Age | Commit message (Collapse) | Author |
|
Commit de47725, first released in 3.2-rc1 removed module.h from some
kernel headers. Include it explicitly now.
Resolves compilation errors like:
error: implicit declaration of function 'try_module_get'
error: 'THIS_MODULE' undeclared (first use in this function)
error: implicit declaration of function 'module_put'
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Origin: http://svnview.digium.com/svn/dahdi?view=rev&rev=10361
Conflicts:
include/dahdi/kernel.h
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/branches/2.5@10363 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
r10066 "wctdm24xxp, wcte12xp: Run the ISR with interrupts disabled."
requested that the interrupt handler be run in "fast" mode (disabled)
but this isn't necessarily guaranteed.
This patch makes the interrupt handler itself disable all the interrupts.
Linux commit 470c66239ef0336429b35345f3f615d47341e13b [1] contains a comment
about why this is necessary.
[1] http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=470c66239ef03364
(closes issue DAHLIN-248)
Reported-and-Tested-by: Vladimir Mikhelson <vlad@mikhelson.com>
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Origin: http://svnview.digium.com/svn/dahdi?view=rev&rev=10118
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/branches/2.5@10119 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
DAHDI currently supports kernels >= 2.6.9. netdev_priv() has been in the
mainline kernel since versions 2.6.6 so it's available in all the
supported kernels. This change is needed to compile against the 3.1 kernel.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Origin: http://svnview.digium.com/svn/dahdi?view=rev&rev=10096
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/branches/2.5@10115 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Revision 9886, "wcte12xp: Use the in-hardirq versions of
dahdi_receive/dahdi_transmit", changed the call into dahdi_receive and
dahdi_transmit to use versions that assume local interrupts are already
disabled. Not all versions of the kernel run interrupt service routines with
all interrupts disabled and therefore it was possible to lock up a CPU with a
recursive grab of the chan_lock.
When LOCKDEP was enabled (on debug kernels) interrupt handlers were run
atomically so this problem would only occur on pre 2.6.35 kernels that did not
have lockdep enabled.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10066 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
The firmware load has been moved into a workqueue to prevent the module load
from blocking for the duration of the firmware upload. This could be up to 40
seconds. Driver prevents configuration until firmware load is finished and
is_initialized() returns true.
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9998 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Support enabled for the vpmoct032 echo cancellation module for the wctdm24xxp
and wcte12xp drivers.
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9997 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Cuts down on the overhead of constantly saving and restoring the
interrupt registers.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9962 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Part of increasing system startup speed.
Splitting these two operations facilitate checking if there is a module
present synchronously on driver load from the actual load of the firmware and
configuration of the channels.
This will allow the presence of the VPM module to be flagged on the span
before registration, but load and configuration can happen in the background.
When the modules are eventually loaded via udev, there will be enough time
from the time the drivers are loaded to when dahdi_cfg will run to complete
the firmware load, eliminating the need to block the driver here.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9951 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
In my opinion naming the VPM workqueues for each board is not worth the
extra complexity.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9950 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
It is only ever accessed in interrupt context anyway. Saves several
hundred nanoseconds from hard interrupt context.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Michael Spiceland <mspiceland@digium.com>
Acked-by: Kinsey Moore <kmoore@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9888 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
No need for the CPU to wait there for that write to flush when all the
writes will be flushed when the interrupt is acknoledged later. Saves
nearly a 1 us from the interrupt handler on a 3Ghz Xeon test system.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Michael Spiceland <mspiceland@digium.com>
Acked-by: Kinsey Moore <kmoore@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9887 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Allows the board drivers to use the regular locks instead of the
irqsave/irqrestore variants.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Michael Spiceland <mspiceland@digium.com>
Acked-by: Kinsey Moore <kmoore@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9885 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
voicebus_quiesce is like voicebus_stop, except that it doesn't wait for any
information to come back from the card. This prevents kexec from blocking
waiting for information from a potentially dead card.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Michael Spiceland <mspiceland@digium.com>
Acked-by: Kinsey Moore <kmoore@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9880 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
r9402 switched to dma_pool for the SFRAMES. This updates the network
debug device to use the dma_pools for SFRAME allocation.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Michael Spiceland <mspiceland@digium.com>
Acked-by: Kinsey Moore <kmoore@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9877 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Since there never was a VPMADT032 module shipped with non-volatile
memory no need to check that bit contantly in interrupt context. Also
remove the unused VPM150M_DTMFDETECT bit.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9781 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9779 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
This is a cleanup patch to make it a little easier to see what the bits
mean.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Kinsey Moore <kmoore@digium.com>
Acked-By: Russ Meyerriecks <rmeyerriecks@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9543 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
There are some platforms where the read-line multiple transaction causes
packets to be dropped in the voicebus pipeline. The only observable
behavior is that packets just go "missing" in the pipeline. This also
only appears to affect PCI cards.
A typical 'symptom' of this problem is you may see IRQ misses increasing
without any corresponding "bumps" in latency in the kernel message log.
Normally, IRQ misses are correlated to latency bumps since that is an
indication that the host was not able to service the card interrupt in a
timely fashion. DAHDI-510 DAHDI-774
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Acked-by: Kinsey Moore <kmoore@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9542 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
rwlock is slower than normal spinlocks and this lock is rarely contended
for. Also noticed that the vpmadt032_module_init function is now (was
already) redundant since all the elements initialized in it were already
initialized.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Kinsey Moore <kmoore@digium.com>
Acked-By: Russ Meyerriecks <rmeyerriecks@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9538 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
This reduces the chance that another interrupt will interfere with the
recovery process. Otherwise it is possible that the hardware advances
past the position that we think it is currently at.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Kinsey Moore <kmoore@digium.com>
Acked-By: Russ Meyerriecks <rmeyerriecks@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9535 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Reading 0xffffffff from the registers is a different error than just not
coming out of reset. Add a little extra debugging information.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Kinsey Moore <kmoore@digium.com>
Acked-By: Russ Meyerriecks <rmeyerriecks@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9534 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
There were some routes through the failure paths in __voicebus_init() where a
registered memory region was not subsequently released. This change closes
those paths.
The result would be on subsequent loads of the driver after hitting the
failure condition you would see "IO Registers are in use by another module."
in dmesg.
request_mem_region/release_mem_region should most likely be converted to
devm_request_region and devm_release_region introduced in 2.6.20
(commit 9ac7849e35f705830f7b016ff272b0ff1f7ff759) which was introduced for
reasons just such as this.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9503 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Using semaphores as mutexes was removed from the kernel in 4882720b267b.
Just use straight semaphores now. 'DECLARE_MUTEX()' -> 'DEFINE_SEMAPHORE()'
and 'init_MUTEX()' -> 'sema_init()'.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
LKML-Reference: <20100907125057.562399240@linutronix.de>
Acked-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9464 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Setting this configuration option would not have had any impact when a
hardware echo canceler was in use.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Kinsey Moore <kmoore@digium.com>
Acked-by: Russ Meyerriecks <rmeyerriecks@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9432 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
The voicebus interface constantly uses the same buffers for moving data
to/from the card. A fixed pool has less overhead in this case than
constantly mapping/unmapping the kmem_cache_alloced buffers.
Saves about ~2.608 us for each run of vb_isr on a 2.40 GHz Xeon test
machine.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Russ Meyerriecks <rmeyerriecks@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9402 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
This change was made in revision 8176.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Russ Meyerriecks <rmeyerriecks@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9401 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
The voicebus library by default configures the PCI interface to poll the
descriptor ring for available buffers. There are some platforms like
the Intel SG3420P motherboard where this does not appear to be
sufficient. Writing to the transmit demand poll register resolves this
problem on these troublesome platforms. DAHDI-700 DAHDI-702.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9397 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
It is possible to softlock if the board stops delivering interrupts in
the middle of a reset.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9332 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Magic number removal.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9327 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
I have yet to personally come across a system where this actually
changes the observable behavior, but it certainly seems like the sane
thing to do and I would rather not let this float around as a patch when
I can just merge it in.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9326 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
These three bits are already set by 'pci_enable_device' and
'pci_set_master' calls.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9325 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9319 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Add compile time option to improve the reference signal provided to
software echo cancelers. The intent here is for this functionality to
become the default behavior but more testing and work on the edge cases
is needed. It's being brought in now as a compile time option since there
have been reports that it helps in some environments.
Instead of using two buffers, which means that at best we're two
milliseconds behind, use a circular buffer where audio data is written
on the transmit side and read on the receive path. In this way high
latency values will not interfere with the operation of software
echo cancelers. DAHDI-291. DAHDI-387.
This work was originally on:
http://svn.asterisk.org/svn/dahdi/linux/team/sruffell/improved_ecreference@9143
and includes a generic kfifo replacement by Matt Fredrickson.
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9144 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@9046 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
The echo canceler will now monitor if the receive signal goes over a
certain threshold, and if so, freezes its adaptation to prevent loss of
convergence. I.e. Fixes conditions where blowing into your handset
could result in echo. Additionally, 1.25 includes improvements for
handling when line conditions change from echo free to containing echo.
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8983 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
In revision 8095, I had moved most of the buffer processing out of the
interrupt handler and into a tasklet. The intended result was to enable
multiple cards to interleave with one another. But once again I was
bitten by the fact that there are some systems that for one reason or
another do not process their tasklets in a timely enough manner for the
real-time nature of TDM processing. This commit moves this processing
back into the interrupt handler by default. It also limits the number
of frames that the interrupt handler will process at any given time
which appears to achieve the same intended result.
(closes issue #17289)
Tested by: alecdavis
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8982 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Will add an option to allow calling the deferred processing callback directly
in the interrupt handler. It appears there are some systems which still are
unable to process their tasklets in a timely fashion, especially if they get
pushed out to the ksoftirqd daemon.
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8981 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Make the drivers a little more kexec friendly.
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8979 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
gcc 4.5.0 generates a warning on the changed lines and
http://gcc.gnu.org/ml/gcc/2004-10/msg00032.html explains why.
Essentially, the only thing guaranteed with the preincrement operator is that
the value will be incremented before the assignment. It's undefined where in
the sequence the mask will be applied.
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8832 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
With kernel version >= 2.6.35-rc1, building failed with:
drivers/dahdi/voicebus/GpakCust.h:114: error: field ‘sem’ has incomplete type
(closes issue #17382)
Reported by: Chainsaw
Patches:
07-semaphore-include-not-optional.diff uploaded by Chainsaw (license 723)
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8758 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
CheckDspReset can return -1 if the DSP is not ready to process any new
commands. In this case we should retry a few times to give the DSP a chance
to become ready. While I'm not ready to say this definitely fixes recently
reported cases when the wcte12xp driver constantly resets, it eliminated
communication failures to the DSP module when under stress (via the
vpm_firmware_version sysfs attribute). However, I haven't let it run long
enough to say that the issue is resolved. DAHDI-603.
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8576 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
If the sleep is ever interrupted, 'up' will still be called in the GpakApi,
essentially making the lock useless after that point.
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8575 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
In revision 8176 I changed register access from I/O space to memory mapped
registers. Unfortunately, when I made that change, I didn't account for
posted writes. This change makes sure all the registers are read back to
ensure that they are posted through any intermediate bridges.
The most readily observable symptom were cards that were taking 2000
interrupts/second. The card reported that it handled an interrupt but the
write to silence the card wasn't flushed through until the second time the
interrupt handler run. DAHDI-602.
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8560 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
WARN_ON_ONCE is not defined in 2.6.9, and this condition would be catastrophic
anyway if it were to occur.
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8494 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Also, fall back to any software echocan configured for this channel for new
calls while we're in the middle of a recovery.
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8473 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
When the transmit descriptor runs out completely, there (appears to be) a
chance for a random command to be sent that results in the VPMADT032 to no
longer respond, typically resulting in one way audio. This change introduces
a poll of the VPM. If it fails the poll, it will be bypassed temporarily
while the driver resets and reprograms it. Also, the VPM is initially
programmed in the spanconfig callback instead of at driver load. This moves
the potential for underruns until later in the boot process. DAHDI-573.
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8468 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
This will make us two behind, which is fine, and eliminates a busy loop in
atomic context.
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8467 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
r8454 and r8460 introduced a change where writes are not retried when other
module/framer commands are retried. This was an error and wasn't what was
actually under test. This commit restores the behavior in wctdm24xxp and
makes sure the vpm writes are retried in the wcte12xp.
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8461 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Closes a memory leak when processing the VPM write commands introduced in
r8454.
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8460 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|
|
Latency conditions could cause the driver to misconfigure the VPM which
would result in one way audio. DAHDI-572.
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@8454 a0bf4364-ded3-4de4-8d8a-66a801d63aff
|