linux/drivers/scsi/ibmvscsi/ibmvscsi.c

1808 lines
53 KiB
C
Raw Normal View History

/* ------------------------------------------------------------
* ibmvscsi.c
* (C) Copyright IBM Corporation 1994, 2004
* Authors: Colin DeVilbiss (devilbis@us.ibm.com)
* Santiago Leon (santil@us.ibm.com)
* Dave Boutcher (sleddog@us.ibm.com)
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
* USA
*
* ------------------------------------------------------------
* Emulation of a SCSI host adapter for Virtual I/O devices
*
* This driver supports the SCSI adapter implemented by the IBM
* Power5 firmware. That SCSI adapter is not a physical adapter,
* but allows Linux SCSI peripheral drivers to directly
* access devices in another logical partition on the physical system.
*
* The virtual adapter(s) are present in the open firmware device
* tree just like real adapters.
*
* One of the capabilities provided on these systems is the ability
* to DMA between partitions. The architecture states that for VSCSI,
* the server side is allowed to DMA to and from the client. The client
* is never trusted to DMA to or from the server directly.
*
* Messages are sent between partitions on a "Command/Response Queue"
* (CRQ), which is just a buffer of 16 byte entries in the receiver's
* Senders cannot access the buffer directly, but send messages by
* making a hypervisor call and passing in the 16 bytes. The hypervisor
* puts the message in the next 16 byte space in round-robbin fashion,
* turns on the high order bit of the message (the valid bit), and
* generates an interrupt to the receiver (if interrupts are turned on.)
* The receiver just turns off the valid bit when they have copied out
* the message.
*
* The VSCSI client builds a SCSI Remote Protocol (SRP) Information Unit
* (IU) (as defined in the T10 standard available at www.t10.org), gets
* a DMA address for the message, and sends it to the server as the
* payload of a CRQ message. The server DMAs the SRP IU and processes it,
* including doing any additional data transfers. When it is done, it
* DMAs the SRP response back to the same address as the request came from,
* and sends a CRQ message back to inform the client that the request has
* completed.
*
* Note that some of the underlying infrastructure is different between
* machines conforming to the "RS/6000 Platform Architecture" (RPA) and
* the older iSeries hypervisor models. To support both, some low level
* routines have been broken out into rpa_vscsi.c and iseries_vscsi.c.
* The Makefile should pick one, not two, not zero, of these.
*
* TODO: This is currently pretty tied to the IBM i/pSeries hypervisor
* interfaces. It would be really nice to abstract this above an RDMA
* layer.
*/
#include <linux/module.h>
#include <linux/moduleparam.h>
#include <linux/dma-mapping.h>
#include <linux/delay.h>
#include <asm/firmware.h>
#include <asm/vio.h>
#include <asm/firmware.h>
#include <scsi/scsi.h>
#include <scsi/scsi_cmnd.h>
#include <scsi/scsi_host.h>
#include <scsi/scsi_device.h>
#include <scsi/scsi_transport_srp.h>
#include "ibmvscsi.h"
/* The values below are somewhat arbitrary default values, but
* OS/400 will use 3 busses (disks, CDs, tapes, I think.)
* Note that there are 3 bits of channel value, 6 bits of id, and
* 5 bits of LUN.
*/
static int max_id = 64;
static int max_channel = 3;
static int init_timeout = 5;
static int max_requests = IBMVSCSI_MAX_REQUESTS_DEFAULT;
static struct scsi_transport_template *ibmvscsi_transport_template;
#define IBMVSCSI_VERSION "1.5.8"
static struct ibmvscsi_ops *ibmvscsi_ops;
MODULE_DESCRIPTION("IBM Virtual SCSI");
MODULE_AUTHOR("Dave Boutcher");
MODULE_LICENSE("GPL");
MODULE_VERSION(IBMVSCSI_VERSION);
module_param_named(max_id, max_id, int, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(max_id, "Largest ID value for each channel");
module_param_named(max_channel, max_channel, int, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(max_channel, "Largest channel value");
module_param_named(init_timeout, init_timeout, int, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(init_timeout, "Initialization timeout in seconds");
module_param_named(max_requests, max_requests, int, S_IRUGO);
MODULE_PARM_DESC(max_requests, "Maximum requests for this adapter");
/* ------------------------------------------------------------
* Routines for the event pool and event structs
*/
/**
* initialize_event_pool: - Allocates and initializes the event pool for a host
* @pool: event_pool to be initialized
* @size: Number of events in pool
* @hostdata: ibmvscsi_host_data who owns the event pool
*
* Returns zero on success.
*/
static int initialize_event_pool(struct event_pool *pool,
int size, struct ibmvscsi_host_data *hostdata)
{
int i;
pool->size = size;
pool->next = 0;
pool->events = kcalloc(pool->size, sizeof(*pool->events), GFP_KERNEL);
if (!pool->events)
return -ENOMEM;
pool->iu_storage =
dma_alloc_coherent(hostdata->dev,
pool->size * sizeof(*pool->iu_storage),
&pool->iu_token, 0);
if (!pool->iu_storage) {
kfree(pool->events);
return -ENOMEM;
}
for (i = 0; i < pool->size; ++i) {
struct srp_event_struct *evt = &pool->events[i];
memset(&evt->crq, 0x00, sizeof(evt->crq));
atomic_set(&evt->free, 1);
evt->crq.valid = 0x80;
evt->crq.IU_length = sizeof(*evt->xfer_iu);
evt->crq.IU_data_ptr = pool->iu_token +
sizeof(*evt->xfer_iu) * i;
evt->xfer_iu = pool->iu_storage + i;
evt->hostdata = hostdata;
evt->ext_list = NULL;
evt->ext_list_token = 0;
}
return 0;
}
/**
* release_event_pool: - Frees memory of an event pool of a host
* @pool: event_pool to be released
* @hostdata: ibmvscsi_host_data who owns the even pool
*
* Returns zero on success.
*/
static void release_event_pool(struct event_pool *pool,
struct ibmvscsi_host_data *hostdata)
{
int i, in_use = 0;
for (i = 0; i < pool->size; ++i) {
if (atomic_read(&pool->events[i].free) != 1)
++in_use;
if (pool->events[i].ext_list) {
dma_free_coherent(hostdata->dev,
SG_ALL * sizeof(struct srp_direct_buf),
pool->events[i].ext_list,
pool->events[i].ext_list_token);
}
}
if (in_use)
dev_warn(hostdata->dev, "releasing event pool with %d "
"events still in use?\n", in_use);
kfree(pool->events);
dma_free_coherent(hostdata->dev,
pool->size * sizeof(*pool->iu_storage),
pool->iu_storage, pool->iu_token);
}
/**
* valid_event_struct: - Determines if event is valid.
* @pool: event_pool that contains the event
* @evt: srp_event_struct to be checked for validity
*
* Returns zero if event is invalid, one otherwise.
*/
static int valid_event_struct(struct event_pool *pool,
struct srp_event_struct *evt)
{
int index = evt - pool->events;
if (index < 0 || index >= pool->size) /* outside of bounds */
return 0;
if (evt != pool->events + index) /* unaligned */
return 0;
return 1;
}
/**
* ibmvscsi_free-event_struct: - Changes status of event to "free"
* @pool: event_pool that contains the event
* @evt: srp_event_struct to be modified
*
*/
static void free_event_struct(struct event_pool *pool,
struct srp_event_struct *evt)
{
if (!valid_event_struct(pool, evt)) {
dev_err(evt->hostdata->dev, "Freeing invalid event_struct %p "
"(not in pool %p)\n", evt, pool->events);
return;
}
if (atomic_inc_return(&evt->free) != 1) {
dev_err(evt->hostdata->dev, "Freeing event_struct %p "
"which is not in use!\n", evt);
return;
}
}
/**
* get_evt_struct: - Gets the next free event in pool
* @pool: event_pool that contains the events to be searched
*
* Returns the next event in "free" state, and NULL if none are free.
* Note that no synchronization is done here, we assume the host_lock
* will syncrhonze things.
*/
static struct srp_event_struct *get_event_struct(struct event_pool *pool)
{
int i;
int poolsize = pool->size;
int offset = pool->next;
for (i = 0; i < poolsize; i++) {
offset = (offset + 1) % poolsize;
if (!atomic_dec_if_positive(&pool->events[offset].free)) {
pool->next = offset;
return &pool->events[offset];
}
}
printk(KERN_ERR "ibmvscsi: found no event struct in pool!\n");
return NULL;
}
/**
* init_event_struct: Initialize fields in an event struct that are always
* required.
* @evt: The event
* @done: Routine to call when the event is responded to
* @format: SRP or MAD format
* @timeout: timeout value set in the CRQ
*/
static void init_event_struct(struct srp_event_struct *evt_struct,
void (*done) (struct srp_event_struct *),
u8 format,
int timeout)
{
evt_struct->cmnd = NULL;
evt_struct->cmnd_done = NULL;
evt_struct->sync_srp = NULL;
evt_struct->crq.format = format;
evt_struct->crq.timeout = timeout;
evt_struct->done = done;
}
/* ------------------------------------------------------------
* Routines for receiving SCSI responses from the hosting partition
*/
/**
* set_srp_direction: Set the fields in the srp related to data
* direction and number of buffers based on the direction in
* the scsi_cmnd and the number of buffers
*/
static void set_srp_direction(struct scsi_cmnd *cmd,
struct srp_cmd *srp_cmd,
int numbuf)
{
u8 fmt;
if (numbuf == 0)
return;
if (numbuf == 1)
fmt = SRP_DATA_DESC_DIRECT;
else {
fmt = SRP_DATA_DESC_INDIRECT;
numbuf = min(numbuf, MAX_INDIRECT_BUFS);
if (cmd->sc_data_direction == DMA_TO_DEVICE)
srp_cmd->data_out_desc_cnt = numbuf;
else
srp_cmd->data_in_desc_cnt = numbuf;
}
if (cmd->sc_data_direction == DMA_TO_DEVICE)
srp_cmd->buf_fmt = fmt << 4;
else
srp_cmd->buf_fmt = fmt;
}
static void unmap_sg_list(int num_entries,
struct device *dev,
struct srp_direct_buf *md)
{
int i;
for (i = 0; i < num_entries; ++i)
dma_unmap_single(dev, md[i].va, md[i].len, DMA_BIDIRECTIONAL);
}
/**
* unmap_cmd_data: - Unmap data pointed in srp_cmd based on the format
* @cmd: srp_cmd whose additional_data member will be unmapped
* @dev: device for which the memory is mapped
*
*/
static void unmap_cmd_data(struct srp_cmd *cmd,
struct srp_event_struct *evt_struct,
struct device *dev)
{
u8 out_fmt, in_fmt;
out_fmt = cmd->buf_fmt >> 4;
in_fmt = cmd->buf_fmt & ((1U << 4) - 1);
if (out_fmt == SRP_NO_DATA_DESC && in_fmt == SRP_NO_DATA_DESC)
return;
else if (out_fmt == SRP_DATA_DESC_DIRECT ||
in_fmt == SRP_DATA_DESC_DIRECT) {
struct srp_direct_buf *data =
(struct srp_direct_buf *) cmd->add_data;
dma_unmap_single(dev, data->va, data->len, DMA_BIDIRECTIONAL);
} else {
struct srp_indirect_buf *indirect =
(struct srp_indirect_buf *) cmd->add_data;
int num_mapped = indirect->table_desc.len /
sizeof(struct srp_direct_buf);
if (num_mapped <= MAX_INDIRECT_BUFS) {
unmap_sg_list(num_mapped, dev, &indirect->desc_list[0]);
return;
}
unmap_sg_list(num_mapped, dev, evt_struct->ext_list);
}
}
static int map_sg_list(struct scsi_cmnd *cmd, int nseg,
struct srp_direct_buf *md)
{
int i;
struct scatterlist *sg;
u64 total_length = 0;
scsi_for_each_sg(cmd, sg, nseg, i) {
struct srp_direct_buf *descr = md + i;
descr->va = sg_dma_address(sg);
descr->len = sg_dma_len(sg);
descr->key = 0;
total_length += sg_dma_len(sg);
}
return total_length;
}
/**
* map_sg_data: - Maps dma for a scatterlist and initializes decriptor fields
* @cmd: Scsi_Cmnd with the scatterlist
* @srp_cmd: srp_cmd that contains the memory descriptor
* @dev: device for which to map dma memory
*
* Called by map_data_for_srp_cmd() when building srp cmd from scsi cmd.
* Returns 1 on success.
*/
static int map_sg_data(struct scsi_cmnd *cmd,
struct srp_event_struct *evt_struct,
struct srp_cmd *srp_cmd, struct device *dev)
{
int sg_mapped;
u64 total_length = 0;
struct srp_direct_buf *data =
(struct srp_direct_buf *) srp_cmd->add_data;
struct srp_indirect_buf *indirect =
(struct srp_indirect_buf *) data;
sg_mapped = scsi_dma_map(cmd);
if (!sg_mapped)
return 1;
else if (sg_mapped < 0)
return 0;
set_srp_direction(cmd, srp_cmd, sg_mapped);
/* special case; we can use a single direct descriptor */
if (sg_mapped == 1) {
map_sg_list(cmd, sg_mapped, data);
return 1;
}
indirect->table_desc.va = 0;
indirect->table_desc.len = sg_mapped * sizeof(struct srp_direct_buf);
indirect->table_desc.key = 0;
if (sg_mapped <= MAX_INDIRECT_BUFS) {
total_length = map_sg_list(cmd, sg_mapped,
&indirect->desc_list[0]);
indirect->len = total_length;
return 1;
}
/* get indirect table */
if (!evt_struct->ext_list) {
evt_struct->ext_list = (struct srp_direct_buf *)
dma_alloc_coherent(dev,
SG_ALL * sizeof(struct srp_direct_buf),
&evt_struct->ext_list_token, 0);
if (!evt_struct->ext_list) {
if (!firmware_has_feature(FW_FEATURE_CMO))
sdev_printk(KERN_ERR, cmd->device,
"Can't allocate memory "
"for indirect table\n");
return 0;
}
}
total_length = map_sg_list(cmd, sg_mapped, evt_struct->ext_list);
indirect->len = total_length;
indirect->table_desc.va = evt_struct->ext_list_token;
indirect->table_desc.len = sg_mapped * sizeof(indirect->desc_list[0]);
memcpy(indirect->desc_list, evt_struct->ext_list,
MAX_INDIRECT_BUFS * sizeof(struct srp_direct_buf));
return 1;
}
/**
* map_data_for_srp_cmd: - Calls functions to map data for srp cmds
* @cmd: struct scsi_cmnd with the memory to be mapped
* @srp_cmd: srp_cmd that contains the memory descriptor
* @dev: dma device for which to map dma memory
*
* Called by scsi_cmd_to_srp_cmd() when converting scsi cmds to srp cmds
* Returns 1 on success.
*/
static int map_data_for_srp_cmd(struct scsi_cmnd *cmd,
struct srp_event_struct *evt_struct,
struct srp_cmd *srp_cmd, struct device *dev)
{
switch (cmd->sc_data_direction) {
case DMA_FROM_DEVICE:
case DMA_TO_DEVICE:
break;
case DMA_NONE:
return 1;
case DMA_BIDIRECTIONAL:
sdev_printk(KERN_ERR, cmd->device,
"Can't map DMA_BIDIRECTIONAL to read/write\n");
return 0;
default:
sdev_printk(KERN_ERR, cmd->device,
"Unknown data direction 0x%02x; can't map!\n",
cmd->sc_data_direction);
return 0;
}
return map_sg_data(cmd, evt_struct, srp_cmd, dev);
}
/**
* purge_requests: Our virtual adapter just shut down. purge any sent requests
* @hostdata: the adapter
*/
static void purge_requests(struct ibmvscsi_host_data *hostdata, int error_code)
{
struct srp_event_struct *tmp_evt, *pos;
unsigned long flags;
spin_lock_irqsave(hostdata->host->host_lock, flags);
list_for_each_entry_safe(tmp_evt, pos, &hostdata->sent, list) {
list_del(&tmp_evt->list);
del_timer(&tmp_evt->timer);
if (tmp_evt->cmnd) {
tmp_evt->cmnd->result = (error_code << 16);
unmap_cmd_data(&tmp_evt->iu.srp.cmd,
tmp_evt,
tmp_evt->hostdata->dev);
if (tmp_evt->cmnd_done)
tmp_evt->cmnd_done(tmp_evt->cmnd);
} else if (tmp_evt->done)
tmp_evt->done(tmp_evt);
free_event_struct(&tmp_evt->hostdata->pool, tmp_evt);
}
spin_unlock_irqrestore(hostdata->host->host_lock, flags);
}
/**
* ibmvscsi_reset_host - Reset the connection to the server
* @hostdata: struct ibmvscsi_host_data to reset
*/
static void ibmvscsi_reset_host(struct ibmvscsi_host_data *hostdata)
{
scsi_block_requests(hostdata->host);
atomic_set(&hostdata->request_limit, 0);
purge_requests(hostdata, DID_ERROR);
if ((ibmvscsi_ops->reset_crq_queue(&hostdata->queue, hostdata)) ||
(ibmvscsi_ops->send_crq(hostdata, 0xC001000000000000LL, 0)) ||
(vio_enable_interrupts(to_vio_dev(hostdata->dev)))) {
atomic_set(&hostdata->request_limit, -1);
dev_err(hostdata->dev, "error after reset\n");
}
scsi_unblock_requests(hostdata->host);
}
/**
* ibmvscsi_timeout - Internal command timeout handler
* @evt_struct: struct srp_event_struct that timed out
*
* Called when an internally generated command times out
*/
static void ibmvscsi_timeout(struct srp_event_struct *evt_struct)
{
struct ibmvscsi_host_data *hostdata = evt_struct->hostdata;
dev_err(hostdata->dev, "Command timed out (%x). Resetting connection\n",
evt_struct->iu.srp.cmd.opcode);
ibmvscsi_reset_host(hostdata);
}
/* ------------------------------------------------------------
* Routines for sending and receiving SRPs
*/
/**
* ibmvscsi_send_srp_event: - Transforms event to u64 array and calls send_crq()
* @evt_struct: evt_struct to be sent
* @hostdata: ibmvscsi_host_data of host
* @timeout: timeout in seconds - 0 means do not time command
*
* Returns the value returned from ibmvscsi_send_crq(). (Zero for success)
* Note that this routine assumes that host_lock is held for synchronization
*/
static int ibmvscsi_send_srp_event(struct srp_event_struct *evt_struct,
struct ibmvscsi_host_data *hostdata,
unsigned long timeout)
{
u64 *crq_as_u64 = (u64 *) &evt_struct->crq;
int request_status = 0;
int rc;
/* If we have exhausted our request limit, just fail this request,
* unless it is for a reset or abort.
* Note that there are rare cases involving driver generated requests
* (such as task management requests) that the mid layer may think we
* can handle more requests (can_queue) when we actually can't
*/
if (evt_struct->crq.format == VIOSRP_SRP_FORMAT) {
request_status =
atomic_dec_if_positive(&hostdata->request_limit);
/* If request limit was -1 when we started, it is now even
* less than that
*/
if (request_status < -1)
goto send_error;
/* Otherwise, we may have run out of requests. */
/* If request limit was 0 when we started the adapter is in the
* process of performing a login with the server adapter, or
* we may have run out of requests.
*/
else if (request_status == -1 &&
evt_struct->iu.srp.login_req.opcode != SRP_LOGIN_REQ)
goto send_busy;
/* Abort and reset calls should make it through.
* Nothing except abort and reset should use the last two
* slots unless we had two or less to begin with.
*/
else if (request_status < 2 &&
evt_struct->iu.srp.cmd.opcode != SRP_TSK_MGMT) {
/* In the case that we have less than two requests
* available, check the server limit as a combination
* of the request limit and the number of requests
* in-flight (the size of the send list). If the
* server limit is greater than 2, return busy so
* that the last two are reserved for reset and abort.
*/
int server_limit = request_status;
struct srp_event_struct *tmp_evt;
list_for_each_entry(tmp_evt, &hostdata->sent, list) {
server_limit++;
}
if (server_limit > 2)
goto send_busy;
}
}
/* Copy the IU into the transfer area */
*evt_struct->xfer_iu = evt_struct->iu;
evt_struct->xfer_iu->srp.rsp.tag = (u64)evt_struct;
/* Add this to the sent list. We need to do this
* before we actually send
* in case it comes back REALLY fast
*/
list_add_tail(&evt_struct->list, &hostdata->sent);
init_timer(&evt_struct->timer);
if (timeout) {
evt_struct->timer.data = (unsigned long) evt_struct;
evt_struct->timer.expires = jiffies + (timeout * HZ);
evt_struct->timer.function = (void (*)(unsigned long))ibmvscsi_timeout;
add_timer(&evt_struct->timer);
}
if ((rc =
ibmvscsi_ops->send_crq(hostdata, crq_as_u64[0], crq_as_u64[1])) != 0) {
list_del(&evt_struct->list);
del_timer(&evt_struct->timer);
/* If send_crq returns H_CLOSED, return SCSI_MLQUEUE_HOST_BUSY.
* Firmware will send a CRQ with a transport event (0xFF) to
* tell this client what has happened to the transport. This
* will be handled in ibmvscsi_handle_crq()
*/
if (rc == H_CLOSED) {
dev_warn(hostdata->dev, "send warning. "
"Receive queue closed, will retry.\n");
goto send_busy;
}
dev_err(hostdata->dev, "send error %d\n", rc);
atomic_inc(&hostdata->request_limit);
goto send_error;
}
return 0;
send_busy:
unmap_cmd_data(&evt_struct->iu.srp.cmd, evt_struct, hostdata->dev);
free_event_struct(&hostdata->pool, evt_struct);
if (request_status != -1)
atomic_inc(&hostdata->request_limit);
return SCSI_MLQUEUE_HOST_BUSY;
send_error:
unmap_cmd_data(&evt_struct->iu.srp.cmd, evt_struct, hostdata->dev);
if (evt_struct->cmnd != NULL) {
evt_struct->cmnd->result = DID_ERROR << 16;
evt_struct->cmnd_done(evt_struct->cmnd);
} else if (evt_struct->done)
evt_struct->done(evt_struct);
free_event_struct(&hostdata->pool, evt_struct);
return 0;
}
/**
* handle_cmd_rsp: - Handle responses from commands
* @evt_struct: srp_event_struct to be handled
*
* Used as a callback by when sending scsi cmds.
* Gets called by ibmvscsi_handle_crq()
*/
static void handle_cmd_rsp(struct srp_event_struct *evt_struct)
{
struct srp_rsp *rsp = &evt_struct->xfer_iu->srp.rsp;
struct scsi_cmnd *cmnd = evt_struct->cmnd;
if (unlikely(rsp->opcode != SRP_RSP)) {
if (printk_ratelimit())
dev_warn(evt_struct->hostdata->dev,
"bad SRP RSP type %d\n", rsp->opcode);
}
if (cmnd) {
cmnd->result |= rsp->status;
if (((cmnd->result >> 1) & 0x1f) == CHECK_CONDITION)
memcpy(cmnd->sense_buffer,
rsp->data,
rsp->sense_data_len);
unmap_cmd_data(&evt_struct->iu.srp.cmd,
evt_struct,
evt_struct->hostdata->dev);
if (rsp->flags & SRP_RSP_FLAG_DOOVER)
scsi_set_resid(cmnd, rsp->data_out_res_cnt);
else if (rsp->flags & SRP_RSP_FLAG_DIOVER)
scsi_set_resid(cmnd, rsp->data_in_res_cnt);
}
if (evt_struct->cmnd_done)
evt_struct->cmnd_done(cmnd);
}
/**
* lun_from_dev: - Returns the lun of the scsi device
* @dev: struct scsi_device
*
*/
static inline u16 lun_from_dev(struct scsi_device *dev)
{
return (0x2 << 14) | (dev->id << 8) | (dev->channel << 5) | dev->lun;
}
/**
* ibmvscsi_queue: - The queuecommand function of the scsi template
* @cmd: struct scsi_cmnd to be executed
* @done: Callback function to be called when cmd is completed
*/
static int ibmvscsi_queuecommand(struct scsi_cmnd *cmnd,
void (*done) (struct scsi_cmnd *))
{
struct srp_cmd *srp_cmd;
struct srp_event_struct *evt_struct;
struct srp_indirect_buf *indirect;
struct ibmvscsi_host_data *hostdata = shost_priv(cmnd->device->host);
u16 lun = lun_from_dev(cmnd->device);
u8 out_fmt, in_fmt;
cmnd->result = (DID_OK << 16);
evt_struct = get_event_struct(&hostdata->pool);
if (!evt_struct)
return SCSI_MLQUEUE_HOST_BUSY;
/* Set up the actual SRP IU */
srp_cmd = &evt_struct->iu.srp.cmd;
memset(srp_cmd, 0x00, SRP_MAX_IU_LEN);
srp_cmd->opcode = SRP_CMD;
[SCSI] Let scsi_cmnd->cmnd use request->cmd buffer - struct scsi_cmnd had a 16 bytes command buffer of its own. This is an unnecessary duplication and copy of request's cmd. It is probably left overs from the time that scsi_cmnd could function without a request attached. So clean that up. - Once above is done, few places, apart from scsi-ml, needed adjustments due to changing the data type of scsi_cmnd->cmnd. - Lots of drivers still use MAX_COMMAND_SIZE. So I have left that #define but equate it to BLK_MAX_CDB. The way I see it and is reflected in the patch below is. MAX_COMMAND_SIZE - means: The longest fixed-length (*) SCSI CDB as per the SCSI standard and is not related to the implementation. BLK_MAX_CDB. - The allocated space at the request level - I have audit all ISA drivers and made sure none use ->cmnd in a DMA Operation. Same audit was done by Andi Kleen. (*)fixed-length here means commands that their size can be determined by their opcode and the CDB does not carry a length specifier, (unlike the VARIABLE_LENGTH_CMD(0x7f) command). This is actually not exactly true and the SCSI standard also defines extended commands and vendor specific commands that can be bigger than 16 bytes. The kernel will support these using the same infrastructure used for VARLEN CDB's. So in effect MAX_COMMAND_SIZE means the maximum size command scsi-ml supports without specifying a cmd_len by ULD's Signed-off-by: Boaz Harrosh <bharrosh@panasas.com> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2008-04-30 08:19:47 +00:00
memcpy(srp_cmd->cdb, cmnd->cmnd, sizeof(srp_cmd->cdb));
srp_cmd->lun = ((u64) lun) << 48;
if (!map_data_for_srp_cmd(cmnd, evt_struct, srp_cmd, hostdata->dev)) {
if (!firmware_has_feature(FW_FEATURE_CMO))
sdev_printk(KERN_ERR, cmnd->device,
"couldn't convert cmd to srp_cmd\n");
free_event_struct(&hostdata->pool, evt_struct);
return SCSI_MLQUEUE_HOST_BUSY;
}
init_event_struct(evt_struct,
handle_cmd_rsp,
VIOSRP_SRP_FORMAT,
cmnd->request->timeout/HZ);
evt_struct->cmnd = cmnd;
evt_struct->cmnd_done = done;
/* Fix up dma address of the buffer itself */
indirect = (struct srp_indirect_buf *) srp_cmd->add_data;
out_fmt = srp_cmd->buf_fmt >> 4;
in_fmt = srp_cmd->buf_fmt & ((1U << 4) - 1);
if ((in_fmt == SRP_DATA_DESC_INDIRECT ||
out_fmt == SRP_DATA_DESC_INDIRECT) &&
indirect->table_desc.va == 0) {
indirect->table_desc.va = evt_struct->crq.IU_data_ptr +
offsetof(struct srp_cmd, add_data) +
offsetof(struct srp_indirect_buf, desc_list);
}
return ibmvscsi_send_srp_event(evt_struct, hostdata, 0);
}
/* ------------------------------------------------------------
* Routines for driver initialization
*/
/**
* adapter_info_rsp: - Handle response to MAD adapter info request
* @evt_struct: srp_event_struct with the response
*
* Used as a "done" callback by when sending adapter_info. Gets called
* by ibmvscsi_handle_crq()
*/
static void adapter_info_rsp(struct srp_event_struct *evt_struct)
{
struct ibmvscsi_host_data *hostdata = evt_struct->hostdata;
dma_unmap_single(hostdata->dev,
evt_struct->iu.mad.adapter_info.buffer,
evt_struct->iu.mad.adapter_info.common.length,
DMA_BIDIRECTIONAL);
if (evt_struct->xfer_iu->mad.adapter_info.common.status) {
dev_err(hostdata->dev, "error %d getting adapter info\n",
evt_struct->xfer_iu->mad.adapter_info.common.status);
} else {
dev_info(hostdata->dev, "host srp version: %s, "
"host partition %s (%d), OS %d, max io %u\n",
hostdata->madapter_info.srp_version,
hostdata->madapter_info.partition_name,
hostdata->madapter_info.partition_number,
hostdata->madapter_info.os_type,
hostdata->madapter_info.port_max_txu[0]);
if (hostdata->madapter_info.port_max_txu[0])
hostdata->host->max_sectors =
hostdata->madapter_info.port_max_txu[0] >> 9;
if (hostdata->madapter_info.os_type == 3 &&
strcmp(hostdata->madapter_info.srp_version, "1.6a") <= 0) {
dev_err(hostdata->dev, "host (Ver. %s) doesn't support large transfers\n",
hostdata->madapter_info.srp_version);
dev_err(hostdata->dev, "limiting scatterlists to %d\n",
MAX_INDIRECT_BUFS);
hostdata->host->sg_tablesize = MAX_INDIRECT_BUFS;
}
}
}
/**
* send_mad_adapter_info: - Sends the mad adapter info request
* and stores the result so it can be retrieved with
* sysfs. We COULD consider causing a failure if the
* returned SRP version doesn't match ours.
* @hostdata: ibmvscsi_host_data of host
*
* Returns zero if successful.
*/
static void send_mad_adapter_info(struct ibmvscsi_host_data *hostdata)
{
struct viosrp_adapter_info *req;
struct srp_event_struct *evt_struct;
unsigned long flags;
dma_addr_t addr;
evt_struct = get_event_struct(&hostdata->pool);
if (!evt_struct) {
dev_err(hostdata->dev,
"couldn't allocate an event for ADAPTER_INFO_REQ!\n");
return;
}
init_event_struct(evt_struct,
adapter_info_rsp,
VIOSRP_MAD_FORMAT,
init_timeout);
req = &evt_struct->iu.mad.adapter_info;
memset(req, 0x00, sizeof(*req));
req->common.type = VIOSRP_ADAPTER_INFO_TYPE;
req->common.length = sizeof(hostdata->madapter_info);
req->buffer = addr = dma_map_single(hostdata->dev,
&hostdata->madapter_info,
sizeof(hostdata->madapter_info),
DMA_BIDIRECTIONAL);
dma-mapping: add the device argument to dma_mapping_error() Add per-device dma_mapping_ops support for CONFIG_X86_64 as POWER architecture does: This enables us to cleanly fix the Calgary IOMMU issue that some devices are not behind the IOMMU (http://lkml.org/lkml/2008/5/8/423). I think that per-device dma_mapping_ops support would be also helpful for KVM people to support PCI passthrough but Andi thinks that this makes it difficult to support the PCI passthrough (see the above thread). So I CC'ed this to KVM camp. Comments are appreciated. A pointer to dma_mapping_ops to struct dev_archdata is added. If the pointer is non NULL, DMA operations in asm/dma-mapping.h use it. If it's NULL, the system-wide dma_ops pointer is used as before. If it's useful for KVM people, I plan to implement a mechanism to register a hook called when a new pci (or dma capable) device is created (it works with hot plugging). It enables IOMMUs to set up an appropriate dma_mapping_ops per device. The major obstacle is that dma_mapping_error doesn't take a pointer to the device unlike other DMA operations. So x86 can't have dma_mapping_ops per device. Note all the POWER IOMMUs use the same dma_mapping_error function so this is not a problem for POWER but x86 IOMMUs use different dma_mapping_error functions. The first patch adds the device argument to dma_mapping_error. The patch is trivial but large since it touches lots of drivers and dma-mapping.h in all the architecture. This patch: dma_mapping_error() doesn't take a pointer to the device unlike other DMA operations. So we can't have dma_mapping_ops per device. Note that POWER already has dma_mapping_ops per device but all the POWER IOMMUs use the same dma_mapping_error function. x86 IOMMUs use device argument. [akpm@linux-foundation.org: fix sge] [akpm@linux-foundation.org: fix svc_rdma] [akpm@linux-foundation.org: build fix] [akpm@linux-foundation.org: fix bnx2x] [akpm@linux-foundation.org: fix s2io] [akpm@linux-foundation.org: fix pasemi_mac] [akpm@linux-foundation.org: fix sdhci] [akpm@linux-foundation.org: build fix] [akpm@linux-foundation.org: fix sparc] [akpm@linux-foundation.org: fix ibmvscsi] Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: Muli Ben-Yehuda <muli@il.ibm.com> Cc: Andi Kleen <andi@firstfloor.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ingo Molnar <mingo@elte.hu> Cc: Avi Kivity <avi@qumranet.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-26 02:44:49 +00:00
if (dma_mapping_error(hostdata->dev, req->buffer)) {
if (!firmware_has_feature(FW_FEATURE_CMO))
dev_err(hostdata->dev,
"Unable to map request_buffer for "
"adapter_info!\n");
free_event_struct(&hostdata->pool, evt_struct);
return;
}
spin_lock_irqsave(hostdata->host->host_lock, flags);
if (ibmvscsi_send_srp_event(evt_struct, hostdata, init_timeout * 2)) {
dev_err(hostdata->dev, "couldn't send ADAPTER_INFO_REQ!\n");
dma_unmap_single(hostdata->dev,
addr,
sizeof(hostdata->madapter_info),
DMA_BIDIRECTIONAL);
}
spin_unlock_irqrestore(hostdata->host->host_lock, flags);
};
/**
* login_rsp: - Handle response to SRP login request
* @evt_struct: srp_event_struct with the response
*
* Used as a "done" callback by when sending srp_login. Gets called
* by ibmvscsi_handle_crq()
*/
static void login_rsp(struct srp_event_struct *evt_struct)
{
struct ibmvscsi_host_data *hostdata = evt_struct->hostdata;
switch (evt_struct->xfer_iu->srp.login_rsp.opcode) {
case SRP_LOGIN_RSP: /* it worked! */
break;
case SRP_LOGIN_REJ: /* refused! */
dev_info(hostdata->dev, "SRP_LOGIN_REJ reason %u\n",
evt_struct->xfer_iu->srp.login_rej.reason);
/* Login failed. */
atomic_set(&hostdata->request_limit, -1);
return;
default:
dev_err(hostdata->dev, "Invalid login response typecode 0x%02x!\n",
evt_struct->xfer_iu->srp.login_rsp.opcode);
/* Login failed. */
atomic_set(&hostdata->request_limit, -1);
return;
}
dev_info(hostdata->dev, "SRP_LOGIN succeeded\n");
if (evt_struct->xfer_iu->srp.login_rsp.req_lim_delta < 0)
dev_err(hostdata->dev, "Invalid request_limit.\n");
/* Now we know what the real request-limit is.
* This value is set rather than added to request_limit because
* request_limit could have been set to -1 by this client.
*/
atomic_set(&hostdata->request_limit,
evt_struct->xfer_iu->srp.login_rsp.req_lim_delta);
/* If we had any pending I/Os, kick them */
scsi_unblock_requests(hostdata->host);
send_mad_adapter_info(hostdata);
return;
}
/**
* send_srp_login: - Sends the srp login
* @hostdata: ibmvscsi_host_data of host
*
* Returns zero if successful.
*/
static int send_srp_login(struct ibmvscsi_host_data *hostdata)
{
int rc;
unsigned long flags;
struct srp_login_req *login;
struct srp_event_struct *evt_struct = get_event_struct(&hostdata->pool);
if (!evt_struct) {
dev_err(hostdata->dev, "couldn't allocate an event for login req!\n");
return FAILED;
}
init_event_struct(evt_struct,
login_rsp,
VIOSRP_SRP_FORMAT,
init_timeout);
login = &evt_struct->iu.srp.login_req;
memset(login, 0x00, sizeof(struct srp_login_req));
login->opcode = SRP_LOGIN_REQ;
login->req_it_iu_len = sizeof(union srp_iu);
login->req_buf_fmt = SRP_BUF_FORMAT_DIRECT | SRP_BUF_FORMAT_INDIRECT;
spin_lock_irqsave(hostdata->host->host_lock, flags);
/* Start out with a request limit of 0, since this is negotiated in
* the login request we are just sending and login requests always
* get sent by the driver regardless of request_limit.
*/
atomic_set(&hostdata->request_limit, 0);
rc = ibmvscsi_send_srp_event(evt_struct, hostdata, init_timeout * 2);
spin_unlock_irqrestore(hostdata->host->host_lock, flags);
dev_info(hostdata->dev, "sent SRP login\n");
return rc;
};
/**
* sync_completion: Signal that a synchronous command has completed
* Note that after returning from this call, the evt_struct is freed.
* the caller waiting on this completion shouldn't touch the evt_struct
* again.
*/
static void sync_completion(struct srp_event_struct *evt_struct)
{
/* copy the response back */
if (evt_struct->sync_srp)
*evt_struct->sync_srp = *evt_struct->xfer_iu;
complete(&evt_struct->comp);
}
/**
* ibmvscsi_abort: Abort a command...from scsi host template
* send this over to the server and wait synchronously for the response
*/
static int ibmvscsi_eh_abort_handler(struct scsi_cmnd *cmd)
{
struct ibmvscsi_host_data *hostdata = shost_priv(cmd->device->host);
struct srp_tsk_mgmt *tsk_mgmt;
struct srp_event_struct *evt;
struct srp_event_struct *tmp_evt, *found_evt;
union viosrp_iu srp_rsp;
int rsp_rc;
unsigned long flags;
u16 lun = lun_from_dev(cmd->device);
unsigned long wait_switch = 0;
/* First, find this command in our sent list so we can figure
* out the correct tag
*/
spin_lock_irqsave(hostdata->host->host_lock, flags);
wait_switch = jiffies + (init_timeout * HZ);
do {
found_evt = NULL;
list_for_each_entry(tmp_evt, &hostdata->sent, list) {
if (tmp_evt->cmnd == cmd) {
found_evt = tmp_evt;
break;
}
}
if (!found_evt) {
spin_unlock_irqrestore(hostdata->host->host_lock, flags);
return SUCCESS;
}
evt = get_event_struct(&hostdata->pool);
if (evt == NULL) {
spin_unlock_irqrestore(hostdata->host->host_lock, flags);
sdev_printk(KERN_ERR, cmd->device,
"failed to allocate abort event\n");
return FAILED;
}
init_event_struct(evt,
sync_completion,
VIOSRP_SRP_FORMAT,
init_timeout);
tsk_mgmt = &evt->iu.srp.tsk_mgmt;
/* Set up an abort SRP command */
memset(tsk_mgmt, 0x00, sizeof(*tsk_mgmt));
tsk_mgmt->opcode = SRP_TSK_MGMT;
tsk_mgmt->lun = ((u64) lun) << 48;
tsk_mgmt->tsk_mgmt_func = SRP_TSK_ABORT_TASK;
tsk_mgmt->task_tag = (u64) found_evt;
evt->sync_srp = &srp_rsp;
init_completion(&evt->comp);
rsp_rc = ibmvscsi_send_srp_event(evt, hostdata, init_timeout * 2);
if (rsp_rc != SCSI_MLQUEUE_HOST_BUSY)
break;
spin_unlock_irqrestore(hostdata->host->host_lock, flags);
msleep(10);
spin_lock_irqsave(hostdata->host->host_lock, flags);
} while (time_before(jiffies, wait_switch));
spin_unlock_irqrestore(hostdata->host->host_lock, flags);
if (rsp_rc != 0) {
sdev_printk(KERN_ERR, cmd->device,
"failed to send abort() event. rc=%d\n", rsp_rc);
return FAILED;
}
sdev_printk(KERN_INFO, cmd->device,
"aborting command. lun 0x%lx, tag 0x%lx\n",
(((u64) lun) << 48), (u64) found_evt);
wait_for_completion(&evt->comp);
/* make sure we got a good response */
if (unlikely(srp_rsp.srp.rsp.opcode != SRP_RSP)) {
if (printk_ratelimit())
sdev_printk(KERN_WARNING, cmd->device, "abort bad SRP RSP type %d\n",
srp_rsp.srp.rsp.opcode);
return FAILED;
}
if (srp_rsp.srp.rsp.flags & SRP_RSP_FLAG_RSPVALID)
rsp_rc = *((int *)srp_rsp.srp.rsp.data);
else
rsp_rc = srp_rsp.srp.rsp.status;
if (rsp_rc) {
if (printk_ratelimit())
sdev_printk(KERN_WARNING, cmd->device,
"abort code %d for task tag 0x%lx\n",
rsp_rc, tsk_mgmt->task_tag);
return FAILED;
}
/* Because we dropped the spinlock above, it's possible
* The event is no longer in our list. Make sure it didn't
* complete while we were aborting
*/
spin_lock_irqsave(hostdata->host->host_lock, flags);
found_evt = NULL;
list_for_each_entry(tmp_evt, &hostdata->sent, list) {
if (tmp_evt->cmnd == cmd) {
found_evt = tmp_evt;
break;
}
}
if (found_evt == NULL) {
spin_unlock_irqrestore(hostdata->host->host_lock, flags);
sdev_printk(KERN_INFO, cmd->device, "aborted task tag 0x%lx completed\n",
tsk_mgmt->task_tag);
return SUCCESS;
}
sdev_printk(KERN_INFO, cmd->device, "successfully aborted task tag 0x%lx\n",
tsk_mgmt->task_tag);
cmd->result = (DID_ABORT << 16);
list_del(&found_evt->list);
unmap_cmd_data(&found_evt->iu.srp.cmd, found_evt,
found_evt->hostdata->dev);
free_event_struct(&found_evt->hostdata->pool, found_evt);
spin_unlock_irqrestore(hostdata->host->host_lock, flags);
atomic_inc(&hostdata->request_limit);
return SUCCESS;
}
/**
* ibmvscsi_eh_device_reset_handler: Reset a single LUN...from scsi host
* template send this over to the server and wait synchronously for the
* response
*/
static int ibmvscsi_eh_device_reset_handler(struct scsi_cmnd *cmd)
{
struct ibmvscsi_host_data *hostdata = shost_priv(cmd->device->host);
struct srp_tsk_mgmt *tsk_mgmt;
struct srp_event_struct *evt;
struct srp_event_struct *tmp_evt, *pos;
union viosrp_iu srp_rsp;
int rsp_rc;
unsigned long flags;
u16 lun = lun_from_dev(cmd->device);
unsigned long wait_switch = 0;
spin_lock_irqsave(hostdata->host->host_lock, flags);
wait_switch = jiffies + (init_timeout * HZ);
do {
evt = get_event_struct(&hostdata->pool);
if (evt == NULL) {
spin_unlock_irqrestore(hostdata->host->host_lock, flags);
sdev_printk(KERN_ERR, cmd->device,
"failed to allocate reset event\n");
return FAILED;
}
init_event_struct(evt,
sync_completion,
VIOSRP_SRP_FORMAT,
init_timeout);
tsk_mgmt = &evt->iu.srp.tsk_mgmt;
/* Set up a lun reset SRP command */
memset(tsk_mgmt, 0x00, sizeof(*tsk_mgmt));
tsk_mgmt->opcode = SRP_TSK_MGMT;
tsk_mgmt->lun = ((u64) lun) << 48;
tsk_mgmt->tsk_mgmt_func = SRP_TSK_LUN_RESET;
evt->sync_srp = &srp_rsp;
init_completion(&evt->comp);
rsp_rc = ibmvscsi_send_srp_event(evt, hostdata, init_timeout * 2);
if (rsp_rc != SCSI_MLQUEUE_HOST_BUSY)
break;
spin_unlock_irqrestore(hostdata->host->host_lock, flags);
msleep(10);
spin_lock_irqsave(hostdata->host->host_lock, flags);
} while (time_before(jiffies, wait_switch));
spin_unlock_irqrestore(hostdata->host->host_lock, flags);
if (rsp_rc != 0) {
sdev_printk(KERN_ERR, cmd->device,
"failed to send reset event. rc=%d\n", rsp_rc);
return FAILED;
}
sdev_printk(KERN_INFO, cmd->device, "resetting device. lun 0x%lx\n",
(((u64) lun) << 48));
wait_for_completion(&evt->comp);
/* make sure we got a good response */
if (unlikely(srp_rsp.srp.rsp.opcode != SRP_RSP)) {
if (printk_ratelimit())
sdev_printk(KERN_WARNING, cmd->device, "reset bad SRP RSP type %d\n",
srp_rsp.srp.rsp.opcode);
return FAILED;
}
if (srp_rsp.srp.rsp.flags & SRP_RSP_FLAG_RSPVALID)
rsp_rc = *((int *)srp_rsp.srp.rsp.data);
else
rsp_rc = srp_rsp.srp.rsp.status;
if (rsp_rc) {
if (printk_ratelimit())
sdev_printk(KERN_WARNING, cmd->device,
"reset code %d for task tag 0x%lx\n",
rsp_rc, tsk_mgmt->task_tag);
return FAILED;
}
/* We need to find all commands for this LUN that have not yet been
* responded to, and fail them with DID_RESET
*/
spin_lock_irqsave(hostdata->host->host_lock, flags);
list_for_each_entry_safe(tmp_evt, pos, &hostdata->sent, list) {
if ((tmp_evt->cmnd) && (tmp_evt->cmnd->device == cmd->device)) {
if (tmp_evt->cmnd)
tmp_evt->cmnd->result = (DID_RESET << 16);
list_del(&tmp_evt->list);
unmap_cmd_data(&tmp_evt->iu.srp.cmd, tmp_evt,
tmp_evt->hostdata->dev);
free_event_struct(&tmp_evt->hostdata->pool,
tmp_evt);
atomic_inc(&hostdata->request_limit);
if (tmp_evt->cmnd_done)
tmp_evt->cmnd_done(tmp_evt->cmnd);
else if (tmp_evt->done)
tmp_evt->done(tmp_evt);
}
}
spin_unlock_irqrestore(hostdata->host->host_lock, flags);
return SUCCESS;
}
/**
* ibmvscsi_eh_host_reset_handler - Reset the connection to the server
* @cmd: struct scsi_cmnd having problems
*/
static int ibmvscsi_eh_host_reset_handler(struct scsi_cmnd *cmd)
{
unsigned long wait_switch = 0;
struct ibmvscsi_host_data *hostdata = shost_priv(cmd->device->host);
dev_err(hostdata->dev, "Resetting connection due to error recovery\n");
ibmvscsi_reset_host(hostdata);
for (wait_switch = jiffies + (init_timeout * HZ);
time_before(jiffies, wait_switch) &&
atomic_read(&hostdata->request_limit) < 2;) {
msleep(10);
}
if (atomic_read(&hostdata->request_limit) <= 0)
return FAILED;
return SUCCESS;
}
/**
* ibmvscsi_handle_crq: - Handles and frees received events in the CRQ
* @crq: Command/Response queue
* @hostdata: ibmvscsi_host_data of host
*
*/
void ibmvscsi_handle_crq(struct viosrp_crq *crq,
struct ibmvscsi_host_data *hostdata)
{
long rc;
unsigned long flags;
struct srp_event_struct *evt_struct =
(struct srp_event_struct *)crq->IU_data_ptr;
switch (crq->valid) {
case 0xC0: /* initialization */
switch (crq->format) {
case 0x01: /* Initialization message */
dev_info(hostdata->dev, "partner initialized\n");
/* Send back a response */
if ((rc = ibmvscsi_ops->send_crq(hostdata,
0xC002000000000000LL, 0)) == 0) {
/* Now login */
send_srp_login(hostdata);
} else {
dev_err(hostdata->dev, "Unable to send init rsp. rc=%ld\n", rc);
}
break;
case 0x02: /* Initialization response */
dev_info(hostdata->dev, "partner initialization complete\n");
/* Now login */
send_srp_login(hostdata);
break;
default:
dev_err(hostdata->dev, "unknown crq message type: %d\n", crq->format);
}
return;
case 0xFF: /* Hypervisor telling us the connection is closed */
scsi_block_requests(hostdata->host);
atomic_set(&hostdata->request_limit, 0);
if (crq->format == 0x06) {
/* We need to re-setup the interpartition connection */
dev_info(hostdata->dev, "Re-enabling adapter!\n");
purge_requests(hostdata, DID_REQUEUE);
if ((ibmvscsi_ops->reenable_crq_queue(&hostdata->queue,
hostdata)) ||
(ibmvscsi_ops->send_crq(hostdata,
0xC001000000000000LL, 0))) {
atomic_set(&hostdata->request_limit,
-1);
dev_err(hostdata->dev, "error after enable\n");
}
} else {
dev_err(hostdata->dev, "Virtual adapter failed rc %d!\n",
crq->format);
purge_requests(hostdata, DID_ERROR);
if ((ibmvscsi_ops->reset_crq_queue(&hostdata->queue,
hostdata)) ||
(ibmvscsi_ops->send_crq(hostdata,
0xC001000000000000LL, 0))) {
atomic_set(&hostdata->request_limit,
-1);
dev_err(hostdata->dev, "error after reset\n");
}
}
scsi_unblock_requests(hostdata->host);
return;
case 0x80: /* real payload */
break;
default:
dev_err(hostdata->dev, "got an invalid message type 0x%02x\n",
crq->valid);
return;
}
/* The only kind of payload CRQs we should get are responses to
* things we send. Make sure this response is to something we
* actually sent
*/
if (!valid_event_struct(&hostdata->pool, evt_struct)) {
dev_err(hostdata->dev, "returned correlation_token 0x%p is invalid!\n",
(void *)crq->IU_data_ptr);
return;
}
if (atomic_read(&evt_struct->free)) {
dev_err(hostdata->dev, "received duplicate correlation_token 0x%p!\n",
(void *)crq->IU_data_ptr);
return;
}
if (crq->format == VIOSRP_SRP_FORMAT)
atomic_add(evt_struct->xfer_iu->srp.rsp.req_lim_delta,
&hostdata->request_limit);
del_timer(&evt_struct->timer);
if ((crq->status != VIOSRP_OK && crq->status != VIOSRP_OK2) && evt_struct->cmnd)
evt_struct->cmnd->result = DID_ERROR << 16;
if (evt_struct->done)
evt_struct->done(evt_struct);
else
dev_err(hostdata->dev, "returned done() is NULL; not running it!\n");
/*
* Lock the host_lock before messing with these structures, since we
* are running in a task context
*/
spin_lock_irqsave(evt_struct->hostdata->host->host_lock, flags);
list_del(&evt_struct->list);
free_event_struct(&evt_struct->hostdata->pool, evt_struct);
spin_unlock_irqrestore(evt_struct->hostdata->host->host_lock, flags);
}
/**
* ibmvscsi_get_host_config: Send the command to the server to get host
* configuration data. The data is opaque to us.
*/
static int ibmvscsi_do_host_config(struct ibmvscsi_host_data *hostdata,
unsigned char *buffer, int length)
{
struct viosrp_host_config *host_config;
struct srp_event_struct *evt_struct;
unsigned long flags;
dma_addr_t addr;
int rc;
evt_struct = get_event_struct(&hostdata->pool);
if (!evt_struct) {
dev_err(hostdata->dev, "couldn't allocate event for HOST_CONFIG!\n");
return -1;
}
init_event_struct(evt_struct,
sync_completion,
VIOSRP_MAD_FORMAT,
init_timeout);
host_config = &evt_struct->iu.mad.host_config;
/* Set up a lun reset SRP command */
memset(host_config, 0x00, sizeof(*host_config));
host_config->common.type = VIOSRP_HOST_CONFIG_TYPE;
host_config->common.length = length;
host_config->buffer = addr = dma_map_single(hostdata->dev, buffer,
length,
DMA_BIDIRECTIONAL);
dma-mapping: add the device argument to dma_mapping_error() Add per-device dma_mapping_ops support for CONFIG_X86_64 as POWER architecture does: This enables us to cleanly fix the Calgary IOMMU issue that some devices are not behind the IOMMU (http://lkml.org/lkml/2008/5/8/423). I think that per-device dma_mapping_ops support would be also helpful for KVM people to support PCI passthrough but Andi thinks that this makes it difficult to support the PCI passthrough (see the above thread). So I CC'ed this to KVM camp. Comments are appreciated. A pointer to dma_mapping_ops to struct dev_archdata is added. If the pointer is non NULL, DMA operations in asm/dma-mapping.h use it. If it's NULL, the system-wide dma_ops pointer is used as before. If it's useful for KVM people, I plan to implement a mechanism to register a hook called when a new pci (or dma capable) device is created (it works with hot plugging). It enables IOMMUs to set up an appropriate dma_mapping_ops per device. The major obstacle is that dma_mapping_error doesn't take a pointer to the device unlike other DMA operations. So x86 can't have dma_mapping_ops per device. Note all the POWER IOMMUs use the same dma_mapping_error function so this is not a problem for POWER but x86 IOMMUs use different dma_mapping_error functions. The first patch adds the device argument to dma_mapping_error. The patch is trivial but large since it touches lots of drivers and dma-mapping.h in all the architecture. This patch: dma_mapping_error() doesn't take a pointer to the device unlike other DMA operations. So we can't have dma_mapping_ops per device. Note that POWER already has dma_mapping_ops per device but all the POWER IOMMUs use the same dma_mapping_error function. x86 IOMMUs use device argument. [akpm@linux-foundation.org: fix sge] [akpm@linux-foundation.org: fix svc_rdma] [akpm@linux-foundation.org: build fix] [akpm@linux-foundation.org: fix bnx2x] [akpm@linux-foundation.org: fix s2io] [akpm@linux-foundation.org: fix pasemi_mac] [akpm@linux-foundation.org: fix sdhci] [akpm@linux-foundation.org: build fix] [akpm@linux-foundation.org: fix sparc] [akpm@linux-foundation.org: fix ibmvscsi] Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: Muli Ben-Yehuda <muli@il.ibm.com> Cc: Andi Kleen <andi@firstfloor.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ingo Molnar <mingo@elte.hu> Cc: Avi Kivity <avi@qumranet.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-26 02:44:49 +00:00
if (dma_mapping_error(hostdata->dev, host_config->buffer)) {
if (!firmware_has_feature(FW_FEATURE_CMO))
dev_err(hostdata->dev,
"dma_mapping error getting host config\n");
free_event_struct(&hostdata->pool, evt_struct);
return -1;
}
init_completion(&evt_struct->comp);
spin_lock_irqsave(hostdata->host->host_lock, flags);
rc = ibmvscsi_send_srp_event(evt_struct, hostdata, init_timeout * 2);
spin_unlock_irqrestore(hostdata->host->host_lock, flags);
if (rc == 0)
wait_for_completion(&evt_struct->comp);
dma_unmap_single(hostdata->dev, addr, length, DMA_BIDIRECTIONAL);
return rc;
}
/**
* ibmvscsi_slave_configure: Set the "allow_restart" flag for each disk.
* @sdev: struct scsi_device device to configure
*
* Enable allow_restart for a device if it is a disk. Adjust the
* queue_depth here also as is required by the documentation for
* struct scsi_host_template.
*/
static int ibmvscsi_slave_configure(struct scsi_device *sdev)
{
struct Scsi_Host *shost = sdev->host;
unsigned long lock_flags = 0;
spin_lock_irqsave(shost->host_lock, lock_flags);
if (sdev->type == TYPE_DISK) {
sdev->allow_restart = 1;
blk_queue_rq_timeout(sdev->request_queue, 60 * HZ);
}
scsi_adjust_queue_depth(sdev, 0, shost->cmd_per_lun);
spin_unlock_irqrestore(shost->host_lock, lock_flags);
return 0;
}
/**
* ibmvscsi_change_queue_depth - Change the device's queue depth
* @sdev: scsi device struct
* @qdepth: depth to set
*
* Return value:
* actual depth set
**/
static int ibmvscsi_change_queue_depth(struct scsi_device *sdev, int qdepth)
{
if (qdepth > IBMVSCSI_MAX_CMDS_PER_LUN)
qdepth = IBMVSCSI_MAX_CMDS_PER_LUN;
scsi_adjust_queue_depth(sdev, 0, qdepth);
return sdev->queue_depth;
}
/* ------------------------------------------------------------
* sysfs attributes
*/
static ssize_t show_host_srp_version(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct Scsi_Host *shost = class_to_shost(dev);
struct ibmvscsi_host_data *hostdata = shost_priv(shost);
int len;
len = snprintf(buf, PAGE_SIZE, "%s\n",
hostdata->madapter_info.srp_version);
return len;
}
static struct device_attribute ibmvscsi_host_srp_version = {
.attr = {
.name = "srp_version",
.mode = S_IRUGO,
},
.show = show_host_srp_version,
};
static ssize_t show_host_partition_name(struct device *dev,
struct device_attribute *attr,
char *buf)
{
struct Scsi_Host *shost = class_to_shost(dev);
struct ibmvscsi_host_data *hostdata = shost_priv(shost);
int len;
len = snprintf(buf, PAGE_SIZE, "%s\n",
hostdata->madapter_info.partition_name);
return len;
}
static struct device_attribute ibmvscsi_host_partition_name = {
.attr = {
.name = "partition_name",
.mode = S_IRUGO,
},
.show = show_host_partition_name,
};
static ssize_t show_host_partition_number(struct device *dev,
struct device_attribute *attr,
char *buf)
{
struct Scsi_Host *shost = class_to_shost(dev);
struct ibmvscsi_host_data *hostdata = shost_priv(shost);
int len;
len = snprintf(buf, PAGE_SIZE, "%d\n",
hostdata->madapter_info.partition_number);
return len;
}
static struct device_attribute ibmvscsi_host_partition_number = {
.attr = {
.name = "partition_number",
.mode = S_IRUGO,
},
.show = show_host_partition_number,
};
static ssize_t show_host_mad_version(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct Scsi_Host *shost = class_to_shost(dev);
struct ibmvscsi_host_data *hostdata = shost_priv(shost);
int len;
len = snprintf(buf, PAGE_SIZE, "%d\n",
hostdata->madapter_info.mad_version);
return len;
}
static struct device_attribute ibmvscsi_host_mad_version = {
.attr = {
.name = "mad_version",
.mode = S_IRUGO,
},
.show = show_host_mad_version,
};
static ssize_t show_host_os_type(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct Scsi_Host *shost = class_to_shost(dev);
struct ibmvscsi_host_data *hostdata = shost_priv(shost);
int len;
len = snprintf(buf, PAGE_SIZE, "%d\n", hostdata->madapter_info.os_type);
return len;
}
static struct device_attribute ibmvscsi_host_os_type = {
.attr = {
.name = "os_type",
.mode = S_IRUGO,
},
.show = show_host_os_type,
};
static ssize_t show_host_config(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct Scsi_Host *shost = class_to_shost(dev);
struct ibmvscsi_host_data *hostdata = shost_priv(shost);
/* returns null-terminated host config data */
if (ibmvscsi_do_host_config(hostdata, buf, PAGE_SIZE) == 0)
return strlen(buf);
else
return 0;
}
static struct device_attribute ibmvscsi_host_config = {
.attr = {
.name = "config",
.mode = S_IRUGO,
},
.show = show_host_config,
};
static struct device_attribute *ibmvscsi_attrs[] = {
&ibmvscsi_host_srp_version,
&ibmvscsi_host_partition_name,
&ibmvscsi_host_partition_number,
&ibmvscsi_host_mad_version,
&ibmvscsi_host_os_type,
&ibmvscsi_host_config,
NULL
};
/* ------------------------------------------------------------
* SCSI driver registration
*/
static struct scsi_host_template driver_template = {
.module = THIS_MODULE,
.name = "IBM POWER Virtual SCSI Adapter " IBMVSCSI_VERSION,
.proc_name = "ibmvscsi",
.queuecommand = ibmvscsi_queuecommand,
.eh_abort_handler = ibmvscsi_eh_abort_handler,
.eh_device_reset_handler = ibmvscsi_eh_device_reset_handler,
.eh_host_reset_handler = ibmvscsi_eh_host_reset_handler,
.slave_configure = ibmvscsi_slave_configure,
.change_queue_depth = ibmvscsi_change_queue_depth,
.cmd_per_lun = IBMVSCSI_CMDS_PER_LUN_DEFAULT,
.can_queue = IBMVSCSI_MAX_REQUESTS_DEFAULT,
.this_id = -1,
.sg_tablesize = SG_ALL,
.use_clustering = ENABLE_CLUSTERING,
.shost_attrs = ibmvscsi_attrs,
};
/**
* ibmvscsi_get_desired_dma - Calculate IO memory desired by the driver
*
* @vdev: struct vio_dev for the device whose desired IO mem is to be returned
*
* Return value:
* Number of bytes of IO data the driver will need to perform well.
*/
static unsigned long ibmvscsi_get_desired_dma(struct vio_dev *vdev)
{
/* iu_storage data allocated in initialize_event_pool */
unsigned long desired_io = max_requests * sizeof(union viosrp_iu);
/* add io space for sg data */
desired_io += (IBMVSCSI_MAX_SECTORS_DEFAULT * 512 *
IBMVSCSI_CMDS_PER_LUN_DEFAULT);
return desired_io;
}
/**
* Called by bus code for each adapter
*/
static int ibmvscsi_probe(struct vio_dev *vdev, const struct vio_device_id *id)
{
struct ibmvscsi_host_data *hostdata;
struct Scsi_Host *host;
struct device *dev = &vdev->dev;
struct srp_rport_identifiers ids;
struct srp_rport *rport;
unsigned long wait_switch = 0;
int rc;
vdev->dev.driver_data = NULL;
driver_template.can_queue = max_requests - 2;
host = scsi_host_alloc(&driver_template, sizeof(*hostdata));
if (!host) {
dev_err(&vdev->dev, "couldn't allocate host data\n");
goto scsi_host_alloc_failed;
}
host->transportt = ibmvscsi_transport_template;
hostdata = shost_priv(host);
memset(hostdata, 0x00, sizeof(*hostdata));
INIT_LIST_HEAD(&hostdata->sent);
hostdata->host = host;
hostdata->dev = dev;
atomic_set(&hostdata->request_limit, -1);
hostdata->host->max_sectors = IBMVSCSI_MAX_SECTORS_DEFAULT;
rc = ibmvscsi_ops->init_crq_queue(&hostdata->queue, hostdata, max_requests);
if (rc != 0 && rc != H_RESOURCE) {
dev_err(&vdev->dev, "couldn't initialize crq. rc=%d\n", rc);
goto init_crq_failed;
}
if (initialize_event_pool(&hostdata->pool, max_requests, hostdata) != 0) {
dev_err(&vdev->dev, "couldn't initialize event pool\n");
goto init_pool_failed;
}
host->max_lun = 8;
host->max_id = max_id;
host->max_channel = max_channel;
if (scsi_add_host(hostdata->host, hostdata->dev))
goto add_host_failed;
/* we don't have a proper target_port_id so let's use the fake one */
memcpy(ids.port_id, hostdata->madapter_info.partition_name,
sizeof(ids.port_id));
ids.roles = SRP_RPORT_ROLE_TARGET;
rport = srp_rport_add(host, &ids);
if (IS_ERR(rport))
goto add_srp_port_failed;
/* Try to send an initialization message. Note that this is allowed
* to fail if the other end is not acive. In that case we don't
* want to scan
*/
if (ibmvscsi_ops->send_crq(hostdata, 0xC001000000000000LL, 0) == 0
|| rc == H_RESOURCE) {
/*
* Wait around max init_timeout secs for the adapter to finish
* initializing. When we are done initializing, we will have a
* valid request_limit. We don't want Linux scanning before
* we are ready.
*/
for (wait_switch = jiffies + (init_timeout * HZ);
time_before(jiffies, wait_switch) &&
atomic_read(&hostdata->request_limit) < 2;) {
msleep(10);
}
/* if we now have a valid request_limit, initiate a scan */
if (atomic_read(&hostdata->request_limit) > 0)
scsi_scan_host(host);
}
vdev->dev.driver_data = hostdata;
return 0;
add_srp_port_failed:
scsi_remove_host(hostdata->host);
add_host_failed:
release_event_pool(&hostdata->pool, hostdata);
init_pool_failed:
ibmvscsi_ops->release_crq_queue(&hostdata->queue, hostdata, max_requests);
init_crq_failed:
scsi_host_put(host);
scsi_host_alloc_failed:
return -1;
}
static int ibmvscsi_remove(struct vio_dev *vdev)
{
struct ibmvscsi_host_data *hostdata = vdev->dev.driver_data;
release_event_pool(&hostdata->pool, hostdata);
ibmvscsi_ops->release_crq_queue(&hostdata->queue, hostdata,
max_requests);
srp_remove_host(hostdata->host);
scsi_remove_host(hostdata->host);
scsi_host_put(hostdata->host);
return 0;
}
/**
* ibmvscsi_device_table: Used by vio.c to match devices in the device tree we
* support.
*/
static struct vio_device_id ibmvscsi_device_table[] __devinitdata = {
{"vscsi", "IBM,v-scsi"},
{ "", "" }
};
MODULE_DEVICE_TABLE(vio, ibmvscsi_device_table);
static struct vio_driver ibmvscsi_driver = {
.id_table = ibmvscsi_device_table,
.probe = ibmvscsi_probe,
.remove = ibmvscsi_remove,
.get_desired_dma = ibmvscsi_get_desired_dma,
.driver = {
.name = "ibmvscsi",
.owner = THIS_MODULE,
}
};
static struct srp_function_template ibmvscsi_transport_functions = {
};
int __init ibmvscsi_module_init(void)
{
int ret;
if (firmware_has_feature(FW_FEATURE_ISERIES))
ibmvscsi_ops = &iseriesvscsi_ops;
else if (firmware_has_feature(FW_FEATURE_VIO))
ibmvscsi_ops = &rpavscsi_ops;
else
return -ENODEV;
ibmvscsi_transport_template =
srp_attach_transport(&ibmvscsi_transport_functions);
if (!ibmvscsi_transport_template)
return -ENOMEM;
ret = vio_register_driver(&ibmvscsi_driver);
if (ret)
srp_release_transport(ibmvscsi_transport_template);
return ret;
}
void __exit ibmvscsi_module_exit(void)
{
vio_unregister_driver(&ibmvscsi_driver);
srp_release_transport(ibmvscsi_transport_template);
}
module_init(ibmvscsi_module_init);
module_exit(ibmvscsi_module_exit);