On Mon, 06.01.14 03:35, Tom Gundersen (tome...@kemper.freedesktop.org) wrote:
> + lease->dns = new0(struct in_addr*, len / 4 + 1); > + if (!lease->dns) > + return -ENOMEM; > + > + for (i = 0; i < len / 4; i++) { > + lease->dns[i] = new0(struct in_addr, 1); > + memcpy(&lease->dns[i]->s_addr, option + 4 * > i, 4); > + } > + > + lease->dns[i + 1] = NULL; Isn't this a bit overkill? Why not just use an array of "struct in_addr" rather than an array of "struct in_addr*"? I mean, even if we'd ignore the overhead of malloc() here, the code is a lot more complex, and on 64bit you use double the memory for storing the pointer to the address than for the actual address stored... ;-) Lennart -- Lennart Poettering, Red Hat _______________________________________________ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel