diff options
Diffstat (limited to 'drivers/firewire')
| -rw-r--r-- | drivers/firewire/fw-cdev.c | 3 | ||||
| -rw-r--r-- | drivers/firewire/fw-device.c | 38 | ||||
| -rw-r--r-- | drivers/firewire/fw-device.h | 12 | ||||
| -rw-r--r-- | drivers/firewire/fw-ohci.c | 498 | ||||
| -rw-r--r-- | drivers/firewire/fw-sbp2.c | 130 | ||||
| -rw-r--r-- | drivers/firewire/fw-topology.c | 6 | ||||
| -rw-r--r-- | drivers/firewire/fw-transaction.c | 4 |
7 files changed, 506 insertions, 185 deletions
diff --git a/drivers/firewire/fw-cdev.c b/drivers/firewire/fw-cdev.c index 60f1a89..7e73cba 100644 --- a/drivers/firewire/fw-cdev.c +++ b/drivers/firewire/fw-cdev.c @@ -206,12 +206,13 @@ fill_bus_reset_event(struct fw_cdev_event_bus_reset *event, event->closure = client->bus_reset_closure; event->type = FW_CDEV_EVENT_BUS_RESET; + event->generation = client->device->generation; + smp_rmb(); /* node_id must not be older than generation */ event->node_id = client->device->node_id; event->local_node_id = card->local_node->node_id; event->bm_node_id = 0; /* FIXME: We don't track the BM. */ event->irm_node_id = card->irm_node->node_id; event->root_node_id = card->root_node->node_id; - event->generation = card->generation; } static void diff --git a/drivers/firewire/fw-device.c b/drivers/firewire/fw-device.c index 56681b3..de9066e 100644 --- a/drivers/firewire/fw-device.c +++ b/drivers/firewire/fw-device.c @@ -27,6 +27,7 @@ #include <linux/idr.h> #include <linux/rwsem.h> #include <asm/semaphore.h> +#include <asm/system.h> #include <linux/ctype.h> #include "fw-transaction.h" #include "fw-topology.h" @@ -182,9 +183,14 @@ static void fw_device_release(struct device *dev) int fw_device_enable_phys_dma(struct fw_device *device) { + int generation = device->generation; + + /* device->node_id, accessed below, must not be older than generation */ + smp_rmb(); + return device->card->driver->enable_phys_dma(device->card, device->node_id, - device->generation); + generation); } EXPORT_SYMBOL(fw_device_enable_phys_dma); @@ -384,17 +390,21 @@ complete_transaction(struct fw_card *card, int rcode, complete(&callback_data->done); } -static int read_rom(struct fw_device *device, int index, u32 * data) +static int +read_rom(struct fw_device *device, int generation, int index, u32 *data) { struct read_quadlet_callback_data callback_data; struct fw_transaction t; u64 offset; + /* device->node_id, accessed below, must not be older than generation */ + smp_rmb(); + init_completion(&callback_data.done); offset = 0xfffff0000400ULL + index * 4; fw_send_request(device->card, &t, TCODE_READ_QUADLET_REQUEST, - device->node_id, device->generation, device->max_speed, + device->node_id, generation, device->max_speed, offset, NULL, 4, complete_transaction, &callback_data); wait_for_completion(&callback_data.done); @@ -404,7 +414,14 @@ static int read_rom(struct fw_device *device, int index, u32 * data) return callback_data.rcode; } -static int read_bus_info_block(struct fw_device *device) +/* + * Read the bus info block, perform a speed probe, and read all of the rest of + * the config ROM. We do all this with a cached bus generation. If the bus + * generation changes under us, read_bus_info_block will fail and get retried. + * It's better to start all over in this case because the node from which we + * are reading the ROM may have changed the ROM during the reset. + */ +static int read_bus_info_block(struct fw_device *device, int generation) { static u32 rom[256]; u32 stack[16], sp, key; @@ -414,7 +431,7 @@ static int read_bus_info_block(struct fw_device *device) /* First read the bus info block. */ for (i = 0; i < 5; i++) { - if (read_rom(device, i, &rom[i]) != RCODE_COMPLETE) + if (read_rom(device, generation, i, &rom[i]) != RCODE_COMPLETE) return -1; /* * As per IEEE1212 7.2, during power-up, devices can @@ -449,7 +466,8 @@ static int read_bus_info_block(struct fw_device *device) device->max_speed = device->card->link_speed; while (device->max_speed > SCODE_100) { - if (read_rom(device, 0, &dummy) == RCODE_COMPLETE) + if (read_rom(device, generation, 0, &dummy) == + RCODE_COMPLETE) break; device->max_speed--; } @@ -482,7 +500,7 @@ static int read_bus_info_block(struct fw_device *device) return -1; /* Read header quadlet for the block to get the length. */ - if (read_rom(device, i, &rom[i]) != RCODE_COMPLETE) + if (read_rom(device, generation, i, &rom[i]) != RCODE_COMPLETE) return -1; end = i + (rom[i] >> 16) + 1; i++; @@ -501,7 +519,8 @@ static int read_bus_info_block(struct fw_device *device) * it references another block, and push it in that case. */ while (i < end) { - if (read_rom(device, i, &rom[i]) != RCODE_COMPLETE) + if (read_rom(device, generation, i, &rom[i]) != + RCODE_COMPLETE) return -1; if ((key >> 30) == 3 && (rom[i] >> 30) > 1 && sp < ARRAY_SIZE(stack)) @@ -648,7 +667,7 @@ static void fw_device_init(struct work_struct *work) * device. */ - if (read_bus_info_block(device) < 0) { + if (read_bus_info_block(device, device->generation) < 0) { if (device->config_rom_retries < MAX_RETRIES) { device->config_rom_retries++; schedule_delayed_work(&device->work, RETRY_DELAY); @@ -801,6 +820,7 @@ void fw_node_event(struct fw_card *card, struct fw_node *node, int event) device = node->data; device->node_id = node->node_id; + smp_wmb(); /* update node_id before generation */ device->generation = card->generation; if (atomic_read(&device->state) == FW_DEVICE_RUNNING) { PREPARE_DELAYED_WORK(&device->work, fw_device_update); diff --git a/drivers/firewire/fw-device.h b/drivers/firewire/fw-device.h index 894d4a9..0854fe2 100644 --- a/drivers/firewire/fw-device.h +++ b/drivers/firewire/fw-device.h @@ -35,6 +35,18 @@ struct fw_attribute_group { struct attribute *attrs[11]; }; +/* + * Note, fw_device.generation always has to be read before fw_device.node_id. + * Use SMP memory barriers to ensure this. Otherwise requests will be sent + * to an outdated node_id if the generation was updated in the meantime due + * to a bus reset. + * + * Likewise, fw-core will take care to update .node_id before .generation so + * that whenever fw_device.generation is current WRT the actual bus generation, + * fw_device.node_id is guaranteed to be current too. + * + * The same applies to fw_device.card->node_id vs. fw_device.generation. + */ struct fw_device { atomic_t state; struct fw_node *node; diff --git a/drivers/firewire/fw-ohci.c b/drivers/firewire/fw-ohci.c index 6758832..7ebad3c 100644 --- a/drivers/firewire/fw-ohci.c +++ b/drivers/firewire/fw-ohci.c @@ -98,17 +98,48 @@ struct context; typedef int (*descriptor_callback_t)(struct context *ctx, struct descriptor *d, struct descriptor *last); + +/* + * A buffer that contains a block of DMA-able coherent memory used for + * storing a portion of a DMA descriptor program. + */ +struct descriptor_buffer { + struct list_head list; + dma_addr_t buffer_bus; + size_t buffer_size; + size_t used; + struct descriptor buffer[0]; +}; + struct context { struct fw_ohci *ohci; u32 regs; + int total_allocation; - struct descriptor *buffer; - dma_addr_t buffer_bus; - size_t buffer_size; - struct descriptor *head_descriptor; - struct descriptor *tail_descriptor; - struct descriptor *tail_descriptor_last; - struct descriptor *prev_descriptor; + /* + * List of page-sized buffers for storing DMA descriptors. + * Head of list contains buffers in use and tail of list contains + * free buffers. + */ + struct list_head buffer_list; + + /* + * Pointer to a buffer inside buffer_list that contains the tail + * end of the current DMA program. + */ + struct descriptor_buffer *buffer_tail; + + /* + * The descriptor containing the branch address of the first + * descriptor that has not yet been filled by the device. + */ + struct descriptor *last; + + /* + * The last descriptor in the DMA program. It contains the branch + * address that must be updated upon appending a new descriptor. + */ + struct descriptor *prev; descriptor_callback_t callback; @@ -125,6 +156,7 @@ struct context { struct iso_context { struct fw_iso_context base; struct context context; + int excess_bytes; void *header; size_t header_length; }; @@ -197,8 +229,6 @@ static inline struct fw_ohci *fw_ohci(struct fw_card *card) #define SELF_ID_BUF_SIZE 0x800 #define OHCI_TCODE_PHY_PACKET 0x0e #define OHCI_VERSION_1_1 0x010010 -#define ISO_BUFFER_SIZE (64 * 1024) -#define AT_BUFFER_SIZE 4096 static char ohci_driver_name[] = KBUILD_MODNAME; @@ -437,74 +467,126 @@ static void ar_context_run(struct ar_context *ctx) flush_writes(ctx->ohci); } +static struct descriptor * +find_branch_descriptor(struct descriptor *d, int z) +{ + int b, key; + + b = (le16_to_cpu(d->control) & DESCRIPTOR_BRANCH_ALWAYS) >> 2; + key = (le16_to_cpu(d->control) & DESCRIPTOR_KEY_IMMEDIATE) >> 8; + + /* figure out which descriptor the branch address goes in */ + if (z == 2 && (b == 3 || key == 2)) + return d; + else + return d + z - 1; +} + static void context_tasklet(unsigned long data) { struct context *ctx = (struct context *) data; - struct fw_ohci *ohci = ctx->ohci; struct descriptor *d, *last; u32 address; int z; + struct descriptor_buffer *desc; - dma_sync_single_for_cpu(ohci->card.device, ctx->buffer_bus, - ctx->buffer_size, DMA_TO_DEVICE); - - d = ctx->tail_descriptor; - last = ctx->tail_descriptor_last; - + desc = list_entry(ctx->buffer_list.next, + struct descriptor_buffer, list); + last = ctx->last; while (last->branch_address != 0) { + struct descriptor_buffer *old_desc = desc; address = le32_to_cpu(last->branch_address); z = address & 0xf; - d = ctx->buffer + (address - ctx->buffer_bus) / sizeof(*d); - last = (z == 2) ? d : d + z - 1; + address &= ~0xf; + + /* If the branch address points to a buffer outside of the + * current buffer, advance to the next buffer. */ + if (address < desc->buffer_bus || + address >= desc->buffer_bus + desc->used) + desc = list_entry(desc->list.next, + struct descriptor_buffer, list); + d = desc->buffer + (address - desc->buffer_bus) / sizeof(*d); + last = find_branch_descriptor(d, z); if (!ctx->callback(ctx, d, last)) break; - ctx->tail_descriptor = d; - ctx->tail_descriptor_last = last; + if (old_desc != desc) { + /* If we've advanced to the next buffer, move the + * previous buffer to the free list. */ + unsigned long flags; + old_desc->used = 0; + spin_lock_irqsave(&ctx->ohci->lock, flags); + list_move_tail(&old_desc->list, &ctx->buffer_list); + spin_unlock_irqrestore(&ctx->ohci->lock, flags); + } + ctx->last = last; } } +/* + * Allocate a new buffer and add it to the list of free buffers for this + * context. Must be called with ohci->lock held. + */ +static int +context_add_buffer(struct context *ctx) +{ + struct descriptor_buffer *desc; + dma_addr_t bus_addr; + int offset; + + /* + * 16MB of descriptors should be far more than enough for any DMA + * program. This will catch run-away userspace or DoS attacks. + */ + if (ctx->total_allocation >= 16*1024*1024) + return -ENOMEM; + + desc = dma_alloc_coherent(ctx->ohci->card.device, PAGE_SIZE, + &bus_addr, GFP_ATOMIC); + if (!desc) + return -ENOMEM; + + offset = (void *)&desc->buffer - (void *)desc; + desc->buffer_size = PAGE_SIZE - offset; + desc->buffer_bus = bus_addr + offset; + desc->used = 0; + + list_add_tail(&desc->list, &ctx->buffer_list); + ctx->total_allocation += PAGE_SIZE; + + return 0; +} + static int context_init(struct context *ctx, struct fw_ohci *ohci, - size_t buffer_size, u32 regs, - descriptor_callback_t callback) + u32 regs, descriptor_callback_t callback) { ctx->ohci = ohci; ctx->regs = regs; - ctx->buffer_size = buffer_size; - ctx->buffer = kmalloc(buffer_size, GFP_KERNEL); - if (ctx->buffer == NULL) + ctx->total_allocation = 0; + + INIT_LIST_HEAD(&ctx->buffer_list); + if (context_add_buffer(ctx) < 0) return -ENOMEM; + ctx->buffer_tail = list_entry(ctx->buffer_list.next, + struct descriptor_buffer, list); + tasklet_init(&ctx->tasklet, context_tasklet, (unsigned long)ctx); ctx->callback = callback; - ctx->buffer_bus = - dma_map_single(ohci->card.device, ctx->buffer, - buffer_size, DMA_TO_DEVICE); - if (dma_mapping_error(ctx->buffer_bus)) { - kfree(ctx->buffer); - return -ENOMEM; - } - - ctx->head_descriptor = ctx->buffer; - ctx->prev_descriptor = ctx->buffer; - ctx->tail_descriptor = ctx->buffer; - ctx->tail_descriptor_last = ctx->buffer; - /* * We put a dummy descriptor in the buffer that has a NULL * branch address and looks like it's been sent. That way we - * have a descriptor to append DMA programs to. Also, the - * ring buffer invariant is that it always has at least one - * element so that head == tail means buffer full. + * have a descriptor to append DMA programs to. */ - - memset(ctx->head_descriptor, 0, sizeof(*ctx->head_descriptor)); - ctx->head_descriptor->control = cpu_to_le16(DESCRIPTOR_OUTPUT_LAST); - ctx->head_descriptor->transfer_status = cpu_to_le16(0x8011); - ctx->head_descriptor++; + memset(ctx->buffer_tail->buffer, 0, sizeof(*ctx->buffer_tail->buffer)); + ctx->buffer_tail->buffer->control = cpu_to_le16(DESCRIPTOR_OUTPUT_LAST); + ctx->buffer_tail->buffer->transfer_status = cpu_to_le16(0x8011); + ctx->buffer_tail->used += sizeof(*ctx->buffer_tail->buffer); + ctx->last = ctx->buffer_tail->buffer; + ctx->prev = ctx->buffer_tail->buffer; return 0; } @@ -513,35 +595,42 @@ static void context_release(struct context *ctx) { struct fw_card *card = &ctx->ohci->card; + struct descriptor_buffer *desc, *tmp; - dma_unmap_single(card->device, ctx->buffer_bus, - ctx->buffer_size, DMA_TO_DEVICE); - kfree(ctx->buffer); + list_for_each_entry_safe(desc, tmp, &ctx->buffer_list, list) + dma_free_coherent(card->device, PAGE_SIZE, desc, + desc->buffer_bus - + ((void *)&desc->buffer - (void *)desc)); } +/* Must be called with ohci->lock held */ static struct descriptor * context_get_descriptors(struct context *ctx, int z, dma_addr_t *d_bus) { - struct descriptor *d, *tail, *end; - - d = ctx->head_descriptor; - tail = ctx->tail_descriptor; - end = ctx->buffer + ctx->buffer_size / sizeof(*d); - - if (d + z <= tail) { - goto has_space; - } else if (d > tail && d + z <= end) { - goto has_space; - } else if (d > tail && ctx->buffer + z <= tail) { - d = ctx->buffer; - goto has_space; - } + struct descriptor *d = NULL; + struct descriptor_buffer *desc = ctx->buffer_tail; + + if (z * sizeof(*d) > desc->buffer_size) + return NULL; + + if (z * sizeof(*d) > desc->buffer_size - desc->used) { + /* No room for the descriptor in this buffer, so advance to the + * next one. */ - return NULL; + if (desc->list.next == &ctx->buffer_list) { + /* If there is no free buffer next in the list, + * allocate one. */ + if (context_add_buffer(ctx) < 0) + return NULL; + } + desc = list_entry(desc->list.next, + struct descriptor_buffer, list); + ctx->buffer_tail = desc; + } - has_space: + d = desc->buffer + desc->used / sizeof(*d); memset(d, 0, z * sizeof(*d)); - *d_bus = ctx->buffer_bus + (d - ctx->buffer) * sizeof(*d); + *d_bus = desc->buffer_bus + desc->used; return d; } @@ -551,7 +640,7 @@ static void context_run(struct context *ctx, u32 extra) struct fw_ohci *ohci = ctx->ohci; reg_write(ohci, COMMAND_PTR(ctx->regs), - le32_to_cpu(ctx->tail_descriptor_last->branch_address)); + le32_to_cpu(ctx->last->branch_address)); reg_write(ohci, CONTROL_CLEAR(ctx->regs), ~0); reg_write(ohci, CONTROL_SET(ctx->regs), CONTEXT_RUN | extra); flush_writes(ohci); @@ -561,15 +650,13 @@ static void context_append(struct context *ctx, struct descriptor *d, int z, int extra) { dma_addr_t d_bus; + struct descriptor_buffer *desc = ctx->buffer_tail; - d_bus = ctx->buffer_bus + (d - ctx->buffer) * sizeof(*d); + d_bus = desc->buffer_bus + (d - desc->buffer) * sizeof(*d); - ctx->head_descriptor = d + z + extra; - ctx->prev_descriptor->branch_address = cpu_to_le32(d_bus | z); - ctx->prev_descriptor = z == 2 ? d : d + z - 1; - - dma_sync_single_for_device(ctx->ohci->card.device, ctx->buffer_bus, - ctx->buffer_size, DMA_TO_DEVICE); + desc->used += (z + extra) * sizeof(*d); + ctx->prev->branch_address = cpu_to_le32(d_bus | z); + ctx->prev = find_branch_descriptor(d, z); reg_write(ctx->ohci, CONTROL_SET(ctx->regs), CONTEXT_WAKE); flush_writes(ctx->ohci); @@ -655,7 +742,7 @@ at_context_queue_packet(struct context *ctx, struct fw_packet *packet) driver_data = (struct driver_data *) &d[3]; driver_data->packet = packet; packet->driver_data = driver_data; - + if (packet->payload_length > 0) { payload_bus = dma_map_single(ohci->card.device, packet->payload, @@ -903,7 +990,7 @@ at_context_transmit(struct context *ctx, struct fw_packet *packet) if (retval < 0) packet->callback(packet, &ctx->ohci->card, packet->ack); - + } static void bus_reset_tasklet(unsigned long data) @@ -984,8 +1071,10 @@ static void bus_reset_tasklet(unsigned long data) */ if (ohci->next_config_rom != NULL) { - free_rom = ohci->config_rom; - free_rom_bus = ohci->config_rom_bus; + if (ohci->next_config_rom != ohci->config_rom) { + free_rom = ohci->config_rom; + free_rom_bus = ohci->config_rom_bus; + } ohci->config_rom = ohci->next_config_rom; ohci->config_rom_bus = ohci->next_config_rom_bus; ohci->next_config_rom = NULL; @@ -1061,6 +1150,13 @@ static irqreturn_t irq_handler(int irq, void *data) if (unlikely(event & OHCI1394_postedWriteErr)) fw_error("PCI posted write error\n"); + if (unlikely(event & OHCI1394_cycleTooLong)) { + if (printk_ratelimit()) + fw_notify("isochronous cycle too long\n"); + reg_write(ohci, OHCI1394_LinkControlSet, + OHCI1394_LinkControl_cycleMaster); + } + if (event & OHCI1394_cycle64Seconds) { cycle_time = reg_read(ohci, OHCI1394_IsochronousCycleTimer); if ((cycle_time & 0x80000000) == 0) @@ -1134,8 +1230,8 @@ static int ohci_enable(struct fw_card *card, u32 *config_rom, size_t length) OHCI1394_RQPkt | OHCI1394_RSPkt | OHCI1394_reqTxComplete | OHCI1394_respTxComplete | OHCI1394_isochRx | OHCI1394_isochTx | - OHCI1394_postedWriteErr | OHCI1394_cycle64Seconds | - OHCI1394_masterIntEnable); + OHCI1394_postedWriteErr | OHCI1394_cycleTooLong | + OHCI1394_cycle64Seconds | OHCI1394_masterIntEnable); /* Activate link_on bit and contender bit in our self ID packets.*/ if (ohci_update_phy_reg(card, 4, 0, @@ -1161,19 +1257,30 @@ static int ohci_enable(struct fw_card *card, u32 *config_rom, size_t length) * the right values in the bus reset tasklet. */ - ohci->next_config_rom = - dma_alloc_coherent(ohci->card.device, CONFIG_ROM_SIZE, - &ohci->next_config_rom_bus, GFP_KERNEL); - if (ohci->next_config_rom == NULL) - return -ENOMEM; + if (config_rom) { + ohci->next_config_rom = + dma_alloc_coherent(ohci->card.device, CONFIG_ROM_SIZE, + &ohci->next_config_rom_bus, + GFP_KERNEL); + if (ohci->next_config_rom == NULL) + return -ENOMEM; - memset(ohci->next_config_rom, 0, CONFIG_ROM_SIZE); - fw_memcpy_to_be32(ohci->next_config_rom, config_rom, length * 4); + memset(ohci->next_config_rom, 0, CONFIG_ROM_SIZE); + fw_memcpy_to_be32(ohci->next_config_rom, config_rom, length * 4); + } else { + /* + * In the suspend case, config_rom is NULL, which + * means that we just reuse the old config rom. + */ + ohci->next_config_rom = ohci->config_rom; + ohci->next_config_rom_bus = ohci->config_rom_bus; + } - ohci->next_header = config_rom[0]; + ohci->next_header = be32_to_cpu(ohci->next_config_rom[0]); ohci->next_config_rom[0] = 0; reg_write(ohci, OHCI1394_ConfigROMhdr, 0); - reg_write(ohci, OHCI1394_BusOptions, config_rom[2]); + reg_write(ohci, OHCI1394_BusOptions, + be32_to_cpu(ohci->next_config_rom[2])); reg_write(ohci, OHCI1394_ConfigROMmap, ohci->next_config_rom_bus); reg_write(ohci, OHCI1394_AsReqFilterHiSet, 0x80000000); @@ -1380,9 +1487,13 @@ static int handle_ir_dualbuffer_packet(struct context *context, void *p, *end; int i; - if (db->first_res_count > 0 && db->second_res_count > 0) - /* This descriptor isn't done yet, stop iteration. */ - return 0; + if (db->first_res_count > 0 && db->second_res_count > 0) { + if (ctx->excess_bytes <= le16_to_cpu(db->second_req_count)) { + /* This descriptor isn't done yet, stop iteration. */ + return 0; + } + ctx->excess_bytes -= le16_to_cpu(db->second_req_count); + } header_length = le16_to_cpu(db->first_req_count) - le16_to_cpu(db->first_res_count); @@ -1401,11 +1512,15 @@ static int handle_ir_dualbuffer_packet(struct context *context, *(u32 *) (ctx->header + i) = __swab32(*(u32 *) (p + 4)); memcpy(ctx->header + i + 4, p + 8, ctx->base.header_size - 4); i += ctx->base.header_size; + ctx->excess_bytes += + (le32_to_cpu(*(u32 *)(p + 4)) >> 16) & 0xffff; p += ctx->base.header_size + 4; } - ctx->header_length = i; + ctx->excess_bytes -= le16_to_cpu(db->second_req_count) - + le16_to_cpu(db->second_res_count); + if (le16_to_cpu(db->control) & DESCRIPTOR_IRQ_ALWAYS) { ir_header = (__le32 *) (db + 1); ctx->base.callback(&ctx->base, @@ -1418,6 +1533,53 @@ static int handle_ir_dualbuffer_packet(struct context *context, return 1; } +static int handle_ir_packet_per_buffer(struct context *context, + struct descriptor *d, + struct descriptor *last) +{ + struct iso_context *ctx = + container_of(context, struct iso_context, context); + struct descriptor *pd; + __le32 *ir_header; + void *p; + int i; + + for (pd = d; pd <= last; pd++) { + if (pd->transfer_status) + break; + } + if (pd > last) + /* Descriptor(s) not done yet, stop iteration */ + return 0; + + i = ctx->header_length; + p = last + 1; + + if (ctx->base.header_size > 0 && + i + ctx->base.header_size <= PAGE_SIZE) { + /* + * The iso header is byteswapped to little endian by + * the controller, but the remaining header quadlets + * are big endian. We want to present all the headers + * as big endian, so we have to swap the first quadlet. + */ + *(u32 *) (ctx->header + i) = __swab32(*(u32 *) (p + 4)); + memcpy(ctx->header + i + 4, p + 8, ctx->base.header_size - 4); + ctx->header_length += ctx->base.header_size; + } + + if (le16_to_cpu(last->control) & DESCRIPTOR_IRQ_ALWAYS) { + ir_header = (__le32 *) p; + ctx->base.callback(&ctx->base, + le32_to_cpu(ir_header[0]) & 0xffff, + ctx->header_length, ctx->header, + ctx->base.callback_data); + ctx->header_length = 0; + } + + return 1; +} + static int handle_it_packet(struct context *context, struct descriptor *d, struct descriptor *last) @@ -1453,14 +1615,12 @@ ohci_allocate_iso_context(struct fw_card *card, int type, size_t header_size) } else { mask = &ohci->ir_context_mask; list = ohci->ir_context_list; - callback = handle_ir_dualbuffer_packet; + if (ohci->version >= OHCI_VERSION_1_1) + callback = handle_ir_dualbuffer_packet; + else + callback = handle_ir_packet_per_buffer; } - /* FIXME: We need a fallback for pre 1.1 OHCI. */ - if (callback == handle_ir_dualbuffer_packet && - ohci->version < OHCI_VERSION_1_1) - return ERR_PTR(-ENOSYS); - spin_lock_irqsave(&ohci->lock, flags); index = ffs(*mask) - 1; if (index >= 0) @@ -1482,8 +1642,7 @@ ohci_allocate_iso_context(struct fw_card *card, int type, size_t header_size) if (ctx->header == NULL) goto out; - retval = context_init(&ctx->context, ohci, ISO_BUFFER_SIZE, - regs, callback); + retval = context_init(&ctx->context, ohci, regs, callback); if (retval < 0) goto out_with_header; @@ -1519,7 +1678,9 @@ static int ohci_start_iso(struct fw_iso_context *base, context_run(&ctx->context, match); } else { index = ctx - ohci->ir_context_list; - control = IR_CONTEXT_DUAL_BUFFER_MODE | IR_CONTEXT_ISOCH_HEADER; + control = IR_CONTEXT_ISOCH_HEADER; + if (ohci->version >= OHCI_VERSION_1_1) + control |= IR_CONTEXT_DUAL_BUFFER_MODE; match = (tags << 28) | (sync << 8) | ctx->base.channel; if (cycle >= 0) { match |= (cycle & 0x07fff) << 12; @@ -1696,19 +1857,6 @@ ohci_queue_iso_receive_dualbuffer(struct fw_iso_context *base, * packet, retransmit or terminate.. */ - if (packet->skip) { - d = context_get_descriptors(&ctx->context, 2, &d_bus); - if (d == NULL) - return -ENOMEM; - - db = (struct db_descriptor *) d; - db->control = cpu_to_le16(DESCRIPTOR_STATUS | - DESCRIPTOR_BRANCH_ALWAYS | - DESCRIPTOR_WAIT); - db->first_size = cpu_to_le16(ctx->base.header_size + 4); - context_append(&ctx->context, d, 2, 0); - } - p = packet; z = 2; @@ -1725,7 +1873,6 @@ ohci_queue_iso_receive_dualbuffer(struct fw_iso_context *base, offset = payload & ~PAGE_MASK; rest = p->payload_length; - /* FIXME: OHCI 1.0 doesn't support dual buffer receive */ /* FIXME: make packet-per-buffer/dual-buffer a context option */ while (rest > 0) { d = context_get_descriptors(&ctx->context, @@ -1737,11 +1884,18 @@ ohci_queue_iso_receive_dualbuffer(struct fw_iso_context *base, db->control = cpu_to_le16(DESCRIPTOR_STATUS | DESCRIPTOR_BRANCH_ALWAYS); db->first_size = cpu_to_le16(ctx->base.header_size + 4); - db->first_req_count = cpu_to_le16(header_size); + if (p->skip && rest == p->payload_length) { + db->control |= cpu_to_le16(DESCRIPTOR_WAIT); + db->first_req_count = db->first_size; + } else { + db->first_req_count = cpu_to_le16(header_size); + } db->first_res_count = db->first_req_count; db->first_buffer = cpu_to_le32(d_bus + sizeof(*db)); - if (offset + rest < PAGE_SIZE) + if (p->skip && rest == p->payload_length) + length = 4; + else if (offset + rest < PAGE_SIZE) length = rest; else length = PAGE_SIZE - offset; @@ -1757,7 +1911,86 @@ ohci_queue_iso_receive_dualbuffer(struct fw_iso_context *base, context_append(&ctx->context, d, z, header_z); offset = (offset + length) & ~PAGE_MASK; rest -= length; - page++; + if (offset == 0) + page++; + } + + return 0; +} + +static int +ohci_queue_iso_receive_packet_per_buffer(struct fw_iso_context *base, + struct fw_iso_packet *packet, + struct fw_iso_buffer *buffer, + unsigned long payload) +{ + struct iso_context *ctx = container_of(base, struct iso_context, base); + struct descriptor *d = NULL, *pd = NULL; + struct fw_iso_packet *p = packet; + dma_addr_t d_bus, page_bus; + u32 z, header_z, rest; + int i, j, length; + int page, offset, packet_count, header_size, payload_per_buffer; + + /* + * The OHCI controller puts the status word in the + * buffer too, so we need 4 extra bytes per packet. + */ + packet_count = p->header_length / ctx->base.header_size; + header_size = ctx->base.header_size + 4; + + /* Get header size in number of descriptors. */ + header_z = DIV_ROUND_UP(header_size, sizeof(*d)); + page = payload >> PAGE_SHIFT; + offset = payload & ~PAGE_MASK; + payload_per_buffer = p->payload_length / packet_count; + + for (i = 0; i < packet_count; i++) { + /* d points to the header descriptor */ + z = DIV_ROUND_UP(payload_per_buffer + offset, PAGE_SIZE) + 1; + d = context_get_descriptors(&ctx->context, + z + header_z, &d_bus); + if (d == NULL) + return -ENOMEM; + + d->control = cpu_to_le16(DESCRIPTOR_STATUS | + DESCRIPTOR_INPUT_MORE); + if (p->skip && i == 0) + d->control |= cpu_to_le16(DESCRIPTOR_WAIT); + d->req_count = cpu_to_le16(header_size); + d->res_count = d->req_count; + d->transfer_status = 0; + d->data_address = cpu_to_le32(d_bus + (z * sizeof(*d))); + + rest = payload_per_buffer; + for (j = 1; j < z; j++) { + pd = d + j; + pd->control = cpu_to_le16(DESCRIPTOR_STATUS | + DESCRIPTOR_INPUT_MORE); + + if (offset + rest < PAGE_SIZE) + length = rest; + else + length = PAGE_SIZE - offset; + pd->req_count = cpu_to_le16(length); + pd->res_count = pd->req_count; + pd->transfer_status = 0; + + page_bus = page_private(buffer->pages[page]); + pd->data_address = cpu_to_le32(page_bus + offset); + + offset = (offset + length) & ~PAGE_MASK; + rest -= length; + if (offset == 0) + page++; + } + pd->control = cpu_to_le16(DESCRIPTOR_STATUS | + DESCRIPTOR_INPUT_LAST | + DESCRIPTOR_BRANCH_ALWAYS); + if (p->interrupt && i == packet_count - 1) + pd->control |= cpu_to_le16(DESCRIPTOR_IRQ_ALWAYS); + + context_append(&ctx->context, d, z, header_z); } return 0; @@ -1770,15 +2003,22 @@ ohci_queue_iso(struct fw_iso_context *base, unsigned long payload) { struct iso_context *ctx = container_of(base, struct iso_context, base); + unsigned long flags; + int retval; + spin_lock_irqsave(&ctx->context.ohci->lock, flags); if (base->type == FW_ISO_CONTEXT_TRANSMIT) - return ohci_queue_iso_transmit(base, packet, buffer, payload); + retval = ohci_queue_iso_transmit(base, packet, buffer, payload); else if (ctx->context.ohci->version >= OHCI_VERSION_1_1) - return ohci_queue_iso_receive_dualbuffer(base, packet, + retval = ohci_queue_iso_receive_dualbuffer(base, packet, buffer, payload); else - /* FIXME: Implement fallback for OHCI 1.0 controllers. */ - return -ENOSYS; + retval = ohci_queue_iso_receive_packet_per_buffer(base, packet, + buffer, + payload); + spin_unlock_irqrestore(&ctx->context.ohci->lock, flags); + + return retval; } static const struct fw_card_driver ohci_driver = { @@ -1850,10 +2090,10 @@ pci_probe(struct pci_dev *dev, const struct pci_device_id *ent) ar_context_init(&ohci->ar_response_ctx, ohci, OHCI1394_AsRspRcvContextControlSet); - context_init(&ohci->at_request_ctx, ohci, AT_BUFFER_SIZE, + context_init(&ohci->at_request_ctx, ohci, OHCI1394_AsReqTrContextControlSet, handle_at_packet); - context_init(&ohci->at_response_ctx, ohci, AT_BUFFER_SIZE, + context_init(&ohci->at_response_ctx, ohci, OHCI1394_AsRspTrContextControlSet, handle_at_packet); reg_write(ohci, OHCI1394_IsoRecvIntMaskSet, ~0); @@ -1898,12 +2138,6 @@ pci_probe(struct pci_dev *dev, const struct pci_device_id *ent) ohci->version = reg_read(ohci, OHCI1394_Version) & 0x00ff00ff; fw_notify("Added fw-ohci device %s, OHCI version %x.%x\n", dev->dev.bus_id, ohci->version >> 16, ohci->version & 0xff); - if (ohci->version < OHCI_VERSION_1_1) { - fw_notify(" Isochronous I/O is not yet implemented for " - "OHCI 1.0 chips.\n"); - fw_notify(" Cameras, audio devices etc. won't work on " - "this controller with this driver version.\n"); - } return 0; fail_self_id: @@ -1984,7 +2218,7 @@ static int pci_resume(struct pci_dev *pdev) return err; } - return ohci_enable(&ohci->card, ohci->config_rom, CONFIG_ROM_SIZE); + return ohci_enable(&ohci->card, NULL, 0); } #endif diff --git a/drivers/firewire/fw-sbp2.c b/drivers/firewire/fw-sbp2.c index 5596df6..19ece9b 100644 --- a/drivers/firewire/fw-sbp2.c +++ b/drivers/firewire/fw-sbp2.c @@ -40,6 +40,7 @@ #include <linux/stringify.h> #include <linux/timer.h> #include <linux/workqueue.h> +#include <asm/system.h> #include <scsi/scsi.h> #include <scsi/scsi_cmnd.h> @@ -148,18 +149,26 @@ struct sbp2_target { unsigned workarounds; struct list_head lu_list; + + unsigned int mgt_orb_timeout; }; -#define SBP2_MAX_SG_ELEMENT_LENGTH 0xf000 -#define SBP2_MAX_SECTORS 255 /* Max sectors supported */ +/* + * Per section 7.4.8 of the SBP-2 spec, a mgt_ORB_timeout value can be + * provided in the config rom. Most devices do provide a value, which + * we'll use for login management orbs, but with some sane limits. + */ +#define SBP2_MIN_LOGIN_ORB_TIMEOUT 5000U /* Timeout in ms */ +#define SBP2_MAX_LOGIN_ORB_TIMEOUT 40000U /* Timeout in ms */ #define SBP2_ORB_TIMEOUT 2000 /* Timeout in ms */ - #define SBP2_ORB_NULL 0x80000000 +#define SBP2_MAX_SG_ELEMENT_LENGTH 0xf000 #define SBP2_DIRECTION_TO_MEDIA 0x0 #define SBP2_DIRECTION_FROM_MEDIA 0x1 /* Unit directory keys */ +#define SBP2_CSR_UNIT_CHARACTERISTICS 0x3a #define SBP2_CSR_FIRMWARE_REVISION 0x3c #define SBP2_CSR_LOGICAL_UNIT_NUMBER 0x14 #define SBP2_CSR_LOGICAL_UNIT_DIRECTORY 0xd4 @@ -489,6 +498,7 @@ sbp2_send_management_orb(struct sbp2_logical_unit *lu, int node_id, { struct fw_device *device = fw_device(lu->tgt->unit->device.parent); struct sbp2_management_orb *orb; + unsigned int timeout; int retval = -ENOMEM; orb = kzalloc(sizeof(*orb), GFP_ATOMIC); @@ -516,9 +526,13 @@ sbp2_send_management_orb(struct sbp2_logical_unit *lu, int node_id, orb->request.status_fifo.low = lu->address_handler.offset; if (function == SBP2_LOGIN_REQUEST) { + /* Ask for 2^2 == 4 seconds reconnect grace period */ orb->request.misc |= - MANAGEMENT_ORB_EXCLUSIVE(sbp2_param_exclusive_login) | - MANAGEMENT_ORB_RECONNECT(0); + MANAGEMENT_ORB_RECONNECT(2) | + MANAGEMENT_ORB_EXCLUSIVE(sbp2_param_exclusive_login); + timeout = lu->tgt->mgt_orb_timeout; + } else { + timeout = SBP2_ORB_TIMEOUT; } fw_memcpy_to_be32(&orb->request, &orb->request, sizeof(orb->request)); @@ -535,8 +549,7 @@ sbp2_send_management_orb(struct sbp2_logical_unit *lu, int node_id, sbp2_send_orb(&orb->base, lu, node_id, generation, lu->tgt->management_agent_address); - wait_for_completion_timeout(&orb->done, - msecs_to_jiffies(SBP2_ORB_TIMEOUT)); + wait_for_completion_timeout(&orb->done, msecs_to_jiffies(timeout)); retval = -EIO; if (sbp2_cancel_orbs(lu) == 0) { @@ -608,13 +621,17 @@ static void sbp2_release_target(struct kref *kref) struct sbp2_logical_unit *lu, *next; struct Scsi_Host *shost = container_of((void *)tgt, struct Scsi_Host, hostdata[0]); + struct fw_device *device = fw_device(tgt->unit->device.parent); list_for_each_entry_safe(lu, next, &tgt->lu_list, link) { if (lu->sdev) scsi_remove_device(lu->sdev); - sbp2_send_management_orb(lu, tgt->node_id, lu->generation, - SBP2_LOGOUT_REQUEST, lu->login_id, NULL); + if (!fw_device_is_shutdown(device)) + sbp2_send_management_orb(lu, tgt->node_id, + lu->generation, SBP2_LOGOUT_REQUEST, + lu->login_id, NULL); + fw_core_remove_address_handler(&lu->address_handler); list_del(&lu->link); kfree(lu); @@ -628,6 +645,21 @@ static void sbp2_release_target(struct kref *kref) static struct workqueue_struct *sbp2_wq; +/* + * Always get the target's kref when scheduling work on one its units. + * Each workqueue job is responsible to call sbp2_target_put() upon return. + */ +static void sbp2_queue_work(struct sbp2_logical_unit *lu, unsigned long delay) +{ + if (queue_delayed_work(sbp2_wq, &lu->work, delay)) + kref_get(&lu->tgt->kref); +} + +static void sbp2_target_put(struct sbp2_target *tgt) +{ + kref_put(&tgt->kref, sbp2_release_target); +} + static void sbp2_reconnect(struct work_struct *work); static void sbp2_login(struct work_struct *work) @@ -643,21 +675,19 @@ static void sbp2_login(struct work_struct *work) struct sbp2_login_response response; int generation, node_id, local_node_id; - generation = device->card->generation; - node_id = device->node->node_id; - local_node_id = device->card->local_node->node_id; + generation = device->generation; + smp_rmb(); /* node_id must not be older than generation */ + node_id = device->node_id; + local_node_id = device->card->node_id; if (sbp2_send_management_orb(lu, node_id, generation, SBP2_LOGIN_REQUEST, lu->lun, &response) < 0) { - if (lu->retries++ < 5) { - queue_delayed_work(sbp2_wq, &lu->work, - DIV_ROUND_UP(HZ, 5)); - } else { + if (lu->retries++ < 5) + sbp2_queue_work(lu, DIV_ROUND_UP(HZ, 5)); + else fw_error("failed to login to %s LUN %04x\n", unit->device.bus_id, lu->lun); - kref_put(&lu->tgt->kref, sbp2_release_target); - } - return; + goto out; } lu->generation = generation; @@ -699,7 +729,8 @@ static void sbp2_login(struct work_struct *work) lu->sdev = sdev; scsi_device_put(sdev); } - kref_put(&lu->tgt->kref, sbp2_release_target); + out: + sbp2_target_put(lu->tgt); } static int sbp2_add_logical_unit(struct sbp2_target *tgt, int lun_entry) @@ -749,6 +780,7 @@ static int sbp2_scan_unit_dir(struct sbp2_target *tgt, u32 *directory, { struct fw_csr_iterator ci; int key, value; + unsigned int timeout; fw_csr_iterator_init(&ci, directory); while (fw_csr_iterator_next(&ci, &key, &value)) { @@ -771,6 +803,21 @@ static int sbp2_scan_unit_dir(struct sbp2_target *tgt, u32 *directory, *firmware_revision = value; break; + case SBP2_CSR_UNIT_CHARACTERISTICS: + /* the timeout value is stored in 500ms units */ + timeout = ((unsigned int) value >> 8 & 0xff) * 500; + timeout = max(timeout, SBP2_MIN_LOGIN_ORB_TIMEOUT); + tgt->mgt_orb_timeout = + min(timeout, SBP2_MAX_LOGIN_ORB_TIMEOUT); + + if (timeout > tgt->mgt_orb_timeout) + fw_notify("%s: config rom contains %ds " + "management ORB timeout, limiting " + "to %ds\n", tgt->unit->device.bus_id, + timeout / 1000, + tgt->mgt_orb_timeout / 1000); + break; + case SBP2_CSR_LOGICAL_UNIT_NUMBER: if (sbp2_add_logical_unit(tgt, value) < 0) return -ENOMEM; @@ -864,18 +911,13 @@ static int sbp2_probe(struct device *dev) get_device(&unit->device); - /* - * We schedule work to do the login so we can easily - * reschedule retries. Always get the ref before scheduling - * work. - */ + /* Do the login in a workqueue so we can easily reschedule retries. */ list_for_each_entry(lu, &tgt->lu_list, link) - if (queue_delayed_work(sbp2_wq, &lu->work, 0)) - kref_get(&tgt->kref); + sbp2_queue_work(lu, 0); return 0; fail_tgt_put: - kref_put(&tgt->kref, sbp2_release_target); + sbp2_target_put(tgt); return -ENOMEM; fail_shost_put: @@ -888,7 +930,7 @@ static int sbp2_remove(struct device *dev) struct fw_unit *unit = fw_unit(dev); struct sbp2_target *tgt = unit->device.driver_data; - kref_put(&tgt->kref, sbp2_release_target); + sbp2_target_put(tgt); return 0; } @@ -900,9 +942,10 @@ static void sbp2_reconnect(struct work_struct *work) struct fw_device *device = fw_device(unit->device.parent); int generation, node_id, local_node_id; - generation = device->card->generation; - node_id = device->node->node_id; - local_node_id = device->card->local_node->node_id; + generation = device->generation; + smp_rmb(); /* node_id must not be older than generation */ + node_id = device->node_id; + local_node_id = device->card->node_id; if (sbp2_send_management_orb(lu, node_id, generation, SBP2_RECONNECT_REQUEST, @@ -914,8 +957,8 @@ static void sbp2_reconnect(struct work_struct *work) lu->retries = 0; PREPARE_DELAYED_WORK(&lu->work, sbp2_login); } - queue_delayed_work(sbp2_wq, &lu->work, DIV_ROUND_UP(HZ, 5)); - return; + sbp2_queue_work(lu, DIV_ROUND_UP(HZ, 5)); + goto out; } lu->generation = generation; @@ -927,8 +970,8 @@ static void sbp2_reconnect(struct work_struct *work) sbp2_agent_reset(lu); sbp2_cancel_orbs(lu); - - kref_put(&lu->tgt->kref, sbp2_release_target); + out: + sbp2_target_put(lu->tgt); } static void sbp2_update(struct fw_unit *unit) @@ -944,8 +987,7 @@ static void sbp2_update(struct fw_unit *unit) */ list_for_each_entry(lu, &tgt->lu_list, link) { lu->retries = 0; - if (queue_delayed_work(sbp2_wq, &lu->work, 0)) - kref_get(&tgt->kref); + sbp2_queue_work(lu, 0); } } @@ -1100,9 +1142,9 @@ sbp2_map_scatterlist(struct sbp2_command_orb *orb, struct fw_device *device, * elements larger than 65535 bytes, some IOMMUs may merge sg elements * during DMA mapping, and Linux currently doesn't prevent this. */ - for (i = 0, j = 0; i < count; i++) { - sg_len = sg_dma_len(sg + i); - sg_addr = sg_dma_address(sg + i); + for (i = 0, j = 0; i < count; i++, sg = sg_next(sg)) { + sg_len = sg_dma_len(sg); + sg_addr = sg_dma_address(sg); while (sg_len) { /* FIXME: This won't get us out of the pinch. */ if (unlikely(j >= ARRAY_SIZE(orb->page_table))) { @@ -1235,6 +1277,12 @@ static int sbp2_scsi_slave_alloc(struct scsi_device *sdev) sdev->allow_restart = 1; + /* + * Update the dma alignment (minimum alignment requirements for + * start and end of DMA transfers) to be a sector + */ + blk_queue_update_dma_alignment(sdev->request_queue, 511); + if (lu->tgt->workarounds & SBP2_WORKAROUND_INQUIRY_36) sdev->inquiry_len = 36; diff --git a/drivers/firewire/fw-topology.c b/drivers/firewire/fw-topology.c index 0fc9b00..172c186 100644 --- a/drivers/firewire/fw-topology.c +++ b/drivers/firewire/fw-topology.c @@ -21,6 +21,7 @@ #include <linux/module.h> #include <linux/wait.h> #include <linux/errno.h> +#include <asm/system.h> #include "fw-transaction.h" #include "fw-topology.h" @@ -518,6 +519,11 @@ fw_core_handle_bus_reset(struct fw_card *card, card->bm_retries = 0; card->node_id = node_id; + /* + * Update node_id before generation to prevent anybody from using + * a stale node_id together with a current generation. + */ + smp_wmb(); card->generation = generation; card->reset_jiffies = jiffies; schedule_delayed_work(&card->work, 0); diff --git a/drivers/firewire/fw-transaction.c b/drivers/firewire/fw-transaction.c index c00d4a9..7fcc59d 100644 --- a/drivers/firewire/fw-transaction.c +++ b/drivers/firewire/fw-transaction.c @@ -153,7 +153,7 @@ fw_fill_request(struct fw_packet *packet, int tcode, int tlabel, int ext_tcode; if (tcode > 0x10) { - ext_tcode = tcode - 0x10; + ext_tcode = tcode & ~0x10; tcode = TCODE_LOCK_REQUEST; } else ext_tcode = 0; @@ -650,7 +650,7 @@ fw_core_handle_request(struct fw_card *card, struct fw_packet *p) HEADER_GET_OFFSET_HIGH(p->header[1]) << 32) | p->header[2]; tcode = HEADER_GET_TCODE(p->header[0]); destination = HEADER_GET_DESTINATION(p->header[0]); - source = HEADER_GET_SOURCE(p->header[0]); + source = HEADER_GET_SOURCE(p->header[1]); spin_lock_irqsave(&address_handler_lock, flags); handler = lookup_enclosing_address_handler(&address_handler_list, |
