On 12/12/20 10:25 AM, Victor Stewart wrote: > On Sat, Dec 12, 2020 at 5:07 PM Jens Axboe <axboe@xxxxxxxxx> wrote: >> >> On 12/12/20 8:31 AM, Victor Stewart wrote: >>> RE our conversation on the "[RFC 0/1] whitelisting UDP GSO and GRO >>> cmsgs" thread... >>> >>> https://lore.kernel.org/io-uring/CAM1kxwi5m6i8hrtkw7nZYoziPTD-Wp03+fcsUwh3CuSc=81kUQ@xxxxxxxxxxxxxx/ >>> >>> here are the patches we discussed. >>> >>> Victor Stewart (3): >>> net/socket.c: add PROTO_CMSG_DATA_ONLY to __sys_sendmsg_sock >>> net/ipv4/af_inet.c: add PROTO_CMSG_DATA_ONLY to inet_dgram_ops >>> net/ipv6/af_inet6.c: add PROTO_CMSG_DATA_ONLY to inet6_dgram_ops >>> >>> net/ipv4/af_inet.c >>> | 1 + >>> net/ipv6/af_inet6.c >>> | 1 + >>> net/socket.c >>> | 8 +- >>> 3 files changed, 7 insertions(+), 3 deletions(-) >> >> Changes look fine to me, but a few comments: >> >> - I'd order 1/3 as 3/3, that ordering makes more sense as at that point it >> could actually be used. > > right that makes sense. > >> >> - For adding it to af_inet/af_inet6, you should write a better commit message >> on the reasoning for the change. Right now it just describes what the >> patch does (which is obvious from the change), not WHY it's done. Really >> goes for current 1/3 as well, commit messages need to be better in >> general. >> > > okay thanks Jens. i would have reiterated the intention but assumed it > were implicit given I linked the initial conversation about enabling > UDP_SEGMENT (GSO) and UDP_GRO through io_uring. > >> I'd also CC Jann Horn on the series, he's the one that found an issue there >> in the past and also acked the previous change on doing PROTO_CMSG_DATA_ONLY. > > I CCed him on this reply. Soheil at the end of the first exchange > thread said he audited the UDP paths and believed this to be safe. > > how/should I resubmit the patch with a proper intention explanation in > the meta and reorder the patches? my first patch and all lol. Just post is as a v2 with the change noted in the cover letter. I'd also ensure that it threads properly, right now it's just coming through as 4 separate emails at my end. If you're using git send-email, make sure you add --thread to the arguments. -- Jens Axboe
