But Linux doesn't behave itself and make sure the PCI devices are always
at different addresses. That's unfortunate.
Gabe
Gabe Black wrote:
I don't disagree with anything you've said, but I think you're missing
my point. When you write 0x to a BAR which expects a 256 byte
large block
* build/X86_SE/tests/fast/quick/00.hello/x86/linux/simple-timing passed.
* build/X86_SE/tests/fast/quick/00.hello/x86/linux/simple-atomic passed.
* build/MIPS_SE/tests/fast/quick/00.hello/mips/linux/simple-timing passed.
*
I'm pretty sure it doesn't temporarily while it's setting them up.
Multiple BARs can be set to 0xFFF0 or 0, particularly 0. That's why
treating 0 and 0x like magic values in pcidev.cc make it work.
Gabe
Ali Saidi wrote:
Umm, Yes, I'm sure it does. If you hard code a PCI device to