From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by smtp.lore.kernel.org (Postfix) with ESMTP id DBCDFD30019 for ; Fri, 18 Oct 2024 15:46:32 +0000 (UTC) Received: by kanga.kvack.org (Postfix) id 6A29A6B0083; Fri, 18 Oct 2024 11:46:32 -0400 (EDT) Received: by kanga.kvack.org (Postfix, from userid 40) id 651FD6B0088; Fri, 18 Oct 2024 11:46:32 -0400 (EDT) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id 519266B0089; Fri, 18 Oct 2024 11:46:32 -0400 (EDT) X-Delivered-To: linux-mm@kvack.org Received: from relay.hostedemail.com (smtprelay0013.hostedemail.com [216.40.44.13]) by kanga.kvack.org (Postfix) with ESMTP id 32B976B0083 for ; Fri, 18 Oct 2024 11:46:32 -0400 (EDT) Received: from smtpin03.hostedemail.com (a10.router.float.18 [10.200.18.1]) by unirelay05.hostedemail.com (Postfix) with ESMTP id B8B44402E8 for ; Fri, 18 Oct 2024 15:46:24 +0000 (UTC) X-FDA: 82687149678.03.D5394C9 Received: from frasgout11.his.huawei.com (frasgout11.his.huawei.com [14.137.139.23]) by imf07.hostedemail.com (Postfix) with ESMTP id 42B8540020 for ; Fri, 18 Oct 2024 15:46:11 +0000 (UTC) Authentication-Results: imf07.hostedemail.com; dkim=none; dmarc=none; spf=pass (imf07.hostedemail.com: domain of roberto.sassu@huaweicloud.com designates 14.137.139.23 as permitted sender) smtp.mailfrom=roberto.sassu@huaweicloud.com ARC-Seal: i=1; s=arc-20220608; d=hostedemail.com; t=1729266341; a=rsa-sha256; cv=none; b=CBJjp2bVcOGJBnFJK/dyQHWYqhwZ6w/4aX8CYgGGpvWVQ+UcM0iZ3+6hMPwiUoMPRnJv+e NFmO7tubh+3JGx5aUEBZkeW/md1jgzeHHhQxz46IcjyJKAn96PKA2tWX6jEENIGt922mSg g8eKZrxkeHIa4GTMsNm+DtglFnst0Sg= ARC-Authentication-Results: i=1; imf07.hostedemail.com; dkim=none; dmarc=none; spf=pass (imf07.hostedemail.com: domain of roberto.sassu@huaweicloud.com designates 14.137.139.23 as permitted sender) smtp.mailfrom=roberto.sassu@huaweicloud.com ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=hostedemail.com; s=arc-20220608; t=1729266341; h=from:from:sender:reply-to:subject:subject:date:date: message-id:message-id:to:to:cc:cc:mime-version:mime-version: content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=shfUB88SgGWDKvhqrWZsoBg0Kt7O7A94nJsviSB05oI=; b=HPanplHVqP2EARzHFt/KZpu8PYYNCej52MyN2pO/X+yn0qu81vXoqrgLu0Z4UfJnF/RUmr qEh9uTFkZT8eWyL5nE7Rv+nbCCFuobRRnzpXoQwl2YepbzyolIB7G4m+fWGrMdo1euc4yk Pok7c9Uk6LajX06pFv9ds74rAAnGrI4= Received: from mail.maildlp.com (unknown [172.18.186.29]) by frasgout11.his.huawei.com (SkyGuard) with ESMTP id 4XVT7M5gdzz9v7J3 for ; Fri, 18 Oct 2024 23:26:07 +0800 (CST) Received: from mail02.huawei.com (unknown [7.182.16.47]) by mail.maildlp.com (Postfix) with ESMTP id 2A363140F00 for ; Fri, 18 Oct 2024 23:46:17 +0800 (CST) Received: from [127.0.0.1] (unknown [10.204.63.22]) by APP1 (Coremail) with SMTP id LxC2BwC3Ry+6ghJn0PUZAw--.23355S2; Fri, 18 Oct 2024 16:46:16 +0100 (CET) Message-ID: <784c68fa023e99c53cd07265f0524e386815b443.camel@huaweicloud.com> Subject: Re: [RFC][PATCH] mm: Split locks in remap_file_pages() From: Roberto Sassu To: Lorenzo Stoakes Cc: akpm@linux-foundation.org, Liam.Howlett@oracle.com, vbabka@suse.cz, jannh@google.com, linux-mm@kvack.org, linux-kernel@vger.kernel.org, ebpqwerty472123@gmail.com, paul@paul-moore.com, zohar@linux.ibm.com, dmitry.kasatkin@gmail.com, eric.snowberg@oracle.com, jmorris@namei.org, serge@hallyn.com, linux-integrity@vger.kernel.org, linux-security-module@vger.kernel.org, bpf@vger.kernel.org, linux-fsdevel@vger.kernel.org, "Kirill A. Shutemov" , stable@vger.kernel.org, syzbot+91ae49e1c1a2634d20c0@syzkaller.appspotmail.com, Roberto Sassu Date: Fri, 18 Oct 2024 17:45:59 +0200 In-Reply-To: References: <20241018144710.3800385-1-roberto.sassu@huaweicloud.com> Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable User-Agent: Evolution 3.44.4-0ubuntu2 MIME-Version: 1.0 X-CM-TRANSID:LxC2BwC3Ry+6ghJn0PUZAw--.23355S2 X-Coremail-Antispam: 1UD129KBjvJXoW3Gr4UWrykZFy7Cr1DWr43GFg_yoW7tr4fpF 95tas8KF4kXFyxZrn2q3WUWFyrtrW8KFyUu3y3tF1rA3sFvF1fKr4fGFy5uF4DArykCF95 ZF4jyr9xGFWDAFJanT9S1TB71UUUUU7qnTZGkaVYY2UrUUUUjbIjqfuFe4nvWSU5nxnvy2 9KBjDU0xBIdaVrnRJUUUvYb4IE77IF4wAFF20E14v26ryj6rWUM7CY07I20VC2zVCF04k2 6cxKx2IYs7xG6rWj6s0DM7CIcVAFz4kK6r1j6r18M28lY4IEw2IIxxk0rwA2F7IY1VAKz4 vEj48ve4kI8wA2z4x0Y4vE2Ix0cI8IcVAFwI0_Jr0_JF4l84ACjcxK6xIIjxv20xvEc7Cj xVAFwI0_Gr0_Cr1l84ACjcxK6I8E87Iv67AKxVW8JVWxJwA2z4x0Y4vEx4A2jsIEc7CjxV AFwI0_Gr1j6F4UJwAS0I0E0xvYzxvE52x082IY62kv0487Mc02F40EFcxC0VAKzVAqx4xG 6I80ewAv7VC0I7IYx2IY67AKxVWUJVWUGwAv7VC2z280aVAFwI0_Jr0_Gr1lOx8S6xCaFV Cjc4AY6r1j6r4UM4x0Y48IcVAKI48JM4IIrI8v6xkF7I0E8cxan2IY04v7MxkF7I0En4kS 14v26r4a6rW5MxAIw28IcxkI7VAKI48JMxC20s026xCaFVCjc4AY6r1j6r4UMI8I3I0E5I 8CrVAFwI0_Jr0_Jr4lx2IqxVCjr7xvwVAFwI0_JrI_JrWlx4CE17CEb7AF67AKxVW8ZVWr XwCIc40Y0x0EwIxGrwCI42IY6xIIjxv20xvE14v26r1j6r1xMIIF0xvE2Ix0cI8IcVCY1x 0267AKxVW8JVWxJwCI42IY6xAIw20EY4v20xvaj40_Jr0_JF4lIxAIcVC2z280aVAFwI0_ Jr0_Gr1lIxAIcVC2z280aVCY1x0267AKxVW8Jr0_Cr1UYxBIdaVFxhVjvjDU0xZFpf9x07 jIksgUUUUU= X-CM-SenderInfo: purev21wro2thvvxqx5xdzvxpfor3voofrz/1tbiAgAABGcRxH8LzQAAsN X-Rspam-User: X-Rspamd-Queue-Id: 42B8540020 X-Rspamd-Server: rspam01 X-Stat-Signature: i7rmcuxoqhf9ny3qwhu1wkxd1mcqhkx5 X-HE-Tag: 1729266371-810941 X-HE-Meta: U2FsdGVkX1+oVryGBTB0xTJq1D+alYXsJ8/hZHoR6qVha+5SBUyJ+mIZxCLSyLc6RuGgtYOW0fTibuBl6jPihSTnHdIDPAQRe4oyMK2m3aaZ9i2iSGH+CiJZnE36J/xgmXt+QrmGUMxJ2intsDz8VKLE61tdaRcq+4vIZJ2QAbcB19kxkjwAdIelhxs7E5xfaWud2r2RNG75jWN1Jvd00EliS7skVFDeXtXJnfozM1P4x95JJWaWI7B2GdGP0pMw8goBlWfMWYQhCFeJgZ4TKu7dom2KfOr9xk6Ftqo0VjFCBObDjjHf7sEkfLaFPpaslr9oKO6gWmcd8bIaVxgl2mOJppwfXG/6H27UtXnlsgJ2X/pluu1uoAvQN79uzAeen/lJSG1nHGXP4+ZG/kG5cJ09D+OSaRv4yYjAf+avQTCVAEcvaNKb2UUDBE3SXZBygmTBb/PbLdQ5zm2OWqKFitsqFUbMT28OXjPuMzavKOawngK/ZYwAjiHfFTP9rLcdWWB5wQMBBesxCQf07qzWIJtoN0MgEFSmAMBMrwrXbrhAbz1Li5VfzeWoofGoEtEvtraRYvKwJvi82xlTJWCyWW+Hpc/l7NyLtz1Dy1RJzj6gQWCD5IesqkS+hBAWakuD+tIo6rJPpfljQUPgCq147HfA2iNtAKR9BFNfvWVzsJM1ZNqxscgAEB9HxZfXPBuZeQez9mjsNrsXsoNOfUCtW8UwUeeSdXBjnH16jxThQu2UL2YmDLWT3LJ2UrWH3dBlsqByGfxE0/N2LFpPc/XrFHmVLNWhR4IfXQrQjvep5L7/M3KWbg8nCVh1Ixgv+zwA+1DRfSYbP620y5yNdFQaQh7+DvkdTwo8FOXh4SJemFwdu8w7Q5G9cI2q4+NnzLj84J/p5Agolm/49H153QIoVbyPtN4NbN6zM1aTMjlDmQpVOFDO0EfgUeMrK1Qb/nawYyQuMMOtw9vr46IYs7B snlgAtix a0ArRsyhzjCiW9jd47gRPDSM2CninwF0QYCHShMviW4XSCoRb+HM5Zf14tuS0axDkzdQKOBZ13AOPXvVpMNemerJZ+yLH81Yjl2FqYaHBbzsV8S/aD9WQzF7Foz6NQiziqAH1CtKABNT/BV4Pz2Ym9TtpxeM0SAS72hC2VwlcLDPQcKZnNkcRWFTvqEH24ZhmK7S4aUtBg+2VY5trIkb+R+aJ5BinCmMuwz1hS3gLwMi2pIU7wt5T9OKzkUpu12fhC6+UAQnGX+Pf5lTHgLuCW3In2tVIH4HXIvrDtrrLJCqjCTtRqQPtr0M3UsLuYFYmgalfSxq0CwTlJhocerSHrPAjmO3vLQqGQdX+3U6rwnqHD0NWh5wdAMZcJGr2AxLpiHprGJdvdyU9AmHbN3CjDDgpy3aBeIkCUIHAdO4rZj8zw4AwKrxztoA61TTYEaizqDfy X-Bogosity: Ham, tests=bogofilter, spamicity=0.000000, version=1.2.4 Sender: owner-linux-mm@kvack.org Precedence: bulk X-Loop: owner-majordomo@kvack.org List-ID: List-Subscribe: List-Unsubscribe: On Fri, 2024-10-18 at 16:42 +0100, Lorenzo Stoakes wrote: > On Fri, Oct 18, 2024 at 04:47:10PM +0200, Roberto Sassu wrote: > > From: "Kirill A. Shutemov" > >=20 > > Commit ea7e2d5e49c0 ("mm: call the security_mmap_file() LSM hook in > > remap_file_pages()") fixed a security issue, it added an LSM check when > > trying to remap file pages, so that LSMs have the opportunity to evalua= te > > such action like for other memory operations such as mmap() and mprotec= t(). > >=20 > > However, that commit called security_mmap_file() inside the mmap_lock l= ock, > > while the other calls do it before taking the lock, after commit > > 8b3ec6814c83 ("take security_mmap_file() outside of ->mmap_sem"). > >=20 > > This caused lock inversion issue with IMA which was taking the mmap_loc= k > > and i_mutex lock in the opposite way when the remap_file_pages() system > > call was called. > >=20 > > Solve the issue by splitting the critical region in remap_file_pages() = in > > two regions: the first takes a read lock of mmap_lock and retrieves the= VMA > > and the file associated, and calculate the 'prot' and 'flags' variable;= the > > second takes a write lock on mmap_lock, checks that the VMA flags and t= he > > VMA file descriptor are the same as the ones obtained in the first crit= ical > > region (otherwise the system call fails), and calls do_mmap(). > >=20 > > In between, after releasing the read lock and taking the write lock, ca= ll > > security_mmap_file(), and solve the lock inversion issue. >=20 > Great description! >=20 > >=20 > > Cc: stable@vger.kernel.org > > Fixes: ea7e2d5e49c0 ("mm: call the security_mmap_file() LSM hook in rem= ap_file_pages()") > > Reported-by: syzbot+91ae49e1c1a2634d20c0@syzkaller.appspotmail.com > > Closes: https://lore.kernel.org/linux-security-module/66f7b10e.050a0220= .46d20.0036.GAE@google.com/ > > Reviewed-by: Roberto Sassu (Calculate prot a= nd flags earlier) > > Signed-off-by: Kirill A. Shutemov >=20 > Other than some nits below: >=20 > Reviewed-by: Lorenzo Stoakes >=20 > I think you're definitely good to un-RFC here. Perfect, will do. Thank you! Roberto > > --- > > mm/mmap.c | 62 ++++++++++++++++++++++++++++++++++++++++--------------- > > 1 file changed, 45 insertions(+), 17 deletions(-) > >=20 > > diff --git a/mm/mmap.c b/mm/mmap.c > > index 9c0fb43064b5..762944427e03 100644 > > --- a/mm/mmap.c > > +++ b/mm/mmap.c > > @@ -1640,6 +1640,7 @@ SYSCALL_DEFINE5(remap_file_pages, unsigned long, = start, unsigned long, size, > > unsigned long populate =3D 0; > > unsigned long ret =3D -EINVAL; > > struct file *file; > > + vm_flags_t vm_flags; > >=20 > > pr_warn_once("%s (%d) uses deprecated remap_file_pages() syscall. See= Documentation/mm/remap_file_pages.rst.\n", > > current->comm, current->pid); > > @@ -1656,12 +1657,53 @@ SYSCALL_DEFINE5(remap_file_pages, unsigned long= , start, unsigned long, size, > > if (pgoff + (size >> PAGE_SHIFT) < pgoff) > > return ret; > >=20 > > - if (mmap_write_lock_killable(mm)) > > + if (mmap_read_lock_killable(mm)) > > + return -EINTR; >=20 > I'm kinda verbose generally, but I'd love a comment like: >=20 > /* > * Look up VMA under read lock first so we can perform the security > * without holding locks (which can be problematic). We reacquire a > * write lock later and check nothing changed underneath us. > */ >=20 > > + > > + vma =3D vma_lookup(mm, start); > > + > > + if (!vma || !(vma->vm_flags & VM_SHARED)) { > > + mmap_read_unlock(mm); > > + return -EINVAL; > > + } > > + > > + prot |=3D vma->vm_flags & VM_READ ? PROT_READ : 0; > > + prot |=3D vma->vm_flags & VM_WRITE ? PROT_WRITE : 0; > > + prot |=3D vma->vm_flags & VM_EXEC ? PROT_EXEC : 0; > > + > > + flags &=3D MAP_NONBLOCK; > > + flags |=3D MAP_SHARED | MAP_FIXED | MAP_POPULATE; > > + if (vma->vm_flags & VM_LOCKED) > > + flags |=3D MAP_LOCKED; > > + > > + /* Save vm_flags used to calculate prot and flags, and recheck later.= */ > > + vm_flags =3D vma->vm_flags; > > + file =3D get_file(vma->vm_file); > > + > > + mmap_read_unlock(mm); > > + >=20 > Maybe worth adding a comment to explain why you're doing this without the > lock so somebody looking at this later can understand the dance? >=20 > > + ret =3D security_mmap_file(file, prot, flags); > > + if (ret) { > > + fput(file); > > + return ret; > > + } > > + > > + ret =3D -EINVAL; > > + >=20 > Again, being verbose, I'd put something here like: >=20 > /* OK security check passed, take write lock + let it rip */ >=20 > > + if (mmap_write_lock_killable(mm)) { > > + fput(file); > > return -EINTR; > > + } > >=20 > > vma =3D vma_lookup(mm, start); > >=20 > > - if (!vma || !(vma->vm_flags & VM_SHARED)) > > + if (!vma) > > + goto out; > > + >=20 > I'd also add something like: >=20 > /* Make sure things didn't change under us. */ >=20 > > + if (vma->vm_flags !=3D vm_flags) > > + goto out; > > + >=20 > And drop this newline to group them together (super nitty I know, sorry!) >=20 > > + if (vma->vm_file !=3D file) > > goto out; > >=20 > > if (start + size > vma->vm_end) { > > @@ -1689,25 +1731,11 @@ SYSCALL_DEFINE5(remap_file_pages, unsigned long= , start, unsigned long, size, > > goto out; > > } > >=20 > > - prot |=3D vma->vm_flags & VM_READ ? PROT_READ : 0; > > - prot |=3D vma->vm_flags & VM_WRITE ? PROT_WRITE : 0; > > - prot |=3D vma->vm_flags & VM_EXEC ? PROT_EXEC : 0; > > - > > - flags &=3D MAP_NONBLOCK; > > - flags |=3D MAP_SHARED | MAP_FIXED | MAP_POPULATE; > > - if (vma->vm_flags & VM_LOCKED) > > - flags |=3D MAP_LOCKED; > > - > > - file =3D get_file(vma->vm_file); > > - ret =3D security_mmap_file(vma->vm_file, prot, flags); > > - if (ret) > > - goto out_fput; > > ret =3D do_mmap(vma->vm_file, start, size, > > prot, flags, 0, pgoff, &populate, NULL); > > -out_fput: > > - fput(file); > > out: > > mmap_write_unlock(mm); > > + fput(file); > > if (populate) > > mm_populate(ret, populate); > > if (!IS_ERR_VALUE(ret)) > > -- > > 2.34.1 > >=20 >=20 > These are just nits, this looks good to me!