Message-ID: <20250206181711.1902989-6-elver@google.com>
Date: Thu,  6 Feb 2025 19:09:59 +0100
From: Marco Elver <elver@...gle.com>
To: elver@...gle.com
Cc: "Paul E. McKenney" <paulmck@...nel.org>, Alexander Potapenko <glider@...gle.com>, 
	Bart Van Assche <bvanassche@....org>, Bill Wendling <morbo@...gle.com>, Boqun Feng <boqun.feng@...il.com>, 
	Dmitry Vyukov <dvyukov@...gle.com>, Frederic Weisbecker <frederic@...nel.org>, 
	Greg Kroah-Hartman <gregkh@...uxfoundation.org>, Ingo Molnar <mingo@...nel.org>, 
	Jann Horn <jannh@...gle.com>, Joel Fernandes <joel@...lfernandes.org>, 
	Jonathan Corbet <corbet@....net>, Josh Triplett <josh@...htriplett.org>, 
	Justin Stitt <justinstitt@...gle.com>, Kees Cook <kees@...nel.org>, 
	Mark Rutland <mark.rutland@....com>, Mathieu Desnoyers <mathieu.desnoyers@...icios.com>, 
	Miguel Ojeda <ojeda@...nel.org>, Nathan Chancellor <nathan@...nel.org>, 
	Neeraj Upadhyay <neeraj.upadhyay@...nel.org>, Nick Desaulniers <ndesaulniers@...gle.com>, 
	Peter Zijlstra <peterz@...radead.org>, Steven Rostedt <rostedt@...dmis.org>, 
	Thomas Gleixner <tglx@...utronix.de>, Uladzislau Rezki <urezki@...il.com>, Waiman Long <longman@...hat.com>, 
	Will Deacon <will@...nel.org>, kasan-dev@...glegroups.com, linux-kernel@...r.kernel.org, 
	llvm@...ts.linux.dev, rcu@...r.kernel.org, linux-crypto@...r.kernel.org
Subject: [PATCH RFC 05/24] Documentation: Add documentation for Compiler-Based
 Capability Analysis

Adds documentation in Documentation/dev-tools/capability-analysis.rst,
and adds it to the index and cross-references from Sparse's document.

Signed-off-by: Marco Elver <elver@...gle.com>
---
 .../dev-tools/capability-analysis.rst         | 147 ++++++++++++++++++
 Documentation/dev-tools/index.rst             |   1 +
 Documentation/dev-tools/sparse.rst            |   4 +
 3 files changed, 152 insertions(+)
 create mode 100644 Documentation/dev-tools/capability-analysis.rst

diff --git a/Documentation/dev-tools/capability-analysis.rst b/Documentation/dev-tools/capability-analysis.rst
new file mode 100644
index 000000000000..2211af90e01b
--- /dev/null
+++ b/Documentation/dev-tools/capability-analysis.rst
@@ -0,0 +1,147 @@
+.. SPDX-License-Identifier: GPL-2.0
+.. Copyright (C) 2025, Google LLC.
+
+.. _capability-analysis:
+
+Compiler-Based Capability Analysis
+==================================
+
+Capability analysis is a C language extension, which enables statically
+checking that user-definable "capabilities" are acquired and released where
+required. An obvious application is lock-safety checking for the kernel's
+various synchronization primitives (each of which represents a "capability"),
+and checking that locking rules are not violated.
+
+The Clang compiler currently supports the full set of capability analysis
+features. To enable for Clang, configure the kernel with::
+
+    CONFIG_WARN_CAPABILITY_ANALYSIS=y
+
+The analysis is *opt-in by default*, and requires declaring which modules and
+subsystems should be analyzed in the respective `Makefile`::
+
+    CAPABILITY_ANALYSIS_mymodule.o := y
+
+Or for all translation units in the directory::
+
+    CAPABILITY_ANALYSIS := y
+
+It is possible to enable the analysis tree-wide, however, which will result in
+numerous false positive warnings currently and is *not* generally recommended::
+
+    CONFIG_WARN_CAPABILITY_ANALYSIS_ALL=y
+
+Independent of the above Clang support, a subset of the analysis is supported
+by :ref:`Sparse <sparse>`, with weaker guarantees (fewer false positives with
+tree-wide analysis, more more false negatives). Compared to Sparse, Clang's
+analysis is more complete.
+
+Programming Model
+-----------------
+
+The below describes the programming model around using capability-enabled
+types.
+
+.. note::
+   Enabling capability analysis can be seen as enabling a dialect of Linux C with
+   a Capability System. Some valid patterns involving complex control-flow are
+   constrained (such as conditional acquisition and later conditional release
+   in the same function, or returning pointers to capabilities from functions.
+
+Capability analysis is a way to specify permissibility of operations to depend
+on capabilities being held (or not held). Typically we are interested in
+protecting data and code by requiring some capability to be held, for example a
+specific lock. The analysis ensures that the caller cannot perform the
+operation without holding the appropriate capability.
+
+Capabilities are associated with named structs, along with functions that
+operate on capability-enabled struct instances to acquire and release the
+associated capability.
+
+Capabilities can be held either exclusively or shared. This mechanism allows
+assign more precise privileges when holding a capability, typically to
+distinguish where a thread may only read (shared) or also write (exclusive) to
+guarded data.
+
+The set of capabilities that are actually held by a given thread at a given
+point in program execution is a run-time concept. The static analysis works by
+calculating an approximation of that set, called the capability environment.
+The capability environment is calculated for every program point, and describes
+the set of capabilities that are statically known to be held, or not held, at
+that particular point. This environment is a conservative approximation of the
+full set of capabilities that will actually held by a thread at run-time.
+
+More details are also documented `here
+<https://clang.llvm.org/docs/ThreadSafetyAnalysis.html>`_.
+
+.. note::
+   Unlike Sparse's context tracking analysis, Clang's analysis explicitly does
+   not infer capabilities acquired or released by inline functions. It requires
+   explicit annotations to (a) assert that it's not a bug if a capability is
+   released or acquired, and (b) to retain consistency between inline and
+   non-inline function declarations.
+
+Supported Kernel Primitives
+~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. Currently the following synchronization primitives are supported:
+
+For capabilities with an initialization function (e.g., `spin_lock_init()`),
+calling this function on the capability instance before initializing any
+guarded members or globals prevents the compiler from issuing warnings about
+unguarded initialization.
+
+Lockdep assertions, such as `lockdep_assert_held()`, inform the compiler's
+capability analysis that the associated synchronization primitive is held after
+the assertion. This avoids false positives in complex control-flow scenarios
+and encourages the use of Lockdep where static analysis is limited. For
+example, this is useful when a function doesn't *always* require a lock, making
+`__must_hold()` inappropriate.
+
+Keywords
+~~~~~~~~
+
+.. kernel-doc:: include/linux/compiler-capability-analysis.h
+   :identifiers: struct_with_capability
+                 token_capability token_capability_instance
+                 __var_guarded_by __ref_guarded_by
+                 __must_hold
+                 __must_not_hold
+                 __acquires
+                 __cond_acquires
+                 __releases
+                 __must_hold_shared
+                 __acquires_shared
+                 __cond_acquires_shared
+                 __releases_shared
+                 __acquire
+                 __release
+                 __cond_acquire
+                 __acquire_shared
+                 __release_shared
+                 __cond_acquire_shared
+                 capability_unsafe
+                 __no_capability_analysis
+                 disable_capability_analysis enable_capability_analysis
+
+Background
+----------
+
+Clang originally called the feature `Thread Safety Analysis
+<https://clang.llvm.org/docs/ThreadSafetyAnalysis.html>`_, with some
+terminology still using the thread-safety-analysis-only names. This was later
+changed and the feature become more flexible, gaining the ability to define
+custom "capabilities".
+
+Indeed, its foundations can be found in `capability systems
+<https://www.cs.cornell.edu/talc/papers/capabilities.pdf>`_, used to specify
+the permissibility of operations to depend on some capability being held (or
+not held).
+
+Because the feature is not just able to express capabilities related to
+synchronization primitives, the naming chosen for the kernel departs from
+Clang's initial "Thread Safety" nomenclature and refers to the feature as
+"Capability Analysis" to avoid confusion. The implementation still makes
+references to the older terminology in some places, such as `-Wthread-safety`
+being the warning enabled option that also still appears in diagnostic
+messages.
diff --git a/Documentation/dev-tools/index.rst b/Documentation/dev-tools/index.rst
index 65c54b27a60b..62ac23f797cd 100644
--- a/Documentation/dev-tools/index.rst
+++ b/Documentation/dev-tools/index.rst
@@ -18,6 +18,7 @@ Documentation/process/debugging/index.rst
    :maxdepth: 2
 
    testing-overview
+   capability-analysis
    checkpatch
    clang-format
    coccinelle
diff --git a/Documentation/dev-tools/sparse.rst b/Documentation/dev-tools/sparse.rst
index dc791c8d84d1..8c2077834b6f 100644
--- a/Documentation/dev-tools/sparse.rst
+++ b/Documentation/dev-tools/sparse.rst
@@ -2,6 +2,8 @@
 .. Copyright 2004 Pavel Machek <pavel@....cz>
 .. Copyright 2006 Bob Copeland <me@...copeland.com>
 
+.. _sparse:
+
 Sparse
 ======
 
@@ -72,6 +74,8 @@ releasing the lock inside the function in a balanced way, no
 annotation is needed.  The three annotations above are for cases where
 sparse would otherwise report a context imbalance.
 
+Also see :ref:`Compiler-Based Capability Analysis <capability-analysis>`.
+
 Getting sparse
 --------------
 
-- 
2.48.1.502.g6dc24dfdaf-goog

Powered by blists - more mailing lists