This makes all of a machine's memory accessible to remote debugging via
FireWire, using the physical response unit (i.e. RDMA) of OHCI-1394 link
layer controllers.
This requires actual support by the controller. The only ones currently
known to support it are Agere/LSI FW643. Most if not all other OHCI-1394
controllers do not implement the optional Physical Upper Bound register.
With them, RDMA will continue to be limited to the lowermost 4 GB.
firewire-ohci's startup message in the kernel log is augmented to tell
whether the controller does expose more than 4 GB to RDMA.
While OHCI-1394 allows for a maximum Physical Upper Bound of
0xffff'0000'0000 (near 256 TB), this implementation sets it to
0x8000'0000'0000 (128 TB) in order to avoid interference with applications
that require interrupt-served asynchronous request reception at
respectively low addresses.
Note, this change does not switch remote DMA on. It only increases the
range of remote access to all memory (instead of just 4 GB) whenever
remote DMA was switched on by other means. The latter is achieved by
setting firewire-ohci's remote_dma parameter, or if the physical DMA
filter is opened through firewire-sbp2.
Derived from patch "firewire: Enable physical DMA above 4GB" by
Peter Hurley <peter@hurleysoftware.com> from March 27, 2013.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
This makes it possible to debug kernel over FireWire without the need to
recompile it.
[Stefan R: changed description from "...0" to "...N"]
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Commit 54b2b50c20 "[SCSI] Disable WRITE SAME for RAID and virtual
host adapter drivers" disabled WRITE SAME support for all SBP-2 attached
targets. But as described in the changelog of commit b0ea5f19d3
"firewire: sbp2: allow WRITE SAME and REPORT SUPPORTED OPERATION CODES",
it is not required to blacklist WRITE SAME.
Bring the feature back by reverting the sbp2.c hunk of commit 54b2b50c20.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Cc: stable@kernel.org
Some host adapters do not pass commands through to the target disk
directly. Instead they provide an emulated target which may or may not
accurately report its capabilities. In some cases the physical device
characteristics are reported even when the host adapter is processing
commands on the device's behalf. This can lead to adapter firmware hangs
or excessive I/O errors.
This patch disables WRITE SAME for devices connected to host adapters
that provide an emulated target. Driver writers can disable WRITE SAME
by setting the no_write_same flag in the host adapter template.
[jejb: fix up rejections due to eh_deadline patch]
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Cc: stable@kernel.org
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Use this new function to make code more comprehensible, since we are
reinitialzing the completion, not initializing.
[akpm@linux-foundation.org: linux-next resyncs]
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Acked-by: Linus Walleij <linus.walleij@linaro.org> (personally at LCE13)
Cc: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Put bus_reset_work into its own workqueue. By doing this, forward
progress of bus_reset_work() is guaranteed if the work is switched over
to a rescuer thread.
Switching work to a rescuer thread happens if a new worker thread could
not be allocated in certain time (MAYDAY_INITIAL_TIMEOUT, typically 10
ms). This might not be possible under high memory pressure or even on a
heavily loaded embedded system running a slow serial console.
The former deadlock occured in the following situation:
The rescuer thread ran
fw_device_init->read_config_rom->read_rom->fw_run_transaction.
fw_run_transaction blocked waiting for the completion object.
This completion object would have been completed in bus_reset_work,
but this work was never executed in the rescuer thread due to its
strictly sequential behaviour.
[Stefan R.: Removed WQ_NON_REENTRANT flag from allocation because
it is no longer needed in current kernels. Add it back if you backport
to kernels older than 3.7, i.e. one which does not contain dbf2576e37
"workqueue: make all workqueues non-reentrant". Swapped order of
destroy_workqueue and pci_unregister_driver.]
Signed-off-by: Stephan Gatzka <stephan.gatzka@gmail.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
a) Sort device IDs by vendor -- device -- revision.
b) Write quirk flags in hexadecimal. This affects the user-visible
output of "modinfo firewire-ohci". Since more flags have been added
recently, it is now easier to cope with them in hexadecimal represen-
tation. Besides, the device-specific combination of quirk flags is
shown in hexadecimal in the kernel log too. (And firewire-sbp2
presents its own quirk flags in modinfo as hexadecimals as well.)
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
We have got
struct descriptor *descriptors;
dma_addr_t descriptors_bus;
dma_addr_t buffer_bus;
struct descriptor buffer[0];
void *misc_buffer;
dma_addr_t misc_buffer_bus;
__be32 *config_rom;
dma_addr_t config_rom_bus;
__be32 *next_config_rom;
dma_addr_t next_config_rom_bus;
But then we have got
__le32 *self_id_cpu;
dma_addr_t self_id_bus;
Better apply the pattern of xyz vs. xyz_bus to self_id vs. self_id_bus
as well. The _cpu suffix looks particularly weird in conversions from
little endian to CPU endian.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
An idr related patch introduced the following sparse warning:
drivers/firewire/core-cdev.c:488:33: warning: incorrect type in initializer (different base types)
drivers/firewire/core-cdev.c:488:33: expected bool [unsigned] [usertype] preload
drivers/firewire/core-cdev.c:488:33: got restricted gfp_t
So let's convert from gfp_t bitfield to Boolean explicitly and safely.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
dbf2576e37 ("workqueue: make all workqueues non-reentrant") made
WQ_NON_REENTRANT no-op and the flag is going away. Remove its usages.
This patch doesn't introduce any behavior changes.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Commit 18d627113b (firewire: prevent dropping of completed iso packet
header data) was intended to be an obvious bug fix, but libdc1394 and
FlyCap2 depend on the old behaviour by ignoring all returned information
and thus not noticing that not all packets have been received yet. The
result was that the video frame buffers would be saved before they
contained the correct data.
Reintroduce the old behaviour for old clients.
Tested-by: Stepan Salenikovich <stepan.salenikovich@gmail.com>
Tested-by: Josep Bosch <jep250@gmail.com>
Cc: <stable@vger.kernel.org> # 3.4+
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
After all IEEE 1394 high-level drivers being converted to bus-specific
.probe/.remove methods, remove support of the obsolete generic methods.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
FireWire upper layer drivers are converted from generic
struct driver.probe() and .remove()
to bus-specific
struct fw_driver.probe() and .remove().
The new .probe() adds a const struct ieee1394_device_id *id argument,
indicating the entry in the driver's device identifiers table which
matched the fw_unit to be probed. This new argument is used by the
snd-firewire-speakers driver to look up device-specific parameters and
methods. There is at least one other FireWire audio driver currently in
development in which this will be useful too.
The new .remove() drops the unused error return code.
Although all in-tree drivers are being converted to the new methods,
support for the old methods is left in place in this commit. This
allows public developer trees to merge this commit and then move to the
new fw_driver methods.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Acked-by: Clemens Ladisch <clemens@ladisch.de> (for sound/firewire/)
Cc: Peter Hurley <peter@hurleysoftware.com> (for drivers/staging/fwserial/)
Pull firewure updates from Stefan Richter:
- fix controller removal when controller is in suspended state
- fix video reception on VIA VT6306 with gstreamer, MythTV, and maybe dv4l
- fix a startup issue with Agere/LSI FW643-e2
- error logging improvements and other small updates
* tag 'firewire-updates' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394:
firewire: ohci: dump_stack() for PHY regs read/write failures
firewire: ohci: Improve bus reset error messages
firewire: ohci: Alias dev_* log functions
firewire: ohci: Fix 'failed to read phy reg' on FW643 rev8
firewire: ohci: fix VIA VT6306 video reception
firewire: ohci: Check LPS before register access on pci removal
firewire: ohci: Fix double free_irq()
firewire: remove unnecessary alloc/OOM messages
firewire: sbp2: replace BUG_ON by WARN_ON
firewire: core: remove an always false test
firewire: Remove two unneeded checks for macros
A stack trace is an invaluable tool in determining the basis
and cause of PHY regs read/write failures.
Include PHY reg addr (and value for writes) in the diagnostic.
[Stefan R: changed whitespace]
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Many of the error messages possible from bus_reset_work() do not
contain enough information to distinguish which error condition
occurred nor enough information to evaluate the error afterwards.
Differentiate all error conditions in bus_reset_work(); add
additional information to make error diagnosis possible.
[Stefan R: fixed self-ID endian conversion]
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Convert dev_xxxx(ohci->card.device, ...) log functions to
ohci_xxxx(ohci, ...).
[Stefan R: Peter argues that this increases readability of the code.]
[Stefan R: changed whitespace]
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Add quirk for VT6306 wake bit behavior.
VT6306 seems to reread the wrong descriptor when the wake bit is
written. work around by putting a copy of the branch address in the
first descriptor of the block.
[Stefan R: This fixes the known broken video reception via gstreamer
on VIA VT6306. 100% repeatable testcase:
$ gst-launch-0.10 dv1394src \! dvdemux \! dvdec \! xvimagesink
with a camcorder or other DV source connected. Likewise for MPEG2-TS
reception via gstreamer, e.g. from TV settop boxes.
Perhaps this also fixes dv4l on VT6306, but this is as yet untested.
Kino, dvgrab or FFADO had not been affected by this chip quirk.
Additional comments from Andy:]
I've looked into some problems with the wake bit on a vt6306 family
chip (1106:3044, rev 46).
I used this firewire card in a mythtv setup (ISO receive MPEG2 stream)
with Debian 2.6.32 kernels for ~2 years without problems.
Since upgrading to 3.2, I've been having problems with the input stream
freezing -- input data stops until I restart mythtv (I expect closing
and reopening the device would be sufficient). This happens
infrequently, maybe one out of 20 recordings. I eventually determined
that the problem is more likely to occur if the system is loaded.
I isolated the kernel version as the triggering SW factor and then
specifically the change from dualbuffer back to packet-per-buffer DMA
mode.
The possibility that the controller does not properly respond to the
wake bit was suggested in
https://bugzilla.redhat.com/show_bug.cgi?id=415841, but not proven.
Based on the fact that dualbuffer mode worked while packet-per-buffer
has trouble, I guessed that upon seeing the wake bit written, the vt6306
controller only checks the branch address in the first descriptor of the
block, even if that is not the correct place to look (because the block
has multiple descriptors).
This theory seems to be correct. When the ISO reception is hung, I am
able to resume it by manually writing the branch address to the first
descriptor in the block, and then writing the wake bit.
I've had luck so far with the attached patch, so I'm including it. It's
probably not a complete solution -- I haven't tested transmit modes to
see whether they have a similar issue.
I doubt that the quirk test is any cheaper than just writing the extra
branch address in all cases, but it does reduce the risk of breaking
other hardware.
[Stefan R: omitted QUIRK_NO_MSI from VT6306 quirks table entry,
changed whitespace]
Signed-off-by: Andy Leiserson <andy@leiserson.org>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
A pci device can be removed while in its suspended state. If the ohci
host controller is suspended, the PHY is also in low-power mode and
LPS is disabled. If LPS is disabled, most of the host registers aren't
accessible, including IntMaskClear. Furthermore, access to these registers
when LPS is disabled can cause hard lockups on some hardware. Since
interrupts are already disabled in this mode, further action is
unnecessary.
Test LPS before attempting to write IntMaskClear to disable interrupts.
[Stefan R: whitespace changes]
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
A pci device can be removed while in its suspended state.
Because the ohci driver freed the irq to suspend, free_irq() is
called twice; once from pci_remove() and again from pci_suspend(),
which issues the warning below [1].
Rather than allocate the irq in the .enable() path, move the
allocation to .probe(). Consequently, the irq is not reallocated
upon pci_resume() and thus is not freed upon pci_suspend().
[1] Warning reported by Mark Einon <mark.einon@gmail.com> when
suspending an MSI MS-1727 GT740 laptop on Ubuntu 3.5.0-22-generic
WARNING: at ./kernel/irq/manage.c:1198 __free_irq+0xa3/0x1e0()
Hardware name: MS-1727
Trying to free already-free IRQ 16
Modules linked in: ip6table_filter ip6_tables ebtable_nat ebtables <...snip...>
Pid: 4, comm: kworker/0:0 Tainted: P O 3.5.0-22-generic #34-Ubuntu
Call Trace:
[<ffffffff81051c1f>] warn_slowpath_common+0x7f/0xc0
[<ffffffff81051d16>] warn_slowpath_fmt+0x46/0x50
[<ffffffff8103fa39>] ? default_spin_lock_flags+0x9/0x10
[<ffffffff810df6b3>] __free_irq+0xa3/0x1e0
[<ffffffff810df844>] free_irq+0x54/0xc0
[<ffffffffa005a27e>] pci_remove+0x6e/0x210 [firewire_ohci]
[<ffffffff8135ae7f>] pci_device_remove+0x3f/0x110
[<ffffffff8141fdbc>] __device_release_driver+0x7c/0xe0
[<ffffffff8141fe4c>] device_release_driver+0x2c/0x40
[<ffffffff8141f5f1>] bus_remove_device+0xe1/0x120
[<ffffffff8141cd1a>] device_del+0x12a/0x1c0
[<ffffffff8141cdc6>] device_unregister+0x16/0x30
[<ffffffff81354784>] pci_stop_bus_device+0x94/0xa0
[<ffffffffa0091c67>] acpiphp_disable_slot+0xb7/0x1a0 [acpiphp]
[<ffffffffa0090716>] ? get_slot_status+0x46/0xc0 [acpiphp]
[<ffffffffa0091d7d>] acpiphp_check_bridge.isra.15+0x2d/0xf0 [acpiphp]
[<ffffffffa0092442>] _handle_hotplug_event_bridge+0x372/0x4d0 [acpiphp]
[<ffffffff81390f8c>] ? acpi_os_execute_deferred+0x2f/0x34
[<ffffffff8116e22d>] ? kfree+0xed/0x110
[<ffffffff8107086a>] process_one_work+0x12a/0x420
[<ffffffffa00920d0>] ? _handle_hotplug_event_func+0x1d0/0x1d0 [acpiphp]
[<ffffffff8107141e>] worker_thread+0x12e/0x2f0
[<ffffffff810712f0>] ? manage_workers.isra.26+0x200/0x200
[<ffffffff81075f13>] kthread+0x93/0xa0
[<ffffffff8168d024>] kernel_thread_helper+0x4/0x10
[<ffffffff81075e80>] ? kthread_freezable_should_stop+0x70/0x70
[<ffffffff8168d020>] ? gs_change+0x13/0x13
Reported-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
No need to crash and burn if S/G element sizes cannot be set to our
liking; just leave a message in the log.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
