On Mon, Sep 17, 2018 at 11:41:06PM +0300, Kimmo Saarela wrote: > 1) In the generic SPI driver documentation, it has been stated that > a drivers setup method should not touch the HW because there might > be ongoing transfer with the same master, but with another CS line. > This driver _seems_ to touch the HW by eventually calling > fsl_spi_change_mode function. Should this be fixed and if so, is here > anyone who has been working on this driver? It should probably be fixed, yes, though it probably affects few systems (this is a fairly frequent issue in drivers). Unfortunately it looks like it's been some considerable time since anyone actively worked on the driver so you might be on your own here :/ > 2) This same spi driver uses only the CPU mode or cpu-qe mode of > operation. On some hardware (mpc8321) it could be possible to use the > QE engine for the transfers to improve performance. In other quicc > SoC's there is better support for this. Has someone tested that mode > of operation with Linux and noticed that it can't be done (some > limitation on QE with other peripherals?) or is it like it is only > because there hasn't been any demand for better driver. No idea on this one, your guess that there's not been much demand is probably right.
Attachment:
signature.asc
Description: PGP signature
