Prepare for different types of commands: replace XFRAME_NEW_CMD with
XFRAME_NEW_REG_CMD in some places.
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
Currently we only have REG (register) commands. Refactor it to allow
MEM (memory) commands.
* A common header
* Don't assume a constant size.
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
* wctdm: set to SPANTYPE_ANALOG_MIXED. Maybe it could be improved if all
slots have modules of the same type.
* wcfxo: type FXO.
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
The xpp modules use time differences for various statistics. In one case
xpp_usb creates a histogram of the time it took the system to send USB
packets (URBs). The time difference is used (after adjustments) as an
index to that array.
However if the clock happens to go back at that exact point in time, we
get to write to an arbitrary negative index in this array.
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
* Drop PCM during sluggish events:
- Drop both rx/tx (was only tx)
- Drop almost all pcm until sluggishness stops (used to drop only
single pcm each time).
- Renamed: "drop_next_pcm" -> "drop_pcm" (to match new semantics)
- Still allow "keep alive" PCM so Astribank will not reset FXS high-voltage.
The "sluggish_pcm_keepalive" parameter set the rate (1/50 by default)
- Added rx/tx drop statistics counters
- Removed "pcm_tx_drops" (replaced by new statistics counters)
* Also improved format of /proc/xpp/XBUS-*/xpp_usb:
- Show cummulative number of sluggish events.
- Clearly show range (in usec) of usb_tx_delay[]
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
Now prints the serial number of the card to the kernel log for cards which
support this feature. This include the wcb23x and wcb43x series.
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Adds support for Digium's new dual span wcb23x series cards. Also other minor
improvements for the new generation cards including the wcb23x and wcb43x
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
A few of the indirect register writes to the A_ST_* indirect registers weren't
being protected by any kind of sequence lock. This could lead to potential race
conditions of two spans were configured simultaneously.
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
The hw init function gets called on a module remove in order to reset the
hardware before shutdown. This was causing "Initializing zarlink echocan" to
print on shutdown which could be confusing.
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
This logic was reading and printing the bri chips automatic selection of sync
source. We always use manual sync selection in the driver so this doesn't need
to exist.
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
When the firmware is reloaded and reset, the packet error count is not.
This can create a condition where the firmware could report an error like
the following if a channel was closed without any new errors:
wctc4xxp 0000:07:08.0: 18446744073709551597 errored receive packets
wcdte.packet_errors is also renamed to wcdte.reported_packet_errors to not
conflict with the packet_errors member of the descriptor lists.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
In the case where two transcoders are loaded in a system and one transcoder
experiences an errored receive packet condition, the logic would ping pong the
error between the two cards, causing a runaway stream of errors in the kernel
log.
This fix moves the global error count var into the wcdte struct to allow for
per-card logic.
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Acked-by: Shaun Ruffell <sruffell@digium.com>
The IRQF_DISABLED flag was removed in 4.1 in commit
"genirq: Remove the deprecated 'IRQF_DISABLED' request_irq() flag entirely" [1].
By default all interrupt handlers are now run with interrupts disabled and
therefore should all be considered "fast interrupts". Any driver that was still
using the flag will now lock interrupts directly in the handler in case running
on an older kernel that is not disabling interrupts when running the handler.
[1] https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=d8bf368d0631d4bc2612d8bf2e4e8e74e620d0cc
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
These flags are direct mappings to the IRQF_[SHARED|DISABLED] flags and no
longer need to be kept separate.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Introduces support for new series b43x cards which make use of the zarlink echo
canceller. Made as few changes to the b410 operation as possible, but the
critical region had some cleanups, so it may be a bit more performant.
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
The cards affected include the TE131/3, TE235/435, A4B, and A8B.
Update all PCIe cards' firmware to increase the incoming and outgoing TDM FIFOs
to 16ms. The FIFOs will only be filled to a depth equal to the driver's latency
setting (ie. 3ms default). The total system latency is not effected. The
firmware and driver now also report the maximum DMA transaction time when in
DEBUG mode to aid in determining if the system is experiencing long PCIe
transactions (ie. TLP completion timeouts).
Decreased the maximum latency to from 20 to 12ms
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
New USB firmware that fix mis-reporting of the number of channels (or
rather: licenses) in the Astribank when a 2FXS6FXO module is used in
conjunction with another module.
USB_FW.hex: rev. 11452
USB_FW.201.hex: rev. 11453
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
Commit d5db139ab3764640e0882a1746e7b9fdee33fd87 "module: make
module_refcount() a signed integer." included in 3.19 makes this
condition slightly more complex.
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
With commit (af3cd13501 "lib/string.c: remove strnicmp()") [1] dahdi can no
longer call strnicmp directly. strncasecmp was added into lib/string.c in kernel
version 2.6.22 so we'll map calls to strncasecmp to strnicmp for any kernel
before that.
This is necessary to compile against kernels >= 4.0.
[1] https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=af3cd13501
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
It was reported that for VLAN interfaces, failing to release this reference
count will prevent a system from rebooting properly, resulting in the need to
power cycle.
Reported-by: Assen Totin
Internal-Issue-ID: DAHLIN-343
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Latest Astribank firmware (as of Rev. 11426 also supports some newer
hardware types, which will have the ID 203. Anyone installing this newer
version will now have 203 as an alias, but older versions will not have
it.
New firmware that includes a better way to identify hardware modules
type that does not require multiplexers.
This is firmware file FPGA_1161.201.hex rev. 11426.
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
Kernel version 3.16+, since upstream commit (7f7f25e82d54870d "replace checking
for ->read/->aio_read presence with check in ->f_mode" )[1], does not like it
when dahdi changes the set of allowed file operations on a file descriptor
outside of the context of an open() system call.
DAHDI changes the available file operations when a channel is opened by first
opening /dev/dahdi/channel and then calling the DAHDI_SPECIFY ioctl to bind it
to a particular DAHDI channel. Until DAHDI_SPECIFY is called there weren't any
read()/write() callbacks implemented and therefore after the initial open, the
kernel was setting not setting FMODE_CAN_{WRITE,READ} on the file descriptor
indicating that those operations were not allowed.
Now define empty shell functions on the general dahdi_fops so the vfs layer will
not mark a file descriptor as unwritteable or unreadable on open.
[1] https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=7f7f25e82d54870df24d415a7007fbd327da027b
Internal-Issue-ID: DAHLIN-340
Reported-and-tested-by: Thomas B. Clark
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
The logic on ENABLE_TASKELETS compiler flag was inverted causing an oops on
normal dahdi_cfg of a dynamic span.
Issue-id: https://issues.asterisk.org/jira/browse/DAHLIN-328
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Acked-by: Shaun Ruffell <sruffell@digium.com>
A fresh modprobe and dahdi_cfg would cause a temporary green alarm state on the
spans for a second while the alarm debounced into RED. New logic keeps the
spans alarm state as NONE during the unconfigured state, but sets the alarm
state to RED after the framer reset in the startup logic.
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
A fresh modprobe and dahdi_cfg would cause a temporary green alarm state on the
driver for a second while the alarm debounced into RED. New logic keeps the
spans alarm state as NONE during the unconfigured state, but sets the alarm
state to RED after the framer reset in the startup logic.
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Acked-by: Shaun Ruffell <sruffell@digium.com>
The firmware/Makefile missed a line that added the firmware for the TE436 to the
download list.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Adds driver support for Digium's new te436 and te236 quad and dual span T1/E1 cards.
[removed whitespace at end of line]
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
If dahdi_span_ops.spanconfig is called multiple times in a row (like when
running dahdi_cfg; dahdi_cfg ) the tx signaling bits would go through a spurious
state that some far side devices would respond to.
Now, if the dahdi_span_ops.spanconfig callback is called, and the configuration
matches the existing configuration, we will not touch the framer.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
If dahdi_span_ops.spanconfig is called multiple times in a row (like when
running dahdi_cfg; dahdi_cfg ) the tx signaling bits would go through a spurious
state that some far side devices would respond to.
Now, if the dahdi_span_ops.spanconfig callback is called, and the configuration
matches the existing configuration, we will not touch the framer.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
If a channel is currently playing a tone when the tone zone is updated, the
existing tone zone could be freed while the channel keeps a reference to the
current tone (curtone) that points into the freed zone.
If the newly freed tone is then modified, there was a window where it was
possible to corrupt 'struct dahdi_chan' (by overrunning swritechunk[])
resulting in a "BUG: unable to handle kernel paging request at virtual address"
panic in the context of __dahdi_transmit_chunk().
Reported-and-Tested-by: Matt Behrens <matt@zigg.com>
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
This fixes a long standing issue where, for CAS signaling, the RX bits were
sometimes misreported after span configuration before the first detected state
change.
The logic in the wct4xxp driver now matches that in the wcte43x driver and
wcte13xp drivers. The wcte12xp driver always polls the sigbits due to how
voicebus works and is not affected by this.
Internal-Issue-ID: DAHDI-1081
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
This is the same change for the wcte13xp driver but applied to the other
xbus-based digital card.
If dahdi_cfg set the DAHDI_CONFIG_NOTOPEN setting on the span, which it does
when the "yellow" flag is added to the span config line, then it was possible
for the span to get stuck with DAHDI_ALARM_NOTOPEN (NOP).
This is because the driver only updates the alarm state when the framer reports
that the span alarm has changed. Therefore, unless the framer goes through an
alarm transition, the fact that channels are opened was never noticed by alarm
handling routine.
Now check the alarm state directly when the first channel is opened, and the
last channel is closed.
Internal-Issue-ID: DAHDI-1103
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
This will facilitate adding another flag for open channels on a span without
needing to add a lock on the span, or taking the global lock. Currently the span
flags are protected by the global reglock. This is not longer required.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
If dahdi_cfg set the DAHDI_CONFIG_NOTOPEN setting on the span, which it does
when the "yellow" flag is added to the span config line, then it was possible
for the span to get stuck with DAHDI_ALARM_NOTOPEN (NOP).
This is because the wcte13xp driver only updates the alarm state when the framer
reports that the span alarm has changed. Therefore, unless the framer goes
through an alarm transition, the fact that channels are opened was never noticed
by alarm handling routine.
Now check the alarm state directly when the first channel is opened, and the
last channel is closed.
Internal-Issue-ID: DAHDI-1103
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
This makes the open symmetrical with the close. It is also considered good
practice to not call through callbacks with spinlocks held.
I modified all the drivers where I could not tell whether it was necessary to
hold the chan->lock with interrupts disabled to simply take the lock inside the
callback.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Fixes inability to reliably get CAS (robbed-bits) when using AMI line encoding
on the TE820 card.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
