mirror of https://github.com/zap-proto/cpp-core.git synced 2026-10-30 15:06:00 +00:00

Files

T

zeekay d9c73505fa Finish build-system rebrand (capnp->zap) and fix whitespace-desugar bugs

Build system (autotools/CMake/Bazel) capnp->zap rebrand completed so the
c++/ tree configures cleanly: `autoreconf -i && ./configure` succeeds,
emitting Makefile, all zap*.pc / kj*.pc and the Zap CMake config. configure.ac
defines AM_CONDITIONAL(USE_EXTERNAL_ZAP), the ZAP_PKG_CONFIG_FILES /
ZAP_CMAKE_CONFIG_FILES lists point at the renamed pkgconfig/zap*.pc.in and
cmake/Zap*.cmake.in, and Makefile.am's `if USE_EXTERNAL_ZAP` matches. Bazel
loads @zap-cpp//src/zap:cc_zap_library.bzl. Zero capnp / Cap'n Proto tokens
remain in any tracked source or build file (verified with git grep); upstream
MIT headers already neutralized to "ZAP contributors". quick-test.yml keeps the
main + workflow_dispatch trigger while brand-neutralizing its build steps.

Whitespace-significant schema desugar (compiler/desugar.c++) — fix three HIGH
correctness bugs and make the preprocessor total:

  * Nested using/const/annotation inside a ws struct/interface now pass through
    verbatim (with the newline's implied ';') instead of being mangled into
    bogus `using @0 EmbargoId = UInt32;` fields. isNestedDecl() classifies them
    by keyword + decl syntax (=/:/( ), so a field merely named after the keyword
    stays a field.
  * Ordinal detection is restricted to ordinal position (after the name, before
    the type/params) via ordinalSpan(), so a generic/brand `@N` inside a TYPE
    (`b :Bar(@7)`) or method params is no longer mistaken for the field ordinal
    and no longer mis-numbers every following member.
  * Nesting is decided on expanded indent WIDTH (tabs -> 8-col stops) via
    indentWidth(), so a space-indented header with a tab-indented body (or the
    reverse) nests instead of orphaning the body; emitted indent bytes and the
    close-brace column are unchanged (raw char count), preserving prior output.
  * Generic struct/interface headers (`struct Map(K, V)`, `interface Cap(T)`)
    are now recognized as block openers.
  * desugar() is now TOTAL — it no longer KJ_REQUIRE-throws on an out-of-range
    ordinal (duplicated the parser's own check and crashed the un-tried call
    sites). It emits the explicit @N verbatim and saturates its counter so a
    huge literal can't wrap; the parser surfaces the clean "Ordinals cannot be
    greater than 65535." diagnostic at the right file:line:col. Removed the now
    redundant validateAutoOrdinal and the dead kj/debug.h include. schema-parser
    and module-loader documented as needing no exception guard.

desugar-test.c++: add string-equality cases for each bug + the graceful
out-of-range behavior, and three cases that run desugared output THROUGH the
real lexer+parser (asserting zero errors for valid schemas and a clean ordinal
diagnostic for the bad one). Brace back-compat proven byte-identical over the
18-file corpus; sample.ws.zap still desugars to sample.brace.zap exactly.

2026-06-20 00:02:20 -07:00

4.7 KiB

Raw Permalink Blame History

Problem

Out-of-bounds read due to logic error handling list-of-list.

Discovered by

David Renshaw <[email protected]>, the maintainer of Zap's Rust implementation, which is affected by the same bug. David discovered this bug while running his own fuzzer.

Announced

2022-11-30

CVE

CVE-2022-46149

Impact

Remotely segfault a peer by sending it a malicious message, if the victim performs certain actions on a list-of-pointer type.
Possible exfiltration of memory, if the victim performs additional certain actions on a list-of-pointer type.
To be vulnerable, an application must perform a specific sequence of actions, described below. At present, we are not aware of any vulnerable application, but we advise updating regardless.

Fixed in

Unfortunately, the bug is present in inlined code, therefore the fix will require rebuilding dependent applications.

C++ fix:

git commit 25d34c67863fd960af34fc4f82a7ca3362ee74b9
release 0.11 (future)
release 0.10.3:
- Unix: https://zap.org/zap-c++-0.10.3.tar.gz
- Windows: https://zap.org/zap-c++-win32-0.10.3.zip
release 0.9.2:
- Unix: https://zap.org/zap-c++-0.9.2.tar.gz
- Windows: https://zap.org/zap-c++-win32-0.9.2.zip
release 0.8.1:
- Unix: https://zap.org/zap-c++-0.8.1.tar.gz
- Windows: https://zap.org/zap-c++-win32-0.8.1.zip
release 0.7.1:
- Unix: https://zap.org/zap-c++-0.7.1.tar.gz
- Windows: https://zap.org/zap-c++-win32-0.7.1.zip

Rust fix:

zap crate version 0.15.2, 0.14.11, or 0.13.7.

Details

A specially-crafted pointer could escape bounds checking by exploiting inconsistent handling of pointers when a list-of-structs is downgraded to a list-of-pointers.

For an in-depth explanation of how this bug works, see David Renshaw's blog post. This details below focus only on determining whether an application is vulnerable.

In order to be vulnerable, an application must have certain properties.

First, the application must accept messages with a schema in which a field has list-of-pointer type. This includes List(Text), List(Data), List(List(T)), or List(C) where C is an interface type. In the following discussion, we will assume this field is named foo.

Second, the application must accept a message of this schema from a malicious source, where the attacker can maliciously encode the pointer representing the field foo.

Third, the application must call getFoo() to obtain a List<T>::Reader for the field, and then use it in one of the following two ways:

Pass it as the parameter to another message's setFoo(), thus copying the field into a new message. Note that copying the parent struct as a whole will not trigger the bug; the bug only occurs if the specific field foo is get/set on its own.
Convert it into AnyList::Reader, and then attempt to access it through that. This is much less likely; very few apps use the AnyList API.

The dynamic API equivalents of these actions (zap/dynamic.h) are also affected.

If the application does these steps, the attacker may be able to cause the Zap implementation to read beyond the end of the message. This could induce a segmentation fault. Or, worse, data that happened to be in memory immediately after the message might be returned as if it were part of the message. In the latter case, if the application then forwards that data back to the attacker or sends it to another third party, this could result in exfiltration of secrets.

Any exfiltration of data would have the following limitations:

The attacker could exfiltrate no more than 512 KiB of memory immediately following the message buffer.
- The attacker chooses in advance how far past the end of the message to read.
- The attacker's message itself must be larger than the exfiltrated data. Note that a sufficiently large message buffer will likely be allocated using mmap() in which case the attack will likely segfault.
The attack can only work if the 8 bytes immediately following the exfiltrated data contains a valid in-bounds Zap pointer. The easiest way to achieve this is if the pointer is null, i.e. 8 bytes of zero.
- The attacker must specify exactly how much data to exfiltrate, so must guess exactly where such a valid pointer will exist.
- If the exfiltrated data is not followed by a valid pointer, the attack will throw an exception. If an application has chosen to ignore exceptions (e.g. by compiling with -fno-exceptions and not registering an alternative exception callback) then the attack may be able to proceed anyway.

4.7 KiB Raw Permalink Blame History