Hi Sai,
On [2019 Oct 17] Thu 15:47:54, Sai Pavan Boddu wrote:
> GQSPI handles chip selects and fifos in a different way compared to
> spips. So skip update of cs and fifos related to spips in gqspi mode.
>
> Signed-off-by: Sai Pavan Boddu <sai.pavan.bo...@xilinx.com>
> ---
> Changes for V2:
> Just skip update of spips cs and fifos
> Update commit message accordingly
>
> hw/ssi/xilinx_spips.c | 7 +++++++
> 1 file changed, 7 insertions(+)
>
> diff --git a/hw/ssi/xilinx_spips.c b/hw/ssi/xilinx_spips.c
> index a309c71..27154b0 100644
> --- a/hw/ssi/xilinx_spips.c
> +++ b/hw/ssi/xilinx_spips.c
> @@ -1022,6 +1022,13 @@ static void xilinx_spips_write(void *opaque, hwaddr
> addr,
> }
> s->regs[addr] = (s->regs[addr] & ~mask) | (value & mask);
> no_reg_update:
> + /* In GQSPI mode skip update of CS and fifo's related to spips */
> + if (object_dynamic_cast(OBJECT(s), TYPE_XLNX_ZYNQMP_QSPIPS)) {
> + XlnxZynqMPQSPIPS *ss = XLNX_ZYNQMP_QSPIPS(s);
> + if (ARRAY_FIELD_EX32(ss->regs, GQSPI_SELECT, GENERIC_QSPI_EN)) {
> + return;
> + }
> + }
Above corrects the issue for the zynqmp but not for the other two models
(below functions shouldn't be called when writing the mentioned config
regs for them either), would it be ok for you to expand to the switch
cases you had in v1 (into the switch in this function and return after
updating the reg values)? (the correction will then spawn all three
models)
Best regards,
Francisco Iglesias
> xilinx_spips_update_cs_lines(s);
> xilinx_spips_check_flush(s);
> xilinx_spips_update_cs_lines(s);
> --
> 2.7.4
>
>
