ieee1394: move init_ohci1394_dma to drivers/firewire/
Subsystems (2)
PATCH 1/2 ieee1394: move init_ohci1394_dma to drivers/firewire/
because drivers/ieee1394/ will be deleted.
Additional changes:
- add some #include directives
- adjust to use firewire/ohci.h instead of ieee1394/ohci1394.h, replace struct ti_ohci by a minimal struct ohci, replace quadlet_t from ieee1394_types.h by u32
- two or three trivial stylistic changes
- __iomem annotation
---
drivers/firewire/Makefile | 1 +
drivers/{ieee1394 => firewire}/init_ohci1394_dma.c | 77 +++++++++++++-------
drivers/ieee1394/Makefile | 2 -
3 files changed, 52 insertions(+), 28 deletions(-)
rename drivers/{ieee1394 => firewire}/init_ohci1394_dma.c (84%)drivers/firewire/Makefile +1 -0
@@ -13,3 +13,4 @@ obj-$(CONFIG_FIREWIRE_OHCI) += firewire-ohci.o obj-$(CONFIG_FIREWIRE_SBP2) += firewire-sbp2.o obj-$(CONFIG_FIREWIRE_NET) += firewire-net.o obj-$(CONFIG_FIREWIRE_NOSY) += nosy.o +obj-$(CONFIG_PROVIDE_OHCI1394_DMA_INIT) += init_ohci1394_dma.o
drivers/firewire/init_ohci1394_dma.c renamed from drivers/ieee1394/init_ohci1394_dma.c +51 -26
185 lines not on this page · raw message
drivers/ieee1394/Makefile +0 -2
@@ -14,5 +14,3 @@ obj-$(CONFIG_IEEE1394_RAWIO) += raw1394.o obj-$(CONFIG_IEEE1394_SBP2) += sbp2.o obj-$(CONFIG_IEEE1394_DV1394) += dv1394.o obj-$(CONFIG_IEEE1394_ETH1394) += eth1394.o - -obj-$(CONFIG_PROVIDE_OHCI1394_DMA_INIT) += init_ohci1394_dma.o
Signature · 7 lines
1.7.2.2 -- Stefan Richter -=====-==-=- =-=- -=--= http://arcgraph.de/sr/
PATCH 2/2 ieee1394: remove the old IEEE 1394 driver stack
The drivers
- ohci1394 (controller driver)
- ieee1394 (core)
- dv1394, raw1394, video1394 (userspace ABI)
- eth1394, sbp2 (protocol drivers)
are replaced by
- firewire-ohci (controller driver)
- firewire-core (core and userspace ABI)
- firewire-net, firewire-sbp2 (protocol drivers)
which are more featureful, better performing, and more secure than the older drivers; all with a smaller and more modern code base.
The driver firedtv in drivers/media/dvb/firewire/ contains backends to both ieee1394 and firewire-core. Its ieee1394 backend code can be removed in an independent commit; firedtv as-is builds and works fine without ieee1394.
The driver pcilynx (another controller driver) is deleted without replacement since PCILynx cards are exteremely rare. Owners of these cards use them with the stand-alone bus sniffer driver nosy instead.
The drivers nosy and init_ohci1394_dma which do not interact with either of the two IEEE 1394 stacks are not affected by the ieee1394 subsystem removal.
There are still some issues with the newer firewire subsystem compared to the older one:
- The rare and quirky controllers ALi M52xx, Apple UniNorth v1, NVIDIA NForce2 are even less well supported by firewire-ohci than by ohci1394. I am looking into the M52xx issue.
- The experimental firewire-net is reportedly less stable than its experimental cousin eth1394.
- Audio playback of a certain group of audio devices (ones based on DICE chipset with EAP; supported by prerelease FFADO code) does not work yet. This issue is still under investigation.
- There were some ieee1394 based out-of-the-mainline drivers. Of them, only lisight, an audio driver for iSight webcams, seems still useful. Work is underway to reimplement it on top of firewire-core.
All these remainig issues are minor; they should not stand in the way of overall better user experience of IEEE 1394 on Linux, together with a reduction in support efforts and maintenance burden. The coexistence of two IEEE 1394 kernel driver stacks in the mainline since 2.6.22 shall end now, as announced earlier this year.
--- Documentation/ABI/obsolete/dv1394 | 9 - Documentation/ABI/removed/dv1394 | 14 + Documentation/ABI/removed/raw1394 | 15 + .../ABI/removed/raw1394_legacy_isochronous | 16 - Documentation/ABI/removed/video1394 | 16 + Documentation/feature-removal-schedule.txt | 10 - drivers/Makefile | 1 - drivers/firewire/Kconfig | 5 - drivers/ieee1394/Kconfig | 182 - drivers/ieee1394/Makefile | 16 - drivers/ieee1394/config_roms.c | 194 -- drivers/ieee1394/config_roms.h | 19 - drivers/ieee1394/csr.c | 843 ----- drivers/ieee1394/csr.h | 99 - drivers/ieee1394/csr1212.c | 1467 -------- drivers/ieee1394/csr1212.h | 383 --- drivers/ieee1394/dma.c | 289 -- drivers/ieee1394/dma.h | 89 - drivers/ieee1394/dv1394-private.h | 587 ---- drivers/ieee1394/dv1394.c | 2584 -------------- drivers/ieee1394/dv1394.h | 305 -- drivers/ieee1394/eth1394.c | 1736 ---------- drivers/ieee1394/eth1394.h | 234 -- drivers/ieee1394/highlevel.c | 691 ---- drivers/ieee1394/highlevel.h | 141 - drivers/ieee1394/hosts.c | 249 -- drivers/ieee1394/hosts.h | 201 -- drivers/ieee1394/ieee1394-ioctl.h | 106 - drivers/ieee1394/ieee1394.h | 220 -- drivers/ieee1394/ieee1394_core.c | 1380 -------- drivers/ieee1394/ieee1394_core.h | 172 - drivers/ieee1394/ieee1394_hotplug.h | 19 - drivers/ieee1394/ieee1394_transactions.c | 595 ---- drivers/ieee1394/ieee1394_transactions.h | 40 - drivers/ieee1394/ieee1394_types.h | 69 - drivers/ieee1394/iso.c | 568 ---- drivers/ieee1394/iso.h | 195 -- drivers/ieee1394/nodemgr.c | 1901 ----------- drivers/ieee1394/nodemgr.h | 186 - drivers/ieee1394/ohci1394.c | 3590 -------------------- drivers/ieee1394/ohci1394.h | 453 --- drivers/ieee1394/pcilynx.c | 1554 --------- drivers/ieee1394/pcilynx.h | 468 --- drivers/ieee1394/raw1394-private.h | 81 - drivers/ieee1394/raw1394.c | 3096 ----------------- drivers/ieee1394/raw1394.h | 191 -- drivers/ieee1394/sbp2.c | 2138 ------------ drivers/ieee1394/sbp2.h | 346 -- drivers/ieee1394/video1394.c | 1528 --------- drivers/ieee1394/video1394.h | 67 - 50 files changed, 45 insertions(+), 29313 deletions(-) delete mode 100644 Documentation/ABI/obsolete/dv1394 create mode 100644 Documentation/ABI/removed/dv1394 create mode 100644 Documentation/ABI/removed/raw1394 delete mode 100644 Documentation/ABI/removed/raw1394_legacy_isochronous create mode 100644 Documentation/ABI/removed/video1394 delete mode 100644 drivers/ieee1394/Kconfig delete mode 100644 drivers/ieee1394/Makefile delete mode 100644 drivers/ieee1394/config_roms.c delete mode 100644 drivers/ieee1394/config_roms.h delete mode 100644 drivers/ieee1394/csr.c delete mode 100644 drivers/ieee1394/csr.h delete mode 100644 drivers/ieee1394/csr1212.c delete mode 100644 drivers/ieee1394/csr1212.h delete mode 100644 drivers/ieee1394/dma.c delete mode 100644 drivers/ieee1394/dma.h delete mode 100644 drivers/ieee1394/dv1394-private.h delete mode 100644 drivers/ieee1394/dv1394.c delete mode 100644 drivers/ieee1394/dv1394.h delete mode 100644 drivers/ieee1394/eth1394.c delete mode 100644 drivers/ieee1394/eth1394.h delete mode 100644 drivers/ieee1394/highlevel.c delete mode 100644 drivers/ieee1394/highlevel.h delete mode 100644 drivers/ieee1394/hosts.c delete mode 100644 drivers/ieee1394/hosts.h delete mode 100644 drivers/ieee1394/ieee1394-ioctl.h delete mode 100644 drivers/ieee1394/ieee1394.h delete mode 100644 drivers/ieee1394/ieee1394_core.c delete mode 100644 drivers/ieee1394/ieee1394_core.h delete mode 100644 drivers/ieee1394/ieee1394_hotplug.h delete mode 100644 drivers/ieee1394/ieee1394_transactions.c delete mode 100644 drivers/ieee1394/ieee1394_transactions.h delete mode 100644 drivers/ieee1394/ieee1394_types.h delete mode 100644 drivers/ieee1394/iso.c delete mode 100644 drivers/ieee1394/iso.h delete mode 100644 drivers/ieee1394/nodemgr.c delete mode 100644 drivers/ieee1394/nodemgr.h delete mode 100644 drivers/ieee1394/ohci1394.c delete mode 100644 drivers/ieee1394/ohci1394.h delete mode 100644 drivers/ieee1394/pcilynx.c delete mode 100644 drivers/ieee1394/pcilynx.h delete mode 100644 drivers/ieee1394/raw1394-private.h delete mode 100644 drivers/ieee1394/raw1394.c delete mode 100644 drivers/ieee1394/raw1394.h delete mode 100644 drivers/ieee1394/sbp2.c delete mode 100644 drivers/ieee1394/sbp2.h delete mode 100644 drivers/ieee1394/video1394.c delete mode 100644 drivers/ieee1394/video1394.h
Documentation/ABI/obsolete/dv1394 deleted +0 -9
@@ -1,9 +0,0 @@ -What: dv1394 (a.k.a. "OHCI-DV I/O support" for FireWire) -Contact: [email protected] -Description: - New application development should use raw1394 + userspace libraries - instead, notably libiec61883 which is functionally equivalent. - -Users: - ffmpeg/libavformat (used by a variety of media players) - dvgrab v1.x (replaced by dvgrab2 on top of raw1394 and resp. libraries)
Documentation/ABI/removed/dv1394 new file +14 -0
@@ -0,0 +1,14 @@ +What: dv1394 (a.k.a. "OHCI-DV I/O support" for FireWire) +Date: May 2010 (scheduled), finally removed in kernel v2.6.37 +Contact: [email protected] +Description: + /dev/dv1394/* were character device files, one for each FireWire + controller and for NTSC and PAL respectively, from which DV data + could be received by read() or transmitted by write(). A few + ioctl()s allowed limited control. + This special-purpose interface has been superseded by libraw1394 + + libiec61883 which are functionally equivalent, support HDV, and + transparently work on top of the newer firewire kernel drivers. + +Users: + ffmpeg/libavformat (if configured for DV1394)
Documentation/ABI/removed/raw1394 new file +15 -0
@@ -0,0 +1,15 @@ +What: raw1394 (a.k.a. "Raw IEEE1394 I/O support" for FireWire) +Date: May 2010 (scheduled), finally removed in kernel v2.6.37 +Contact: [email protected] +Description: + /dev/raw1394 was a character device file that allowed low-level + access to FireWire buses. Its major drawbacks were its inability + to implement sensible device security policies, and its low level + of abstraction that required userspace clients do duplicate much + of the kernel's ieee1394 core functionality. + Replaced by /dev/fw*, i.e. the <linux/firewire-cdev.h> ABI of + firewire-core. + +Users: + libraw1394 (works with firewire-cdev too, transparent to library ABI + users)
Documentation/ABI/removed/raw1394_legacy_isochronous deleted +0 -16
@@ -1,16 +0,0 @@ -What: legacy isochronous ABI of raw1394 (1st generation iso ABI) -Date: June 2007 (scheduled), removed in kernel v2.6.23 -Contact: [email protected] -Description: - The two request types RAW1394_REQ_ISO_SEND, RAW1394_REQ_ISO_LISTEN have - been deprecated for quite some time. They are very inefficient as they - come with high interrupt load and several layers of callbacks for each - packet. Because of these deficiencies, the video1394 and dv1394 drivers - and the 3rd-generation isochronous ABI in raw1394 (rawiso) were created. - -Users: - libraw1394 users via the long deprecated API raw1394_iso_write, - raw1394_start_iso_write, raw1394_start_iso_rcv, raw1394_stop_iso_rcv - - libdc1394, which optionally uses these old libraw1394 calls - alternatively to the more efficient video1394 ABI
Documentation/ABI/removed/video1394 new file +16 -0
@@ -0,0 +1,16 @@ +What: video1394 (a.k.a. "OHCI-1394 Video support" for FireWire) +Date: May 2010 (scheduled), finally removed in kernel v2.6.37 +Contact: [email protected] +Description: + /dev/video1394/* were character device files, one for each FireWire + controller, which were used for isochronous I/O. It was added as an + alternative to raw1394's isochronous I/O functionality which had + performance issues in its first generation. Any video1394 user had + to use raw1394 + libraw1394 too because video1394 did not provide + asynchronous I/O for device discovery and configuration. + Replaced by /dev/fw*, i.e. the <linux/firewire-cdev.h> ABI of + firewire-core. + +Users: + libdc1394 (works with firewire-cdev too, transparent to library ABI + users)
Documentation/feature-removal-schedule.txt +0 -10
@@ -637,16 +637,6 @@ Who: Thomas Gleixner <[email protected]> ---------------------------- -What: old ieee1394 subsystem (CONFIG_IEEE1394) -When: 2.6.37 -Files: drivers/ieee1394/ except init_ohci1394_dma.c -Why: superseded by drivers/firewire/ (CONFIG_FIREWIRE) which offers more - features, better performance, and better security, all with smaller - and more modern code base -Who: Stefan Richter <[email protected]> - ----------------------------- - What: The acpi_sleep=s4_nonvs command line option When: 2.6.37 Files: arch/x86/kernel/acpi/sleep.c
drivers/Makefile +0 -1
@@ -51,7 +51,6 @@ obj-y += net/ obj-$(CONFIG_ATM) += atm/ obj-$(CONFIG_FUSION) += message/ obj-y += firewire/ -obj-y += ieee1394/ obj-$(CONFIG_UIO) += uio/ obj-y += cdrom/ obj-y += auxdisplay/
drivers/firewire/Kconfig +0 -5
@@ -3,9 +3,6 @@ menu "IEEE 1394 (FireWire) support" # firewire-core does not depend on PCI but is # not useful without PCI controller driver -comment "You can enable one or both FireWire driver stacks." -comment "The newer stack is recommended." - config FIREWIRE tristate "FireWire driver stack" select CRC_ITU_T
@@ -64,8 +61,6 @@ config FIREWIRE_NET To compile this driver as a module, say M here: The module will be called firewire-net. -source "drivers/ieee1394/Kconfig" - config FIREWIRE_NOSY tristate "Nosy - a FireWire traffic sniffer for PCILynx cards" depends on PCI
drivers/ieee1394/Kconfig deleted +0 -0
[...useless all-minus diff hunks deleted...]
Wasn't there a git option to create a shorter diff without all this noise?
Signature · 3 lines
Stefan Richter -=====-==-=- =-=- -=-=- http://arcgraph.de/sr/
Re: [PATCH 2/2] ieee1394: remove the old IEEE 1394 driver stack
Quoted 11 lines · Stefan Richter
The drivers
- ohci1394 (controller driver)
- ieee1394 (core)
- dv1394, raw1394, video1394 (userspace ABI)
- eth1394, sbp2 (protocol drivers)
are replaced by
- firewire-ohci (controller driver)
- firewire-core (core and userspace ABI)
- firewire-net, firewire-sbp2 (protocol drivers)
which are more featureful, better performing, and more secure than the older drivers; all with a smaller and more modern code base.
But new stack doesn't have working networking support...
Best regards,
Maxim Levitsky
Re: [PATCH 2/2] ieee1394: remove the old IEEE 1394 driver stack
Quoted 12 lines · Maxim Levitsky
Quoted 11 lines · Stefan Richter
The drivers
- ohci1394 (controller driver)
- ieee1394 (core)
- dv1394, raw1394, video1394 (userspace ABI)
- eth1394, sbp2 (protocol drivers)
are replaced by
- firewire-ohci (controller driver)
- firewire-core (core and userspace ABI)
- firewire-net, firewire-sbp2 (protocol drivers)
which are more featureful, better performing, and more secure than the older drivers; all with a smaller and more modern code base.
Quoted 1 line
But new stack doesn't have working networking support...
firewire-net did work for me to /some/ degree when I tested it last: Transfer s via FTP or SCP command line clients was OK, even with huge files, but an attempt to use FTP via desktop file manager ended in kernel crash.
In contrast, eth1394 worked to /some/ other degree for me when I tested it
last: It never crashed, but it had irregular performance (due to DMA context program overflow, besides ieee1394's tlabel recycling in a wrong context) when paired with {Linux,Windows,OS X}/x86, whereas transfers from and to an OS X/PPC peer always failed due to data corruption.
That's what I meant with
Quoted 2 lines
Quoted 2 lines
- The experimental firewire-net is reportedly less stable than its experimental cousin eth1394.
Maybe these words were too kind to firewire-net.
eth1394 was never stabilized because there was nobody there to do so. Will somebody be there to stabilize firewire-net? A while ago I spent the time to fix an SMP bug in firewire-net. Lately I looked a bit into the currently known crash bug but obviously did not resolve it yet. At the moment I suspect a race between fw_card_driver.send_request and fw_card_driver.cancel_packet but did not find a hole in that code yet.
Overall I am not convinced that this or any of the other open issues that I mentioned warrants to stretch out the coexistence of two 1394 kernel stacks further. Many distributors still enable only the old, buggy, virtually unmaintained stack. Some distributors enable both stacks but are not aware that they then should also provide a modprobe blacklist file, and their users end up running a random set of drivers. Until 2.6.36-rc4 inclusive that usually seemed to be the old stack; from 2.6.36-rc5 onwards that randomness will probably be skewed towards the new stack. There is only one way to fix that only seemingly trivial deployment issue.
Signature · 3 lines
Stefan Richter -=====-==-=- =-=- -===- http://arcgraph.de/sr/
Re: [PATCH 2/2] ieee1394: remove the old IEEE 1394 driver stack
Quoted 19 lines · Stefan Richter
Quoted 12 lines · Maxim Levitsky
Quoted 11 lines · Stefan Richter
The drivers
- ohci1394 (controller driver)
- ieee1394 (core)
- dv1394, raw1394, video1394 (userspace ABI)
- eth1394, sbp2 (protocol drivers)
are replaced by
- firewire-ohci (controller driver)
- firewire-core (core and userspace ABI)
- firewire-net, firewire-sbp2 (protocol drivers)
which are more featureful, better performing, and more secure than the older drivers; all with a smaller and more modern code base.Quoted 1 line
But new stack doesn't have working networking support...
firewire-net did work for me to /some/ degree when I tested it last: Transfer s via FTP or SCP command line clients was OK, even with huge files, but an attempt to use FTP via desktop file manager ended in kernel crash.
Quoted 5 lines
In contrast, eth1394 worked to /some/ other degree for me when I tested it
last: It never crashed, but it had irregular performance (due to DMA context program overflow, besides ieee1394's tlabel recycling in a wrong context) when paired with {Linux,Windows,OS X}/x86, whereas transfers from and to an OS X/PPC peer always failed due to data corruption.
Here transfers new stack <-> new stack fail at around 250 KB due to corrupted packets. (tested via SFTP and GUI).
But anyway, I can't agree more with what you said.
So go ahead and remove the old stack.
But maybe you need to move the firewire-net driver to staging? It really doesn't work at all, this way or around.
EXPEREMENTAL today doesn't mean much.
Best regards,
Maxim Levitsky
Quoted 4 lines · Stefan Richter
Additional changes:
- add some #include directives
- adjust to usefirewire/ohci.hinstead ofieee1394/ohci1394.h, replace structti_ohciby a minimal struct ohci,
[...]
drivers/firewire/init_ohci1394_dma.c line 52
Quoted 4 lines
drivers/firewire/init_ohci1394_dma.c renamed from drivers/ieee1394/init_ohci1394_dma.c +4
-0+struct ohci { +struct pci_dev *dev; +void __iomem *registers; +};
The dev member is actually unused to. I'll remove it but keep the now rather useless struct ohci so that the code stays similar to ohci.c.
Signature · 3 lines
Stefan Richter -=====-==-=- =-=- -=-=- http://arcgraph.de/sr/