From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paolo Valente Subject: Re: usercopy whitelist woe in scsi_sense_cache Date: Thu, 19 Apr 2018 11:32:34 +0200 Message-ID: <13DBFC76-4849-4DDA-AC44-B2C1257912E7@linaro.org> References: <10360653.ov98egbaqx@natalenko.name> <8473f909-2123-0cfc-43b1-beba0b1aef9b@kernel.dk> <07f263ff-cea6-ac3c-944b-0f36fee8ba25@kernel.dk> <8b32e079-d4e6-3fea-a89d-ff856e4e13b1@kernel.dk> <0fbf2b13-8bae-c7c5-d930-ebaafdc72202@kernel.dk> <011EF7D1-B095-4B8D-AD2A-993048932C49@linaro.org> Mime-Version: 1.0 (Mac OS X Mail 11.2 \(3445.5.20\)) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: quoted-printable Return-path: In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org To: Jens Axboe Cc: Kees Cook , Oleksandr Natalenko , Bart Van Assche , David Windsor , "James E.J. Bottomley" , "Martin K. Petersen" , linux-scsi@vger.kernel.org, LKML , Christoph Hellwig , Hannes Reinecke , Johannes Thumshirn , linux-block , Ulf Hansson , Mark Brown , Linus Walleij List-Id: linux-scsi@vger.kernel.org > Il giorno 18 apr 2018, alle ore 16:30, Jens Axboe ha = scritto: >=20 > On 4/18/18 3:08 AM, Paolo Valente wrote: >>=20 >>=20 >>> Il giorno 18 apr 2018, alle ore 00:57, Jens Axboe = ha scritto: >>>=20 >>> On 4/17/18 3:48 PM, Jens Axboe wrote: >>>> On 4/17/18 3:47 PM, Kees Cook wrote: >>>>> On Tue, Apr 17, 2018 at 2:39 PM, Jens Axboe = wrote: >>>>>> On 4/17/18 3:25 PM, Kees Cook wrote: >>>>>>> On Tue, Apr 17, 2018 at 1:46 PM, Kees Cook = wrote: >>>>>>>> I see elv.priv[1] assignments made in a few places -- is it = possible >>>>>>>> there is some kind of uninitialized-but-not-NULL state that can = leak >>>>>>>> in there? >>>>>>>=20 >>>>>>> Got it. This fixes it for me: >>>>>>>=20 >>>>>>> diff --git a/block/blk-mq.c b/block/blk-mq.c >>>>>>> index 0dc9e341c2a7..859df3160303 100644 >>>>>>> --- a/block/blk-mq.c >>>>>>> +++ b/block/blk-mq.c >>>>>>> @@ -363,7 +363,7 @@ static struct request = *blk_mq_get_request(struct >>>>>>> request_queue *q, >>>>>>>=20 >>>>>>> rq =3D blk_mq_rq_ctx_init(data, tag, op); >>>>>>> if (!op_is_flush(op)) { >>>>>>> - rq->elv.icq =3D NULL; >>>>>>> + memset(&rq->elv, 0, sizeof(rq->elv)); >>>>>>> if (e && e->type->ops.mq.prepare_request) { >>>>>>> if (e->type->icq_cache && rq_ioc(bio)) >>>>>>> blk_mq_sched_assign_ioc(rq, bio); >>>>>>> @@ -461,7 +461,7 @@ void blk_mq_free_request(struct request *rq) >>>>>>> e->type->ops.mq.finish_request(rq); >>>>>>> if (rq->elv.icq) { >>>>>>> put_io_context(rq->elv.icq->ioc); >>>>>>> - rq->elv.icq =3D NULL; >>>>>>> + memset(&rq->elv, 0, sizeof(rq->elv)); >>>>>>> } >>>>>>> } >>>>>>=20 >>>>>> This looks like a BFQ problem, this should not be necessary. = Paolo, >>>>>> you're calling your own prepare request handler from the insert >>>>>> as well, and your prepare request does nothing if rq->elv.icq =3D=3D= NULL. >>>>>=20 >>>>> I sent the patch anyway, since it's kind of a robustness = improvement, >>>>> I'd hope. If you fix BFQ also, please add: >>>>=20 >>>> It's also a memset() in the hot path, would prefer to avoid that... >>>> The issue here is really the convoluted bfq usage of = insert/prepare, >>>> I'm sure Paolo can take it from here. >>>=20 >>=20 >> Hi, >> I'm very sorry for tuning in very late, but, at the same time, very >> glad to find the problem probably already solved ;) (in this respect, = I swear, >> my delay was not intentional) >>=20 >>> Does this fix it? >>>=20 >>> diff --git a/block/bfq-iosched.c b/block/bfq-iosched.c >>> index f0ecd98509d8..d883469a1582 100644 >>> --- a/block/bfq-iosched.c >>> +++ b/block/bfq-iosched.c >>> @@ -4934,8 +4934,11 @@ static void bfq_prepare_request(struct = request *rq, struct bio *bio) >>> bool new_queue =3D false; >>> bool bfqq_already_existing =3D false, split =3D false; >>>=20 >>> - if (!rq->elv.icq) >>> + if (!rq->elv.icq) { >>> + rq->elv.priv[0] =3D rq->elv.priv[1] =3D NULL; >>> return; >>> + } >>> + >>=20 >> This does solve the problem at hand. But it also arouses a question, >> related to a possible subtle bug. >>=20 >> For BFQ, !rq->elv.icq basically means "this request is not for me, as >> I am an icq-based scheduler". But, IIUC the main points in this >> thread, then this assumption is false. If it is actually false, then >> I hope that all requests with !rq->elv.icq that are sent to BFQ do >> verify the condition (at_head || blk_rq_is_passthrough(rq)). In = fact, >> requests that do not verify that condition are those that BFQ must = put >> in a bfq_queue. So, even if this patch makes the crash disappear, we >> drive BFQ completely crazy (and we may expect other strange failures) >> if we send BFQ a request with !((at_head || = blk_rq_is_passthrough(rq)) >> and !rq->elv.icq. BFQ has to put that rq into a bfq_queue, but = simply >> cannot. >>=20 >> Jens, or any other, could you please shed a light on this, and = explain >> how things are exactly? >=20 First, thanks for summing up the problem. > Your assumption is correct, however you set ->priv[0] and ->priv[1] = for > requests, but only for ->elv.icq !=3D NULL. So let's assume you get a > request and assign those two, request completes. Later on, you get > the same request, bypass insert it. BFQ doesn't clear the bic/bfqq > pointers in the request, since ->elv.icq =3D=3D NULL. I'm missing something here. When the request gets completed in the first place, the hook bfq_finish_requeue_request gets called, and that hook clears both ->elv.priv elements (as the request has a non-null elv.icq). So, when bfq gets the same request again, those elements must be NULL. What am I getting wrong? I have some more concern on this point, but I'll stick to this for the moment, to not create more confusion. Thanks, Paolo > It gets inserted > into the dispatch list. >=20 > Then when __bfq_dispatch_request() is called, you do: >=20 > bfqq =3D RQ_BFQQ(rq); > if (bfqq) > bfqq->dispatched++; > [...] >=20 > which is wrong, since you don't know if you assigned a bfqq for this > request. The memory that bfqq points to could be long gone, if that > queue is freed. So you could either guard any bfqq/bic retrieval > with ->elv.icq !=3D NULL, or you could just clear the pointers for > the case where the values aren't valid. >=20 > --=20 > Jens Axboe