Ben Cheatham wrote: [..] > So, it seems like a bad idea to provide legacy access for CXL 1.0/1.1 ports > due to this issue. CXL 2.0+ ports don't suffer from this issue though since > they are specified by an SBDF, not a MMIO address. And looking at the code, > it looks like EINJ error types that use an SBDF already get through the > filter. So it doesn't look like there's actually anything to be done for > CXL 2.0+ ports and the new interface is just a convenience feature for 2.0+ > ports. Ok, I am on board. Continue the status quo of blocking CXL 1.1 error injection, and let CXL 2.0 error injection be accomplished through either interface.
