On 2014-04-15 18:37, H Hartley Sweeten wrote:
Step 5 of the (*do_cmdtest) validates that the chanlist, chan/range/aref/etc,
is compatible with the actual hardware. At this point in the (*do_cmdtest)
the chanlist is valid, due to checks in the core, so the sanity check is not
needed.
For aesthetics, factor the step 5 code into a helper function. Tidy up the
factored out code.
Signed-off-by: H Hartley Sweeten <hsweeten@xxxxxxxxxxxxxxxxxxx>
Cc: Ian Abbott <abbotti@xxxxxxxxx>
Cc: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
---
drivers/staging/comedi/drivers/amplc_pci230.c | 306 ++++++++++++--------------
1 file changed, 135 insertions(+), 171 deletions(-)
diff --git a/drivers/staging/comedi/drivers/amplc_pci230.c b/drivers/staging/comedi/drivers/amplc_pci230.c
index 99e6083..20765f1 100644
--- a/drivers/staging/comedi/drivers/amplc_pci230.c
+++ b/drivers/staging/comedi/drivers/amplc_pci230.c
@@ -953,6 +953,37 @@ static int pci230_ao_rinsn(struct comedi_device *dev,
return i;
}
+static int pci230_ao_check_chanlist(struct comedi_device *dev,
+ struct comedi_subdevice *s,
+ struct comedi_cmd *cmd)
+{
+ unsigned int prev_chan = CR_CHAN(cmd->chanlist[0]);
+ unsigned int range0 = CR_RANGE(cmd->chanlist[0]);
+ int i;
+
+ for (i = 1; i < cmd->chanlist_len; i++) {
+ unsigned int chan = CR_CHAN(cmd->chanlist[i]);
+ unsigned int range = CR_RANGE(cmd->chanlist[i]);
+
+ if (chan < prev_chan) {
+ dev_dbg(dev->class_dev,
+ "%s: channel numbers must increase\n",
+ __func__);
+ return -EINVAL;
+ }
+
+ if (range != range0) {
+ dev_dbg(dev->class_dev,
+ "%s: channels must have the same range\n",
+ __func__);
+ return -EINVAL;
+ }
+
+ prev_chan = chan;
+ }
+ return 0;
+}
+
static int pci230_ao_cmdtest(struct comedi_device *dev,
struct comedi_subdevice *s, struct comedi_cmd *cmd)
{
@@ -1065,48 +1096,9 @@ static int pci230_ao_cmdtest(struct comedi_device *dev,
if (err)
return 4;
- /* Step 5: check channel list if it exists. */
-
- if (cmd->chanlist && cmd->chanlist_len > 0) {
- enum {
- seq_err = (1 << 0),
- range_err = (1 << 1)
- };
- unsigned int errors;
- unsigned int n;
- unsigned int chan, prev_chan;
- unsigned int range, first_range;
-
- prev_chan = CR_CHAN(cmd->chanlist[0]);
- first_range = CR_RANGE(cmd->chanlist[0]);
- errors = 0;
- for (n = 1; n < cmd->chanlist_len; n++) {
- chan = CR_CHAN(cmd->chanlist[n]);
- range = CR_RANGE(cmd->chanlist[n]);
- /* Channel numbers must strictly increase. */
- if (chan < prev_chan)
- errors |= seq_err;
-
- /* Ranges must be the same. */
- if (range != first_range)
- errors |= range_err;
-
- prev_chan = chan;
- }
- if (errors != 0) {
- err++;
- if ((errors & seq_err) != 0) {
- dev_dbg(dev->class_dev,
- "%s: channel numbers must increase\n",
- __func__);
- }
- if ((errors & range_err) != 0) {
- dev_dbg(dev->class_dev,
- "%s: channels must have the same range\n",
- __func__);
- }
- }
- }
+ /* Step 5: check channel list */
+
+ err |= pci230_ao_check_chanlist(dev, s, cmd);
if (err)
return 5;
@@ -1552,6 +1544,105 @@ static int pci230_ai_check_scan_period(struct comedi_cmd *cmd)
return !err;
}
As for the remarks in patch 02, it's possible for chanlist to be NULL or
for chanlist_len to be 0, so pci230_ao_check_chanlist() should only be
called if chanlist is non-NULL and chanlist_len is non-0.
But if chanlist is forced to be NULL by __comedi_get_user_chanlist()
when chanlist_len is 0, it would be sufficient to check that chanlist is
non-NULL before calling pci230_ao_check_chanlist().
+static int pci230_ai_check_chanlist(struct comedi_device *dev,
+ struct comedi_subdevice *s,
+ struct comedi_cmd *cmd)
+{
+ struct pci230_private *devpriv = dev->private;
+ unsigned int max_diff_chan = (s->n_chan / 2) - 1;
+ unsigned int prev_chan = 0;
+ unsigned int prev_range = 0;
+ unsigned int prev_aref = 0;
+ unsigned int prev_polarity = 0;
+ unsigned int subseq_len = 0;
+ int i;
+
+ for (i = 0; i < cmd->chanlist_len; i++) {
+ unsigned int chanspec = cmd->chanlist[i];
+ unsigned int chan = CR_CHAN(chanspec);
+ unsigned int range = CR_RANGE(chanspec);
+ unsigned int aref = CR_AREF(chanspec);
+ unsigned int polarity = pci230_ai_bipolar[range];
+
+ if (aref == AREF_DIFF && chan > max_diff_chan) {
+ dev_dbg(dev->class_dev,
+ "%s: differential channel number out of range 0 to %u\n",
+ __func__, max_diff_chan);
+ return -EINVAL;
+ }
+
+ if (i > 0) {
+ if (chan <= prev_chan && subseq_len == 0)
+ subseq_len = i;
+
+ if (subseq_len > 0 &&
+ cmd->chanlist[i % subseq_len] != chanspec) {
+ dev_dbg(dev->class_dev,
+ "%s: channel numbers must increase or sequence must repeat exactly\n",
+ __func__);
+ return -EINVAL;
+ }
+
+ if (aref != prev_aref) {
+ dev_dbg(dev->class_dev,
+ "%s: channel sequence analogue references must be all the same (single-ended or differential)\n",
+ __func__);
+ return -EINVAL;
+ }
+
+ if (polarity != prev_polarity) {
+ dev_dbg(dev->class_dev,
+ "%s: channel sequence ranges must be all bipolar or all unipolar\n",
+ __func__);
+ return -EINVAL;
+ }
+
+ if (aref != AREF_DIFF && range != prev_range &&
+ ((chan ^ prev_chan) & ~1) == 0) {
+ dev_dbg(dev->class_dev,
+ "%s: single-ended channel pairs must have the same range\n",
+ __func__);
+ return -EINVAL;
+ }
+ }
+ prev_chan = chan;
+ prev_range = range;
+ prev_aref = aref;
+ prev_polarity = polarity;
+ }
+
+ if (subseq_len == 0)
+ subseq_len = cmd->chanlist_len;
+
+ if ((cmd->chanlist_len % subseq_len) != 0) {
+ dev_dbg(dev->class_dev,
+ "%s: channel numbers must increase or sequence must repeat exactly\n",
+ __func__);
+ return -EINVAL;
+ }
+
+ /*
+ * Buggy PCI230+ or PCI260+ requires channel 0 to be (first) in the
+ * sequence if the sequence contains more than one channel. Hardware
+ * versions 1 and 2 have the bug. There is no hardware version 3.
+ *
+ * Actually, there are two firmwares that report themselves as
+ * hardware version 1 (the boards have different ADC chips with
+ * slightly different timing requirements, which was supposed to
+ * be invisible to software). The first one doesn't seem to have
+ * the bug, but the second one does, and we can't tell them apart!
+ */
+ if (devpriv->hwver > 0 && devpriv->hwver < 4) {
+ if (subseq_len > 1 && CR_CHAN(cmd->chanlist[0]) != 0) {
+ dev_info(dev->class_dev,
+ "%s: Buggy PCI230+/260+ h/w version %u requires first channel of multi-channel sequence to be 0 (corrected in h/w version 4)\n",
+ __func__, devpriv->hwver);
+ return -EINVAL;
+ }
+ }
+ return 0;
+}
+
static int pci230_ai_cmdtest(struct comedi_device *dev,
struct comedi_subdevice *s, struct comedi_cmd *cmd)
{
@@ -1740,136 +1831,9 @@ static int pci230_ai_cmdtest(struct comedi_device *dev,
if (err)
return 4;
- /* Step 5: check channel list if it exists. */
-
- if (cmd->chanlist && cmd->chanlist_len > 0) {
- enum {
- seq_err = 1 << 0,
- rangepair_err = 1 << 1,
- polarity_err = 1 << 2,
- aref_err = 1 << 3,
- diffchan_err = 1 << 4,
- buggy_chan0_err = 1 << 5
- };
- unsigned int errors;
- unsigned int chan, prev_chan;
- unsigned int range, prev_range;
- unsigned int polarity, prev_polarity;
- unsigned int aref, prev_aref;
- unsigned int subseq_len;
- unsigned int n;
-
- subseq_len = 0;
- errors = 0;
- prev_chan = prev_aref = prev_range = prev_polarity = 0;
- for (n = 0; n < cmd->chanlist_len; n++) {
- chan = CR_CHAN(cmd->chanlist[n]);
- range = CR_RANGE(cmd->chanlist[n]);
- aref = CR_AREF(cmd->chanlist[n]);
- polarity = pci230_ai_bipolar[range];
- /* Only the first half of the channels are available if
- * differential. (These are remapped in software. In
- * hardware, only the even channels are available.) */
- if ((aref == AREF_DIFF)
- && (chan >= (s->n_chan / 2))) {
- errors |= diffchan_err;
- }
- if (n > 0) {
- /* Channel numbers must strictly increase or
- * subsequence must repeat exactly. */
- if ((chan <= prev_chan)
- && (subseq_len == 0)) {
- subseq_len = n;
- }
- if ((subseq_len > 0)
- && (cmd->chanlist[n] !=
- cmd->chanlist[n % subseq_len])) {
- errors |= seq_err;
- }
- /* Channels must have same AREF. */
- if (aref != prev_aref)
- errors |= aref_err;
-
- /* Channel ranges must have same polarity. */
- if (polarity != prev_polarity)
- errors |= polarity_err;
-
- /* Single-ended channel pairs must have same
- * range. */
- if ((aref != AREF_DIFF)
- && (((chan ^ prev_chan) & ~1) == 0)
- && (range != prev_range)) {
- errors |= rangepair_err;
- }
- }
- prev_chan = chan;
- prev_range = range;
- prev_aref = aref;
- prev_polarity = polarity;
- }
- if (subseq_len == 0) {
- /* Subsequence is whole sequence. */
- subseq_len = n;
- }
- /* If channel list is a repeating subsequence, need a whole
- * number of repeats. */
- if ((n % subseq_len) != 0)
- errors |= seq_err;
+ /* Step 5: check channel list */
- if ((devpriv->hwver > 0) && (devpriv->hwver < 4)) {
- /*
- * Buggy PCI230+ or PCI260+ requires channel 0 to be
- * (first) in the sequence if the sequence contains
- * more than one channel. Hardware versions 1 and 2
- * have the bug. There is no hardware version 3.
- *
- * Actually, there are two firmwares that report
- * themselves as hardware version 1 (the boards
- * have different ADC chips with slightly different
- * timing requirements, which was supposed to be
- * invisible to software). The first one doesn't
- * seem to have the bug, but the second one
- * does, and we can't tell them apart!
- */
- if ((subseq_len > 1)
- && (CR_CHAN(cmd->chanlist[0]) != 0)) {
- errors |= buggy_chan0_err;
- }
- }
- if (errors != 0) {
- err++;
- if ((errors & seq_err) != 0) {
- dev_dbg(dev->class_dev,
- "%s: channel numbers must increase or sequence must repeat exactly\n",
- __func__);
- }
- if ((errors & rangepair_err) != 0) {
- dev_dbg(dev->class_dev,
- "%s: single-ended channel pairs must have the same range\n",
- __func__);
- }
- if ((errors & polarity_err) != 0) {
- dev_dbg(dev->class_dev,
- "%s: channel sequence ranges must be all bipolar or all unipolar\n",
- __func__);
- }
- if ((errors & aref_err) != 0) {
- dev_dbg(dev->class_dev,
- "%s: channel sequence analogue references must be all the same (single-ended or differential)\n",
- __func__);
- }
- if ((errors & diffchan_err) != 0) {
- dev_dbg(dev->class_dev,
- "%s: differential channel number out of range 0 to %u\n",
- __func__, (s->n_chan / 2) - 1);
- }
- if ((errors & buggy_chan0_err) != 0) {
- dev_info(dev->class_dev,
- "amplc_pci230: ai_cmdtest: Buggy PCI230+/260+ h/w version %u requires first channel of multi-channel sequence to be 0 (corrected in h/w version 4)\n",
- devpriv->hwver);
- }
- }
- }
+ err |= pci230_ai_check_chanlist(dev, s, cmd);
if (err)
return 5;
Ditto. pci230_ai_check_chanlist() should only be called if
cmd->chanlist is non-NULL and cmd->chanlist_len is non-0. But if
chanlist is forced to be NULL by __comedi_get_user_chanlist() when
chanlist_len is 0, it would be sufficient to check that chanlist is
non-NULL before calling pci230_ai_check_chanlist().
--
-=( Ian Abbott @ MEV Ltd. E-mail: <abbotti@xxxxxxxxx> )=-
-=( Tel: +44 (0)161 477 1898 FAX: +44 (0)161 718 3587 )=-
_______________________________________________
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxx
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
