From mboxrd@z Thu Jan 1 00:00:00 1970 From: Pavel Begunkov Date: Tue, 22 Sep 2020 07:57:11 +0000 Subject: Re: [PATCH 1/9] kernel: add a PF_FORCE_COMPAT flag Message-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit List-Id: References: <563138b5-7073-74bc-f0c5-b2bad6277e87@gmail.com> <486c92d0-0f2e-bd61-1ab8-302524af5e08@gmail.com> In-Reply-To: To: Arnd Bergmann Cc: Andy Lutomirski , Christoph Hellwig , Al Viro , Andrew Morton , Jens Axboe , David Howells , linux-arm-kernel , X86 ML , LKML , "open list:MIPS" , Parisc List , linuxppc-dev , linux-s390 , sparclinux , linux-block , Linux SCSI List , Linux FS Devel , linux-aio , io-uring@vger.kernel.org, linux-arch , Linux-MM , Network Development , keyrings@vger.kernel.org, LSM List On 22/09/2020 10:23, Arnd Bergmann wrote: > On Tue, Sep 22, 2020 at 8:32 AM Pavel Begunkov wrote: >> On 22/09/2020 03:58, Andy Lutomirski wrote: >>> On Mon, Sep 21, 2020 at 5:24 PM Pavel Begunkov wrote: >>> I may be looking at a different kernel than you, but aren't you >>> preventing creating an io_uring regardless of whether SQPOLL is >>> requested? >> >> I diffed a not-saved file on a sleepy head, thanks for noticing. >> As you said, there should be an SQPOLL check. >> >> ... >> if (ctx->compat && (p->flags & IORING_SETUP_SQPOLL)) >> goto err; > > Wouldn't that mean that now 32-bit containers behave differently > between compat and native execution? > > I think if you want to prevent 32-bit applications from using SQPOLL, > it needs to be done the same way on both to be consistent: The intention was to disable only compat not native 32-bit. > > if ((!IS_ENABLED(CONFIG_64BIT) || ctx->compat) && > (p->flags & IORING_SETUP_SQPOLL)) > goto err; > > I don't really see how taking away SQPOLL from 32-bit tasks is > any better than just preventing access to the known-broken files > as Al suggested, or adding the hack to make it work as in > Christoph's original patch. That's why I'm hoping that Christoph's work and the discussion will reach consensus, but the bug should be patched in the end. IMHO, it's a good and easy enough fallback option (temporal?). > > Can we expect all existing and future user space to have a sane > fallback when IORING_SETUP_SQPOLL fails? SQPOLL has a few differences with non-SQPOLL modes, but it's easy to convert between them. Anyway, SQPOLL is a privileged special case that's here for performance/latency reasons, I don't think there will be any non-accidental users of it. -- Pavel Begunkov