxQMgrQId)0,
(char *) NULL,
(IxQMgrCallback) NULL,
(IxQMgrCallbackId) 0,
0,
0,
0,
0,
0,
0
},
{ /* PlaceHolder for rx queues
* depending on the QoS configured
*/
(IxQMgrQId)0,
(char *) NULL,
(IxQMgrCallback) NULL,
(IxQMgrCallbackId) 0,
0,
0,
0,
0,
0,
0
},
{ /* PlaceHolder for rx queues
* depending on the QoS configured
*/
(IxQMgrQId)0,
(char *) NULL,
(IxQMgrCallback) NULL,
(IxQMgrCallbackId) 0,
0,
0,
0,
0,
0,
0
},
{ /* PlaceHolder for rx queues
* depending on the QoS configured
*/
(IxQMgrQId)0,
(char *) NULL,
(IxQMgrCallback) NULL,
(IxQMgrCallbackId) 0,
0,
0,
0,
0,
0,
0
},
{ /* Null Termination entry
*/
(IxQMgrQId)0,
(char *) NULL,
(IxQMgrCallback) NULL,
(IxQMgrCallbackId) 0,
0,
0,
0,
0,
0,
0
}
};
/* forward declarations */
IX_ETH_ACC_PRIVATE IxEthAccStatus
ixEthAccQMgrQueueSetup(IxEthAccQregInfo *qInfoDes);
/**
* @fn ixEthAccQMgrQueueSetup(void)
*
* @brief Setup one queue and its event, and register the callback required
* by this component to the QMgr
*
* @internal
*/
IX_ETH_ACC_PRIVATE IxEthAccStatus
ixEthAccQMgrQueueSetup(IxEthAccQregInfo *qInfoDes)
{
/*
* Configure each Q.
*/
if ( ixQMgrQConfig( qInfoDes->qName,
qInfoDes->qId,
qInfoDes->qSize,
qInfoDes->qWords) != IX_SUCCESS)
{
return IX_ETH_ACC_FAIL;
}
if ( ixQMgrWatermarkSet( qInfoDes->qId,
qInfoDes->AlmostEmptyThreshold,
qInfoDes->AlmostFullThreshold
) != IX_SUCCESS)
{
return IX_ETH_ACC_FAIL;
}
/*
* Set dispatcher priority.
*/
if ( ixQMgrDispatcherPrioritySet( qInfoDes->qId,
IX_ETH_ACC_QM_QUEUE_DISPATCH_PRIORITY)
!= IX_SUCCESS)
{
return IX_ETH_ACC_FAIL;
}
/*
* Register callbacks for each Q.
*/
if ( ixQMgrNotificationCallbackSet(qInfoDes->qId,
qInfoDes->qCallback,
qInfoDes->callbackTag)
!= IX_SUCCESS )
{
return IX_ETH_ACC_FAIL;
}
/*
* Set notification condition for Q
*/
if ( qInfoDes->qNotificationEnableAtStartup == TRUE )
{
if ( ixQMgrNotificationEnable(qInfoDes->qId,
qInfoDes->qConditionSource)
!= IX_SUCCESS )
{
return IX_ETH_ACC_FAIL;
}
}
return(IX_ETH_ACC_SUCCESS);
}
/**
* @fn ixEthAccQMgrQueuesConfig(void)
*
* @brief Setup all the queues and register all callbacks required
* by this component to the QMgr
*
* The RxFree queues, tx queues, rx queues are configured statically
*
* Rx queues configuration is driven by QoS setup.
* Many Rx queues may be required when QoS is enabled (this depends
* on IxEthDB setup and the images being downloaded). The configuration
* of the rxQueues is done in many steps as follows:
*
* @li select all Rx queues as configured by ethDB for all ports
* @li sort the queues by traffic class
* @li build the priority dependency for all queues
* @li fill the configuration for all rx queues
* @li configure all statically configured queues
* @li configure all dynamically configured queues
*
* @param none
*
* @return IxEthAccStatus
*
* @internal
*/
IX_ETH_ACC_PUBLIC
IxEthAccStatus ixEthAccQMgrQueuesConfig(void)
{
struct
{
int npeCount;
UINT32 npeId;
IxQMgrQId qId;
IxEthDBProperty trafficClass;
} rxQueues[IX_ETHACC_MAX_RX_QUEUES];
UINT32 rxQueue = 0;
UINT32 rxQueueCount = 0;
IxQMgrQId ixQId =IX_QMGR_MAX_NUM_QUEUES;
IxEthDBStatus ixEthDBStatus = IX_ETH_DB_SUCCESS;
IxEthDBPortId ixEthDbPortId = 0;
IxEthAccPortId ixEthAccPortId = 0;
UINT32 ixNpeId = 0;
UINT32 ixHighestNpeId = 0;
UINT32 sortIterations = 0;
IxEthAccStatus ret = IX_ETH_ACC_SUCCESS;
IxEthAccQregInfo *qInfoDes = NULL;
IxEthDBProperty ixEthDBTrafficClass = IX_ETH_DB_QOS_TRAFFIC_CLASS_0_RX_QUEUE_PROPERTY;
IxEthDBPropertyType ixEthDBPropertyType = IX_ETH_DB_INTEGER_PROPERTY;
UINT32 ixEthDBParameter = 0;
BOOL completelySorted = FALSE;
/* Fill the corspondance between ports and queues
* This defines the mapping from port to queue Ids.
*/
ixEthAccPortData[IX_ETH_PORT_1].ixEthAccRxData.rxFreeQueue
= IX_ETH_ACC_RX_FREE_BUFF_ENET0_Q;
ixEthAccPortData[IX_ETH_PORT_2].ixEthAccRxData.rxFreeQueue
= IX_ETH_ACC_RX_FREE_BUFF_ENET1_Q;
#ifdef __ixp46X
ixEthAccPortData[IX_ETH_PORT_3].ixEthAccRxData.rxFreeQueue
= IX_ETH_ACC_RX_FREE_BUFF_ENET2_Q;
#endif
ixEthAccPortData[IX_ETH_PORT_1].ixEthAccTxData.txQueue
= IX_ETH_ACC_TX_FRAME_ENET0_Q;
ixEthAccPortData[IX_ETH_PORT_2].ixEthAccTxData.txQueue
= IX_ETH_ACC_TX_FRAME_ENET1_Q;
#ifdef __ixp46X
ixEthAccPortData[IX_ETH_PORT_3].ixEthAccTxData.txQueue
= IX_ETH_ACC_TX_FRAME_ENET2_Q;
#endif
/* Fill the corspondance between ports and NPEs
* This defines the mapping from port to npeIds.
*/
ixEthAccPortData[IX_ETH_PORT_1].npeId = IX_NPEMH_NPEID_NPEB;
ixEthAccPortData[IX_ETH_PORT_2].npeId = IX_NPEMH_NPEID_NPEC;
#ifdef __ixp46X
ixEthAccPortData[IX_ETH_PORT_3].npeId = IX_NPEMH_NPEID_NPEA;
#endif
/* set the default rx scheduling discipline */
ixEthAccDataInfo.schDiscipline = FIFO_NO_PRIORITY;
/*
* Queue Selection step:
*
* The following code selects all the queues and build
* a temporary array which contains for each queue
* - the queue Id,
* - the highest traffic class (in case of many
* priorities configured for the same queue on different
* ports)
* - the number of different Npes which are
* configured to write to this queue.
*
* The output of this loop is a temporary array of RX queues
* in any order.
*
*/
#ifdef CONFIG_IXP425_COMPONENT_ETHDB
for (ixEthAccPortId = 0;
(ixEthAccPortId < IX_ETH_ACC_NUMBER_OF_PORTS)
&& (ret == IX_ETH_ACC_SUCCESS);
ixEthAccPortId++)
{
/* map between ethDb and ethAcc port Ids */
ixEthDbPortId = (IxEthDBPortId)ixEthAccPortId;
/* map between npeId and ethAcc port Ids */
ixNpeId = IX_ETH_ACC_PORT_TO_NPE_ID(ixEthAccPortId);
/* Iterate thru the different priorities */
for (ixEthDBTrafficClass = IX_ETH_DB_QOS_TRAFFIC_CLASS_0_RX_QUEUE_PROPERTY;
ixEthDBTrafficClass <= IX_ETH_DB_QOS_TRAFFIC_CLASS_7_RX_QUEUE_PROPERTY;
ixEthDBTrafficClass++)
{
ixEthDBStatus = ixEthDBFeaturePropertyGet(
ixEthDbPortId,
IX_ETH_DB_VLAN_QOS,
ixEthDBTrafficClass,
&ixEthDBPropertyType,
(void *)&ixEthDBParameter);
if (ixEthDBStatus == IX_ETH_DB_SUCCESS)
{
/* This port and QoS class are mapped to
* a RX queue.
*/
if (ixEthDBPropertyType == IX_ETH_DB_INTEGER_PROPERTY)
{
/* remember the highest npe Id supporting ethernet */
if (ixNpeId > ixHighestNpeId)
{
ixHighestNpeId = ixNpeId;
}
/* search the queue in the list of queues
* already used by an other port or QoS
*/
for (rxQueue = 0;
rxQueue < rxQueueCount;
rxQueue++)
{
if (rxQueues[rxQueue].qId == (IxQMgrQId)ixEthDBParameter)
{
/* found an existing setup, update the number of ports
* for this queue if the port maps to
* a different NPE.
*/
if (rxQueues[rxQueue].npeId != ixNpeId)
{
rxQueues[rxQueue].npeCount++;
rxQueues[rxQueue].npeId = ixNpeId;
}
/* get the highest traffic class for this queue */
if (rxQueues[rxQueue].trafficClass > ixEthDBTrafficClass)
{
rxQueues[rxQueue].trafficClass = ixEthDBTrafficClass;
}
break;
}
}
if (rxQueue == rxQueueCount)
{
/* new queue not found in the current list,
* add a new entry.
*/
IX_OSAL_ASSERT(rxQueueCount < IX_ETHACC_MAX_RX_QUEUES);
rxQueues[rxQueueCount].qId = ixEthDBParameter;
rxQueues[rxQueueCount].npeCount = 1;
rxQueues[rxQueueCount].npeId = ixNpeId;
rxQueues[rxQueueCount].trafficClass = ixEthDBTrafficClass;
rxQueueCount++;
}
}
else
{
/* unexpected property type (not Integer) */
ret = IX_ETH_ACC_FAIL;
IX_ETH_ACC_WARNING_LOG("ixEthAccQMgrQueuesConfig: unexpected property type returned by EthDB\n", 0, 0, 0, 0, 0, 0);
/* no point to continue to iterate */
break;
}
}
else
{
/* No Rx queue configured for this port
* and this traffic class. Do nothing.
*/
}
}
/* notify EthDB that queue initialization is complete and traffic class allocation is frozen */
ixEthDBFeaturePropertySet(ixEthDbPortId,
IX_ETH_DB_VLAN_QOS,
IX_ETH_DB_QOS_QUEUE_CONFIGURATION_COMPLETE,
NULL /* ignored */);
}
#else
ixNpeId = IX_ETH_ACC_PORT_TO_NPE_ID(ixEthAccPortId);
rxQueues[0].qId = 4;
rxQueues[0].npeCount = 1;
rxQueues[0].npeId = ixNpeId;
rxQueues[0].trafficClass = IX_ETH_DB_QOS_TRAFFIC_CLASS_0_RX_QUEUE_PROPERTY;
rxQueueCount++;
#endif
/* check there is at least 1 rx queue : there is no point
* to continue if there is no rx queue configured
*/
if ((rxQueueCount == 0) || (ret == IX_ETH_ACC_FAIL))
{
IX_ETH_ACC_WARNING_LOG("ixEthAccQMgrQueuesConfig: no queues configured, bailing out\n", 0, 0, 0, 0, 0, 0);
return (IX_ETH_ACC_FAIL);
}
/* Queue sort step:
*
* Re-order the array of queues by decreasing traffic class
* using a bubble sort. (trafficClass 0 is the lowest
* priority traffic, trafficClass 7 is the highest priority traffic)
*
* Primary sort order is traffic class
* Secondary sort order is npeId
*
* Note that a bubble sort algorithm is not very efficient when
* the number of queues grows . However, this is not a very bad choice
* considering the very small number of entries to sort. Also, bubble
* sort is extremely fast when the list is already sorted.
*
* The output of this loop is a sorted array of queues.
*
*/
sortIterations = 0;
do
{
sortIterations++;
completelySorted = TRUE;
for (rxQueue = 0;
rxQueue < rxQueueCount - sortIterations;
rxQueue++)
{
/* compare adjacent elements */
if ((rxQueues[rxQueue].trafficClass <
rxQueues[rxQueue+1].trafficClass)
|| ((rxQueues[rxQueue].trafficClass ==
rxQueues[rxQueue+1].trafficClass)
&&(rxQueues[rxQueue].npeId <
rxQueues[rxQueue+1].npeId)))
{
/* swap adjacent elements */
int npeCount = rxQueues[rxQueue].npeCount;
UINT32 npeId = rxQueues[rxQueue].npeId;
IxQMgrQId qId = rxQueues[rxQueue].qId;
IxEthDBProperty trafficClass = rxQueues[rxQueue].trafficClass;
rxQueues[rxQueue].npeCount = rxQueues[rxQueue+1].npeCount;
rxQueues[rxQueue].npeId = rxQueues[rxQueue+1].npeId;
rxQueues[rxQueue].qId = rxQueues[rxQueue+1].qId;
rxQueues[rxQueue].trafficClass = rxQueues[rxQueue+1].trafficClass;
rxQueues[rxQueue+1].npeCount = npeCount;
rxQueues[rxQueue+1].npeId = npeId;
rxQueues[rxQueue+1].qId = qId;
rxQueues[rxQueue+1].trafficClass = trafficClass;
completelySorted = FALSE;
}
}
}
while (!completelySorted);
/* Queue traffic class list:
*
* Fill an array of rx queues linked by ascending traffic classes.
*
* If the queues are configured as follows
* qId 6 -> traffic class 0 (lowest)
* qId 7 -> traffic class 0
* qId 8 -> traffic class 6
* qId 12 -> traffic class 7 (highest)
*
* Then the output of this loop will be
*
* higherPriorityQueue[6] = 8
* higherPriorityQueue[7] = 8
* higherPriorityQueue[8] = 12
* higherPriorityQueue[12] = Invalid queueId
* higherPriorityQueue[...] = Invalid queueId
*
* Note that this queue ordering does not handle all possibilities
* that could result from different rules associated with different
* ports, and inconsistencies in the rules. In all cases, the
* output of this algorithm is a simple linked list of queues,
* without closed circuit.
* This list is implemented as an array with invalid values initialized
* with an "invalid" queue id which is the maximum number of queues.
*
*/
/*
* Initialise the rx queue list.
*/
for (rxQueue = 0; rxQueue < IX_QMGR_MAX_NUM_QUEUES; rxQueue++)
{
ixEthAccDataInfo.higherPriorityQueue[rxQueue] = IX_QMGR_MAX_NUM_QUEUES;
}
/* build the linked list for this NPE.
*/
for (ixNpeId = 0;
ixNpeId <= ixHighestNpeId;
ixNpeId++)
{
/* iterate thru the sorted list of queues
*/
ixQId = IX_QMGR_MAX_NUM_QUEUES;
for (rxQueue = 0;
rxQueue < rxQueueCount;
rxQueue++)
{
if (rxQueues[rxQueue].npeId == ixNpeId)
{
ixEthAccDataInfo.higherPriorityQueue[rxQueues[rxQueue].qId] = ixQId;
/* iterate thru queues with the same traffic class
* than the current queue. (queues are ordered by descending
* traffic classes and npeIds).
*/
while ((rxQueue < rxQueueCount - 1)
&& (rxQueues[rxQueue].trafficClass
== rxQueues[rxQueue+1].trafficClass)
&& (ixNpeId == rxQueues[rxQueue].npeId))
{
rxQueue++;
ixEthAccDataInfo.higherPriorityQueue[rxQueues[rxQueue].qId] = ixQId;
}
ixQId = rxQueues[rxQueue].qId;
}
}
}
/* point on the first dynamic queue description */
qInfoDes = ixEthAccQmgrRxQueuesInfo;
/* update the list of queues with the rx queues */
for (rxQueue = 0;
(rxQueue < rxQueueCount) && (ret == IX_ETH_ACC_SUCCESS);
rxQueue++)
{
/* Don't utilize more than IX_ETHACC_MAX_LARGE_RX_QUEUES queues
* with the full 128 entries. For the lower priority queues, use
* a smaller number of entries. This ensures queue resources
* remain available for other components.
*/
if( (rxQueueCount > IX_ETHACC_MAX_LARGE_RX_QUEUES) &&
(rxQueue < rxQueueCount - IX_ETHACC_MAX_LARGE_RX_QUEUES) )
{
/* add the small RX Queue setup template to the list of queues */
memcpy(qInfoDes, &ixEthAccQmgrRxSmallTemplate, sizeof(*qInfoDes));
} else {
/* add the default RX Queue setup template to the list of queues */
memcpy(qInfoDes, &ixEthAccQmgrRxDefaultTemplate, sizeof(*qInfoDes));
}
/* setup the RxQueue ID */
qInfoDes->qId = rxQueues[rxQueue].qId;
/* setup the RxQueue watermark level
*
* Each queue can be filled by many NPEs. To avoid the
* NPEs to write to a full queue, need to set the
* high watermark level for nearly full condition.
* (the high watermark level are a power of 2
* starting from the top of the queue)
*
* Number of watermark
* ports level
* 1 0
* 2 1
* 3 2
* 4 4
* 5 4
* 6 8
* n approx. 2**ceil(log2(n))
*/
if (rxQueues[rxQueue].npeCount == 1)
{
qInfoDes->AlmostFullThreshold = IX_QMGR_Q_WM_LEVEL0;
}
else if (rxQueues[rxQueue].npeCount == 2)
{
qInfoDes->AlmostFullThreshold = IX_QMGR_Q_WM_LEVEL1;
}
else if (rxQueues[rxQueue].npeCount == 3)
{
qInfoDes->AlmostFullThreshold = IX_QMGR_Q_WM_LEVEL2;
}
else
{
/* reach the maximum number for CSR 2.0 */
IX_ETH_ACC_WARNING_LOG("ixEthAccQMgrQueuesConfig: maximum number of NPEs per queue reached, bailing out\n", 0, 0, 0, 0, 0, 0);
ret = IX_ETH_ACC_FAIL;
break;
}
/* move to next queue entry */
++qInfoDes;
}
/* configure the static list (RxFree, Tx and TxDone queues) */
for (qInfoDes = ixEthAccQmgrStaticInfo;
(qInfoDes->qCallback != (IxQMgrCallback) NULL )
&& (ret == IX_ETH_ACC_SUCCESS);
++qInfoDes)
{
ret = ixEthAccQMgrQueueSetup(qInfoDes);
}
/* configure the dynamic list (Rx queues) */
for (qInfoDes = ixEthAccQmgrRxQueuesInfo;
(qInfoDes->qCallback != (IxQMgrCallback) NULL )
&& (ret == IX_ETH_ACC_SUCCESS);
++qInfoDes)
{
ret = ixEthAccQMgrQueueSetup(qInfoDes);
}
return(ret);
}
/**
* @fn ixEthAccQMgrRxQEntryGet(UINT32 *rxQueueEntries)
*
* @brief Add and return the total number of entries in all Rx queues
*
* @param UINT32 rxQueueEntries[in] number of entries in all queues
*
* @return void
*
* @note Rx queues configuration is driven by Qos Setup. There is a
* variable number of rx queues which are set at initialisation.
*
* @internal
*/
IX_ETH_ACC_PUBLIC
void ixEthAccQMgrRxQEntryGet(UINT32 *numRxQueueEntries)
{
UINT32 rxQueueLevel;
IxEthAccQregInfo *qInfoDes;;
*numRxQueueEntries = 0;
/* iterate thru rx queues */
for (qInfoDes = ixEthAccQmgrRxQueuesInfo;
qInfoDes->qCallback != (IxQMgrCallback)NULL;
++qInfoDes)
{
/* retrieve the rx queue level */
rxQueueLevel = 0;
ixQMgrQNumEntriesGet(qInfoDes->qId, &rxQueueLevel);
(*numRxQueueEntries) += rxQueueLevel;
}
}
/**
* @fn ixEthAccQMgrRxCallbacksRegister(IxQMgrCallback ixQMgrCallback)
*
* @brief Change the callback registered to all rx queues.
*
* @param IxQMgrCallback ixQMgrCallback[in] QMgr callback to register
*
* @return IxEthAccStatus
*
* @note The user may decide to use different Rx mechanisms
* (e.g. receive many frames at the same time , or receive
* one frame at a time, depending on the overall application
* performances). A different QMgr callback is registered. This
* way, there is no excessive pointer checks in the datapath.
*
* @internal
*/
IX_ETH_ACC_PUBLIC
IxEthAccStatus ixEthAccQMgrRxCallbacksRegister(IxQMgrCallback ixQMgrCallback)
{
IxEthAccQregInfo *qInfoDes;
IxEthAccStatus ret = IX_ETH_ACC_SUCCESS;
/* parameter check */
if (NULL == ixQMgrCallback)
{
ret = IX_ETH_ACC_FAIL;
}
/* iterate thru rx queues */
for (qInfoDes = ixEthAccQmgrRxQueuesInfo;
(qInfoDes->qCallback != (IxQMgrCallback) NULL )
&& (ret == IX_ETH_ACC_SUCCESS);
++qInfoDes)
{
/* register the rx callback for all queues */
if (ixQMgrNotificationCallbackSet(qInfoDes->qId,
ixQMgrCallback,
qInfoDes->callbackTag
) != IX_SUCCESS)
{
ret = IX_ETH_ACC_FAIL;
}
}
return(ret);
}
/**
* @fn ixEthAccSingleEthNpeCheck(IxEthAccPortId portId)
*
* @brief Check the npe exists for this port
*
* @param IxEthAccPortId portId[in] port
*
* @return IxEthAccStatus
*
* @internal
*/
IX_ETH_ACC_PUBLIC
IxEthAccStatus ixEthAccSingleEthNpeCheck(IxEthAccPortId portId)
{
/* If not IXP42X A0 stepping, proceed to check for existence of coprocessors */
if ((IX_FEATURE_CTRL_SILICON_TYPE_A0 !=
(ixFeatureCtrlProductIdRead() & IX_FEATURE_CTRL_SILICON_STEPPING_MASK))
|| (IX_FEATURE_CTRL_DEVICE_TYPE_IXP42X != ixFeatureCtrlDeviceRead ()))
{
if ((IX_ETH_PORT_1 == portId) &&
(ixFeatureCtrlComponentCheck(IX_FEATURECTRL_ETH0) ==
IX_FEATURE_CTRL_COMPONENT_ENABLED))
{
return IX_ETH_ACC_SUCCESS;
}
if ((IX_ETH_PORT_2 == portId) &&
(ixFeatureCtrlComponentCheck(IX_FEATURECTRL_ETH1) ==
IX_FEATURE_CTRL_COMPONENT_ENABLED))
{
return IX_ETH_ACC_SUCCESS;
}
if ((IX_ETH_PORT_3 == portId) &&
(ixFeatureCtrlComponentCheck(IX_FEATURECTRL_NPEA_ETH) ==
IX_FEATURE_CTRL_COMPONENT_ENABLED))
{
return IX_ETH_ACC_SUCCESS;
}
return IX_ETH_ACC_FAIL;
}
return IX_ETH_ACC_SUCCESS;
}
/**
* @fn ixEthAccStatsShow(void)
*
* @brief Displays all EthAcc stats
*
* @return void
*
*/
void ixEthAccStatsShow(IxEthAccPortId portId)
{
ixEthAccMdioShow();
printf("\nPort %u\nUnicast MAC : ", portId);
ixEthAccPortUnicastAddressShow(portId);
ixEthAccPortMulticastAddressShow(portId);
printf("\n");
ixEthAccDataPlaneShow();
}