This closes a reference and memory leak when multiple CPUs are enabling echocan
on a single channel in parallel.
The essential problem is that the call to try_module_get() is not serialized.
Two separate threads can come into ioctl_echocan() on the same channel, they
coordinate via the dahdi_chan.lock to release any current echocan, but then both
create a new echocan state, bump the reference on the module, and the last one
through will actually attach the new state to the channel. The earlier reference
/ memory is leaked.
I tried to conceive of a way to fix this leak without adding a new lock, but the
choices where calling throught the function pointers with dahdi_chan.lock.
Otherwise I needed to change the semantics of echocan_create /free which would
ripple through the hardware echocan modules.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Normally the board drivers should define this, but if they do not we will
provide a suitable default. This allows compilation with vanilla Linux 2.6.18.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Commit 74e949c33a exported the module
variable dahdi.auto_assigned_spans. However this is not a good name.
This commit reverts the export and replaces it with a new function to
get the value of the variable.
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
Acked-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Add a new sysfs attribute to dahdi_device: registration_time
* Records the time of the device's registration with DAHDI.
* Used by dahdi_auto_assign_compat to assign spans by device
registration order when backward compatibility needed.
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
Acked-by: Shaun Ruffell <sruffell@digium.com>
Instead, use the inverse of dahdi.auto_assign_spans value.
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
Acked-by: Russ Meyerriecks <rmeyerriecks@digium.com>
The previous commit from earlier today to fix the backport of PDE_DATA was
wrong in that it would not then process the other defines for older kernels if
it detected it was running on a redhat release.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
This will fix the "error: redefinition of 'PDE_DATA'" error when compiling.
Internal-Issue-ID: DAHLIN-330
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Fixes the following warning when loading the driver:
dahdi: Warning: Span DYN/eth/eth1/00:50:c2:97:92:1d/1 didn't specify a spantype. Please fix driver!
The spantype is intended to be used by the auto configuration tools
(dahdi_genconf, pinned spans, etc..) but dynamic spans, since they are created
directly by dahdi_cfg, never take part in the pre-registration auto
configuration, therefore the span type was never used.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
* This changeset adds master_span as an attribute of the span's driver:
/sys/bus/dahdi_spans/drivers/generic_lowlevel/master_span
* This is mainly used for debugging.
* Reading from it the master span number (or 0 if there is no master
span).
* Writing a number to it, force specified span to be master
* Existing Alternatives:
- grep "MASTER" from /proc/dahdi/*
- cat /sys/bus/dahdi_spans/devices/span-*/is_sync_master
(Note: commit d8fe2af23d is also Acked-By
Tzafrir Cohen <tzafrir.cohen@xorcom.com>)
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
Acked-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Rename as terminology has changed. No change in kernel interface.
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
Acked-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Some older versions of Kbuild will not pass KBUILD_MODNAME to a
compilation unit that is linked to multiple drivers. This change works
around this issue by globally modifing dev_dbg in this case.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
There are not any drivers in the tree that make reference to this definition
anymore, so it can be removed.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
create_proc_entry() was deprecated and replace with proc_create_data() since
it's open to a race condition where the proc entry is visible before the file
operations have been set for it.
The PDE() macro also is no longer available as of Linux 3.10 and is replaced
with PDE_DATA() to get the data member from a proc entry. This is due to the
fact that 'struct proc_dir_entry' is now private to the proc_fs.
This commit changes the core of DAHDI and also introduces proc_create_data() and
PDE_DATA() for older kernels.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
Cc: Russ Meyerriecks <rmeyerriecks@digium.com>
Cc: Oron Peled <oron.peled@xorcom.com>
This was an accidental commit that slipped in as part of (a682401 "dahdi: Expose
dahdi devices in sysfs.")
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
This is mostly a revert of commit r9463. If you need to use DAHDI_CONFLINK
ioctl, make sure to define CONFIG_DAHDI_CONFLINK in
include/dahdi/dahdi_config.h. Apparently there were some users of CONFLINK out
there still.
It's a compile time option now since most users won't need to run the test for
conflinks in the hot-path that is the process_masterspan function.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Tested-by: Ted Gerold <ted@twg.org>
The __dev* directives and functions were removed in 3.8, as they
are no-ops. We still have use of them for older versions, thus
we should define them (as noops) if they don't exist.
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
struct dahdi_count is not directly exposed to user space, so we can use the
native u32 type instead of __u32 to clarify that.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
This allows timingslips to be reset along with the other counters and clarifies
the intended use.
This came up when Doug Bailey asked why he couldn't use dahdi_maint to clear
timing slips in addition to the other counters.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
* Added minimal infrastructure:
- A 'chan_device' member to struct dahdi_chan
- An empty 'device_attribute' array
- A 'bus_type' with its methods
- A 'device_driver' with its methods
- Initialization/Cleanup code
Signed-off-by: Oron Peled <oron.peled@xorcom.com>
Since r5021 [1], first released in DAHDI-Linux 2.2.0, it's been impossible for
user space to change the rxbufpolicy. This hasn't caused any problems and it's
safe to remove a few more of the vestiges of the rxbufpolicy from the driver.
This streamlines the code path in a few places and saves 8 bytes from the size
of struct dahdi_chan.
The user visible parts are maintained and will indicate
DAHDI_POLICY_IMMEDIATE, like it has since 2.2.0.
[1] http://svnview.digium.com/svn/dahdi?view=revision&revision=5021
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10730 a0bf4364-ded3-4de4-8d8a-66a801d63aff
When DAHDI is bridging channels between two different cards, or a user is trying
to record from a channel on a card that is not using the same timing source as
the master span, it is possible to miss audio.
For example, given the following scenario:
<T1 Span> <---> <ISDN CARD> <---> <ANALOG CARD> <---> <local handset>
If DAHDI was set to natively bridge the call from the T1 span to the local
handset it is possible that system conditions could result in one of the cards
servicing their interrupt three times before the other card has a chance to.
Since there are currently only two conference buffers to pass audio between the
cards, one chunk (1ms) of audio will be dropped since the card servicing it's
interrupt more frequently will run out of space to copy audio for the other card
and will have emptied the audio buffer from the other card.
Increasing the number of conference buffers to eight greatly reduces the
probability of audio from being dropped under these conditions at cost of an
additional 72 (32-bit) or 96 (64-bit) bytes per DAHDI channel.
Internal-Issue-ID: DAHLIN-159, DAHDI-976
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Russ Meyerriecks <rmeyerriecks@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10726 a0bf4364-ded3-4de4-8d8a-66a801d63aff
When compiling against kernels 2.6.25, you could get the following error.
error: linux/pci-aspm.h: No such file or directory
This fixes a build regression introduced in r10556 "dahdi: Add
dahdi_pci_disable_link_state for kernel < 2.6.25." [1]
[1] http://svnview.digium.com/svn/dahdi?view=revision&revision=10556
Reported-by: Jean-Philippe Lord
Internal-Issue-ID: DAHLIN-299
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10705 a0bf4364-ded3-4de4-8d8a-66a801d63aff
* This is a minimal convertion -- everything compiles and looks OK.
* We print a warning for spans registering without a spantype.
* Low-level drivers may later want (but not required)
to fold their internal representations to this canonical
representation -- it will save code and make it more readable.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10683 a0bf4364-ded3-4de4-8d8a-66a801d63aff
Instead of registering a function pointer, register a dahdi_dynamic_ops
structure that contains the owner as well as the ioctl callback. This way
dahdi.ko can bump up the reference count on dahdi_dynamic.ko before calling
the ioctl callback.
Also, use the registration mutex to guard against the module being unloaded
between the time the structure pointer was checked, and the module reference
is taken.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10623 a0bf4364-ded3-4de4-8d8a-66a801d63aff
This in conjunction with r10449 "A parent-less device should not crash dahdi",
this allows dahdi_dynamic spans to work post the dahdi_devices changes in
2.6.0.
The full address of the device is not used since kernels prior to 2.6.31 limit
the length of a devicename to 20 characters. The full address of the device
can be pulled out of the "hardware_id" and "type" fields of the span.
This patch is just to get things working again. dahdi_dynamic devices *may*
still have issues if the auto_assign_spans module parameter is 0.
Internal-Issue-ID: DAHLIN-280
Reported-by: Pavel Selivanov
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10563 a0bf4364-ded3-4de4-8d8a-66a801d63aff
This is needed because dev_name() is mapped to kobject_name() in a backport,
but the kobject name isn't set until after device_add(). The result would be
parentless devices would fail since dahdi would not think a name was set for
these devices.
For these systems, we'll set both the bus_id string and the underlying
kobject_name.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10561 a0bf4364-ded3-4de4-8d8a-66a801d63aff
In addition to updating the year, this also adds some boilerplate to
dahdi-sysfs.c and dahdi-sysfs-chan.c that wasn't there previously.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10560 a0bf4364-ded3-4de4-8d8a-66a801d63aff
Will allow the ASPM (Active State Power Management) state to be disabled on
PCIe devices before kernel version 2.6.25.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10556 a0bf4364-ded3-4de4-8d8a-66a801d63aff
This patch removes support for kernel versions < 2.6.18, as those are
not actively supported in any major Linux distribution (except RHEL4,
which is in the "extended" support level of the product life cycle).
This removes much of the more #ifdef-rich parts of the code.
Signed-off-by: Oron Peled <oron.peled@xorcom.com>
Acked-by: Acked-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10472 a0bf4364-ded3-4de4-8d8a-66a801d63aff
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>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10361 a0bf4364-ded3-4de4-8d8a-66a801d63aff
* Added dahdi_dev_dbg() macro to print when we don't (yet)
have a span number.
* Added a new debug category: DAHDI_DBG_ASSIGN
* Made sure error return code paths prints helpfull messages
* Promote error messages from INFO to NOTICE
* Change some errno values from EINVAL to EFAULT (internal
errors not caused by user input)
Signed-off-by: Oron Peled <oron.peled@xorcom.com>
Acked-by Tzafrir Cohen <tzafrir.cohen@xorcom.com>
Acked-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10328 a0bf4364-ded3-4de4-8d8a-66a801d63aff
For some boards, the linemode (E1/T1/J1) is software selectable but
needs to be configured before the spans were historically registered
since the line mode determines the channel count available on the span.
This change exports a "spantype" attribute from the dahdi_device that
can be used to set E1/T1/J1 before the spans are assigned. When
userspace writes to this attribute (in a <span offset>:<span type
string> format), and if the board driver has implemented a set_spantype
function in it's dahdi_span_ops, then the board driver can optionally
change it's mode before registration.
Also part of this change is breaking out the raw data structure
initialization of the spans / channels via the dahdi_init_device_spans
function since the board drivers may need to reallocate channels / spans
as part of this callback. For example, changing from T1 to E1 mode will
require allocating 7 new channels.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10277 a0bf4364-ded3-4de4-8d8a-66a801d63aff
'irqmisses' is more a function of the device and there are better ways
to get to IRQ for a device than storing it in any DAHDI structures.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10276 a0bf4364-ded3-4de4-8d8a-66a801d63aff
This exposes dahdi devices in sysfs and also exposes attributes that
will allow user space to control the registration order in spans. This
facilitates loading drivers out of order yet keeping consistent
span/channel numbering, which in turn will eventually allow the
blacklist for DAHDI drivers to be removed. The default behavior,
controlled with the auto_register module parameter on dahdi is to number
the spans / channels in order like is currently done. So this change
does not introduce any new behavior by default.
* Writing (anything) to this attribute returns the span to its
unassigned state
* Fix dahdi_chan_unreg() echocan refcount
* Add safeguard against duplicate unassignment to _dahdi_unregister_span()
* Remove the span from device_node list, only in dahdi_unregister_device()
and not in dahdi_unregister_span()
* Free allocated span->span_device in span_sysfs_remove()
[is it safe?, didn't cause problem so far...]
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Oron Peled <oron.peled@xorcom.com>
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
dahdi: Add "hardware_id" dahdi_device attribute.
- The "hardware_id" does not change with device location (e.g: when a PCI
card is moved from one slot to another).
- Not all devices have this attribute. It is legal for it to be NULL (that
is the default for all low-level drivers that do not set it explicitly).
- When "hardware_id" is NULL, the sysfs attribute value is "\n"
Signed-off-by: Oron Peled <oron.peled@xorcom.com>
Acked-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10275 a0bf4364-ded3-4de4-8d8a-66a801d63aff
This change will facilitate creating rules that will allow spans and channels
to be accessed by named device files instead of by numbers.
Signed-off-by: Oron Peled <oron.peled@xorcom.com>
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10274 a0bf4364-ded3-4de4-8d8a-66a801d63aff
Increasingly, spans are implemented by devices that support more than a
single span. Introduce a 'struct dahdi_device' object which explicitly
contains multiple spans. This will allow a cleaner representation of
spans and devices in sysfs since order of arrival will not determine the
layout of the devices. This also gives the core of dahdi a way to know
the relationship between spans.
This generalizes similar concepts that were previously xpp specific. The
conversion of the xpp code was almost entirely done by Oron and Tzafrir.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Signed-off-by: Oron Peled <oron.peled@xorcom.com>
Signed-off-by: Tzafrir Cohen <tzafrir.cohen@xorcom.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10273 a0bf4364-ded3-4de4-8d8a-66a801d63aff
I only generally test on RHEL 4 when testing against kernels older
than 2.6.18. Apparently OpenSUSE 10.1 runs with 2.6.16 and doesn't
have WARN_ON_ONCE backported. I took the patch Richard Miller
originally attached to the issue and moved it to
include/dahdi/kernel.h so it would be available for all the board
drivers in the future.
Internal-Issue-ID: DAHLIN-260
Reported-by: Richard Miller
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10252 a0bf4364-ded3-4de4-8d8a-66a801d63aff
Uses the linecompat member to determine what type of span it is. This will
allow removing T1/E1 flags from other places where the span type is stored.
This function also changes the return value from bool to int for the
inlines defined in include/dahdi/kernel.h. This is because not all
kernel versions include stdbool.h in the headers and it will conflict
with boolean values that are exported via module parameters on some
older kernels if dahdi included it globally.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Michael Spiceland <mspiceland@digium.com>
Acked-by: Russ Meyerriecks <rmeyerriecks@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10227 a0bf4364-ded3-4de4-8d8a-66a801d63aff
The pr_ macros are the recommended way for subsystems to print messages
but not all kernel versions DAHDI support has them defined.
Signed-off-by: Shaun Ruffell <sruffell@digium.com>
Acked-by: Michael Spiceland <mspiceland@digium.com>
Acked-by: Russ Meyerriecks <rmeyerriecks@digium.com>
git-svn-id: http://svn.asterisk.org/svn/dahdi/linux/trunk@10226 a0bf4364-ded3-4de4-8d8a-66a801d63aff
