| Message ID | 1586942266-21480-1-git-send-email-jprakash@codeaurora.org |
|---|---|
| Headers | show |
| Series | iio: adc: Add support for QCOM SPMI PMIC7 ADC | expand |
On Thu, Apr 16, 2020 at 1:48 AM Jishnu Prakash <jprakash@codeaurora.org> wrote: > > The ADC architecture on PMIC7 is changed as compared to PMIC5. The > major change from PMIC5 is that all SW communication to ADC goes through > PMK8350, which communicates with other PMICs through PBS when the ADC > on PMK8350 works in master mode. The SID register is used to identify the > PMICs with which the PBS needs to communicate. Add support for the same. Please, split pr_*() -> dev_*() to separate patch. Also think about other logical pieces you may split out. ... > +static const struct adc5_data adc7_data_pmic; Global variable? Hmm... ... > + int ret; > + u8 conv_req = 0, buf[4]; > + > + ret = adc5_masked_write(adc, ADC_APP_SID, ADC_APP_SID_MASK, prop->sid); > + if (ret) > + return ret; > + > + ret = adc5_read(adc, ADC5_USR_DIG_PARAM, buf, sizeof(buf)); > + if (ret < 0) Does > 0 have a meaning? > + return ret; > + > + /* Digital param selection */ > + adc5_update_dig_param(adc, prop, &buf[0]); > + > + /* Update fast average sample value */ > + buf[1] &= 0xff & ~ADC5_USR_FAST_AVG_CTL_SAMPLES_MASK; What the point of 0xff & part? > + buf[1] |= prop->avg_samples; > + > + /* Select ADC channel */ > + buf[2] = prop->channel; > + > + /* Select HW settle delay for channel */ > + buf[3] &= 0xff & ~ADC5_USR_HW_SETTLE_DELAY_MASK; Ditto. > + buf[3] |= prop->hw_settle_time; ... > +static int adc7_do_conversion(struct adc5_chip *adc, > + struct adc5_channel_prop *prop, > + struct iio_chan_spec const *chan, > + u16 *data_volt, u16 *data_cur) > +{ > + int ret; > + u8 status = 0; Redundant assignment. > + mutex_lock(&adc->lock); > + > + ret = adc7_configure(adc, prop); > + if (ret) { > + dev_err(adc->dev, "ADC configure failed with %d\n", ret); > + goto unlock; > + } > + > + /* No support for polling mode at present*/ Missed space > + wait_for_completion_timeout(&adc->complete, ADC7_CONV_TIMEOUT); > + > + ret = adc5_read(adc, ADC5_USR_STATUS1, &status, 1); > + if (ret < 0) Remove all ' < 0' where it is not needed. > + goto unlock; > + > + if (status & ADC5_USR_STATUS1_CONV_FAULT) { > + dev_err(adc->dev, "Unexpected conversion fault\n"); > + ret = -EIO; > + goto unlock; > + } > + > + ret = adc5_read_voltage_data(adc, data_volt); > + > +unlock: > + mutex_unlock(&adc->lock); ... > + for (i = 0; i < adc->nchannels; i++) { > + v_channel = (adc->chan_props[i].sid << ADC_CHANNEL_OFFSET | > + adc->chan_props[i].channel); Too many parentheses or they are in a wrong position. I don't remember operator precedence by heart. > + if (v_channel == iiospec->args[0]) > + return i; > + } ... > + /* > + * Value read from "reg" is virtual channel number > + * virtual channel number = (sid << 8 | channel number). Too many parentheses. And perhaps formulas better to have on a separate line. > + */ ... > +static const struct vadc_map_pt adcmap7_100k[] = { > + { 4250657, -40960 }, > + { 3962085, -39936 }, > + { 419448, -3072 }, > + { 396851, -2048 }, > + { 375597, -1024 }, > + { 355598, 0 }, > + { 336775, 1024 }, > + { 319052, 2048 }, > + { 302359, 3072 }, > + { 2560, 128000 }, > + { 2489, 129024 }, > + { 2420, 130048 } > +}; I'm wondering why you have second column here? Can't you derive it from index? Seems to me pretty easy calculus. ... > + int ret, result = 0; Redundant assignment. > + if (adc_code >= RATIO_MAX_ADC7) > + return -EINVAL; > + > + /* (ADC code * R_PULLUP (100Kohm)) / (full_scale_code - ADC code)*/ > + resistance *= R_PU_100K; > + resistance = div64_s64(resistance, RATIO_MAX_ADC7 - adc_code); > + > + ret = qcom_vadc_map_voltage_temp(adcmap7_100k, > + ARRAY_SIZE(adcmap7_100k), > + resistance, &result); > + if (ret) > + return ret; > + > + *result_mdec = result; ... > + for (i = 0; i < ARRAY_SIZE(adcmap7_die_temp); i++) > + if (adcmap7_die_temp[i].x > voltage) > + break; > + > + if (i == 0) { > + *result_mdec = DIE_TEMP_ADC7_SCALE_1; > + } else if (i == ARRAY_SIZE(adcmap7_die_temp)) { > + *result_mdec = DIE_TEMP_ADC7_MAX; I think you can done these checks before loop, and return immediately. > + } else { > + vtemp0 = adcmap7_die_temp[i - 1].x; > + voltage = voltage - vtemp0; > + temp = div64_s64(voltage * DIE_TEMP_ADC7_SCALE_FACTOR, > + adcmap7_die_temp[i - 1].y); > + temp += DIE_TEMP_ADC7_SCALE_1 + (DIE_TEMP_ADC7_SCALE_2 * (i - 1)); > + *result_mdec = temp; > + }
On Mon, Apr 27, 2020 at 3:56 PM Jishnu Prakash <jprakash@codeaurora.org> wrote: > On 4/17/2020 3:51 PM, Andy Shevchenko wrote: > On Thu, Apr 16, 2020 at 1:48 AM Jishnu Prakash <jprakash@codeaurora.org> wrote: Stop using HTML. It breaks badly the reply and discussion. ... > +static const struct adc5_data adc7_data_pmic; > > Global variable? Hmm... > > adc7_data_pmic is referenced twice before its actual definition (which was added along with corresponding adc5_data struct for PMIC5 ADC), so I have given the initial declaration here. Maybe you can realize how to avoid global variable at all? ... > + buf[1] &= 0xff & ~ADC5_USR_FAST_AVG_CTL_SAMPLES_MASK; > > What the point of 0xff & part? > > This was something you suggested in my first post: > > > + buf[1] &= (u8) ~ADC5_USR_FAST_AVG_CTL_SAMPLES_MASK; > > Use '0xFF ^ _MASK' instead of casting. > > ... > > > + buf[3] &= (u8) ~ADC5_USR_HW_SETTLE_DELAY_MASK; > > Ditto. > > I think "0xff &" works as intended here in place of casting to (u8)... Does it work without casting? (Note, I suggested slightly different expression) I.o.w. what the problem casting solves? > + buf[1] |= prop->avg_samples; > + > + /* Select ADC channel */ > + buf[2] = prop->channel; > + > + /* Select HW settle delay for channel */ > + buf[3] &= 0xff & ~ADC5_USR_HW_SETTLE_DELAY_MASK; > > Ditto. > > + buf[3] |= prop->hw_settle_time;
Hi Andy, On 4/27/2020 6:58 PM, Andy Shevchenko wrote: > On Mon, Apr 27, 2020 at 3:56 PM Jishnu Prakash <jprakash@codeaurora.org> wrote: >> On 4/17/2020 3:51 PM, Andy Shevchenko wrote: >> On Thu, Apr 16, 2020 at 1:48 AM Jishnu Prakash <jprakash@codeaurora.org> wrote: > Stop using HTML. It breaks badly the reply and discussion. > > ... > >> +static const struct adc5_data adc7_data_pmic; >> >> Global variable? Hmm... >> >> adc7_data_pmic is referenced twice before its actual definition (which was added along with corresponding adc5_data struct for PMIC5 ADC), so I have given the initial declaration here. > Maybe you can realize how to avoid global variable at all? There is a way to remove this, I'll make this change with some other changes in the fifth patch of my latest post. > > ... > >> + buf[1] &= 0xff & ~ADC5_USR_FAST_AVG_CTL_SAMPLES_MASK; >> >> What the point of 0xff & part? >> >> This was something you suggested in my first post: >> >>> + buf[1] &= (u8) ~ADC5_USR_FAST_AVG_CTL_SAMPLES_MASK; >> Use '0xFF ^ _MASK' instead of casting. >> >> ... >> >>> + buf[3] &= (u8) ~ADC5_USR_HW_SETTLE_DELAY_MASK; >> Ditto. >> >> I think "0xff &" works as intended here in place of casting to (u8)... > Does it work without casting? (Note, I suggested slightly different expression) > I.o.w. what the problem casting solves? I checked this part again. It looks like casting is not strictly required here, I'll remove it in my latest post. > >> + buf[1] |= prop->avg_samples; >> + >> + /* Select ADC channel */ >> + buf[2] = prop->channel; >> + >> + /* Select HW settle delay for channel */ >> + buf[3] &= 0xff & ~ADC5_USR_HW_SETTLE_DELAY_MASK; >> >> Ditto. >> >> + buf[3] |= prop->hw_settle_time; >