[PATCH 14/16] staging: comedi: comedi_test: handle partial scans in timer routine

From: Ian Abbott
Date: Tue Oct 27 2015 - 13:01:25 EST


For asynchronous command handling on the analog input subdevice, a
kernel timer routine is used to generate the fake waveform data. A
"scan" consists of a number of conversions separated in time by a
conversion period. Successive scans are separated in time by a scan
period, which is at least the conversion period multiplied by the number
of conversions per scan. Currently, the timer routine does not generate
any data until the end of a scan period, generating whole scans of data
at a time. Change it to generate data at the end of each conversion
period, with an extra delay after the final conversion in each scan if
necessary. Use new member `ai_convert_time` in the private data
structure `struct waveform_private` to keep track of when the next
conversion is due. This replaces the old member `ai_last_scan_time`
which kept track of the time of the previous scan.

Signed-off-by: Ian Abbott <abbotti@xxxxxxxxx>
---
drivers/staging/comedi/drivers/comedi_test.c | 85 ++++++++++++++++++----------
1 file changed, 54 insertions(+), 31 deletions(-)

diff --git a/drivers/staging/comedi/drivers/comedi_test.c b/drivers/staging/comedi/drivers/comedi_test.c
index 468847a..318340c 100644
--- a/drivers/staging/comedi/drivers/comedi_test.c
+++ b/drivers/staging/comedi/drivers/comedi_test.c
@@ -63,7 +63,7 @@ enum waveform_state_bits {
/* Data unique to this driver */
struct waveform_private {
struct timer_list ai_timer; /* timer for AI commands */
- u64 ai_last_scan_time; /* time of last AI scan in usec */
+ u64 ai_convert_time; /* time of next AI conversion in usec */
unsigned int wf_amplitude; /* waveform amplitude in microvolts */
unsigned int wf_period; /* waveform period in microseconds */
unsigned int wf_current; /* current time in waveform period */
@@ -183,46 +183,51 @@ static void waveform_ai_interrupt(unsigned long arg)
struct comedi_subdevice *s = dev->read_subdev;
struct comedi_async *async = s->async;
struct comedi_cmd *cmd = &async->cmd;
- unsigned int i, j;
- unsigned long elapsed_time;
- unsigned int num_scans;
+ u64 now;
+ unsigned int nsamples;
+ unsigned int time_increment;

/* check command is still active */
if (!test_bit(WAVEFORM_AI_RUNNING, &devpriv->state_bits))
return;

- elapsed_time = ktime_to_us(ktime_get()) - devpriv->ai_last_scan_time;
- num_scans = elapsed_time / devpriv->ai_scan_period;
-
- num_scans = comedi_nscans_left(s, num_scans);
- for (i = 0; i < num_scans; i++) {
- unsigned int scan_remain_period = devpriv->ai_scan_period;
-
- for (j = 0; j < cmd->chanlist_len; j++) {
- unsigned short sample;
-
- if (devpriv->wf_current >= devpriv->wf_period)
- devpriv->wf_current %= devpriv->wf_period;
- sample = fake_waveform(dev, CR_CHAN(cmd->chanlist[j]),
- CR_RANGE(cmd->chanlist[j]),
- devpriv->wf_current);
- comedi_buf_write_samples(s, &sample, 1);
- devpriv->wf_current += devpriv->ai_convert_period;
- scan_remain_period -= devpriv->ai_convert_period;
+ now = ktime_to_us(ktime_get());
+ nsamples = comedi_nsamples_left(s, UINT_MAX);
+
+ while (nsamples && devpriv->ai_convert_time < now) {
+ unsigned int chanspec = cmd->chanlist[async->cur_chan];
+ unsigned short sample;
+
+ sample = fake_waveform(dev, CR_CHAN(chanspec),
+ CR_RANGE(chanspec), devpriv->wf_current);
+ if (comedi_buf_write_samples(s, &sample, 1) == 0)
+ goto overrun;
+ time_increment = devpriv->ai_convert_period;
+ if (async->scan_progress == 0) {
+ /* done last conversion in scan, so add dead time */
+ time_increment += devpriv->ai_scan_period -
+ devpriv->ai_convert_period *
+ cmd->scan_end_arg;
}
- devpriv->wf_current += scan_remain_period;
- devpriv->ai_last_scan_time += devpriv->ai_scan_period;
+ devpriv->wf_current += time_increment;
+ if (devpriv->wf_current >= devpriv->wf_period)
+ devpriv->wf_current %= devpriv->wf_period;
+ devpriv->ai_convert_time += time_increment;
+ nsamples--;
}
- if (devpriv->wf_current >= devpriv->wf_period)
- devpriv->wf_current %= devpriv->wf_period;

if (cmd->stop_src == TRIG_COUNT && async->scans_done >= cmd->stop_arg) {
async->events |= COMEDI_CB_EOA;
} else {
+ if (devpriv->ai_convert_time > now)
+ time_increment = devpriv->ai_convert_time - now;
+ else
+ time_increment = 1;
mod_timer(&devpriv->ai_timer,
- jiffies + usecs_to_jiffies(devpriv->ai_scan_period));
+ jiffies + usecs_to_jiffies(time_increment));
}

+overrun:
comedi_handle_events(dev, s);
}

@@ -332,6 +337,7 @@ static int waveform_ai_cmd(struct comedi_device *dev,
{
struct waveform_private *devpriv = dev->private;
struct comedi_cmd *cmd = &s->async->cmd;
+ unsigned int first_convert_time;
u64 wf_current;

if (cmd->flags & CMDF_PRIORITY) {
@@ -352,13 +358,30 @@ static int waveform_ai_cmd(struct comedi_device *dev,
devpriv->ai_scan_period = cmd->scan_begin_arg / NSEC_PER_USEC;
}

- devpriv->ai_last_scan_time = ktime_to_us(ktime_get());
- /* Determine time within waveform period. */
- wf_current = devpriv->ai_last_scan_time;
+ /*
+ * Simulate first conversion to occur at convert period after
+ * conversion timer starts. If scan_begin_src is TRIG_FOLLOW, assume
+ * the conversion timer starts immediately. If scan_begin_src is
+ * TRIG_TIMER, assume the conversion timer starts after the scan
+ * period.
+ */
+ first_convert_time = devpriv->ai_convert_period;
+ if (cmd->scan_begin_src == TRIG_TIMER)
+ first_convert_time += devpriv->ai_scan_period;
+ devpriv->ai_convert_time = ktime_to_us(ktime_get()) +
+ first_convert_time;
+
+ /* Determine time within waveform period at time of conversion. */
+ wf_current = devpriv->ai_convert_time;
devpriv->wf_current = do_div(wf_current, devpriv->wf_period);

+ /*
+ * Schedule timer to expire just after first conversion time.
+ * Seem to need an extra jiffy here, otherwise timer expires slightly
+ * early!
+ */
devpriv->ai_timer.expires =
- jiffies + usecs_to_jiffies(devpriv->ai_scan_period);
+ jiffies + usecs_to_jiffies(devpriv->ai_convert_period) + 1;

/* mark command as active */
smp_mb__before_atomic();
--
2.6.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/