--- /dev/null
+/*
+ * Copyright (c) 2000-2006,2011-2012,2014 Apple Inc. All Rights Reserved.
+ *
+ * @APPLE_LICENSE_HEADER_START@
+ *
+ * This file contains Original Code and/or Modifications of Original Code
+ * as defined in and that are subject to the Apple Public Source License
+ * Version 2.0 (the 'License'). You may not use this file except in
+ * compliance with the License. Please obtain a copy of the License at
+ * http://www.opensource.apple.com/apsl/ and read it before using this
+ * file.
+ *
+ * The Original Code and all software distributed under the License are
+ * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
+ * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
+ * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
+ * Please see the License for the specific language governing rights and
+ * limitations under the License.
+ *
+ * @APPLE_LICENSE_HEADER_END@
+ */
+
+
+//
+// acl_comment - "ignore" ACL subject type.
+//
+// CommentAclSubjects were a bad idea, badly implemented. The code below
+// exists solely to keep existing (external) ACL forms from blowing up the
+// ACL reader machinery and crashing the evaluation host.
+// The original serialization code was not architecture independent - for either
+// pointer sizes(!) or byte ordering. Yes, that was a stupid mistake.
+// The following code is intentionally, wilfully violating the layer separation
+// of the ACL reader/writer machine to deduce enough information about the
+// originating architecture to cleanly consume (just) the bytes making up this
+// ACL's external representation. We make no use of the bytes read; thankfully,
+// the semantics of a CommentAclSubject have always been "never matches."
+// We do not preserve them on write-out; a newly-written ACL will contain no data
+// (and will read cleanly).
+// If you use this code as a template for anything (other than a how-not-to-write-code
+// seminar), your backups shall rot right after your main harddrive crashes, and
+// you have only yourself to blame.
+//
+#include <security_cdsa_utilities/acl_comment.h>
+#include <security_cdsa_utilities/cssmwalkers.h>
+#include <security_cdsa_utilities/cssmlist.h>
+#include <algorithm>
+
+using namespace DataWalkers;
+
+
+//
+// The COMMENT subject matches nothing, no matter how pretty.
+//
+bool CommentAclSubject::validate(const AclValidationContext &) const
+{
+ return false;
+}
+
+
+//
+// The list form has no values.
+//
+CssmList CommentAclSubject::toList(Allocator &alloc) const
+{
+ return TypedList(Allocator::standard(), CSSM_ACL_SUBJECT_TYPE_COMMENT);
+}
+
+
+//
+// We completely disregard any data contained in CSSM form COMMENT ACLs.
+//
+CommentAclSubject *CommentAclSubject::Maker::make(const TypedList &list) const
+{
+ return new CommentAclSubject();
+}
+
+
+//
+// This is the nasty code. We don't really care what data was originally baked
+// into this ACL's external (stream) form, but since there's no external framing
+// to delimit it, we need to figure out how many bytes to consume to keep the
+// reader from going out of sync. And that's not pretty, since the external form
+// contains (stupidly!) a pointer, so we have all permutations of byte order and
+// pointer size to worry about.
+//
+CommentAclSubject *CommentAclSubject::Maker::make(Version, Reader &pub, Reader &) const
+{
+ //
+ // At this point, the Reader is positioned at data that was once written using
+ // this code:
+ // pub(ptr); // yes, that's a pointer
+ // pub.countedData(ptr, size);
+ // We know ptr was a non-NULL pointer (4 or 8 bytes, alas).
+ // CountedData writes a 4-byte NBO length followed by that many bytes.
+ // The data written starts with a CSSM_LIST structure in native architecture.
+ // That in turn begins with a CSSM_LIST_TYPE (4 bytes, native, 0<=type<=2).
+ // So to summarize (h=host byte order, n=network byte order), we might be looking at:
+ // 32 bits: | P4h | L4n | T4h | (L-4 bytes) |
+ // 64 bits: | P8h | L4n | (L bytes) |
+ // It's the T4h-or-L4n bytes that save our day, since we know that
+ // 0 <= T <= 2 (definition of CSSM_LIST_TYPE)
+ // 16M > L >= sizeof(CSSM_LIST) >= 12
+ // Phew. I'd rather be lucky than good...
+ //
+ // So let's get started:
+#ifndef NDEBUG
+ static const size_t minCssmList = 12; // min(sizeof(CSSM_LIST)) of all architectures
+#endif
+ pub.get<void>(4); // skip first 4 bytes
+ uint32_t lop; pub(lop); // read L4n-or-(bottom of)P8h
+ uint32_t tol; pub(tol); // read T4h-or-L4n
+ if (tol <= 2 || flip(tol) <= 2) { // 32 bits
+ // the latter can't be a very big (flipped) L because we know 12 < L < 16M,
+ // and you'd have to be a multiple of 2^24 to pass that test
+ size_t length = n2h(lop);
+ assert(length >= minCssmList);
+ pub.get<void>(length - sizeof(tol)); // skip L-4 bytes
+ } else { // 64 bits
+ size_t length = n2h(tol);
+ assert(length >= minCssmList);
+ pub.get<void>(length); // skip L bytes
+ }
+
+ // we've successfully thrown out the garbage. What's left is a data-less subject
+ return new CommentAclSubject(); // no data
+}
+
+
+//
+// Export to blob form.
+// This simply writes the smallest form consistent with the heuristic above.
+//
+void CommentAclSubject::exportBlob(Writer::Counter &pub, Writer::Counter &)
+{
+ uint32_t zero = 0;
+ Endian<uint32_t> length = 12;
+ pub(zero); pub(length); pub(zero); pub(zero); pub(zero);
+}
+
+void CommentAclSubject::exportBlob(Writer &pub, Writer &)
+{
+ uint32_t zero = 0;
+ Endian<uint32_t> length = 12;
+ pub(zero); pub(length); pub(zero); pub(zero); pub(zero);
+}
+
+
+#ifdef DEBUGDUMP
+
+void CommentAclSubject::debugDump() const
+{
+ Debug::dump("Comment[never]");
+}
+
+#endif //DEBUGDUMP
projects / apple / security.git / blobdiff