On 5/28/21 10:02 AM, Paul Moore wrote: > On Wed, May 26, 2021 at 4:19 PM Paul Moore <paul@xxxxxxxxxxxxxx> wrote: >> ... If we moved the _entry >> and _exit calls into the individual operation case blocks (quick >> openat example below) so that only certain operations were able to be >> audited would that be acceptable assuming the high frequency ops were >> untouched? My initial gut feeling was that this would involve >50% of >> the ops, but Steve Grubb seems to think it would be less; it may be >> time to look at that a bit more seriously, but if it gets a NACK >> regardless it isn't worth the time - thoughts? >> >> case IORING_OP_OPENAT: >> audit_uring_entry(req->opcode); >> ret = io_openat(req, issue_flags); >> audit_uring_exit(!ret, ret); >> break; > > I wanted to pose this question again in case it was lost in the > thread, I suspect this may be the last option before we have to "fix" > things at the Kconfig level. I definitely don't want to have to go > that route, and I suspect most everyone on this thread feels the same, > so I'm hopeful we can find a solution that is begrudgingly acceptable > to both groups. Sorry for the lack of response here, but to sum up my order of preference: 1) It's probably better to just make the audit an opt-out in io_op_defs for each opcode, and avoid needing boiler plate code for each op handler. The opt-out would ensure that new opcodes get it by default it someone doesn't know what it is, and the io_op_defs addition would mean that it's in generic code rather then in the handlers. Yes it's a bit slower, but it's saner imho. 2) With the above, I'm fine with adding this to io_uring. I don't think going the route of mutual exclusion in kconfig helps anyone, it'd be counter productive to both sides. Hope that works and helps move this forward. I'll be mostly out of touch the next week and a half, but wanted to ensure that I sent out my (brief) thoughts before going away. -- Jens Axboe
