On 02/11/2020 10:25, Ian Abbott wrote:
On 29/10/2020 14:18, Ian Abbott wrote:Commit eddd2a4c675c ("staging: comedi: cb_pcidas: refactor write_calibration_bitstream()") inadvertently removed one of the `udelay(1)` calls when writing to the calibration register in `cb_pcidas_calib_write()`. Reinstate the delay. It may seem strange that the delay is placed before the register write, but this function is called in a loop so the extra delay can make a difference. This _might_ solve reported issues reading analog inputs on a PCIe-DAS1602/16 card where the analog input values "were scaled in a strange way that didn't make sense". On the same hardware running a system with a 3.13 kernel, and then a system with a 4.4 kernel, but with the same application software, the system with the 3.13 kernel was fine, but the one with the 4.4 kernel exhibited the problem. Of the 90 changes to the driver between those kernel versions, this change looked like the most likely culprit.Actually, I've realized that this patch will have no effect on the PCIe-DAS1602/16 card because it uses a different driver - cb_pcimdas, not cb_pcidas.
But that's also confusing because PCIe-DAS1602/16 was not supported until the 3.19 kernel! I know the reported has both PCI-DAS1602/16 and PCIe-DAS1602/16 cards (supported by cb_pcidas and cb_pcimdas respectively), so there could have been some mix-up in the reporting.
Greg, you might as well drop this patch if you haven't already applied it, since it was only a hunch that it fixed a problem.
-- -=( Ian Abbott <abbotti@xxxxxxxxx> || MEV Ltd. is a company )=- -=( registered in England & Wales. Regd. number: 02862268. )=- -=( Regd. addr.: S11 & 12 Building 67, Europa Business Park, )=- -=( Bird Hall Lane, STOCKPORT, SK3 0XA, UK. || www.mev.co.uk )=-
