... > pidfd_info overwrites the request_mask with what is supported by the > kernel. I don't think userspace setting random stuff in the request_mask > is a problem. It would already be a problem with statx() and we haven't > seen that so far. I don't think it is wise/necessary for the kernel to set bits that weren't set in the request (for values that aren't being returned). That would let you add some mutually exclusive options that use the same part of the buffer area. > If userspace happens to set a some random bit in the request_mask and > that bit ends up being used a few kernel releases later to e.g., > retrieve additional information then all that happens is that userspace > would now receive information they didn't need. That's not a problem. Especially since the buffer is unlikely to be large enough. David - Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
