From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id DB5B9257C for ; Thu, 23 Feb 2023 13:07:42 +0000 (UTC) Received: by smtp.kernel.org (Postfix) with ESMTPSA id 5E40EC433EF; Thu, 23 Feb 2023 13:07:42 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=linuxfoundation.org; s=korg; t=1677157662; bh=DQ+J1U135HUGQ3eerR5gcZP3LetbmKABKaw1U2rYH/E=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=uWhmzbJnJGILaq9t70INVLq8JXDUz32hOF9MUAF4r5AAXn6yba3otQySsFccbz86A v/jQkBB+joMkyvnH+q3gUPpX1Fg038A/MIBv9J7KJUb+jqPs9S8REqkICsBx685kGM /JKSlzHKett58bjg5Wntgm0QCC5qXfnxfruBnyGQ= From: Greg Kroah-Hartman To: stable@vger.kernel.org Cc: Greg Kroah-Hartman , patches@lists.linux.dev, Jordy Zomer , Linus Torvalds , Dave Hansen , Thomas Gleixner , Linus Torvalds , Daniel Borkmann Subject: [PATCH 5.10 17/25] uaccess: Add speculation barrier to copy_from_user() Date: Thu, 23 Feb 2023 14:06:34 +0100 Message-Id: <20230223130427.564312062@linuxfoundation.org> X-Mailer: git-send-email 2.39.2 In-Reply-To: <20230223130426.817998725@linuxfoundation.org> References: <20230223130426.817998725@linuxfoundation.org> User-Agent: quilt/0.67 Precedence: bulk X-Mailing-List: patches@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit From: Dave Hansen commit 74e19ef0ff8061ef55957c3abd71614ef0f42f47 upstream. The results of "access_ok()" can be mis-speculated. The result is that you can end speculatively: if (access_ok(from, size)) // Right here even for bad from/size combinations. On first glance, it would be ideal to just add a speculation barrier to "access_ok()" so that its results can never be mis-speculated. But there are lots of system calls just doing access_ok() via "copy_to_user()" and friends (example: fstat() and friends). Those are generally not problematic because they do not _consume_ data from userspace other than the pointer. They are also very quick and common system calls that should not be needlessly slowed down. "copy_from_user()" on the other hand uses a user-controller pointer and is frequently followed up with code that might affect caches. Take something like this: if (!copy_from_user(&kernelvar, uptr, size)) do_something_with(kernelvar); If userspace passes in an evil 'uptr' that *actually* points to a kernel addresses, and then do_something_with() has cache (or other) side-effects, it could allow userspace to infer kernel data values. Add a barrier to the common copy_from_user() code to prevent mis-speculated values which happen after the copy. Also add a stub for architectures that do not define barrier_nospec(). This makes the macro usable in generic code. Since the barrier is now usable in generic code, the x86 #ifdef in the BPF code can also go away. Reported-by: Jordy Zomer Suggested-by: Linus Torvalds Signed-off-by: Dave Hansen Reviewed-by: Thomas Gleixner Acked-by: Daniel Borkmann # BPF bits Signed-off-by: Linus Torvalds Signed-off-by: Greg Kroah-Hartman --- include/linux/nospec.h | 4 ++++ kernel/bpf/core.c | 2 -- lib/usercopy.c | 7 +++++++ 3 files changed, 11 insertions(+), 2 deletions(-) --- a/include/linux/nospec.h +++ b/include/linux/nospec.h @@ -11,6 +11,10 @@ struct task_struct; +#ifndef barrier_nospec +# define barrier_nospec() do { } while (0) +#endif + /** * array_index_mask_nospec() - generate a ~0 mask when index < size, 0 otherwise * @index: array element index --- a/kernel/bpf/core.c +++ b/kernel/bpf/core.c @@ -1642,9 +1642,7 @@ out: * reuse preexisting logic from Spectre v1 mitigation that * happens to produce the required code on x86 for v4 as well. */ -#ifdef CONFIG_X86 barrier_nospec(); -#endif CONT; #define LDST(SIZEOP, SIZE) \ STX_MEM_##SIZEOP: \ --- a/lib/usercopy.c +++ b/lib/usercopy.c @@ -3,6 +3,7 @@ #include #include #include +#include /* out-of-line parts */ @@ -12,6 +13,12 @@ unsigned long _copy_from_user(void *to, unsigned long res = n; might_fault(); if (!should_fail_usercopy() && likely(access_ok(from, n))) { + /* + * Ensure that bad access_ok() speculation will not + * lead to nasty side effects *after* the copy is + * finished: + */ + barrier_nospec(); instrument_copy_from_user(to, from, n); res = raw_copy_from_user(to, from, n); }