On 09/03/2016 09:26, Peter Xu wrote: >>> > > if ((rp_cmd_args[header_type].len != -1 && >>> > > header_len != rp_cmd_args[header_type].len) || >>> > > - header_len > max_len) { >>> > > + header_len > sizeof(buf)) { >> > >> > sizeof works fine because buf is an array of bytes, but ARRAY_SIZE is >> > better because it works for any type of buffer. > I carefully chose "sizeof" out of "ARRAY_SIZE" as commented by Eric > that, we'd better use sizeof() for char typed buffers. > > Will this be related to how header_len is defined? That's something > I do not know... :( E.g., if it's defined as "length in bytes", then > shall we better use sizeof() in all cases?
Oops, I was wrong. It's related to how header_len is used. header_len here is clearly a size in bytes (it's passed to qemu_get_buffer), so the patch is okay. Paolo > Anyway, I take my above question as trivial since current buffer is > char typed.