On 07/20/2017 12:06 PM, Daniel P. Berrange wrote: >>> IMHO your patch is ok, or we could be alittle more explicit about >>> catching just the case where you pass -1 for bytes, and have >>> >>> && bytes != -1 >> >> This seems bizarre to me since bytes is size_t bytes and size_t >> is unsigned, so I'd have sympathy for a compiler that warned that >> bytes != -1 was always true. > > Could be paranoid and do "&& bytes != (size_t)-1"
Any compiler that treats 'bytes != -1' differently from 'bytes != (size_t)-1' is broken, since those two are equivalent per C99 promotion rules (unsigned op signed produces unsigned, and conversion of signed to unsigned is well-defined). I prefer the form without a cast. -- Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3266 Virtualization: qemu.org | libvirt.org
signature.asc
Description: OpenPGP digital signature
