We are experiencing problems with our USB host driver in DMA mode.
The information available seems inconsistent. In the Linux SDk a bug is reported where the Host fails if EP0 is used while a DMA transfer is ongoing - but this does not appear in the chip errata. Is this a real problem of the silicon? should DMA use be avoided if EP0 could be active during a transfer?
There is a second issue - that is the data toggle in DMA mode that is in the errata. But a workaround is recommended for this.
We need to make a design decision about the driver - and whether to revert to non-DMA mode. But really need to understand the real situation to do that.
Thanks