On 17/8/26 14:44, Peter Xu wrote:
On Mon, Aug 17, 2026 at 06:05:45AM +0200, Philippe Mathieu-Daudé wrote:
I'd rather avoid kvm_enabled() checks in kvm-specific API (accel/kvm/).
If ram_block_rebind() is KVM-specific, why expose it as RAMBlock API?
I only started to remember why this patch existed.. after this series
applied, kvm_create_guest_memfd() can be invoked from hostmem code, so we
need this patch to fail on old kernels that do not support init shared (old
kernel case, user triggerable).
OK, then please add a comment in the code, otherwise after also
forgetting the why I'd be tempted to remove it, the pattern is odd :o)
