- 18 Mar, 2014 5 commits
-
-
Juan David González Cobas authored
We need to know the programmed delay, hence this extends the API of the library to recover it in ps units. The ensuing arithmetic is also added
-
Juan David González Cobas authored
-
Juan David González Cobas authored
A channel start, end and delta time registers are exported read-only through sysfs to allow easy readback of the configured values.
-
Juan David González Cobas authored
Now we can read back the programmed values of output registers, so we allow its readout from user space
-
Alessandro Rubini authored
-
- 11 Mar, 2014 2 commits
-
-
Tomasz Wlostowski authored
-
Tomasz Wlostowski authored
This patch unifies VIC interrupt handling by removing VIC support code from the driver and using the one provided by the carrier. Since the SVEC now supports IRQs too, both carriers can benefit from interrupts.
-
- 21 Feb, 2014 12 commits
-
-
Alessandro Rubini authored
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
Alessandro Rubini authored
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
Alessandro Rubini authored
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
Alessandro Rubini authored
Meanwhile, turn ZIO to @sc{zio} in the whole manual. Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
Alessandro Rubini authored
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
Alessandro Rubini authored
moreover, I was reading the hardware fifo both at interrupt level and at tasklet level. The tasklet read-out is mandatory if we are using a timer instead of an interrupt (the timer is the tasklet function), but should be avoided if we have interrupts. Sometimes I got duplicate samples because the tasklet reading the hardware fifo was preempted by the interrupt handler. Now I can stamp 50kHz bursts withouth data loss, provided the zio buffer is big enough. Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
Alessandro Rubini authored
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
Alessandro Rubini authored
Now that we are self-timed, the only reason white the active input block is null may be a zio buffer-full situation. We need to keep the loop running anyways (by discarding one sample), because we miss notification when the buffer has new space. We can't change zio to this aim, as it would be extra code in a hot path: we'd better loose one block when the buffer is full, we would loose it soon after anyways, since the system is overloaded. Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
Alessandro Rubini authored
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
Alessandro Rubini authored
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
Alessandro Rubini authored
This was not possible with the older zio, but now we have support for that. Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
Alessandro Rubini authored
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
- 18 Jan, 2014 1 commit
-
-
Alessandro Rubini authored
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
- 17 Jan, 2014 2 commits
-
-
Alessandro Rubini authored
Changing critical trigger parameters forces an abort of pending blocks. This means that a running reader will loos its own active block when lib/fdelay-pulse is called (the incoming samples are not lost, but the reader must be restarted). The previous code would work with the new self-timed feature in recent zio master, but by now this patch allows remaining at zio-1.0 -- in the "fixes" branch. Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
Alessandro Rubini authored
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
- 16 Jan, 2014 1 commit
-
-
Alessandro Rubini authored
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
- 15 Sep, 2013 3 commits
-
-
Tomasz Wlostowski authored
Signed-off-by: Juan David Gonzalez Cobas <dcobas@cern.ch>
-
Juan David González Cobas authored
Although the standard include path in the build tree is declared in the makefile, deployment of the library in a separate environment will rely on coexistent lib and kernel .h files, hence the double quotes
-
Juan David González Cobas authored
-
- 10 Sep, 2013 1 commit
-
-
Alessandro Rubini authored
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
-
- 27 Aug, 2013 1 commit
-
-
Tomasz Wlostowski authored
Signed-off-by: Juan David Gonzalez Cobas <dcobas@cern.ch>
-
- 21 Aug, 2013 1 commit
-
-
Tomasz Wlostowski authored
-
- 12 Aug, 2013 1 commit
-
-
Tomasz Wlostowski authored
doc: added some info about SVEC, documented WR options in fdelay-board-time and mentioned fdelay-pulse-tom. VME-specific documentation still to be done.
-
- 23 Jul, 2013 2 commits
-
-
Tomasz Wlostowski authored
-
Tomasz Wlostowski authored
-
- 12 Jul, 2013 3 commits
-
-
Juan David González Cobas authored
This is below spec, and the driver must ignore it Amending 289428fe, delay_h is not taken into account, given that the specification excludes delays > 200s. Conflicts: kernel/fd-zio.c
-
Juan David González Cobas authored
This patch addresses the performance problem reported by Christophe Chanavat (a single call to fdelay_config_pulse takes 2.1ms to complete). The time is spent in driving the GPIO enable line that controls the output stage decoupling. This is done via SPI, and each spi_xfer takes 420us, which add up to the reported delay. The fix leaves the enable pins always on, so the output relay is not used to control whether the outputs are enabled/disabled; it is the DCR enable bit who takes this role instead. This results in an average 8us time for output delay configuration, 3us for output disable. Conflicts: kernel/fd-zio.c
-
Juan David González Cobas authored
This is far more informative about what specific mezzanine originates an error, e.g., low-level traces like gpio_writel_with_retry: succeded after 66 retries gpio_writel_with_retry: succeded after 65 retries gpio_writel_with_retry: succeded after 97 retries ... coming from nameless devices require identification of the originator.
-
- 07 Jun, 2013 3 commits
-
-
Juan David González Cobas authored
This is below spec, and the driver must ignore it
-
Juan David González Cobas authored
If fdelay is programmed with -1 or some other random value for the repetition rate, a drifting delayed pulse will appear. Only rep = 1 makes sense
-
Juan David González Cobas authored
This avoids locking the board by programming a bogus delay plus disable repeatedly
-
- 30 May, 2013 2 commits
-
-
Juan David González Cobas authored
For traceability purposes, we include commit descriptions as MODULE_VERSION information in drivers that are deployed in CERN control system. Ideally, the commit description should be added to all the binary blobs installed, but the kernel module is critical in this connection, as it is crucial to trace back quickly the whereabouts of installed .ko sources.
-
Juan David González Cobas authored
These additions were overlooked in b7b800ae
-