Skip to contentSkip to conversation

What's in linux1394-2.6.git?

PATCH Stefan Richter 1 participant 1 message Nov 24, 2009 Firewire

lore
Share & tools
lore
#0 lore raw

Here is a summary of what is currently queued up in

    git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394-2.6.git

and also visible in linux-next. Gitweb:

    http://git.kernel.org/?p=linux/kernel/git/ieee1394/linux1394-2.6.git

fixes branch
============

These are two fixes which make the new firewire driver stack interoperable with FFADO, the userspace drivers for FireWire audio interfaces. I will ask Linus to pull this branch later this week, in hope to get these fixes released in 2.6.32.

In addition to these two kernel fixes, there are also some updates to the firewire-core backend of libraw1394 necessary to make it work with FFADO. At the moment these proposed updates are queued up at the "testing" branch of

    git://git.kernel.org/pub/scm/libs/ieee1394/libraw1394.git testing

Here is the shortlog and diffstat of the fixes branch; full log and diff follow at the end of this message:

Jay Fenlason (2):
firewire: ohci: Make cycleMatch ISO transmission work
firewire: ohci: pass correct iso xmit timestamps to core

 drivers/firewire/ohci.c |   41 +++++++++++++++++++++++++++++++++-----
 1 files changed, 35 insertions(+), 6 deletions(-)

Many thanks to Jay for debugging and fixing these issues.

Some more kernel work will be neessary for fully satisfactory support of FireWire audio interfaces: Multiplex FCP requests as necessary for multiple device support; implement variable split transaction time-out;
longer-term: implement a kernelspace streaming component for FFADO (some groundwork has been done in a GSoC project).

master branch
=============

This is for the post 2.6.32 merge window. We have got minor fixes and tweaks, among them reduced stack usage in ioctls and and a considerable reduction of stack usage in the bus reset handler tasklet.

Akinobu Mita (1):

      ieee1394: Use hweight32

Stefan Richter (9):
firewire: cdev: fix memory leak in an error path
firewire: normalize style of queue_work wrappers
firewire: cdev: normalize variable names
firewire: optimize config ROM creation
firewire: core: clarify generate_config_rom usage
firewire: core: optimize Topology Map creation
firewire: core: WARN on wrong usage of core transaction functions
firewire: ohci: 0 may be a valid DMA address
firewire: cdev: reduce stack usage by ioctl_dispatch

 drivers/firewire/core-card.c        |   75 ++++++++----------
 drivers/firewire/core-cdev.c        |  113 ++++++++++++++++-----------
 drivers/firewire/core-topology.c    |   17 +++--
 drivers/firewire/core-transaction.c |   19 ++---
 drivers/firewire/core.h             |    9 +-
 drivers/firewire/ohci.c             |   39 ++++++----
 drivers/firewire/sbp2.c             |    9 ++-
 drivers/ieee1394/ohci1394.c         |    8 +--
 include/linux/firewire.h            |   17 +----
 9 files changed, 159 insertions(+), 147 deletions(-)

firedtv branch
==============

This is only a temporary branch with updates of the FireDTV DVB driver. These updates are currently picked up by Mauro and will go into the mainline via the V4L-DVB tree. Most notably, the firedtv driver is now finally ported from the old ieee1394 stack to the new firewire stack. Depending on which one or two of the stacks are chosen in Kconfig, the firedtv driver is built with an interface to one or both of the stacks.

Stefan Richter (11):
firedtv: move remote control workqueue handling into rc source file
firedtv: reform lock transaction backend call
firedtv: add missing include, rename a constant
firedtv: port to new firewire core
firedtv: shrink buffer pointer table
firedtv: packet requeuing is likely to succeed
firedtv: remove an unnecessary function argument
Merge tag 'v2.6.32-rc8' into firedtv
firedtv: do not DMA-map stack addresses
firedtv: remove check for interrupting signal
firedtv: reduce memset()s

 drivers/media/dvb/firewire/Kconfig        |    7 +-
 drivers/media/dvb/firewire/Makefile       |    1 +
 drivers/media/dvb/firewire/firedtv-1394.c |   42 +-
 drivers/media/dvb/firewire/firedtv-avc.c  |  566 +++++++++++----------
 drivers/media/dvb/firewire/firedtv-dvb.c  |   16 +-
 drivers/media/dvb/firewire/firedtv-fw.c   |  376 ++++++++++++++
 drivers/media/dvb/firewire/firedtv-rc.c   |    2 +
 drivers/media/dvb/firewire/firedtv.h      |   23 +-
 8 files changed, 746 insertions(+), 287 deletions(-)
 create mode 100644 drivers/media/dvb/firewire/firedtv-fw.c

And here are the two patches from the "fixes" branch because they weren't posted to LKML yet:

commit 31769cef2e973544164aa7d0db2e2024660d5e21

Author: Jay Fenlason <[email protected]>
Date:   Sat Nov 21 00:05:56 2009 +0100

firewire: ohci: pass correct iso xmit timestamps to core

Here is the final set of patches I used to get ffado to work with the new firewire stack. With these patches, I was able to start ardour and record from and playback to my PreSonus Inspire1394 from a (mostly) Fedora 12 system.

Until now, firewire-ohci exposed only the transmit cycle of the last transmitted packet at each isochronous transmit complete event. This made it impossible for FFADO (FireWire audio drivers in userspace) to synchronize audio-out streams. The fix is to store the timestamp of each packet in the iso xmit event. As a bonus, the transfer status is stored too.

drivers/firewire/ohci.c +20 -5
index c07cfad..94260aa 100644
@@ -1904,15 +1904,30 @@ static int handle_it_packet(struct context *context,
 {
 	struct iso_context *ctx =
 		container_of(context, struct iso_context, context);
+	int i;
+	struct descriptor *pd;
 
-	if (last->transfer_status == 0)
-		/* This descriptor isn't done yet, stop iteration. */
+	for (pd = d; pd <= last; pd++)
+		if (pd->transfer_status)
+			break;
+	if (pd > last)
+		/* Descriptor(s) not done yet, stop iteration */
 		return 0;
 
-	if (le16_to_cpu(last->control) & DESCRIPTOR_IRQ_ALWAYS)
+	i = ctx->header_length;
+	if (i + 4 < PAGE_SIZE) {
+		/* Present this value as big-endian to match the receive code */
+		*(__be32 *)(ctx->header + i) = cpu_to_be32(
+				((u32)le16_to_cpu(pd->transfer_status) << 16) |
+				le16_to_cpu(pd->res_count));
+		ctx->header_length += 4;
+	}
+	if (le16_to_cpu(last->control) & DESCRIPTOR_IRQ_ALWAYS) {
 		ctx->base.callback(&ctx->base, le16_to_cpu(last->res_count),
-				   0, NULL, ctx->base.callback_data);
-
+				   ctx->header_length, ctx->header,
+				   ctx->base.callback_data);
+		ctx->header_length = 0;
+	}
 	return 1;
 }
 

commit 5ed1f321a71b8549cc2eea26c94fe7943ed01d31

Author: Jay Fenlason <[email protected]>
Date:   Tue Nov 17 12:29:17 2009 -0500

firewire: ohci: Make cycleMatch ISO transmission work

Calling the START_ISO ioctl with a nonnegative cycle paramater has never worked. Last night I got around to figuring out why. Most of this patch is a big comment explaining why we enable an interrupt source then don't actually do anything when we get one. As the comment says, we should do more, but we don't have a way to tell userspace what happened. . .

drivers/firewire/ohci.c +15 -1
index 5d52425..c07cfad 100644
@@ -275,7 +275,7 @@ static void log_irqs(u32 evt)
 	    !(evt & OHCI1394_busReset))
 		return;
 
-	fw_notify("IRQ %08x%s%s%s%s%s%s%s%s%s%s%s%s%s\n", evt,
+	fw_notify("IRQ %08x%s%s%s%s%s%s%s%s%s%s%s%s%s%s\n", evt,
 	    evt & OHCI1394_selfIDComplete	? " selfID"		: "",
 	    evt & OHCI1394_RQPkt		? " AR_req"		: "",
 	    evt & OHCI1394_RSPkt		? " AR_resp"		: "",
@@ -286,6 +286,7 @@ static void log_irqs(u32 evt)
 	    evt & OHCI1394_postedWriteErr	? " postedWriteErr"	: "",
 	    evt & OHCI1394_cycleTooLong		? " cycleTooLong"	: "",
 	    evt & OHCI1394_cycle64Seconds	? " cycle64Seconds"	: "",
+	    evt & OHCI1394_cycleInconsistent	? " cycleInconsistent"	: "",
 	    evt & OHCI1394_regAccessFail	? " regAccessFail"	: "",
 	    evt & OHCI1394_busReset		? " busReset"		: "",
 	    evt & ~(OHCI1394_selfIDComplete | OHCI1394_RQPkt |
@@ -293,6 +294,7 @@ static void log_irqs(u32 evt)
 		    OHCI1394_respTxComplete | OHCI1394_isochRx |
 		    OHCI1394_isochTx | OHCI1394_postedWriteErr |
 		    OHCI1394_cycleTooLong | OHCI1394_cycle64Seconds |
+		    OHCI1394_cycleInconsistent |
 		    OHCI1394_regAccessFail | OHCI1394_busReset)
 						? " ?"			: "");
 }
@@ -1439,6 +1441,17 @@ static irqreturn_t irq_handler(int irq, void *data)
 			  OHCI1394_LinkControl_cycleMaster);
 	}
 
+	if (unlikely(event & OHCI1394_cycleInconsistent)) {
+		/*
+		 * We need to clear this event bit in order to make
+		 * cycleMatch isochronous I/O work.  In theory we should
+		 * stop active cycleMatch iso contexts now and restart
+		 * them at least two cycles later.  (FIXME?)
+		 */
+		if (printk_ratelimit())
+			fw_notify("isochronous cycle inconsistent\n");
+	}
+
 	if (event & OHCI1394_cycle64Seconds) {
 		cycle_time = reg_read(ohci, OHCI1394_IsochronousCycleTimer);
 		if ((cycle_time & 0x80000000) == 0)
@@ -1528,6 +1541,7 @@ static int ohci_enable(struct fw_card *card, u32 *config_rom, size_t length)
 		  OHCI1394_reqTxComplete | OHCI1394_respTxComplete |
 		  OHCI1394_isochRx | OHCI1394_isochTx |
 		  OHCI1394_postedWriteErr | OHCI1394_cycleTooLong |
+		  OHCI1394_cycleInconsistent |
 		  OHCI1394_cycle64Seconds | OHCI1394_regAccessFail |
 		  OHCI1394_masterIntEnable);
 	if (param_debug & OHCI_PARAM_DEBUG_BUSRESETS)
Signature · 3 lines
Stefan Richter
-=====-==--= =-== ==---
http://arcgraph.de/sr/