Linus, please pull from the for-linus branch at
git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394-2.6.git for-linus
to receive the following FireWire subsystem update.
This considerably enhances compatibility of the new firewire-ohci driver
with a number of controllers. It shrinks the list of chips with trouble
with isochronous reception to VIA VT6306 and some variants of VT6307.
The patch is somewhat big for this late -rc phase, and it has so far only
surfaced in 2.6.24-rc4-mm1 and in recent Fedora test kernels. But the
author and I did a lot of tests with as much previously working chips as
we could get our hands on (some more than listed below) to make sure that
there is no regression.
drivers/firewire/fw-ohci.c | 175 +++++++++++++++++++++++++++++++++++++++-----
1 files changed, 155 insertions(+), 20 deletions(-)
Jarod Wilson (1):
firewire: OHCI 1.0 Isochronous Receive support
Full log and diff:
commit a186b4a6b22fdc96a1ed63da483d267b5d00839e
Author: Jarod Wilson <[email protected]>
Date: Mon Dec 3 13:43:12 2007 -0500
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <[email protected]>
Signed-off-by: Stefan Richter <[email protected]>
diff --git a/drivers/firewire/fw-ohci.c b/drivers/firewire/fw-ohci.c
index c9b9081..436a855 100644
--- a/drivers/firewire/fw-ohci.c
+++ b/drivers/firewire/fw-ohci.c
@@ -437,6 +437,21 @@ 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;
@@ -455,7 +470,7 @@ static void context_tasklet(unsigned long data)
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;
+ last = find_branch_descriptor(d, z);
if (!ctx->callback(ctx, d, last))
break;
@@ -566,7 +581,7 @@ static void context_append(struct context *ctx,
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;
+ ctx->prev_descriptor = find_branch_descriptor(d, z);
dma_sync_single_for_device(ctx->ohci->card.device, ctx->buffer_bus,
ctx->buffer_size, DMA_TO_DEVICE);
@@ -655,7 +670,7 @@ 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 +918,7 @@ 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)
@@ -1431,6 +1446,57 @@ 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 = d + 1;
+ __le32 *ir_header;
+ size_t header_length;
+ void *p, *end;
+ int i, z;
+
+ if (pd->res_count == pd->req_count)
+ /* Descriptor(s) not done yet, stop iteration */
+ return 0;
+
+ header_length = le16_to_cpu(d->req_count);
+
+ i = ctx->header_length;
+ z = le32_to_cpu(pd->branch_address) & 0xf;
+ p = d + z;
+ end = p + header_length;
+
+ while (p < end && 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);
+ i += ctx->base.header_size;
+ p += ctx->base.header_size + 4;
+ }
+
+ ctx->header_length = i;
+
+ if (le16_to_cpu(pd->control) & DESCRIPTOR_IRQ_ALWAYS) {
+ ir_header = (__le32 *) (d + z);
+ 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)
@@ -1466,14 +1532,12 @@ 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)
@@ -1532,7 +1596,9 @@ 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;
@@ -1738,7 +1804,6 @@ 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,
@@ -1777,6 +1842,81 @@ ohci_queue_iso_receive_dualbuffer(struct fw_iso_context *base,
}
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;
+ dma_addr_t d_bus, page_bus;
+ u32 z, header_z, rest;
+ int i, page, offset, packet_count, header_size;
+
+ if (packet->skip) {
+ d = context_get_descriptors(&ctx->context, 1, &d_bus);
+ if (d == NULL)
+ return -ENOMEM;
+
+ d->control = cpu_to_le16(DESCRIPTOR_STATUS |
+ DESCRIPTOR_INPUT_LAST |
+ DESCRIPTOR_BRANCH_ALWAYS |
+ DESCRIPTOR_WAIT);
+ context_append(&ctx->context, d, 1, 0);
+ }
+
+ /* one descriptor for header, one for payload */
+ /* FIXME: handle cases where we need multiple desc. for payload */
+ z = 2;
+ p = packet;
+
+ /*
+ * 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 = packet_count * (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;
+ rest = p->payload_length;
+
+ for (i = 0; i < packet_count; i++) {
+ /* d points to the header descriptor */
+ d = context_get_descriptors(&ctx->context,
+ z + header_z, &d_bus);
+ if (d == NULL)
+ return -ENOMEM;
+
+ d->control = cpu_to_le16(DESCRIPTOR_INPUT_MORE);
+ d->req_count = cpu_to_le16(header_size);
+ d->res_count = d->req_count;
+ d->data_address = cpu_to_le32(d_bus + (z * sizeof(*d)));
+
+ /* pd points to the payload descriptor */
+ pd = d + 1;
+ pd->control = cpu_to_le16(DESCRIPTOR_STATUS |
+ DESCRIPTOR_INPUT_LAST |
+ DESCRIPTOR_BRANCH_ALWAYS);
+ if (p->interrupt)
+ pd->control |= cpu_to_le16(DESCRIPTOR_IRQ_ALWAYS);
+
+ pd->req_count = cpu_to_le16(rest);
+ pd->res_count = pd->req_count;
+
+ page_bus = page_private(buffer->pages[page]);
+ pd->data_address = cpu_to_le32(page_bus + offset);
+
+ context_append(&ctx->context, d, z, header_z);
+ }
+
+ return 0;
+}
+
+static int
ohci_queue_iso(struct fw_iso_context *base,
struct fw_iso_packet *packet,
struct fw_iso_buffer *buffer,
@@ -1790,8 +1930,9 @@ ohci_queue_iso(struct fw_iso_context *base,
return ohci_queue_iso_receive_dualbuffer(base, packet,
buffer, payload);
else
- /* FIXME: Implement fallback for OHCI 1.0 controllers. */
- return -ENOSYS;
+ return ohci_queue_iso_receive_packet_per_buffer(base, packet,
+ buffer,
+ payload);
}
static const struct fw_card_driver ohci_driver = {
@@ -1911,12 +2052,6 @@ 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:
--
Stefan Richter
-=====-=-=== ==-- -=-=-
http://arcgraph.de/sr/
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[Index of Archives]
[Kernel Newbies]
[Netfilter]
[Bugtraq]
[Photo]
[Stuff]
[Gimp]
[Yosemite News]
[MIPS Linux]
[ARM Linux]
[Linux Security]
[Linux RAID]
[Video 4 Linux]
[Linux for the blind]
[Linux Resources]
