linux-arch.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Mark Rutland <mark.rutland@arm.com>
To: linux-kernel@vger.kernel.org
Cc: linux-arch@vger.kernel.org, Mark Rutland <mark.rutland@arm.com>,
	Will Deacon <will.deacon@arm.com>
Subject: [RFC PATCH 2/4] Documentation: document nospec helpers
Date: Wed,  3 Jan 2018 22:38:25 +0000	[thread overview]
Message-ID: <20180103223827.39601-3-mark.rutland@arm.com> (raw)
In-Reply-To: <20180103223827.39601-1-mark.rutland@arm.com>

Document the rationale and usage of the new nospec*() helpers.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
---
 Documentation/speculation.txt | 99 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 99 insertions(+)
 create mode 100644 Documentation/speculation.txt

diff --git a/Documentation/speculation.txt b/Documentation/speculation.txt
new file mode 100644
index 000000000000..0bec4ed5ac29
--- /dev/null
+++ b/Documentation/speculation.txt
@@ -0,0 +1,99 @@
+This document explains potential effects of speculation, and how undesirable
+effects can be mitigated portably using common APIs.
+
+===========
+Speculation
+===========
+
+To improve performance and minimize average latencies, many contemporary CPUs
+employ speculative execution techniques such as branch prediction, performing
+work which may be discarded at a later stage.
+
+Typically speculative execution cannot be observed from architectural state,
+such as the contents of registers. However, in some cases it is possible to
+observe its impact on microarchitectural state, such as the presence or
+absence of data in caches. Such state may form side-channels which can be
+observed to extract secret information.
+
+For example, in the presence of branch prediction, it is possible for bounds
+checks to be ignored by code which is speculatively executed. Consider the
+following code:
+
+	int load_array(int *array, unsigned int idx) {
+		if (idx >= MAX_ARRAY_ELEMS)
+			return 0;
+		else
+			return array[idx];
+	}
+
+Which, on arm64, may be compiled to an assembly sequence such as:
+
+	CMP	<idx>, #MAX_ARRAY_ELEMS
+	B.LT	less
+	MOV	<returnval>, #0
+	RET
+  less:
+	LDR	<returnval>, [<array>, <idx>]
+	RET
+
+It is possible that a CPU mis-predicts the conditional branch, and
+speculatively loads array[idx], even if idx >= MAX_ARRAY_ELEMS. This value
+will subsequently be discarded, but the speculated load may affect
+microarchitectural state which can be subsequently measured.
+
+More complex sequences involving multiple dependent memory accesses may result
+in sensitive information being leaked. Consider the following code, building on
+the prior example:
+
+	int load_dependent_arrays(int *arr1, int *arr2, int idx) {
+		int val1, val2,
+
+		val1 = load_array(arr1, idx);
+		val2 = load_array(arr2, val1);
+
+		return val2;
+	}
+
+Under speculation, the first call to load_array() may return the value of an
+out-of-bounds address, while the second call will influence microarchitectural
+state dependent on this value. This may provide an arbitrary read primitive.
+
+====================================
+Mitigating speculation side-channels
+====================================
+
+The kernel provides a generic API to ensure that bounds checks are respected
+even under speculation. Architectures which are affected by speculation-based
+side-channels are expected to implement these primitives.
+
+The following helpers found in <asm/barrier.h> can be used to prevent
+information from being leaked via side-channels.
+
+* nospec_load(ptr, lo, hi)
+
+  Returns the data at *ptr only if ptr falls in the [lo, hi) interval. When
+  ptr < lo or ptr >= hi, typeof(*ptr)0 is returned, even under speculation.
+
+  This does not prevent an out-of-bounds load from being speculated, but does
+  prevent its value from influencing code which is subsequently speculated,
+  preventing the value from being leaked.
+
+* nospec_array_load(arr, idx, sz)
+
+  Returns the data at arr[idx] only if idx falls in the [0, sz) interval. When
+  idx < 0 or idx > sz, typeof(*arr)0 is returned, even under speculation.
+
+  This is a wrapper around nospec_load() provided for convenience.
+
+* nospec_ptr(ptr, lo, hi)
+
+  Returns a sanitized pointer that is bounded by the [lo, hi) interval, even
+  under speculation. If ptr < lo, or ptr >= hi, NULL is returned.
+
+  This is expected to be used by code which computes a pointer to an element
+  of a data structure, or where multiple fields of a data structure will be
+  accessed.
+
+  Note that it is not safe to compare the returned value to the original
+  pointer, as compiler optimizations may infer that the original unsanitized
+  pointer is safe to use when the two compare equal.
-- 
2.11.0

  parent reply	other threads:[~2018-01-03 22:38 UTC|newest]

Thread overview: 82+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-01-03 22:38 [RFC PATCH 0/4] API for inhibiting speculative arbitrary read primitives Mark Rutland
2018-01-03 22:38 ` [RFC PATCH 1/4] asm-generic/barrier: add generic nospec helpers Mark Rutland
2018-01-03 22:38   ` Mark Rutland
2018-01-04 12:00   ` Mark Rutland
2018-01-05  4:21     ` Dan Williams
2018-01-05  9:15       ` Mark Rutland
2018-01-03 22:38 ` Mark Rutland [this message]
2018-01-03 22:38   ` [RFC PATCH 2/4] Documentation: document " Mark Rutland
2018-01-03 22:38 ` [RFC PATCH 3/4] arm64: implement nospec_{load,ptr}() Mark Rutland
2018-01-03 22:38   ` Mark Rutland
2018-01-03 22:38 ` [RFC PATCH 4/4] bpf: inhibit speculated out-of-bounds pointers Mark Rutland
2018-01-03 22:38   ` Mark Rutland
2018-01-03 23:45   ` Peter Zijlstra
2018-01-03 23:45     ` Peter Zijlstra
2018-01-04 10:59     ` Mark Rutland
2018-01-04  0:15 ` [RFC PATCH] asm/generic: introduce if_nospec and nospec_barrier Dan Williams
2018-01-04  0:15   ` Dan Williams
2018-01-04  0:39   ` Linus Torvalds
2018-01-04  1:07     ` Alan Cox
2018-01-04  1:13       ` Dan Williams
2018-01-04  1:13         ` Dan Williams
2018-01-04  6:28         ` Julia Lawall
2018-01-04 17:58           ` Dan Williams
2018-01-04 19:26             ` Pavel Machek
2018-01-04 19:26               ` Pavel Machek
2018-01-04 21:43               ` Dan Williams
2018-01-04 22:20                 ` Linus Torvalds
2018-01-04 22:23                   ` Linus Torvalds
2018-01-04 22:55                   ` Alan Cox
2018-01-04 23:06                     ` Linus Torvalds
2018-01-04 23:11                       ` Alan Cox
2018-01-04 23:11                         ` Alan Cox
2018-01-05  0:24                       ` Dan Williams
2018-01-04 22:44                 ` Pavel Machek
2018-01-04 23:12                   ` Dan Williams
2018-01-04 23:12                     ` Dan Williams
2018-01-04 23:21                     ` Alan Cox
2018-01-04 23:33                     ` Pavel Machek
2018-01-05  8:11                       ` Julia Lawall
2018-01-04  1:27       ` Jiri Kosina
2018-01-04  1:27         ` Jiri Kosina
2018-01-04  1:41         ` Alan Cox
2018-01-04  1:47           ` Jiri Kosina
2018-01-04  1:47             ` Jiri Kosina
2018-01-04 19:39             ` Pavel Machek
2018-01-04 20:32               ` Alan Cox
2018-01-04 20:32                 ` Alan Cox
2018-01-04 20:39                 ` Jiri Kosina
2018-01-04 21:23                   ` Alan Cox
2018-01-04 21:23                     ` Alan Cox
2018-01-04 21:48                     ` Pavel Machek
2018-01-04  1:51         ` Dan Williams
2018-01-04  1:51           ` Dan Williams
2018-01-04  1:54           ` Linus Torvalds
2018-01-04  1:54             ` Linus Torvalds
2018-01-04  3:10             ` Williams, Dan J
2018-01-04  4:44               ` Al Viro
2018-01-04  5:44                 ` Dan Williams
2018-01-04  5:49                   ` Dave Hansen
2018-01-04  5:49                     ` Dave Hansen
2018-01-04  5:50                   ` Al Viro
2018-01-04  5:55                     ` Al Viro
2018-01-04  6:42                       ` Dan Williams
2018-01-04  5:01               ` Eric W. Biederman
2018-01-04  6:32                 ` Dan Williams
2018-01-04 14:54                   ` Eric W. Biederman
2018-01-04 16:39                     ` Mark Rutland
2018-01-04 20:56                     ` Pavel Machek
2018-01-04 20:56                       ` Pavel Machek
2018-01-04 11:47               ` Mark Rutland
2018-01-04 11:47                 ` Mark Rutland
2018-01-04 22:09                 ` Dan Williams
2018-01-05 14:40                   ` Mark Rutland
2018-01-05 16:44                     ` Dan Williams
2018-01-05 18:05                       ` Dan Williams
2018-01-04  1:59           ` Jiri Kosina
2018-01-04  1:59             ` Jiri Kosina
2018-01-04  2:15             ` Alan Cox
2018-01-04  3:12               ` Alexei Starovoitov
2018-01-04  9:16                 ` Reshetova, Elena
2018-01-04  9:16                   ` Reshetova, Elena
2018-01-04 20:40             ` Pavel Machek

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20180103223827.39601-3-mark.rutland@arm.com \
    --to=mark.rutland@arm.com \
    --cc=linux-arch@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=will.deacon@arm.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).