1 comment:
Patchset:
In current implementation fast_spi uses memcpy() and it is clear that the change in memcpy() broke it.
We can make memcpy() to any no. of bytes, but it shouldn't break existing code that depend on memcpy() and cause regression on existing platforms. Also saying using memcpy() is "wrong" is wrong. It's a valid approach.
It's really not a valid approach to rely on specific internals of memcpy if your hardware specifically needs 32bit MMIO. And your "shouldn't break existing code" is also inappropriate: that fast_spi code is wrong, not memcpy...
To view, visit change 81959. To unsubscribe, or for help writing mail filters, visit settings.