This patch applies spinlock to protect access to rx / tx structs in
certain call sites, which fixes the following crash in gdm_suspend.
It also fixes usb_set_intfdata() in gdm_usb_probe to avoid setting an
already freed phy_dev.
<5>[ 4996.815018] [<7f0074b0>] (gdm_suspend+0x1c/0x2b4 [gdmwm]) from [<803020a4>] (usb_suspend_both+0x80/0x1a0)
<5>[ 4996.815055] [<803020a4>] (usb_suspend_both+0x80/0x1a0) from [<80302c84>] (usb_runtime_suspend+0x38/0x64)
<5>[ 4996.815089] [<80302c84>] (usb_runtime_suspend+0x38/0x64) from [<802becc0>] (__rpm_callback+0x48/0x78)
<5>[ 4996.815118] [<802becc0>] (__rpm_callback+0x48/0x78) from [<802bf8dc>] (rpm_suspend+0x394/0x5ec)
<5>[ 4996.815145] [<802bf8dc>] (rpm_suspend+0x394/0x5ec) from [<802c0550>] (pm_runtime_work+0x8c/0xa4)
<5>[ 4996.815177] [<802c0550>] (pm_runtime_work+0x8c/0xa4) from [<800456cc>] (process_one_work+0x264/0x438)
<5>[ 4996.815209] [<800456cc>] (process_one_work+0x264/0x438) from [<80045acc>] (worker_thread+0x22c/0x3b8)
<5>[ 4996.815239] [<80045acc>] (worker_thread+0x22c/0x3b8) from [<8004a43c>] (kthread+0x9c/0xa8)
<5>[ 4996.815270] [<8004a43c>] (kthread+0x9c/0xa8) from [<8000f160>] (kernel_thread_exit+0x0/0x8)
<0>[ 4996.815295] Code: e92d4000 e8bd4000 e2800020 eb4ab9a1 (e5905000)
Signed-off-by: Ben Chan <benchan@chromium.org>
Signed-off-by: Sameer Nanda <snanda@chromium.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
as ret is every time assigned to EINVAL in fail cases, initialise it at
the beginning.
Signed-off-by: Devendra Naga <devendra.aaru@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The line6 drive no longer requests MIDI dumps from the device so
dumprequest.c is not needed.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The pod startup procedure dumps model data. This is no longer useful
since the sysfs attrs which expose this information to userspace have
been removed. The dump request wasn't been processed anymore.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The variax startup procedure dumps model data. This is no longer useful
since the sysfs attrs which expose this information to userspace have
been removed. The dump request wasn't been processed anymore.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The sysfs attrs to access the model dump have been removed so it's safe
to drop the model_data field. The next step will be to simplify the
startup process since we no longer need to fetch this data via dump
requests, but that will be done in a later patch.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The driver no longer keeps state of MIDI-accessible device parameters.
Drop the buffers but be careful to keep the device startup procedure
working even though we no longer store the contents of the dump request.
The startup procedure will be simplified in a later patch.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Both pod.c and variax.c expose the device's MIDI parameters as sysfs
attrs. Knowledge of MIDI constants should be in userspace, not in the
driver. Drop the sysfs attrs and let userspace interpret parameters it
cares about instead.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since the driver no longer needs to keep state of device parameters it
is no longer necessary to refresh state when messages that affect other
parameters are sent. Drop the code to trigger a dump when amp/effects
are changed.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Now that both pod.c and variax.c have removed their raw sysfs attrs, the
CONFIG_LINE6_USB_RAW config option can be dropped. The drivers are
being stripped down to just PCM and MIDI I/O so there is less state and
filtering. Therefore it's no longer necessary to have a special sysfs
attr to bypass the filtering ("raw").
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Note that VARIAX_SYSEX_ACTIVATE was never used so drop it along with the
other VARIAX_SYSEX_* constants which are now unused.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The get_string() function has no other callers so remove it too.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The get_string() function is no longer used so drop it too.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The pod_set_system_param_int() helper function is only used to set the
monitor level. Previously it was also used to control the tuner and has
special checks. These checks can now be dropped, along with the tuner
constants.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The raw sysfs attr transmits MIDI messages with less filtering. This
was a useful debugging tool while the staging driver still had a lot of
state and filtering. It is not useful anymore since the filtering is
being dropped.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
It is no longer necessary to trigger on MIDI transmit messages that will
change the state of the device since:
1. We've dropped the midi_postprocess sysfs attr in a previous commit so
it is not possible to activate this feature anymore.
2. The other sysfs attrs that could inspect the state after a dump
request have been dropped.
Therefore we can safely remove this dead code.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The midi_mask_transmit sysfs attribute selects the MIDI channels on
which to transmit messages. If ALSA wants to transmit a message that
does not match an allowed channel, the message will be dropped. This
driver feature is not really used and applications don't know how to
take advantage of it. Therefore we drop it and rely on applications or
MIDI sequencers to select the channels used for communication.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The midi_mask_receive sysfs attribute selects the MIDI channels on which
to receive messages. There is no need to do this at the driver level,
instead the MIDI application endpoints decide which channels to use.
Other drivers don't implement midi_mask_receive so applications cannot
make use of this feature. Therefore we drop it as part of the effort to
clean up the staging driver.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Dropping this final tuner sysfs attr also leaves the
POD_GET_SYSTEM_PARAM() macro, pod_get_system_param_string(), and
pod_get_system_param_int() unused. Remove them immediately to avoid
compiler warnings.
POD_monitor_level is now the only SYSEX value that needs to be watched
by the driver since the ALSA volume control should reflect the current
monitor volume level used by the device. Therefore, drop the switch
statement entirely and just check for POD_monitor_level, ignoring
unknown system parameter changes.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The monitor_level sysfs attr has been dropped and we never need to wait
on the monitor_level value to change. Switch to a plain int and drop
the wait queue.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The monitor level can be adjusted using an ALSA volume control.
Therefore the sysfs attribute is unnecessary. Note that we leave the
monitor_level driver variable so that the ALSA volume control still
works.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Once further sysfs attrs have been removed it will also be possible to
drop prog_data and other pieces which we keep for now.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The param_dirty bitmap tracks which parameters have been modified since
saving a preset. The bitmap is never used though so we can drop this
deadcode.
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>