Merge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild.git

Documentation/kbuild/gendwarfksyms.rst

@@ -0,0 +1,308 @@
+=======================
+DWARF module versioning
+=======================
+
+1. Introduction
+===============
+
+When CONFIG_MODVERSIONS is enabled, symbol versions for modules
+are typically calculated from preprocessed source code using the
+**genksyms** tool.  However, this is incompatible with languages such
+as Rust, where the source code has insufficient information about
+the resulting ABI. With CONFIG_GENDWARFKSYMS (and CONFIG_DEBUG_INFO)
+selected, **gendwarfksyms** is used instead to calculate symbol versions
+from the DWARF debugging information, which contains the necessary
+details about the final module ABI.
+
+1.1. Usage
+==========
+
+gendwarfksyms accepts a list of object files on the command line, and a
+list of symbol names (one per line) in standard input::
+
+        Usage: gendwarfksyms [options] elf-object-file ... < symbol-list
+
+        Options:
+          -d, --debug          Print debugging information
+              --dump-dies      Dump DWARF DIE contents
+              --dump-die-map   Print debugging information about die_map changes
+              --dump-types     Dump type strings
+              --dump-versions  Dump expanded type strings used for symbol versions
+          -s, --stable         Support kABI stability features
+          -T, --symtypes file  Write a symtypes file
+          -h, --help           Print this message
+
+
+2. Type information availability
+================================
+
+While symbols are typically exported in the same translation unit (TU)
+where they're defined, it's also perfectly fine for a TU to export
+external symbols. For example, this is done when calculating symbol
+versions for exports in stand-alone assembly code.
+
+To ensure the compiler emits the necessary DWARF type information in the
+TU where symbols are actually exported, gendwarfksyms adds a pointer
+to exported symbols in the `EXPORT_SYMBOL()` macro using the following
+macro::
+
+        #define __GENDWARFKSYMS_EXPORT(sym)                             \
+                static typeof(sym) *__gendwarfksyms_ptr_##sym __used    \
+                        __section(".discard.gendwarfksyms") = &sym;
+
+
+When a symbol pointer is found in DWARF, gendwarfksyms can use its
+type for calculating symbol versions even if the symbol is defined
+elsewhere. The name of the symbol pointer is expected to start with
+`__gendwarfksyms_ptr_`, followed by the name of the exported symbol.
+
+3. Symtypes output format
+=========================
+
+Similarly to genksyms, gendwarfksyms supports writing a symtypes
+file for each processed object that contain types for exported
+symbols and each referenced type that was used in calculating symbol
+versions. These files can be useful when trying to determine what
+exactly caused symbol versions to change between builds. To generate
+symtypes files during a kernel build, set `KBUILD_SYMTYPES=1`.
+
+Matching the existing format, the first column of each line contains
+either a type reference or a symbol name. Type references have a
+one-letter prefix followed by "#" and the name of the type. Four
+reference types are supported::
+
+        e#<type> = enum
+        s#<type> = struct
+        t#<type> = typedef
+        u#<type> = union
+
+Type names with spaces in them are wrapped in single quotes, e.g.::
+
+        s#'core::result::Result<u8, core::num::error::ParseIntError>'
+
+The rest of the line contains a type string. Unlike with genksyms that
+produces C-style type strings, gendwarfksyms uses the same simple parsed
+DWARF format produced by **--dump-dies**, but with type references
+instead of fully expanded strings.
+
+4. Maintaining a stable kABI
+============================
+
+Distribution maintainers often need the ability to make ABI compatible
+changes to kernel data structures due to LTS updates or backports. Using
+the traditional `#ifndef __GENKSYMS__` to hide these changes from symbol
+versioning won't work when processing object files. To support this
+use case, gendwarfksyms provides kABI stability features designed to
+hide changes that won't affect the ABI when calculating versions. These
+features are all gated behind the **--stable** command line flag and are
+not used in the mainline kernel. To use stable features during a kernel
+build, set `KBUILD_GENDWARFKSYMS_STABLE=1`.
+
+Examples for using these features are provided in the
+**scripts/gendwarfksyms/examples** directory, including helper macros
+for source code annotation. Note that as these features are only used to
+transform the inputs for symbol versioning, the user is responsible for
+ensuring that their changes actually won't break the ABI.
+
+4.1. kABI rules
+===============
+
+kABI rules allow distributions to fine-tune certain parts
+of gendwarfksyms output and thus control how symbol
+versions are calculated. These rules are defined in the
+`.discard.gendwarfksyms.kabi_rules` section of the object file and
+consist of simple null-terminated strings with the following structure::
+
+	version\0type\0target\0value\0
+
+This string sequence is repeated as many times as needed to express all
+the rules. The fields are as follows:
+
+- `version`: Ensures backward compatibility for future changes to the
+  structure. Currently expected to be "1".
+- `type`: Indicates the type of rule being applied.
+- `target`: Specifies the target of the rule, typically the fully
+  qualified name of the DWARF Debugging Information Entry (DIE).
+- `value`: Provides rule-specific data.
+
+The following helper macro, for example, can be used to specify rules
+in the source code::
+
+	#define __KABI_RULE(hint, target, value)                             \
+		static const char __PASTE(__gendwarfksyms_rule_,             \
+					  __COUNTER__)[] __used __aligned(1) \
+			__section(".discard.gendwarfksyms.kabi_rules") =     \
+				"1\0" #hint "\0" #target "\0" #value
+
+
+Currently, only the rules discussed in this section are supported, but
+the format is extensible enough to allow further rules to be added as
+need arises.
+
+4.1.1. Managing definition visibility
+=====================================
+
+A declaration can change into a full definition when additional includes
+are pulled into the translation unit. This changes the versions of any
+symbol that references the type even if the ABI remains unchanged. As
+it may not be possible to drop includes without breaking the build, the
+`declonly` rule can be used to specify a type as declaration-only, even
+if the debugging information contains the full definition.
+
+The rule fields are expected to be as follows:
+
+- `type`: "declonly"
+- `target`: The fully qualified name of the target data structure
+  (as shown in **--dump-dies** output).
+- `value`: This field is ignored.
+
+Using the `__KABI_RULE` macro, this rule can be defined as::
+
+	#define KABI_DECLONLY(fqn) __KABI_RULE(declonly, fqn, )
+
+Example usage::
+
+	struct s {
+		/* definition */
+	};
+
+	KABI_DECLONLY(s);
+
+4.1.2. Adding enumerators
+=========================
+
+For enums, all enumerators and their values are included in calculating
+symbol versions, which becomes a problem if we later need to add more
+enumerators without changing symbol versions. The `enumerator_ignore`
+rule allows us to hide named enumerators from the input.
+
+The rule fields are expected to be as follows:
+
+- `type`: "enumerator_ignore"
+- `target`: The fully qualified name of the target enum
+  (as shown in **--dump-dies** output) and the name of the
+  enumerator field separated by a space.
+- `value`: This field is ignored.
+
+Using the `__KABI_RULE` macro, this rule can be defined as::
+
+	#define KABI_ENUMERATOR_IGNORE(fqn, field) \
+		__KABI_RULE(enumerator_ignore, fqn field, )
+
+Example usage::
+
+	enum e {
+		A, B, C, D,
+	};
+
+	KABI_ENUMERATOR_IGNORE(e, B);
+	KABI_ENUMERATOR_IGNORE(e, C);
+
+If the enum additionally includes an end marker and new values must
+be added in the middle, we may need to use the old value for the last
+enumerator when calculating versions. The `enumerator_value` rule allows
+us to override the value of an enumerator for version calculation:
+
+- `type`: "enumerator_value"
+- `target`: The fully qualified name of the target enum
+  (as shown in **--dump-dies** output) and the name of the
+  enumerator field separated by a space.
+- `value`: Integer value used for the field.
+
+Using the `__KABI_RULE` macro, this rule can be defined as::
+
+	#define KABI_ENUMERATOR_VALUE(fqn, field, value) \
+		__KABI_RULE(enumerator_value, fqn field, value)
+
+Example usage::
+
+	enum e {
+		A, B, C, LAST,
+	};
+
+	KABI_ENUMERATOR_IGNORE(e, C);
+	KABI_ENUMERATOR_VALUE(e, LAST, 2);
+
+4.3. Adding structure members
+=============================
+
+Perhaps the most common ABI compatible change is adding a member to a
+kernel data structure. When changes to a structure are anticipated,
+distribution maintainers can pre-emptively reserve space in the
+structure and take it into use later without breaking the ABI. If
+changes are needed to data structures without reserved space, existing
+alignment holes can potentially be used instead. While kABI rules could
+be added for these type of changes, using unions is typically a more
+natural method. This section describes gendwarfksyms support for using
+reserved space in data structures and hiding members that don't change
+the ABI when calculating symbol versions.
+
+4.3.1. Reserving space and replacing members
+============================================
+
+Space is typically reserved for later use by appending integer types, or
+arrays, to the end of the data structure, but any type can be used. Each
+reserved member needs a unique name, but as the actual purpose is usually
+not known at the time the space is reserved, for convenience, names that
+start with `__kabi_` are left out when calculating symbol versions::
+
+        struct s {
+                long a;
+                long __kabi_reserved_0; /* reserved for future use */
+        };
+
+The reserved space can be taken into use by wrapping the member in a
+union, which includes the original type and the replacement member::
+
+        struct s {
+                long a;
+                union {
+                        long __kabi_reserved_0; /* original type */
+                        struct b b; /* replaced field */
+                };
+        };
+
+If the `__kabi_` naming scheme was used when reserving space, the name
+of the first member of the union must start with `__kabi_reserved`. This
+ensures the original type is used when calculating versions, but the name
+is again left out. The rest of the union is ignored.
+
+If we're replacing a member that doesn't follow this naming convention,
+we also need to preserve the original name to avoid changing versions,
+which we can do by changing the first union member's name to start with
+`__kabi_renamed` followed by the original name.
+
+The examples include `KABI_(RESERVE|USE|REPLACE)*` macros that help
+simplify the process and also ensure the replacement member is correctly
+aligned and its size won't exceed the reserved space.
+
+4.3.2. Hiding members
+=====================
+
+Predicting which structures will require changes during the support
+timeframe isn't always possible, in which case one might have to resort
+to placing new members into existing alignment holes::
+
+        struct s {
+                int a;
+                /* a 4-byte alignment hole */
+                unsigned long b;
+        };
+
+
+While this won't change the size of the data structure, one needs to
+be able to hide the added members from symbol versioning. Similarly
+to reserved fields, this can be accomplished by wrapping the added
+member to a union where one of the fields has a name starting with
+`__kabi_ignored`::
+
+        struct s {
+                int a;
+                union {
+                        char __kabi_ignored_0;
+                        int n;
+                };
+                unsigned long b;
+        };
+
+With **--stable**, both versions produce the same symbol version.

Documentation/kbuild/index.rst

@@ -21,6 +21,7 @@ Kernel Build System
     reproducible-builds
     gcc-plugins
     llvm
+    gendwarfksyms
 
 .. only::  subproject and html
 

Documentation/kbuild/modules.rst

@@ -423,6 +423,26 @@ Symbols From the Kernel (vmlinux + modules)
 	1) It lists all exported symbols from vmlinux and all modules.
 	2) It lists the CRC if CONFIG_MODVERSIONS is enabled.
 
+Version Information Formats
+---------------------------
+
+	Exported symbols have information stored in __ksymtab or __ksymtab_gpl
+	sections. Symbol names and namespaces are stored in __ksymtab_strings,
+	using a format similar to the string table used for ELF. If
+	CONFIG_MODVERSIONS is enabled, the CRCs corresponding to exported
+	symbols will be added to the __kcrctab or __kcrctab_gpl.
+
+	If CONFIG_BASIC_MODVERSIONS is enabled (default with
+	CONFIG_MODVERSIONS), imported symbols will have their symbol name and
+	CRC stored in the __versions section of the importing module. This
+	mode only supports symbols of length up to 64 bytes.
+
+	If CONFIG_EXTENDED_MODVERSIONS is enabled (required to enable both
+	CONFIG_MODVERSIONS and CONFIG_RUST at the same time), imported symbols
+	will have their symbol name recorded in the __version_ext_names
+	section as a series of concatenated, null-terminated strings. CRCs for
+	these symbols will be recorded in the __version_ext_crcs section.
+
 Symbols and External Modules
 ----------------------------
 

Documentation/process/changes.rst

@@ -59,7 +59,6 @@ iptables               1.4.2            iptables -V
 openssl & libcrypto    1.0.0            openssl version
 bc                     1.06.95          bc --version
 Sphinx\ [#f1]_         2.4.4            sphinx-build --version
-cpio                   any              cpio --version
 GNU tar                1.28             tar --version
 gtags (optional)       6.6.5            gtags --version
 mkimage (optional)     2017.01          mkimage --version
@@ -536,11 +535,6 @@ mcelog
 
 - <https://www.mcelog.org/>
 
-cpio
-----
-
-- <https://www.gnu.org/software/cpio/>
-
 Networking
 **********
 

MAINTAINERS

@@ -9548,6 +9548,13 @@ W:	https://linuxtv.org
 T:	git git://linuxtv.org/media.git
 F:	drivers/media/radio/radio-gemtek*
 
+GENDWARFKSYMS
+M:	Sami Tolvanen <samitolvanen@google.com>
+L:	linux-modules@vger.kernel.org
+L:	linux-kbuild@vger.kernel.org
+S:	Maintained
+F:	scripts/gendwarfksyms/
+
 GENERIC ARCHITECTURE TOPOLOGY
 M:	Sudeep Holla <sudeep.holla@arm.com>
 L:	linux-kernel@vger.kernel.org

arch/powerpc/kernel/module_64.c

@@ -369,6 +369,24 @@ static void dedotify_versions(struct modversion_info *vers,
 		}
 }
 
+/* Same as normal versions, remove a leading dot if present. */
+static void dedotify_ext_version_names(char *str_seq, unsigned long size)
+{
+	unsigned long out = 0;
+	unsigned long in;
+	char last = '\0';
+
+	for (in = 0; in < size; in++) {
+		/* Skip one leading dot */
+		if (last == '\0' && str_seq[in] == '.')
+			in++;
+		last = str_seq[in];
+		str_seq[out++] = last;
+	}
+	/* Zero the trailing portion of the names table for robustness */
+	memset(&str_seq[out], 0, size - out);
+}
+
 /*
  * Undefined symbols which refer to .funcname, hack to funcname. Make .TOC.
  * seem to be defined (value set later).
@@ -438,10 +456,12 @@ int module_frob_arch_sections(Elf64_Ehdr *hdr,
 			me->arch.toc_section = i;
 			if (sechdrs[i].sh_addralign < 8)
 				sechdrs[i].sh_addralign = 8;
-		}
-		else if (strcmp(secstrings+sechdrs[i].sh_name,"__versions")==0)
+		} else if (strcmp(secstrings + sechdrs[i].sh_name, "__versions") == 0)
 			dedotify_versions((void *)hdr + sechdrs[i].sh_offset,
 					  sechdrs[i].sh_size);
+		else if (strcmp(secstrings + sechdrs[i].sh_name, "__version_ext_names") == 0)
+			dedotify_ext_version_names((void *)hdr + sechdrs[i].sh_offset,
+						   sechdrs[i].sh_size);
 
 		if (sechdrs[i].sh_type == SHT_SYMTAB)
 			dedotify((void *)hdr + sechdrs[i].sh_offset,

include/linux/export.h

@@ -52,9 +52,24 @@
 
 #else
 
+#ifdef CONFIG_GENDWARFKSYMS
+/*
+ * With CONFIG_GENDWARFKSYMS, ensure the compiler emits debugging
+ * information for all exported symbols, including those defined in
+ * different TUs, by adding a __gendwarfksyms_ptr_<symbol> pointer
+ * that's discarded during the final link.
+ */
+#define __GENDWARFKSYMS_EXPORT(sym)				\
+	static typeof(sym) *__gendwarfksyms_ptr_##sym __used	\
+		__section(".discard.gendwarfksyms") = &sym;
+#else
+#define __GENDWARFKSYMS_EXPORT(sym)
+#endif
+
 #define __EXPORT_SYMBOL(sym, license, ns)			\
 	extern typeof(sym) sym;					\
 	__ADDRESSABLE(sym)					\
+	__GENDWARFKSYMS_EXPORT(sym)				\
 	asm(__stringify(___EXPORT_SYMBOL(sym, license, ns)))
 
 #endif

include/linux/module.h

@@ -430,7 +430,7 @@ struct module {
 
 	/* Exported symbols */
 	const struct kernel_symbol *syms;
-	const s32 *crcs;
+	const u32 *crcs;
 	unsigned int num_syms;
 
 #ifdef CONFIG_ARCH_USES_CFI_TRAPS
@@ -448,7 +448,7 @@ struct module {
 	/* GPL-only exported symbols. */
 	unsigned int num_gpl_syms;
 	const struct kernel_symbol *gpl_syms;
-	const s32 *gpl_crcs;
+	const u32 *gpl_crcs;
 	bool using_gplonly_symbols;
 
 #ifdef CONFIG_MODULE_SIG

init/Kconfig

@@ -1959,7 +1959,8 @@ config RUST
 	bool "Rust support"
 	depends on HAVE_RUST
 	depends on RUST_IS_AVAILABLE
-	depends on !MODVERSIONS
+	select EXTENDED_MODVERSIONS if MODVERSIONS
+	depends on !MODVERSIONS || GENDWARFKSYMS
 	depends on !GCC_PLUGIN_RANDSTRUCT
 	depends on !RANDSTRUCT
 	depends on !DEBUG_INFO_BTF || PAHOLE_HAS_LANG_EXCLUDE

kernel/gen_kheaders.sh

@@ -7,20 +7,13 @@ set -e
 sfile="$(readlink -f "$0")"
 outdir="$(pwd)"
 tarfile=$1
-cpio_dir=$outdir/${tarfile%/*}/.tmp_cpio_dir
+tmpdir=$outdir/${tarfile%/*}/.tmp_dir
 
 dir_list="
 include/
 arch/$SRCARCH/include/
 "
 
-if ! command -v cpio >/dev/null; then
-	echo >&2 "***"
-	echo >&2 "*** 'cpio' could not be found."
-	echo >&2 "***"
-	exit 1
-fi
-
 # Support incremental builds by skipping archive generation
 # if timestamps of files being archived are not changed.
 
@@ -48,9 +41,9 @@ all_dirs="$all_dirs $dir_list"
 # check include/generated/autoconf.h explicitly.
 #
 # Ignore them for md5 calculation to avoid pointless regeneration.
-headers_md5="$(find $all_dirs -name "*.h"			|
-		grep -v "include/generated/utsversion.h"	|
-		grep -v "include/generated/autoconf.h"	|
+headers_md5="$(find $all_dirs -name "*.h" -a			\
+		! -path include/generated/utsversion.h -a	\
+		! -path include/generated/autoconf.h		|
 		xargs ls -l | md5sum | cut -d ' ' -f1)"
 
 # Any changes to this script will also cause a rebuild of the archive.
@@ -65,36 +58,43 @@ fi
 
 echo "  GEN     $tarfile"
 
-rm -rf $cpio_dir
-mkdir $cpio_dir
+rm -rf "${tmpdir}"
+mkdir "${tmpdir}"
 
 if [ "$building_out_of_srctree" ]; then
 	(
 		cd $srctree
 		for f in $dir_list
 			do find "$f" -name "*.h";
-		done | cpio --quiet -pd $cpio_dir
+		done | tar -c -f - -T - | tar -xf - -C "${tmpdir}"
 	)
 fi
 
-# The second CPIO can complain if files already exist which can happen with out
-# of tree builds having stale headers in srctree. Just silence CPIO for now.
 for f in $dir_list;
 	do find "$f" -name "*.h";
-done | cpio --quiet -pdu $cpio_dir >/dev/null 2>&1
+done | tar -c -f - -T - | tar -xf - -C "${tmpdir}"
+
+# Always exclude include/generated/utsversion.h
+# Otherwise, the contents of the tarball may vary depending on the build steps.
+rm -f "${tmpdir}/include/generated/utsversion.h"
 
 # Remove comments except SDPX lines
-find $cpio_dir -type f -print0 |
-	xargs -0 -P8 -n1 perl -pi -e 'BEGIN {undef $/;}; s/\/\*((?!SPDX).)*?\*\///smg;'
+# Use a temporary file to store directory contents to prevent find/xargs from
+# seeing temporary files created by perl.
+find "${tmpdir}" -type f -print0 > "${tmpdir}.contents.txt"
+xargs -0 -P8 -n1 \
+	perl -pi -e 'BEGIN {undef $/;}; s/\/\*((?!SPDX).)*?\*\///smg;' \
+	< "${tmpdir}.contents.txt"
+rm -f "${tmpdir}.contents.txt"
 
 # Create archive and try to normalize metadata for reproducibility.
 tar "${KBUILD_BUILD_TIMESTAMP:+--mtime=$KBUILD_BUILD_TIMESTAMP}" \
     --exclude=".__afs*" --exclude=".nfs*" \
     --owner=0 --group=0 --sort=name --numeric-owner --mode=u=rw,go=r,a+X \
-    -I $XZ -cf $tarfile -C $cpio_dir/ . > /dev/null
+    -I $XZ -cf $tarfile -C "${tmpdir}/" . > /dev/null
 
 echo $headers_md5 > kernel/kheaders.md5
 echo "$this_file_md5" >> kernel/kheaders.md5
 echo "$(md5sum $tarfile | cut -d ' ' -f1)" >> kernel/kheaders.md5
 
-rm -rf $cpio_dir
+rm -rf "${tmpdir}"

kernel/module/Kconfig

@@ -169,6 +169,36 @@ config MODVERSIONS
 	  make them incompatible with the kernel you are running.  If
 	  unsure, say N.
 
+choice
+	prompt "Module versioning implementation"
+	depends on MODVERSIONS
+	help
+	  Select the tool used to calculate symbol versions for modules.
+
+	  If unsure, select GENKSYMS.
+
+config GENKSYMS
+	bool "genksyms (from source code)"
+	help
+	  Calculate symbol versions from pre-processed source code using
+	  genksyms.
+
+	  If unsure, say Y.
+
+config GENDWARFKSYMS
+	bool "gendwarfksyms (from debugging information)"
+	depends on DEBUG_INFO
+	# Requires full debugging information, split DWARF not supported.
+	depends on !DEBUG_INFO_REDUCED && !DEBUG_INFO_SPLIT
+	# Requires ELF object files.
+	depends on !LTO
+	help
+	  Calculate symbol versions from DWARF debugging information using
+	  gendwarfksyms. Requires DEBUG_INFO to be enabled.
+
+	  If unsure, say N.
+endchoice
+
 config ASM_MODVERSIONS
 	bool
 	default HAVE_ASM_MODVERSIONS && MODVERSIONS
@@ -177,6 +207,31 @@ config ASM_MODVERSIONS
 	  assembly. This can be enabled only when the target architecture
 	  supports it.
 
+config EXTENDED_MODVERSIONS
+	bool "Extended Module Versioning Support"
+	depends on MODVERSIONS
+	help
+	  This enables extended MODVERSIONs support, allowing long symbol
+	  names to be versioned.
+
+	  The most likely reason you would enable this is to enable Rust
+	  support. If unsure, say N.
+
+config BASIC_MODVERSIONS
+	bool "Basic Module Versioning Support"
+	depends on MODVERSIONS
+	default y
+	help
+	  This enables basic MODVERSIONS support, allowing older tools or
+	  kernels to potentially load modules.
+
+	  Disabling this may cause older `modprobe` or `kmod` to be unable
+	  to read MODVERSIONS information from built modules. With this
+	  disabled, older kernels may treat this module as unversioned.
+
+	  This is enabled by default when MODVERSIONS are enabled.
+	  If unsure, say Y.
+
 config MODULE_SRCVERSION_ALL
 	bool "Source checksum for all modules"
 	help

kernel/module/internal.h

@@ -55,8 +55,8 @@ extern const struct kernel_symbol __start___ksymtab[];
 extern const struct kernel_symbol __stop___ksymtab[];
 extern const struct kernel_symbol __start___ksymtab_gpl[];
 extern const struct kernel_symbol __stop___ksymtab_gpl[];
-extern const s32 __start___kcrctab[];
-extern const s32 __start___kcrctab_gpl[];
+extern const u32 __start___kcrctab[];
+extern const u32 __start___kcrctab_gpl[];
 
 struct load_info {
 	const char *name;
@@ -86,6 +86,8 @@ struct load_info {
 		unsigned int vers;
 		unsigned int info;
 		unsigned int pcpu;
+		unsigned int vers_ext_crc;
+		unsigned int vers_ext_name;
 	} index;
 };
 
@@ -102,7 +104,7 @@ struct find_symbol_arg {
 
 	/* Output */
 	struct module *owner;
-	const s32 *crc;
+	const u32 *crc;
 	const struct kernel_symbol *sym;
 	enum mod_license license;
 };
@@ -384,16 +386,25 @@ static inline void init_param_lock(struct module *mod) { }
 
 #ifdef CONFIG_MODVERSIONS
 int check_version(const struct load_info *info,
-		  const char *symname, struct module *mod, const s32 *crc);
+		  const char *symname, struct module *mod, const u32 *crc);
 void module_layout(struct module *mod, struct modversion_info *ver, struct kernel_param *kp,
 		   struct kernel_symbol *ks, struct tracepoint * const *tp);
 int check_modstruct_version(const struct load_info *info, struct module *mod);
 int same_magic(const char *amagic, const char *bmagic, bool has_crcs);
+struct modversion_info_ext {
+	size_t remaining;
+	const u32 *crc;
+	const char *name;
+};
+void modversion_ext_start(const struct load_info *info, struct modversion_info_ext *ver);
+void modversion_ext_advance(struct modversion_info_ext *ver);
+#define for_each_modversion_info_ext(ver, info) \
+	for (modversion_ext_start(info, &ver); ver.remaining > 0; modversion_ext_advance(&ver))
 #else /* !CONFIG_MODVERSIONS */
 static inline int check_version(const struct load_info *info,
 				const char *symname,
 				struct module *mod,
-				const s32 *crc)
+				const u32 *crc)
 {
 	return 1;
 }

kernel/module/main.c

@@ -86,7 +86,7 @@ struct mod_tree_root mod_tree __cacheline_aligned = {
 
 struct symsearch {
 	const struct kernel_symbol *start, *stop;
-	const s32 *crcs;
+	const u32 *crcs;
 	enum mod_license license;
 };
 
@@ -2073,6 +2073,82 @@ static int elf_validity_cache_index_str(struct load_info *info)
 	return 0;
 }
 
+/**
+ * elf_validity_cache_index_versions() - Validate and cache version indices
+ * @info:  Load info to cache version indices in.
+ *         Must have &load_info->sechdrs and &load_info->secstrings populated.
+ * @flags: Load flags, relevant to suppress version loading, see
+ *         uapi/linux/module.h
+ *
+ * If we're ignoring modversions based on @flags, zero all version indices
+ * and return validity. Othewrise check:
+ *
+ * * If "__version_ext_crcs" is present, "__version_ext_names" is present
+ * * There is a name present for every crc
+ *
+ * Then populate:
+ *
+ * * &load_info->index.vers
+ * * &load_info->index.vers_ext_crc
+ * * &load_info->index.vers_ext_names
+ *
+ * if present.
+ *
+ * Return: %0 if valid, %-ENOEXEC on failure.
+ */
+static int elf_validity_cache_index_versions(struct load_info *info, int flags)
+{
+	unsigned int vers_ext_crc;
+	unsigned int vers_ext_name;
+	size_t crc_count;
+	size_t remaining_len;
+	size_t name_size;
+	char *name;
+
+	/* If modversions were suppressed, pretend we didn't find any */
+	if (flags & MODULE_INIT_IGNORE_MODVERSIONS) {
+		info->index.vers = 0;
+		info->index.vers_ext_crc = 0;
+		info->index.vers_ext_name = 0;
+		return 0;
+	}
+
+	vers_ext_crc = find_sec(info, "__version_ext_crcs");
+	vers_ext_name = find_sec(info, "__version_ext_names");
+
+	/* If we have one field, we must have the other */
+	if (!!vers_ext_crc != !!vers_ext_name) {
+		pr_err("extended version crc+name presence does not match");
+		return -ENOEXEC;
+	}
+
+	/*
+	 * If we have extended version information, we should have the same
+	 * number of entries in every section.
+	 */
+	if (vers_ext_crc) {
+		crc_count = info->sechdrs[vers_ext_crc].sh_size / sizeof(u32);
+		name = (void *)info->hdr +
+			info->sechdrs[vers_ext_name].sh_offset;
+		remaining_len = info->sechdrs[vers_ext_name].sh_size;
+
+		while (crc_count--) {
+			name_size = strnlen(name, remaining_len) + 1;
+			if (name_size > remaining_len) {
+				pr_err("more extended version crcs than names");
+				return -ENOEXEC;
+			}
+			remaining_len -= name_size;
+			name += name_size;
+		}
+	}
+
+	info->index.vers = find_sec(info, "__versions");
+	info->index.vers_ext_crc = vers_ext_crc;
+	info->index.vers_ext_name = vers_ext_name;
+	return 0;
+}
+
 /**
  * elf_validity_cache_index() - Resolve, validate, cache section indices
  * @info:  Load info to read from and update.
@@ -2087,9 +2163,7 @@ static int elf_validity_cache_index_str(struct load_info *info)
  * * elf_validity_cache_index_mod()
  * * elf_validity_cache_index_sym()
  * * elf_validity_cache_index_str()
- *
- * If versioning is not suppressed via flags, load the version index from
- * a section called "__versions" with no validation.
+ * * elf_validity_cache_index_versions()
  *
  * If CONFIG_SMP is enabled, load the percpu section by name with no
  * validation.
@@ -2112,11 +2186,9 @@ static int elf_validity_cache_index(struct load_info *info, int flags)
 	err = elf_validity_cache_index_str(info);
 	if (err < 0)
 		return err;
-
-	if (flags & MODULE_INIT_IGNORE_MODVERSIONS)
-		info->index.vers = 0; /* Pretend no __versions section! */
-	else
-		info->index.vers = find_sec(info, "__versions");
+	err = elf_validity_cache_index_versions(info, flags);
+	if (err < 0)
+		return err;
 
 	info->index.pcpu = find_pcpusec(info);
 
@@ -2327,6 +2399,10 @@ static int rewrite_section_headers(struct load_info *info, int flags)
 
 	/* Track but don't keep modinfo and version sections. */
 	info->sechdrs[info->index.vers].sh_flags &= ~(unsigned long)SHF_ALLOC;
+	info->sechdrs[info->index.vers_ext_crc].sh_flags &=
+		~(unsigned long)SHF_ALLOC;
+	info->sechdrs[info->index.vers_ext_name].sh_flags &=
+		~(unsigned long)SHF_ALLOC;
 	info->sechdrs[info->index.info].sh_flags &= ~(unsigned long)SHF_ALLOC;
 
 	return 0;

kernel/module/version.c

@@ -13,17 +13,34 @@
 int check_version(const struct load_info *info,
 		  const char *symname,
 			 struct module *mod,
-			 const s32 *crc)
+			 const u32 *crc)
 {
 	Elf_Shdr *sechdrs = info->sechdrs;
 	unsigned int versindex = info->index.vers;
 	unsigned int i, num_versions;
 	struct modversion_info *versions;
+	struct modversion_info_ext version_ext;
 
 	/* Exporting module didn't supply crcs?  OK, we're already tainted. */
 	if (!crc)
 		return 1;
 
+	/* If we have extended version info, rely on it */
+	if (info->index.vers_ext_crc) {
+		for_each_modversion_info_ext(version_ext, info) {
+			if (strcmp(version_ext.name, symname) != 0)
+				continue;
+			if (*version_ext.crc == *crc)
+				return 1;
+			pr_debug("Found checksum %X vs module %X\n",
+				 *crc, *version_ext.crc);
+			goto bad_version;
+		}
+		pr_warn_once("%s: no extended symbol version for %s\n",
+			     info->name, symname);
+		return 1;
+	}
+
 	/* No versions at all?  modprobe --force does this. */
 	if (versindex == 0)
 		return try_to_force_load(mod, symname) == 0;
@@ -87,6 +104,34 @@ int same_magic(const char *amagic, const char *bmagic,
 	return strcmp(amagic, bmagic) == 0;
 }
 
+void modversion_ext_start(const struct load_info *info,
+			  struct modversion_info_ext *start)
+{
+	unsigned int crc_idx = info->index.vers_ext_crc;
+	unsigned int name_idx = info->index.vers_ext_name;
+	Elf_Shdr *sechdrs = info->sechdrs;
+
+	/*
+	 * Both of these fields are needed for this to be useful
+	 * Any future fields should be initialized to NULL if absent.
+	 */
+	if (crc_idx == 0 || name_idx == 0) {
+		start->remaining = 0;
+		return;
+	}
+
+	start->crc = (const u32 *)sechdrs[crc_idx].sh_addr;
+	start->name = (const char *)sechdrs[name_idx].sh_addr;
+	start->remaining = sechdrs[crc_idx].sh_size / sizeof(*start->crc);
+}
+
+void modversion_ext_advance(struct modversion_info_ext *vers)
+{
+	vers->remaining--;
+	vers->crc++;
+	vers->name += strlen(vers->name) + 1;
+}
+
 /*
  * Generate the signature for all relevant module structures here.
  * If these change, we don't want to try to parse the module.

rust/Makefile

@@ -329,10 +329,11 @@ $(obj)/bindings/bindings_helpers_generated.rs: private bindgen_target_extra = ;
 $(obj)/bindings/bindings_helpers_generated.rs: $(src)/helpers/helpers.c FORCE
 	$(call if_changed_dep,bindgen)
 
+rust_exports = $(NM) -p --defined-only $(1) | awk '$$2~/(T|R|D|B)/ && $$3!~/__cfi/ { printf $(2),$(3) }'
+
 quiet_cmd_exports = EXPORTS $@
       cmd_exports = \
-	$(NM) -p --defined-only $< \
-		| awk '$$2~/(T|R|D|B)/ && $$3!~/__cfi/ {printf "EXPORT_SYMBOL_RUST_GPL(%s);\n",$$3}' > $@
+	$(call rust_exports,$<,"EXPORT_SYMBOL_RUST_GPL(%s);\n",$$3) > $@
 
 $(obj)/exports_core_generated.h: $(obj)/core.o FORCE
 	$(call if_changed,exports)
@@ -401,11 +402,36 @@ ifneq ($(or $(CONFIG_ARM64),$(and $(CONFIG_RISCV),$(CONFIG_64BIT))),)
 		__ashlti3 __lshrti3
 endif
 
+ifdef CONFIG_MODVERSIONS
+cmd_gendwarfksyms = $(if $(skip_gendwarfksyms),, \
+	$(call rust_exports,$@,"%s\n",$$3) | \
+	scripts/gendwarfksyms/gendwarfksyms \
+		$(if $(KBUILD_GENDWARFKSYMS_STABLE), --stable) \
+		$(if $(KBUILD_SYMTYPES), --symtypes $(@:.o=.symtypes),) \
+		$@ >> $(dot-target).cmd)
+endif
+
 define rule_rustc_library
 	$(call cmd_and_fixdep,rustc_library)
 	$(call cmd,gen_objtooldep)
+	$(call cmd,gendwarfksyms)
 endef
 
+define rule_rust_cc_library
+	$(call if_changed_rule,cc_o_c)
+	$(call cmd,force_checksrc)
+	$(call cmd,gendwarfksyms)
+endef
+
+# helpers.o uses the same export mechanism as Rust libraries, so ensure symbol
+# versions are calculated for the helpers too.
+$(obj)/helpers/helpers.o: $(src)/helpers/helpers.c $(recordmcount_source) FORCE
+	+$(call if_changed_rule,rust_cc_library)
+
+# Disable symbol versioning for exports.o to avoid conflicts with the actual
+# symbol versions generated from Rust objects.
+$(obj)/exports.o: private skip_gendwarfksyms = 1
+
 $(obj)/core.o: private skip_clippy = 1
 $(obj)/core.o: private skip_flags = -Wunreachable_pub
 $(obj)/core.o: private rustc_objcopy = $(foreach sym,$(redirect-intrinsics),--redefine-sym $(sym)=__rust$(sym))
@@ -417,13 +443,16 @@ ifneq ($(or $(CONFIG_X86_64),$(CONFIG_X86_32)),)
 $(obj)/core.o: scripts/target.json
 endif
 
+$(obj)/compiler_builtins.o: private skip_gendwarfksyms = 1
 $(obj)/compiler_builtins.o: private rustc_objcopy = -w -W '__*'
 $(obj)/compiler_builtins.o: $(src)/compiler_builtins.rs $(obj)/core.o FORCE
 	+$(call if_changed_rule,rustc_library)
 
+$(obj)/build_error.o: private skip_gendwarfksyms = 1
 $(obj)/build_error.o: $(src)/build_error.rs $(obj)/compiler_builtins.o FORCE
 	+$(call if_changed_rule,rustc_library)
 
+$(obj)/ffi.o: private skip_gendwarfksyms = 1
 $(obj)/ffi.o: $(src)/ffi.rs $(obj)/compiler_builtins.o FORCE
 	+$(call if_changed_rule,rustc_library)
 
@@ -435,6 +464,7 @@ $(obj)/bindings.o: $(src)/bindings/lib.rs \
 	+$(call if_changed_rule,rustc_library)
 
 $(obj)/uapi.o: private rustc_target_flags = --extern ffi
+$(obj)/uapi.o: private skip_gendwarfksyms = 1
 $(obj)/uapi.o: $(src)/uapi/lib.rs \
     $(obj)/ffi.o \
     $(obj)/uapi/uapi_generated.rs FORCE

scripts/Makefile

@@ -53,7 +53,8 @@ hostprogs += unifdef
 targets += module.lds
 
 subdir-$(CONFIG_GCC_PLUGINS) += gcc-plugins
-subdir-$(CONFIG_MODVERSIONS) += genksyms
+subdir-$(CONFIG_GENKSYMS) += genksyms
+subdir-$(CONFIG_GENDWARFKSYMS) += gendwarfksyms
 subdir-$(CONFIG_SECURITY_SELINUX) += selinux
 subdir-$(CONFIG_SECURITY_IPE) += ipe
 

scripts/Makefile.build

@@ -107,13 +107,24 @@ cmd_cpp_i_c       = $(CPP) $(c_flags) -o $@ $<
 $(obj)/%.i: $(obj)/%.c FORCE
 	$(call if_changed_dep,cpp_i_c)
 
+getexportsymbols = $(NM) $@ | sed -n 's/.* __export_symbol_\(.*\)/$(1)/p'
+
+gendwarfksyms = $(objtree)/scripts/gendwarfksyms/gendwarfksyms	\
+	$(if $(KBUILD_SYMTYPES), --symtypes $(@:.o=.symtypes))	\
+	$(if $(KBUILD_GENDWARFKSYMS_STABLE), --stable)
+
 genksyms = $(objtree)/scripts/genksyms/genksyms		\
 	$(if $(KBUILD_SYMTYPES), -T $(@:.o=.symtypes))	\
 	$(if $(KBUILD_PRESERVE), -p)			\
 	$(addprefix -r , $(wildcard $(@:.o=.symref)))
 
 # These mirror gensymtypes_S and co below, keep them in synch.
+ifdef CONFIG_GENDWARFKSYMS
+cmd_gensymtypes_c = $(if $(skip_gendwarfksyms),,	\
+	$(call getexportsymbols,\1) | $(gendwarfksyms) $@)
+else
 cmd_gensymtypes_c = $(CPP) -D__GENKSYMS__ $(c_flags) $< | $(genksyms)
+endif # CONFIG_GENDWARFKSYMS
 
 # LLVM assembly
 # Generate .ll files from .c
@@ -286,14 +297,26 @@ $(obj)/%.rs: $(obj)/%.rs.S FORCE
 # This is convoluted. The .S file must first be preprocessed to run guards and
 # expand names, then the resulting exports must be constructed into plain
 # EXPORT_SYMBOL(symbol); to build our dummy C file, and that gets preprocessed
-# to make the genksyms input.
+# to make the genksyms input or compiled into an object for gendwarfksyms.
 #
 # These mirror gensymtypes_c and co above, keep them in synch.
-cmd_gensymtypes_S =                                                         \
-   { echo "\#include <linux/kernel.h>" ;                                    \
-     echo "\#include <asm/asm-prototypes.h>" ;                              \
-     $(NM) $@ | sed -n 's/.* __export_symbol_\(.*\)/EXPORT_SYMBOL(\1);/p' ; } | \
-    $(CPP) -D__GENKSYMS__ $(c_flags) -xc - | $(genksyms)
+getasmexports =								\
+   { echo "\#include <linux/kernel.h>" ;				\
+     echo "\#include <linux/string.h>" ;				\
+     echo "\#include <asm/asm-prototypes.h>" ;				\
+     $(call getexportsymbols,EXPORT_SYMBOL(\1);) ; }
+
+ifdef CONFIG_GENDWARFKSYMS
+cmd_gensymtypes_S =							\
+	$(getasmexports) |						\
+	$(CC) $(c_flags) -c -o $(@:.o=.gendwarfksyms.o) -xc -;		\
+	$(call getexportsymbols,\1) |					\
+	$(gendwarfksyms) $(@:.o=.gendwarfksyms.o)
+else
+cmd_gensymtypes_S =							\
+	$(getasmexports) |						\
+	$(CPP) -D__GENKSYMS__ $(c_flags) -xc - | $(genksyms)
+endif # CONFIG_GENDWARFKSYMS
 
 quiet_cmd_cpp_s_S = CPP $(quiet_modtag) $@
 cmd_cpp_s_S       = $(CPP) $(a_flags) -o $@ $<

scripts/Makefile.defconf

@@ -1,6 +1,11 @@
 # SPDX-License-Identifier: GPL-2.0
 # Configuration heplers
 
+cmd_merge_fragments = \
+	$(srctree)/scripts/kconfig/merge_config.sh \
+	$4 -m -O $(objtree) $(srctree)/arch/$(SRCARCH)/configs/$2 \
+	$(foreach config,$3,$(srctree)/arch/$(SRCARCH)/configs/$(config).config)
+
 # Creates 'merged defconfigs'
 # ---------------------------------------------------------------------------
 # Usage:
@@ -8,9 +13,7 @@
 #
 # Input config fragments without '.config' suffix
 define merge_into_defconfig
-	$(Q)$(CONFIG_SHELL) $(srctree)/scripts/kconfig/merge_config.sh \
-		-m -O $(objtree) $(srctree)/arch/$(SRCARCH)/configs/$(1) \
-		$(foreach config,$(2),$(srctree)/arch/$(SRCARCH)/configs/$(config).config)
+	$(call cmd,merge_fragments,$1,$2)
 	+$(Q)$(MAKE) -f $(srctree)/Makefile olddefconfig
 endef
 
@@ -22,8 +25,6 @@ endef
 #
 # Input config fragments without '.config' suffix
 define merge_into_defconfig_override
-	$(Q)$(CONFIG_SHELL) $(srctree)/scripts/kconfig/merge_config.sh \
-		-Q -m -O $(objtree) $(srctree)/arch/$(SRCARCH)/configs/$(1) \
-		$(foreach config,$(2),$(srctree)/arch/$(SRCARCH)/configs/$(config).config)
+	$(call cmd,merge_fragments,$1,$2,-Q)
 	+$(Q)$(MAKE) -f $(srctree)/Makefile olddefconfig
 endef

scripts/Makefile.modpost

@@ -43,6 +43,8 @@ MODPOST = $(objtree)/scripts/mod/modpost
 modpost-args =										\
 	$(if $(CONFIG_MODULES),-M)							\
 	$(if $(CONFIG_MODVERSIONS),-m)							\
+	$(if $(CONFIG_BASIC_MODVERSIONS),-b)						\
+	$(if $(CONFIG_EXTENDED_MODVERSIONS),-x)						\
 	$(if $(CONFIG_MODULE_SRCVERSION_ALL),-a)					\
 	$(if $(CONFIG_SECTION_MISMATCH_WARN_ONLY),,-E)					\
 	$(if $(KBUILD_MODPOST_WARN),-w)							\

scripts/gendwarfksyms/.gitignore

@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0
+/gendwarfksyms

scripts/gendwarfksyms/Makefile

@@ -0,0 +1,12 @@
+# SPDX-License-Identifier: GPL-2.0
+hostprogs-always-y += gendwarfksyms
+
+gendwarfksyms-objs += gendwarfksyms.o
+gendwarfksyms-objs += cache.o
+gendwarfksyms-objs += die.o
+gendwarfksyms-objs += dwarf.o
+gendwarfksyms-objs += kabi.o
+gendwarfksyms-objs += symbols.o
+gendwarfksyms-objs += types.o
+
+HOSTLDLIBS_gendwarfksyms := -ldw -lelf -lz

scripts/gendwarfksyms/cache.c

@@ -0,0 +1,51 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2024 Google LLC
+ */
+
+#include "gendwarfksyms.h"
+
+struct cache_item {
+	unsigned long key;
+	int value;
+	struct hlist_node hash;
+};
+
+void cache_set(struct cache *cache, unsigned long key, int value)
+{
+	struct cache_item *ci;
+
+	ci = xmalloc(sizeof(struct cache_item));
+	ci->key = key;
+	ci->value = value;
+	hash_add(cache->cache, &ci->hash, hash_32(key));
+}
+
+int cache_get(struct cache *cache, unsigned long key)
+{
+	struct cache_item *ci;
+
+	hash_for_each_possible(cache->cache, ci, hash, hash_32(key)) {
+		if (ci->key == key)
+			return ci->value;
+	}
+
+	return -1;
+}
+
+void cache_init(struct cache *cache)
+{
+	hash_init(cache->cache);
+}
+
+void cache_free(struct cache *cache)
+{
+	struct hlist_node *tmp;
+	struct cache_item *ci;
+
+	hash_for_each_safe(cache->cache, ci, tmp, hash) {
+		free(ci);
+	}
+
+	hash_init(cache->cache);
+}

scripts/gendwarfksyms/die.c

@@ -0,0 +1,166 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2024 Google LLC
+ */
+
+#include <string.h>
+#include "gendwarfksyms.h"
+
+#define DIE_HASH_BITS 15
+
+/* {die->addr, state} -> struct die * */
+static HASHTABLE_DEFINE(die_map, 1 << DIE_HASH_BITS);
+
+static unsigned int map_hits;
+static unsigned int map_misses;
+
+static inline unsigned int die_hash(uintptr_t addr, enum die_state state)
+{
+	return hash_32(addr_hash(addr) ^ (unsigned int)state);
+}
+
+static void init_die(struct die *cd)
+{
+	cd->state = DIE_INCOMPLETE;
+	cd->mapped = false;
+	cd->fqn = NULL;
+	cd->tag = -1;
+	cd->addr = 0;
+	INIT_LIST_HEAD(&cd->fragments);
+}
+
+static struct die *create_die(Dwarf_Die *die, enum die_state state)
+{
+	struct die *cd;
+
+	cd = xmalloc(sizeof(struct die));
+	init_die(cd);
+	cd->addr = (uintptr_t)die->addr;
+
+	hash_add(die_map, &cd->hash, die_hash(cd->addr, state));
+	return cd;
+}
+
+int __die_map_get(uintptr_t addr, enum die_state state, struct die **res)
+{
+	struct die *cd;
+
+	hash_for_each_possible(die_map, cd, hash, die_hash(addr, state)) {
+		if (cd->addr == addr && cd->state == state) {
+			*res = cd;
+			return 0;
+		}
+	}
+
+	return -1;
+}
+
+struct die *die_map_get(Dwarf_Die *die, enum die_state state)
+{
+	struct die *cd;
+
+	if (__die_map_get((uintptr_t)die->addr, state, &cd) == 0) {
+		map_hits++;
+		return cd;
+	}
+
+	map_misses++;
+	return create_die(die, state);
+}
+
+static void reset_die(struct die *cd)
+{
+	struct die_fragment *tmp;
+	struct die_fragment *df;
+
+	list_for_each_entry_safe(df, tmp, &cd->fragments, list) {
+		if (df->type == FRAGMENT_STRING)
+			free(df->data.str);
+		free(df);
+	}
+
+	if (cd->fqn && *cd->fqn)
+		free(cd->fqn);
+	init_die(cd);
+}
+
+void die_map_for_each(die_map_callback_t func, void *arg)
+{
+	struct hlist_node *tmp;
+	struct die *cd;
+
+	hash_for_each_safe(die_map, cd, tmp, hash) {
+		func(cd, arg);
+	}
+}
+
+void die_map_free(void)
+{
+	struct hlist_node *tmp;
+	unsigned int stats[DIE_LAST + 1];
+	struct die *cd;
+	int i;
+
+	memset(stats, 0, sizeof(stats));
+
+	hash_for_each_safe(die_map, cd, tmp, hash) {
+		stats[cd->state]++;
+		reset_die(cd);
+		free(cd);
+	}
+	hash_init(die_map);
+
+	if (map_hits + map_misses > 0)
+		debug("hits %u, misses %u (hit rate %.02f%%)", map_hits,
+		      map_misses,
+		      (100.0f * map_hits) / (map_hits + map_misses));
+
+	for (i = 0; i <= DIE_LAST; i++)
+		debug("%s: %u entries", die_state_name(i), stats[i]);
+}
+
+static struct die_fragment *append_item(struct die *cd)
+{
+	struct die_fragment *df;
+
+	df = xmalloc(sizeof(struct die_fragment));
+	df->type = FRAGMENT_EMPTY;
+	list_add_tail(&df->list, &cd->fragments);
+	return df;
+}
+
+void die_map_add_string(struct die *cd, const char *str)
+{
+	struct die_fragment *df;
+
+	if (!cd)
+		return;
+
+	df = append_item(cd);
+	df->data.str = xstrdup(str);
+	df->type = FRAGMENT_STRING;
+}
+
+void die_map_add_linebreak(struct die *cd, int linebreak)
+{
+	struct die_fragment *df;
+
+	if (!cd)
+		return;
+
+	df = append_item(cd);
+	df->data.linebreak = linebreak;
+	df->type = FRAGMENT_LINEBREAK;
+}
+
+void die_map_add_die(struct die *cd, struct die *child)
+{
+	struct die_fragment *df;
+
+	if (!cd)
+		return;
+
+	df = append_item(cd);
+	df->data.addr = child->addr;
+	df->type = FRAGMENT_DIE;
+}

scripts/gendwarfksyms/examples/kabi.h

@@ -0,0 +1,157 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2024 Google LLC
+ *
+ * Example macros for maintaining kABI stability.
+ *
+ * This file is based on android_kabi.h, which has the following notice:
+ *
+ * Heavily influenced by rh_kabi.h which came from the RHEL/CENTOS kernel
+ * and was:
+ *	Copyright (c) 2014 Don Zickus
+ *	Copyright (c) 2015-2018 Jiri Benc
+ *	Copyright (c) 2015 Sabrina Dubroca, Hannes Frederic Sowa
+ *	Copyright (c) 2016-2018 Prarit Bhargava
+ *	Copyright (c) 2017 Paolo Abeni, Larry Woodman
+ */
+
+#ifndef __KABI_H__
+#define __KABI_H__
+
+/* Kernel macros for userspace testing. */
+#ifndef __aligned
+#define __aligned(x) __attribute__((__aligned__(x)))
+#endif
+#ifndef __used
+#define __used __attribute__((__used__))
+#endif
+#ifndef __section
+#define __section(section) __attribute__((__section__(section)))
+#endif
+#ifndef __PASTE
+#define ___PASTE(a, b) a##b
+#define __PASTE(a, b) ___PASTE(a, b)
+#endif
+#ifndef __stringify
+#define __stringify_1(x...) #x
+#define __stringify(x...) __stringify_1(x)
+#endif
+
+#define __KABI_RULE(hint, target, value)                             \
+	static const char __PASTE(__gendwarfksyms_rule_,             \
+				  __COUNTER__)[] __used __aligned(1) \
+		__section(".discard.gendwarfksyms.kabi_rules") =     \
+			"1\0" #hint "\0" #target "\0" #value
+
+#define __KABI_NORMAL_SIZE_ALIGN(_orig, _new)                                             \
+	union {                                                                           \
+		_Static_assert(                                                           \
+			sizeof(struct { _new; }) <= sizeof(struct { _orig; }),            \
+			__FILE__ ":" __stringify(__LINE__) ": " __stringify(              \
+				_new) " is larger than " __stringify(_orig));             \
+		_Static_assert(                                                           \
+			__alignof__(struct { _new; }) <=                                  \
+				__alignof__(struct { _orig; }),                           \
+			__FILE__ ":" __stringify(__LINE__) ": " __stringify(              \
+				_orig) " is not aligned the same as " __stringify(_new)); \
+	}
+
+#define __KABI_REPLACE(_orig, _new)                    \
+	union {                                        \
+		_new;                                  \
+		struct {                               \
+			_orig;                         \
+		};                                     \
+		__KABI_NORMAL_SIZE_ALIGN(_orig, _new); \
+	}
+
+/*
+ * KABI_DECLONLY(fqn)
+ *   Treat the struct/union/enum fqn as a declaration, i.e. even if
+ *   a definition is available, don't expand the contents.
+ */
+#define KABI_DECLONLY(fqn) __KABI_RULE(declonly, fqn, )
+
+/*
+ * KABI_ENUMERATOR_IGNORE(fqn, field)
+ *   When expanding enum fqn, skip the provided field. This makes it
+ *   possible to hide added enum fields from versioning.
+ */
+#define KABI_ENUMERATOR_IGNORE(fqn, field) \
+	__KABI_RULE(enumerator_ignore, fqn field, )
+
+/*
+ * KABI_ENUMERATOR_VALUE(fqn, field, value)
+ *   When expanding enum fqn, use the provided value for the
+ *   specified field. This makes it possible to override enumerator
+ *   values when calculating versions.
+ */
+#define KABI_ENUMERATOR_VALUE(fqn, field, value) \
+	__KABI_RULE(enumerator_value, fqn field, value)
+
+/*
+ * KABI_RESERVE
+ *   Reserve some "padding" in a structure for use by LTS backports.
+ *   This is normally placed at the end of a structure.
+ *   number: the "number" of the padding variable in the structure.  Start with
+ *   1 and go up.
+ */
+#define KABI_RESERVE(n) unsigned long __kabi_reserved##n
+
+/*
+ * KABI_RESERVE_ARRAY
+ *   Same as _BACKPORT_RESERVE but allocates an array with the specified
+ *   size in bytes.
+ */
+#define KABI_RESERVE_ARRAY(n, s) \
+	unsigned char __aligned(8) __kabi_reserved##n[s]
+
+/*
+ * KABI_IGNORE
+ *   Add a new field that's ignored in versioning.
+ */
+#define KABI_IGNORE(n, _new)                     \
+	union {                                  \
+		_new;                            \
+		unsigned char __kabi_ignored##n; \
+	}
+
+/*
+ * KABI_REPLACE
+ *   Replace a field with a compatible new field.
+ */
+#define KABI_REPLACE(_oldtype, _oldname, _new) \
+	__KABI_REPLACE(_oldtype __kabi_renamed##_oldname, struct { _new; })
+
+/*
+ * KABI_USE(number, _new)
+ *   Use a previous padding entry that was defined with KABI_RESERVE
+ *   number: the previous "number" of the padding variable
+ *   _new: the variable to use now instead of the padding variable
+ */
+#define KABI_USE(number, _new) __KABI_REPLACE(KABI_RESERVE(number), _new)
+
+/*
+ * KABI_USE2(number, _new1, _new2)
+ *   Use a previous padding entry that was defined with KABI_RESERVE for
+ *   two new variables that fit into 64 bits.  This is good for when you do not
+ *   want to "burn" a 64bit padding variable for a smaller variable size if not
+ *   needed.
+ */
+#define KABI_USE2(number, _new1, _new2)        \
+	__KABI_REPLACE(                        \
+		KABI_RESERVE(number), struct { \
+			_new1;                 \
+			_new2;                 \
+		})
+/*
+ * KABI_USE_ARRAY(number, bytes, _new)
+ *   Use a previous padding entry that was defined with KABI_RESERVE_ARRAY
+ *   number: the previous "number" of the padding variable
+ *   bytes: the size in bytes reserved for the array
+ *   _new: the variable to use now instead of the padding variable
+ */
+#define KABI_USE_ARRAY(number, bytes, _new) \
+	__KABI_REPLACE(KABI_RESERVE_ARRAY(number, bytes), _new)
+
+#endif /* __KABI_H__ */

scripts/gendwarfksyms/examples/kabi_ex.c

@@ -0,0 +1,30 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * kabi_ex.c
+ *
+ * Copyright (C) 2024 Google LLC
+ *
+ * Examples for kABI stability features with --stable. See kabi_ex.h
+ * for details.
+ */
+
+#include "kabi_ex.h"
+
+struct s e0;
+enum e e1;
+
+struct ex0a ex0a;
+struct ex0b ex0b;
+struct ex0c ex0c;
+
+struct ex1a ex1a;
+struct ex1b ex1b;
+struct ex1c ex1c;
+
+struct ex2a ex2a;
+struct ex2b ex2b;
+struct ex2c ex2c;
+
+struct ex3a ex3a;
+struct ex3b ex3b;
+struct ex3c ex3c;

scripts/gendwarfksyms/examples/kabi_ex.h

@@ -0,0 +1,263 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * kabi_ex.h
+ *
+ * Copyright (C) 2024 Google LLC
+ *
+ * Examples for kABI stability features with --stable.
+ */
+
+/*
+ * The comments below each example contain the expected gendwarfksyms
+ * output, which can be verified using LLVM's FileCheck tool:
+ *
+ * https://llvm.org/docs/CommandGuide/FileCheck.html
+ *
+ * Usage:
+ *
+ * $ gcc -g -c examples/kabi_ex.c -o examples/kabi_ex.o
+ *
+ * $ nm examples/kabi_ex.o | awk '{ print $NF }' | \
+ * 	./gendwarfksyms --stable --dump-dies \
+ * 		examples/kabi_ex.o 2>&1 >/dev/null | \
+ * 	FileCheck examples/kabi_ex.h --check-prefix=STABLE
+ */
+
+#ifndef __KABI_EX_H__
+#define __KABI_EX_H__
+
+#include "kabi.h"
+
+/*
+ * Example: kABI rules
+ */
+
+struct s {
+	int a;
+};
+
+KABI_DECLONLY(s);
+
+/*
+ * STABLE:      variable structure_type s {
+ * STABLE-NEXT: }
+ */
+
+enum e {
+	A,
+	B,
+	C,
+	D,
+};
+
+KABI_ENUMERATOR_IGNORE(e, B);
+KABI_ENUMERATOR_IGNORE(e, C);
+KABI_ENUMERATOR_VALUE(e, D, 123456789);
+
+/*
+ * STABLE:      variable enumeration_type e {
+ * STABLE-NEXT:   enumerator A = 0 ,
+ * STABLE-NEXT:   enumerator D = 123456789
+ * STABLE-NEXT: } byte_size(4)
+*/
+
+/*
+ * Example: Reserved fields
+ */
+struct ex0a {
+	int a;
+	KABI_RESERVE(0);
+	KABI_RESERVE(1);
+};
+
+/*
+ * STABLE:      variable structure_type ex0a {
+ * STABLE-NEXT:   member base_type int byte_size(4) encoding(5) a data_member_location(0) ,
+ * STABLE-NEXT:   member base_type [[ULONG:long unsigned int|unsigned long]] byte_size(8) encoding(7) data_member_location(8) ,
+ * STABLE-NEXT:   member base_type [[ULONG]] byte_size(8) encoding(7) data_member_location(16)
+ * STABLE-NEXT: } byte_size(24)
+ */
+
+struct ex0b {
+	int a;
+	KABI_RESERVE(0);
+	KABI_USE2(1, int b, int c);
+};
+
+/*
+ * STABLE:      variable structure_type ex0b {
+ * STABLE-NEXT:   member base_type int byte_size(4) encoding(5) a data_member_location(0) ,
+ * STABLE-NEXT:   member base_type [[ULONG]] byte_size(8) encoding(7) data_member_location(8) ,
+ * STABLE-NEXT:   member base_type [[ULONG]] byte_size(8) encoding(7) data_member_location(16)
+ * STABLE-NEXT: } byte_size(24)
+ */
+
+struct ex0c {
+	int a;
+	KABI_USE(0, void *p);
+	KABI_USE2(1, int b, int c);
+};
+
+/*
+ * STABLE:      variable structure_type ex0c {
+ * STABLE-NEXT:   member base_type int byte_size(4) encoding(5) a data_member_location(0) ,
+ * STABLE-NEXT:   member base_type [[ULONG]] byte_size(8) encoding(7) data_member_location(8) ,
+ * STABLE-NEXT:   member base_type [[ULONG]] byte_size(8) encoding(7) data_member_location(16)
+ * STABLE-NEXT: } byte_size(24)
+ */
+
+/*
+ * Example: A reserved array
+ */
+
+struct ex1a {
+	unsigned int a;
+	KABI_RESERVE_ARRAY(0, 64);
+};
+
+/*
+ * STABLE:      variable structure_type ex1a {
+ * STABLE-NEXT:   member base_type unsigned int byte_size(4) encoding(7) a data_member_location(0) ,
+ * STABLE-NEXT:   member array_type[64] {
+ * STABLE-NEXT:     base_type unsigned char byte_size(1) encoding(8)
+ * STABLE-NEXT:   } data_member_location(8)
+ * STABLE-NEXT: } byte_size(72)
+ */
+
+struct ex1b {
+	unsigned int a;
+	KABI_USE_ARRAY(
+		0, 64, struct {
+			void *p;
+			KABI_RESERVE_ARRAY(1, 56);
+		});
+};
+
+/*
+ * STABLE:      variable structure_type ex1b {
+ * STABLE-NEXT:   member base_type unsigned int byte_size(4) encoding(7) a data_member_location(0) ,
+ * STABLE-NEXT:   member array_type[64] {
+ * STABLE-NEXT:     base_type unsigned char byte_size(1) encoding(8)
+ * STABLE-NEXT:   } data_member_location(8)
+ * STABLE-NEXT: } byte_size(72)
+ */
+
+struct ex1c {
+	unsigned int a;
+	KABI_USE_ARRAY(0, 64, void *p[8]);
+};
+
+/*
+ * STABLE:      variable structure_type ex1c {
+ * STABLE-NEXT:   member base_type unsigned int byte_size(4) encoding(7) a data_member_location(0) ,
+ * STABLE-NEXT:   member array_type[64] {
+ * STABLE-NEXT:     base_type unsigned char byte_size(1) encoding(8)
+ * STABLE-NEXT:   } data_member_location(8)
+ * STABLE-NEXT: } byte_size(72)
+ */
+
+/*
+ * Example: An ignored field added to an alignment hole
+ */
+
+struct ex2a {
+	int a;
+	unsigned long b;
+	int c;
+	unsigned long d;
+};
+
+/*
+ * STABLE:      variable structure_type ex2a {
+ * STABLE-NEXT:   member base_type int byte_size(4) encoding(5) a data_member_location(0) ,
+ * STABLE-NEXT:   member base_type [[ULONG:long unsigned int|unsigned long]] byte_size(8) encoding(7) b data_member_location(8)
+ * STABLE-NEXT:   member base_type int byte_size(4) encoding(5) c data_member_location(16) ,
+ * STABLE-NEXT:   member base_type [[ULONG]] byte_size(8) encoding(7) d data_member_location(24)
+ * STABLE-NEXT: } byte_size(32)
+ */
+
+struct ex2b {
+	int a;
+	KABI_IGNORE(0, unsigned int n);
+	unsigned long b;
+	int c;
+	unsigned long d;
+};
+
+_Static_assert(sizeof(struct ex2a) == sizeof(struct ex2b), "ex2a size doesn't match ex2b");
+
+/*
+ * STABLE:      variable structure_type ex2b {
+ * STABLE-NEXT:   member base_type int byte_size(4) encoding(5) a data_member_location(0) ,
+ * STABLE-NEXT:   member base_type [[ULONG]] byte_size(8) encoding(7) b data_member_location(8)
+ * STABLE-NEXT:   member base_type int byte_size(4) encoding(5) c data_member_location(16) ,
+ * STABLE-NEXT:   member base_type [[ULONG]] byte_size(8) encoding(7) d data_member_location(24)
+ * STABLE-NEXT: } byte_size(32)
+ */
+
+struct ex2c {
+	int a;
+	KABI_IGNORE(0, unsigned int n);
+	unsigned long b;
+	int c;
+	KABI_IGNORE(1, unsigned int m);
+	unsigned long d;
+};
+
+_Static_assert(sizeof(struct ex2a) == sizeof(struct ex2c), "ex2a size doesn't match ex2c");
+
+/*
+ * STABLE:      variable structure_type ex2c {
+ * STABLE-NEXT:   member base_type int byte_size(4) encoding(5) a data_member_location(0) ,
+ * STABLE-NEXT:   member base_type [[ULONG]] byte_size(8) encoding(7) b data_member_location(8)
+ * STABLE-NEXT:   member base_type int byte_size(4) encoding(5) c data_member_location(16) ,
+ * STABLE-NEXT:   member base_type [[ULONG]] byte_size(8) encoding(7) d data_member_location(24)
+ * STABLE-NEXT: } byte_size(32)
+ */
+
+
+/*
+ * Example: A replaced field
+ */
+
+struct ex3a {
+	unsigned long a;
+	unsigned long unused;
+};
+
+/*
+ * STABLE:      variable structure_type ex3a {
+ * STABLE-NEXT:   member base_type [[ULONG:long unsigned int|unsigned long]] byte_size(8) encoding(7) a data_member_location(0)
+ * STABLE-NEXT:   member base_type [[ULONG]] byte_size(8) encoding(7) unused data_member_location(8)
+ * STABLE-NEXT: } byte_size(16)
+ */
+
+struct ex3b {
+	unsigned long a;
+	KABI_REPLACE(unsigned long, unused, unsigned long renamed);
+};
+
+_Static_assert(sizeof(struct ex3a) == sizeof(struct ex3b), "ex3a size doesn't match ex3b");
+
+/*
+ * STABLE:      variable structure_type ex3b {
+ * STABLE-NEXT:   member base_type [[ULONG]] byte_size(8) encoding(7) a data_member_location(0)
+ * STABLE-NEXT:   member base_type [[ULONG]] byte_size(8) encoding(7) unused data_member_location(8)
+ * STABLE-NEXT: } byte_size(16)
+ */
+
+struct ex3c {
+	unsigned long a;
+	KABI_REPLACE(unsigned long, unused, long replaced);
+};
+
+_Static_assert(sizeof(struct ex3a) == sizeof(struct ex3c), "ex3a size doesn't match ex3c");
+
+/*
+ * STABLE:      variable structure_type ex3c {
+ * STABLE-NEXT:   member base_type [[ULONG]] byte_size(8) encoding(7) a data_member_location(0)
+ * STABLE-NEXT:   member base_type [[ULONG]] byte_size(8) encoding(7) unused data_member_location(8)
+ * STABLE-NEXT: } byte_size(16)
+ */
+
+#endif /* __KABI_EX_H__ */

scripts/gendwarfksyms/examples/symbolptr.c

@@ -0,0 +1,33 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2024 Google LLC
+ *
+ * Example for symbol pointers. When compiled with Clang, gendwarfkyms
+ * uses a symbol pointer for `f`.
+ *
+ * $ clang -g -c examples/symbolptr.c -o examples/symbolptr.o
+ * $ echo -e "f\ng\np" | ./gendwarfksyms -d examples/symbolptr.o
+ */
+
+/* Kernel macros for userspace testing. */
+#ifndef __used
+#define __used __attribute__((__used__))
+#endif
+#ifndef __section
+#define __section(section) __attribute__((__section__(section)))
+#endif
+
+#define __GENDWARFKSYMS_EXPORT(sym)				\
+	static typeof(sym) *__gendwarfksyms_ptr_##sym __used	\
+		__section(".discard.gendwarfksyms") = &sym;
+
+extern void f(unsigned int arg);
+void g(int *arg);
+void g(int *arg) {}
+
+struct s;
+extern struct s *p;
+
+__GENDWARFKSYMS_EXPORT(f);
+__GENDWARFKSYMS_EXPORT(g);
+__GENDWARFKSYMS_EXPORT(p);

scripts/gendwarfksyms/gendwarfksyms.c

@@ -0,0 +1,187 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2024 Google LLC
+ */
+
+#include <fcntl.h>
+#include <getopt.h>
+#include <errno.h>
+#include <stdarg.h>
+#include <string.h>
+#include <unistd.h>
+#include "gendwarfksyms.h"
+
+/*
+ * Options
+ */
+
+/* Print debugging information to stderr */
+int debug;
+/* Dump DIE contents */
+int dump_dies;
+/* Print debugging information about die_map changes */
+int dump_die_map;
+/* Print out type strings (i.e. type_map) */
+int dump_types;
+/* Print out expanded type strings used for symbol versions */
+int dump_versions;
+/* Support kABI stability features */
+int stable;
+/* Write a symtypes file */
+int symtypes;
+static const char *symtypes_file;
+
+static void usage(void)
+{
+	fputs("Usage: gendwarfksyms [options] elf-object-file ... < symbol-list\n\n"
+	      "Options:\n"
+	      "  -d, --debug          Print debugging information\n"
+	      "      --dump-dies      Dump DWARF DIE contents\n"
+	      "      --dump-die-map   Print debugging information about die_map changes\n"
+	      "      --dump-types     Dump type strings\n"
+	      "      --dump-versions  Dump expanded type strings used for symbol versions\n"
+	      "  -s, --stable         Support kABI stability features\n"
+	      "  -T, --symtypes file  Write a symtypes file\n"
+	      "  -h, --help           Print this message\n"
+	      "\n",
+	      stderr);
+}
+
+static int process_module(Dwfl_Module *mod, void **userdata, const char *name,
+			  Dwarf_Addr base, void *arg)
+{
+	Dwarf_Addr dwbias;
+	Dwarf_Die cudie;
+	Dwarf_CU *cu = NULL;
+	Dwarf *dbg;
+	FILE *symfile = arg;
+	int res;
+
+	debug("%s", name);
+	dbg = dwfl_module_getdwarf(mod, &dwbias);
+
+	/*
+	 * Look for exported symbols in each CU, follow the DIE tree, and add
+	 * the entries to die_map.
+	 */
+	do {
+		res = dwarf_get_units(dbg, cu, &cu, NULL, NULL, &cudie, NULL);
+		if (res < 0)
+			error("dwarf_get_units failed: no debugging information?");
+		if (res == 1)
+			break; /* No more units */
+
+		process_cu(&cudie);
+	} while (cu);
+
+	/*
+	 * Use die_map to expand type strings, write them to `symfile`, and
+	 * calculate symbol versions.
+	 */
+	generate_symtypes_and_versions(symfile);
+	die_map_free();
+
+	return DWARF_CB_OK;
+}
+
+static const Dwfl_Callbacks callbacks = {
+	.section_address = dwfl_offline_section_address,
+	.find_debuginfo = dwfl_standard_find_debuginfo,
+};
+
+int main(int argc, char **argv)
+{
+	FILE *symfile = NULL;
+	unsigned int n;
+	int opt;
+
+	static const struct option opts[] = {
+		{ "debug", 0, NULL, 'd' },
+		{ "dump-dies", 0, &dump_dies, 1 },
+		{ "dump-die-map", 0, &dump_die_map, 1 },
+		{ "dump-types", 0, &dump_types, 1 },
+		{ "dump-versions", 0, &dump_versions, 1 },
+		{ "stable", 0, NULL, 's' },
+		{ "symtypes", 1, NULL, 'T' },
+		{ "help", 0, NULL, 'h' },
+		{ 0, 0, NULL, 0 }
+	};
+
+	while ((opt = getopt_long(argc, argv, "dsT:h", opts, NULL)) != EOF) {
+		switch (opt) {
+		case 0:
+			break;
+		case 'd':
+			debug = 1;
+			break;
+		case 's':
+			stable = 1;
+			break;
+		case 'T':
+			symtypes = 1;
+			symtypes_file = optarg;
+			break;
+		case 'h':
+			usage();
+			return 0;
+		default:
+			usage();
+			return 1;
+		}
+	}
+
+	if (dump_die_map)
+		dump_dies = 1;
+
+	if (optind >= argc) {
+		usage();
+		error("no input files?");
+	}
+
+	symbol_read_exports(stdin);
+
+	if (symtypes_file) {
+		symfile = fopen(symtypes_file, "w");
+		if (!symfile)
+			error("fopen failed for '%s': %s", symtypes_file,
+			      strerror(errno));
+	}
+
+	for (n = optind; n < argc; n++) {
+		Dwfl *dwfl;
+		int fd;
+
+		fd = open(argv[n], O_RDONLY);
+		if (fd == -1)
+			error("open failed for '%s': %s", argv[n],
+			      strerror(errno));
+
+		symbol_read_symtab(fd);
+		kabi_read_rules(fd);
+
+		dwfl = dwfl_begin(&callbacks);
+		if (!dwfl)
+			error("dwfl_begin failed for '%s': %s", argv[n],
+			      dwarf_errmsg(-1));
+
+		if (!dwfl_report_offline(dwfl, argv[n], argv[n], fd))
+			error("dwfl_report_offline failed for '%s': %s",
+			      argv[n], dwarf_errmsg(-1));
+
+		dwfl_report_end(dwfl, NULL, NULL);
+
+		if (dwfl_getmodules(dwfl, &process_module, symfile, 0))
+			error("dwfl_getmodules failed for '%s'", argv[n]);
+
+		dwfl_end(dwfl);
+		kabi_free();
+	}
+
+	if (symfile)
+		check(fclose(symfile));
+
+	symbol_print_versions();
+	symbol_free();
+
+	return 0;
+}

scripts/gendwarfksyms/gendwarfksyms.h

@@ -0,0 +1,296 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2024 Google LLC
+ */
+
+#include <dwarf.h>
+#include <elfutils/libdw.h>
+#include <elfutils/libdwfl.h>
+#include <stdlib.h>
+#include <stdio.h>
+
+#include <hash.h>
+#include <hashtable.h>
+#include <xalloc.h>
+
+#ifndef __GENDWARFKSYMS_H
+#define __GENDWARFKSYMS_H
+
+/*
+ * Options -- in gendwarfksyms.c
+ */
+extern int debug;
+extern int dump_dies;
+extern int dump_die_map;
+extern int dump_types;
+extern int dump_versions;
+extern int stable;
+extern int symtypes;
+
+/*
+ * Output helpers
+ */
+#define __PREFIX "gendwarfksyms: "
+#define __println(prefix, format, ...)                                \
+	fprintf(stderr, prefix __PREFIX "%s: " format "\n", __func__, \
+		##__VA_ARGS__)
+
+#define debug(format, ...)                                    \
+	do {                                                  \
+		if (debug)                                    \
+			__println("", format, ##__VA_ARGS__); \
+	} while (0)
+
+#define warn(format, ...) __println("warning: ", format, ##__VA_ARGS__)
+#define error(format, ...)                                   \
+	do {                                                 \
+		__println("error: ", format, ##__VA_ARGS__); \
+		exit(1);                                     \
+	} while (0)
+
+#define __die_debug(color, format, ...)                                 \
+	do {                                                            \
+		if (dump_dies && dump_die_map)                          \
+			fprintf(stderr,                                 \
+				"\033[" #color "m<" format ">\033[39m", \
+				__VA_ARGS__);                           \
+	} while (0)
+
+#define die_debug_r(format, ...) __die_debug(91, format, __VA_ARGS__)
+#define die_debug_g(format, ...) __die_debug(92, format, __VA_ARGS__)
+#define die_debug_b(format, ...) __die_debug(94, format, __VA_ARGS__)
+
+/*
+ * Error handling helpers
+ */
+#define __check(expr, test)                                     \
+	({                                                      \
+		int __res = expr;                               \
+		if (test)                                       \
+			error("`%s` failed: %d", #expr, __res); \
+		__res;                                          \
+	})
+
+/* Error == non-zero values */
+#define check(expr) __check(expr, __res)
+/* Error == negative values */
+#define checkp(expr) __check(expr, __res < 0)
+
+/* Consistent aliases (DW_TAG_<type>_type) for DWARF tags */
+#define DW_TAG_enumerator_type DW_TAG_enumerator
+#define DW_TAG_formal_parameter_type DW_TAG_formal_parameter
+#define DW_TAG_member_type DW_TAG_member
+#define DW_TAG_template_type_parameter_type DW_TAG_template_type_parameter
+#define DW_TAG_typedef_type DW_TAG_typedef
+#define DW_TAG_variant_part_type DW_TAG_variant_part
+#define DW_TAG_variant_type DW_TAG_variant
+
+/*
+ * symbols.c
+ */
+
+/* See symbols.c:is_symbol_ptr */
+#define SYMBOL_PTR_PREFIX "__gendwarfksyms_ptr_"
+#define SYMBOL_PTR_PREFIX_LEN (sizeof(SYMBOL_PTR_PREFIX) - 1)
+
+static inline unsigned int addr_hash(uintptr_t addr)
+{
+	return hash_ptr((const void *)addr);
+}
+
+enum symbol_state {
+	SYMBOL_UNPROCESSED,
+	SYMBOL_MAPPED,
+	SYMBOL_PROCESSED
+};
+
+struct symbol_addr {
+	uint32_t section;
+	Elf64_Addr address;
+};
+
+struct symbol {
+	const char *name;
+	struct symbol_addr addr;
+	struct hlist_node addr_hash;
+	struct hlist_node name_hash;
+	enum symbol_state state;
+	uintptr_t die_addr;
+	uintptr_t ptr_die_addr;
+	unsigned long crc;
+};
+
+typedef void (*symbol_callback_t)(struct symbol *, void *arg);
+
+bool is_symbol_ptr(const char *name);
+void symbol_read_exports(FILE *file);
+void symbol_read_symtab(int fd);
+struct symbol *symbol_get(const char *name);
+void symbol_set_ptr(struct symbol *sym, Dwarf_Die *ptr);
+void symbol_set_die(struct symbol *sym, Dwarf_Die *die);
+void symbol_set_crc(struct symbol *sym, unsigned long crc);
+void symbol_for_each(symbol_callback_t func, void *arg);
+void symbol_print_versions(void);
+void symbol_free(void);
+
+/*
+ * die.c
+ */
+
+enum die_state {
+	DIE_INCOMPLETE,
+	DIE_UNEXPANDED,
+	DIE_COMPLETE,
+	DIE_SYMBOL,
+	DIE_LAST = DIE_SYMBOL
+};
+
+enum die_fragment_type {
+	FRAGMENT_EMPTY,
+	FRAGMENT_STRING,
+	FRAGMENT_LINEBREAK,
+	FRAGMENT_DIE
+};
+
+struct die_fragment {
+	enum die_fragment_type type;
+	union {
+		char *str;
+		int linebreak;
+		uintptr_t addr;
+	} data;
+	struct list_head list;
+};
+
+#define CASE_CONST_TO_STR(name) \
+	case name:              \
+		return #name;
+
+static inline const char *die_state_name(enum die_state state)
+{
+	switch (state) {
+	CASE_CONST_TO_STR(DIE_INCOMPLETE)
+	CASE_CONST_TO_STR(DIE_UNEXPANDED)
+	CASE_CONST_TO_STR(DIE_COMPLETE)
+	CASE_CONST_TO_STR(DIE_SYMBOL)
+	}
+
+	error("unexpected die_state: %d", state);
+}
+
+struct die {
+	enum die_state state;
+	bool mapped;
+	char *fqn;
+	int tag;
+	uintptr_t addr;
+	struct list_head fragments;
+	struct hlist_node hash;
+};
+
+typedef void (*die_map_callback_t)(struct die *, void *arg);
+
+int __die_map_get(uintptr_t addr, enum die_state state, struct die **res);
+struct die *die_map_get(Dwarf_Die *die, enum die_state state);
+void die_map_add_string(struct die *pd, const char *str);
+void die_map_add_linebreak(struct die *pd, int linebreak);
+void die_map_for_each(die_map_callback_t func, void *arg);
+void die_map_add_die(struct die *pd, struct die *child);
+void die_map_free(void);
+
+/*
+ * cache.c
+ */
+
+#define CACHE_HASH_BITS 10
+
+/* A cache for addresses we've already seen. */
+struct cache {
+	HASHTABLE_DECLARE(cache, 1 << CACHE_HASH_BITS);
+};
+
+void cache_set(struct cache *cache, unsigned long key, int value);
+int cache_get(struct cache *cache, unsigned long key);
+void cache_init(struct cache *cache);
+void cache_free(struct cache *cache);
+
+static inline void __cache_mark_expanded(struct cache *cache, uintptr_t addr)
+{
+	cache_set(cache, addr, 1);
+}
+
+static inline bool __cache_was_expanded(struct cache *cache, uintptr_t addr)
+{
+	return cache_get(cache, addr) == 1;
+}
+
+static inline void cache_mark_expanded(struct cache *cache, void *addr)
+{
+	__cache_mark_expanded(cache, (uintptr_t)addr);
+}
+
+static inline bool cache_was_expanded(struct cache *cache, void *addr)
+{
+	return __cache_was_expanded(cache, (uintptr_t)addr);
+}
+
+/*
+ * dwarf.c
+ */
+
+struct expansion_state {
+	bool expand;
+	const char *current_fqn;
+};
+
+struct kabi_state {
+	int members;
+	Dwarf_Die placeholder;
+	const char *orig_name;
+};
+
+struct state {
+	struct symbol *sym;
+	Dwarf_Die die;
+
+	/* List expansion */
+	bool first_list_item;
+
+	/* Structure expansion */
+	struct expansion_state expand;
+	struct cache expansion_cache;
+
+	/* Reserved or ignored members */
+	struct kabi_state kabi;
+};
+
+typedef int (*die_callback_t)(struct state *state, struct die *cache,
+			      Dwarf_Die *die);
+typedef bool (*die_match_callback_t)(Dwarf_Die *die);
+bool match_all(Dwarf_Die *die);
+
+int process_die_container(struct state *state, struct die *cache,
+			  Dwarf_Die *die, die_callback_t func,
+			  die_match_callback_t match);
+
+void process_cu(Dwarf_Die *cudie);
+
+/*
+ * types.c
+ */
+
+void generate_symtypes_and_versions(FILE *file);
+
+/*
+ * kabi.c
+ */
+
+bool kabi_is_enumerator_ignored(const char *fqn, const char *field);
+bool kabi_get_enumerator_value(const char *fqn, const char *field,
+			       unsigned long *value);
+bool kabi_is_declonly(const char *fqn);
+
+void kabi_read_rules(int fd);
+void kabi_free(void);
+
+#endif /* __GENDWARFKSYMS_H */

scripts/gendwarfksyms/kabi.c

@@ -0,0 +1,336 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2024 Google LLC
+ */
+
+#define _GNU_SOURCE
+#include <errno.h>
+#include <stdio.h>
+
+#include "gendwarfksyms.h"
+
+#define KABI_RULE_SECTION ".discard.gendwarfksyms.kabi_rules"
+#define KABI_RULE_VERSION "1"
+
+/*
+ * The rule section consists of four null-terminated strings per
+ * entry:
+ *
+ *   1. version
+ *      Entry format version. Must match KABI_RULE_VERSION.
+ *
+ *   2. type
+ *      Type of the kABI rule. Must be one of the tags defined below.
+ *
+ *   3. target
+ *      Rule-dependent target, typically the fully qualified name of
+ *      the target DIE.
+ *
+ *   4. value
+ *      Rule-dependent value.
+ */
+#define KABI_RULE_MIN_ENTRY_SIZE                                  \
+	(/* version\0 */ 2 + /* type\0 */ 2 + /* target\0" */ 1 + \
+	 /* value\0 */ 1)
+#define KABI_RULE_EMPTY_VALUE ""
+
+/*
+ * Rule: declonly
+ * - For the struct/enum/union in the target field, treat it as a
+ *   declaration only even if a definition is available.
+ */
+#define KABI_RULE_TAG_DECLONLY "declonly"
+
+/*
+ * Rule: enumerator_ignore
+ * - For the enum_field in the target field, ignore the enumerator.
+ */
+#define KABI_RULE_TAG_ENUMERATOR_IGNORE "enumerator_ignore"
+
+/*
+ * Rule: enumerator_value
+ * - For the fqn_field in the target field, set the value to the
+ *   unsigned integer in the value field.
+ */
+#define KABI_RULE_TAG_ENUMERATOR_VALUE "enumerator_value"
+
+enum kabi_rule_type {
+	KABI_RULE_TYPE_UNKNOWN,
+	KABI_RULE_TYPE_DECLONLY,
+	KABI_RULE_TYPE_ENUMERATOR_IGNORE,
+	KABI_RULE_TYPE_ENUMERATOR_VALUE,
+};
+
+#define RULE_HASH_BITS 7
+
+struct rule {
+	enum kabi_rule_type type;
+	const char *target;
+	const char *value;
+	struct hlist_node hash;
+};
+
+/* { type, target } -> struct rule */
+static HASHTABLE_DEFINE(rules, 1 << RULE_HASH_BITS);
+
+static inline unsigned int rule_values_hash(enum kabi_rule_type type,
+					    const char *target)
+{
+	return hash_32(type) ^ hash_str(target);
+}
+
+static inline unsigned int rule_hash(const struct rule *rule)
+{
+	return rule_values_hash(rule->type, rule->target);
+}
+
+static inline const char *get_rule_field(const char **pos, ssize_t *left)
+{
+	const char *start = *pos;
+	size_t len;
+
+	if (*left <= 0)
+		error("unexpected end of kABI rules");
+
+	len = strnlen(start, *left) + 1;
+	*pos += len;
+	*left -= len;
+
+	return start;
+}
+
+void kabi_read_rules(int fd)
+{
+	GElf_Shdr shdr_mem;
+	GElf_Shdr *shdr;
+	Elf_Data *rule_data = NULL;
+	Elf_Scn *scn;
+	Elf *elf;
+	size_t shstrndx;
+	const char *rule_str;
+	ssize_t left;
+	int i;
+
+	const struct {
+		enum kabi_rule_type type;
+		const char *tag;
+	} rule_types[] = {
+		{
+			.type = KABI_RULE_TYPE_DECLONLY,
+			.tag = KABI_RULE_TAG_DECLONLY,
+		},
+		{
+			.type = KABI_RULE_TYPE_ENUMERATOR_IGNORE,
+			.tag = KABI_RULE_TAG_ENUMERATOR_IGNORE,
+		},
+		{
+			.type = KABI_RULE_TYPE_ENUMERATOR_VALUE,
+			.tag = KABI_RULE_TAG_ENUMERATOR_VALUE,
+		},
+	};
+
+	if (!stable)
+		return;
+
+	if (elf_version(EV_CURRENT) != EV_CURRENT)
+		error("elf_version failed: %s", elf_errmsg(-1));
+
+	elf = elf_begin(fd, ELF_C_READ_MMAP, NULL);
+	if (!elf)
+		error("elf_begin failed: %s", elf_errmsg(-1));
+
+	if (elf_getshdrstrndx(elf, &shstrndx) < 0)
+		error("elf_getshdrstrndx failed: %s", elf_errmsg(-1));
+
+	scn = elf_nextscn(elf, NULL);
+
+	while (scn) {
+		const char *sname;
+
+		shdr = gelf_getshdr(scn, &shdr_mem);
+		if (!shdr)
+			error("gelf_getshdr failed: %s", elf_errmsg(-1));
+
+		sname = elf_strptr(elf, shstrndx, shdr->sh_name);
+		if (!sname)
+			error("elf_strptr failed: %s", elf_errmsg(-1));
+
+		if (!strcmp(sname, KABI_RULE_SECTION)) {
+			rule_data = elf_getdata(scn, NULL);
+			if (!rule_data)
+				error("elf_getdata failed: %s", elf_errmsg(-1));
+			break;
+		}
+
+		scn = elf_nextscn(elf, scn);
+	}
+
+	if (!rule_data) {
+		debug("kABI rules not found");
+		check(elf_end(elf));
+		return;
+	}
+
+	rule_str = rule_data->d_buf;
+	left = shdr->sh_size;
+
+	if (left < KABI_RULE_MIN_ENTRY_SIZE)
+		error("kABI rule section too small: %zd bytes", left);
+
+	if (rule_str[left - 1] != '\0')
+		error("kABI rules are not null-terminated");
+
+	while (left > KABI_RULE_MIN_ENTRY_SIZE) {
+		enum kabi_rule_type type = KABI_RULE_TYPE_UNKNOWN;
+		const char *field;
+		struct rule *rule;
+
+		/* version */
+		field = get_rule_field(&rule_str, &left);
+
+		if (strcmp(field, KABI_RULE_VERSION))
+			error("unsupported kABI rule version: '%s'", field);
+
+		/* type */
+		field = get_rule_field(&rule_str, &left);
+
+		for (i = 0; i < ARRAY_SIZE(rule_types); i++) {
+			if (!strcmp(field, rule_types[i].tag)) {
+				type = rule_types[i].type;
+				break;
+			}
+		}
+
+		if (type == KABI_RULE_TYPE_UNKNOWN)
+			error("unsupported kABI rule type: '%s'", field);
+
+		rule = xmalloc(sizeof(struct rule));
+
+		rule->type = type;
+		rule->target = xstrdup(get_rule_field(&rule_str, &left));
+		rule->value = xstrdup(get_rule_field(&rule_str, &left));
+
+		hash_add(rules, &rule->hash, rule_hash(rule));
+
+		debug("kABI rule: type: '%s', target: '%s', value: '%s'", field,
+		      rule->target, rule->value);
+	}
+
+	if (left > 0)
+		warn("unexpected data at the end of the kABI rules section");
+
+	check(elf_end(elf));
+}
+
+bool kabi_is_declonly(const char *fqn)
+{
+	struct rule *rule;
+
+	if (!stable)
+		return false;
+	if (!fqn || !*fqn)
+		return false;
+
+	hash_for_each_possible(rules, rule, hash,
+			       rule_values_hash(KABI_RULE_TYPE_DECLONLY, fqn)) {
+		if (rule->type == KABI_RULE_TYPE_DECLONLY &&
+		    !strcmp(fqn, rule->target))
+			return true;
+	}
+
+	return false;
+}
+
+static char *get_enumerator_target(const char *fqn, const char *field)
+{
+	char *target = NULL;
+
+	if (asprintf(&target, "%s %s", fqn, field) < 0)
+		error("asprintf failed for '%s %s'", fqn, field);
+
+	return target;
+}
+
+static unsigned long get_ulong_value(const char *value)
+{
+	unsigned long result = 0;
+	char *endptr = NULL;
+
+	errno = 0;
+	result = strtoul(value, &endptr, 10);
+
+	if (errno || *endptr)
+		error("invalid unsigned value '%s'", value);
+
+	return result;
+}
+
+bool kabi_is_enumerator_ignored(const char *fqn, const char *field)
+{
+	bool match = false;
+	struct rule *rule;
+	char *target;
+
+	if (!stable)
+		return false;
+	if (!fqn || !*fqn || !field || !*field)
+		return false;
+
+	target = get_enumerator_target(fqn, field);
+
+	hash_for_each_possible(
+		rules, rule, hash,
+		rule_values_hash(KABI_RULE_TYPE_ENUMERATOR_IGNORE, target)) {
+		if (rule->type == KABI_RULE_TYPE_ENUMERATOR_IGNORE &&
+		    !strcmp(target, rule->target)) {
+			match = true;
+			break;
+		}
+	}
+
+	free(target);
+	return match;
+}
+
+bool kabi_get_enumerator_value(const char *fqn, const char *field,
+			       unsigned long *value)
+{
+	bool match = false;
+	struct rule *rule;
+	char *target;
+
+	if (!stable)
+		return false;
+	if (!fqn || !*fqn || !field || !*field)
+		return false;
+
+	target = get_enumerator_target(fqn, field);
+
+	hash_for_each_possible(rules, rule, hash,
+			       rule_values_hash(KABI_RULE_TYPE_ENUMERATOR_VALUE,
+						target)) {
+		if (rule->type == KABI_RULE_TYPE_ENUMERATOR_VALUE &&
+		    !strcmp(target, rule->target)) {
+			*value = get_ulong_value(rule->value);
+			match = true;
+			break;
+		}
+	}
+
+	free(target);
+	return match;
+}
+
+void kabi_free(void)
+{
+	struct hlist_node *tmp;
+	struct rule *rule;
+
+	hash_for_each_safe(rules, rule, tmp, hash) {
+		free((void *)rule->target);
+		free((void *)rule->value);
+		free(rule);
+	}
+
+	hash_init(rules);
+}

scripts/gendwarfksyms/symbols.c

@@ -0,0 +1,341 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2024 Google LLC
+ */
+
+#include "gendwarfksyms.h"
+
+#define SYMBOL_HASH_BITS 12
+
+/* struct symbol_addr -> struct symbol */
+static HASHTABLE_DEFINE(symbol_addrs, 1 << SYMBOL_HASH_BITS);
+/* name -> struct symbol */
+static HASHTABLE_DEFINE(symbol_names, 1 << SYMBOL_HASH_BITS);
+
+static inline unsigned int symbol_addr_hash(const struct symbol_addr *addr)
+{
+	return hash_32(addr->section ^ addr_hash(addr->address));
+}
+
+static unsigned int __for_each_addr(struct symbol *sym, symbol_callback_t func,
+				    void *data)
+{
+	struct hlist_node *tmp;
+	struct symbol *match = NULL;
+	unsigned int processed = 0;
+
+	hash_for_each_possible_safe(symbol_addrs, match, tmp, addr_hash,
+				    symbol_addr_hash(&sym->addr)) {
+		if (match == sym)
+			continue; /* Already processed */
+
+		if (match->addr.section == sym->addr.section &&
+		    match->addr.address == sym->addr.address) {
+			func(match, data);
+			++processed;
+		}
+	}
+
+	return processed;
+}
+
+/*
+ * For symbols without debugging information (e.g. symbols defined in other
+ * TUs), we also match __gendwarfksyms_ptr_<symbol_name> symbols, which the
+ * kernel uses to ensure type information is present in the TU that exports
+ * the symbol. A __gendwarfksyms_ptr pointer must have the same type as the
+ * exported symbol, e.g.:
+ *
+ *   typeof(symname) *__gendwarf_ptr_symname = &symname;
+ */
+bool is_symbol_ptr(const char *name)
+{
+	return name && !strncmp(name, SYMBOL_PTR_PREFIX, SYMBOL_PTR_PREFIX_LEN);
+}
+
+static unsigned int for_each(const char *name, symbol_callback_t func,
+			     void *data)
+{
+	struct hlist_node *tmp;
+	struct symbol *match;
+
+	if (!name || !*name)
+		return 0;
+	if (is_symbol_ptr(name))
+		name += SYMBOL_PTR_PREFIX_LEN;
+
+	hash_for_each_possible_safe(symbol_names, match, tmp, name_hash,
+				    hash_str(name)) {
+		if (strcmp(match->name, name))
+			continue;
+
+		/* Call func for the match, and all address matches */
+		if (func)
+			func(match, data);
+
+		if (match->addr.section != SHN_UNDEF)
+			return __for_each_addr(match, func, data) + 1;
+
+		return 1;
+	}
+
+	return 0;
+}
+
+static void set_crc(struct symbol *sym, void *data)
+{
+	unsigned long *crc = data;
+
+	if (sym->state == SYMBOL_PROCESSED && sym->crc != *crc)
+		warn("overriding version for symbol %s (crc %lx vs. %lx)",
+		     sym->name, sym->crc, *crc);
+
+	sym->state = SYMBOL_PROCESSED;
+	sym->crc = *crc;
+}
+
+void symbol_set_crc(struct symbol *sym, unsigned long crc)
+{
+	if (for_each(sym->name, set_crc, &crc) == 0)
+		error("no matching symbols: '%s'", sym->name);
+}
+
+static void set_ptr(struct symbol *sym, void *data)
+{
+	sym->ptr_die_addr = (uintptr_t)((Dwarf_Die *)data)->addr;
+}
+
+void symbol_set_ptr(struct symbol *sym, Dwarf_Die *ptr)
+{
+	if (for_each(sym->name, set_ptr, ptr) == 0)
+		error("no matching symbols: '%s'", sym->name);
+}
+
+static void set_die(struct symbol *sym, void *data)
+{
+	sym->die_addr = (uintptr_t)((Dwarf_Die *)data)->addr;
+	sym->state = SYMBOL_MAPPED;
+}
+
+void symbol_set_die(struct symbol *sym, Dwarf_Die *die)
+{
+	if (for_each(sym->name, set_die, die) == 0)
+		error("no matching symbols: '%s'", sym->name);
+}
+
+static bool is_exported(const char *name)
+{
+	return for_each(name, NULL, NULL) > 0;
+}
+
+void symbol_read_exports(FILE *file)
+{
+	struct symbol *sym;
+	char *line = NULL;
+	char *name = NULL;
+	size_t size = 0;
+	int nsym = 0;
+
+	while (getline(&line, &size, file) > 0) {
+		if (sscanf(line, "%ms\n", &name) != 1)
+			error("malformed input line: %s", line);
+
+		if (is_exported(name)) {
+			/* Ignore duplicates */
+			free(name);
+			continue;
+		}
+
+		sym = xcalloc(1, sizeof(struct symbol));
+		sym->name = name;
+		sym->addr.section = SHN_UNDEF;
+		sym->state = SYMBOL_UNPROCESSED;
+
+		hash_add(symbol_names, &sym->name_hash, hash_str(sym->name));
+		++nsym;
+
+		debug("%s", sym->name);
+	}
+
+	free(line);
+	debug("%d exported symbols", nsym);
+}
+
+static void get_symbol(struct symbol *sym, void *arg)
+{
+	struct symbol **res = arg;
+
+	if (sym->state == SYMBOL_UNPROCESSED)
+		*res = sym;
+}
+
+struct symbol *symbol_get(const char *name)
+{
+	struct symbol *sym = NULL;
+
+	for_each(name, get_symbol, &sym);
+	return sym;
+}
+
+void symbol_for_each(symbol_callback_t func, void *arg)
+{
+	struct hlist_node *tmp;
+	struct symbol *sym;
+
+	hash_for_each_safe(symbol_names, sym, tmp, name_hash) {
+		func(sym, arg);
+	}
+}
+
+typedef void (*elf_symbol_callback_t)(const char *name, GElf_Sym *sym,
+				      Elf32_Word xndx, void *arg);
+
+static void elf_for_each_global(int fd, elf_symbol_callback_t func, void *arg)
+{
+	size_t sym_size;
+	GElf_Shdr shdr_mem;
+	GElf_Shdr *shdr;
+	Elf_Data *xndx_data = NULL;
+	Elf_Scn *scn;
+	Elf *elf;
+
+	if (elf_version(EV_CURRENT) != EV_CURRENT)
+		error("elf_version failed: %s", elf_errmsg(-1));
+
+	elf = elf_begin(fd, ELF_C_READ_MMAP, NULL);
+	if (!elf)
+		error("elf_begin failed: %s", elf_errmsg(-1));
+
+	scn = elf_nextscn(elf, NULL);
+
+	while (scn) {
+		shdr = gelf_getshdr(scn, &shdr_mem);
+		if (!shdr)
+			error("gelf_getshdr failed: %s", elf_errmsg(-1));
+
+		if (shdr->sh_type == SHT_SYMTAB_SHNDX) {
+			xndx_data = elf_getdata(scn, NULL);
+			if (!xndx_data)
+				error("elf_getdata failed: %s", elf_errmsg(-1));
+			break;
+		}
+
+		scn = elf_nextscn(elf, scn);
+	}
+
+	sym_size = gelf_fsize(elf, ELF_T_SYM, 1, EV_CURRENT);
+	scn = elf_nextscn(elf, NULL);
+
+	while (scn) {
+		shdr = gelf_getshdr(scn, &shdr_mem);
+		if (!shdr)
+			error("gelf_getshdr failed: %s", elf_errmsg(-1));
+
+		if (shdr->sh_type == SHT_SYMTAB) {
+			unsigned int nsyms;
+			unsigned int n;
+			Elf_Data *data = elf_getdata(scn, NULL);
+
+			if (!data)
+				error("elf_getdata failed: %s", elf_errmsg(-1));
+
+			if (shdr->sh_entsize != sym_size)
+				error("expected sh_entsize (%lu) to be %zu",
+				      shdr->sh_entsize, sym_size);
+
+			nsyms = shdr->sh_size / shdr->sh_entsize;
+
+			for (n = 1; n < nsyms; ++n) {
+				const char *name = NULL;
+				Elf32_Word xndx = 0;
+				GElf_Sym sym_mem;
+				GElf_Sym *sym;
+
+				sym = gelf_getsymshndx(data, xndx_data, n,
+						       &sym_mem, &xndx);
+				if (!sym)
+					error("gelf_getsymshndx failed: %s",
+					      elf_errmsg(-1));
+
+				if (GELF_ST_BIND(sym->st_info) == STB_LOCAL)
+					continue;
+
+				if (sym->st_shndx != SHN_XINDEX)
+					xndx = sym->st_shndx;
+
+				name = elf_strptr(elf, shdr->sh_link,
+						  sym->st_name);
+				if (!name)
+					error("elf_strptr failed: %s",
+					      elf_errmsg(-1));
+
+				/* Skip empty symbol names */
+				if (*name)
+					func(name, sym, xndx, arg);
+			}
+		}
+
+		scn = elf_nextscn(elf, scn);
+	}
+
+	check(elf_end(elf));
+}
+
+static void set_symbol_addr(struct symbol *sym, void *arg)
+{
+	struct symbol_addr *addr = arg;
+
+	if (sym->addr.section == SHN_UNDEF) {
+		sym->addr = *addr;
+		hash_add(symbol_addrs, &sym->addr_hash,
+			 symbol_addr_hash(&sym->addr));
+
+		debug("%s -> { %u, %lx }", sym->name, sym->addr.section,
+		      sym->addr.address);
+	} else if (sym->addr.section != addr->section ||
+		   sym->addr.address != addr->address) {
+		warn("multiple addresses for symbol %s?", sym->name);
+	}
+}
+
+static void elf_set_symbol_addr(const char *name, GElf_Sym *sym,
+				Elf32_Word xndx, void *arg)
+{
+	struct symbol_addr addr = { .section = xndx, .address = sym->st_value };
+
+	/* Set addresses for exported symbols */
+	if (addr.section != SHN_UNDEF)
+		for_each(name, set_symbol_addr, &addr);
+}
+
+void symbol_read_symtab(int fd)
+{
+	elf_for_each_global(fd, elf_set_symbol_addr, NULL);
+}
+
+void symbol_print_versions(void)
+{
+	struct hlist_node *tmp;
+	struct symbol *sym;
+
+	hash_for_each_safe(symbol_names, sym, tmp, name_hash) {
+		if (sym->state != SYMBOL_PROCESSED)
+			warn("no information for symbol %s", sym->name);
+
+		printf("#SYMVER %s 0x%08lx\n", sym->name, sym->crc);
+	}
+}
+
+void symbol_free(void)
+{
+	struct hlist_node *tmp;
+	struct symbol *sym;
+
+	hash_for_each_safe(symbol_names, sym, tmp, name_hash) {
+		free((void *)sym->name);
+		free(sym);
+	}
+
+	hash_init(symbol_addrs);
+	hash_init(symbol_names);
+}

scripts/gendwarfksyms/types.c

@@ -0,0 +1,481 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2024 Google LLC
+ */
+
+#define _GNU_SOURCE
+#include <inttypes.h>
+#include <stdio.h>
+#include <zlib.h>
+
+#include "gendwarfksyms.h"
+
+static struct cache expansion_cache;
+
+/*
+ * A simple linked list of shared or owned strings to avoid copying strings
+ * around when not necessary.
+ */
+struct type_list_entry {
+	const char *str;
+	void *owned;
+	struct list_head list;
+};
+
+static void type_list_free(struct list_head *list)
+{
+	struct type_list_entry *entry;
+	struct type_list_entry *tmp;
+
+	list_for_each_entry_safe(entry, tmp, list, list) {
+		if (entry->owned)
+			free(entry->owned);
+		free(entry);
+	}
+
+	INIT_LIST_HEAD(list);
+}
+
+static int type_list_append(struct list_head *list, const char *s, void *owned)
+{
+	struct type_list_entry *entry;
+
+	if (!s)
+		return 0;
+
+	entry = xmalloc(sizeof(struct type_list_entry));
+	entry->str = s;
+	entry->owned = owned;
+	list_add_tail(&entry->list, list);
+
+	return strlen(entry->str);
+}
+
+static void type_list_write(struct list_head *list, FILE *file)
+{
+	struct type_list_entry *entry;
+
+	list_for_each_entry(entry, list, list) {
+		if (entry->str)
+			checkp(fputs(entry->str, file));
+	}
+}
+
+/*
+ * An expanded type string in symtypes format.
+ */
+struct type_expansion {
+	char *name;
+	size_t len;
+	struct list_head expanded;
+	struct hlist_node hash;
+};
+
+static void type_expansion_init(struct type_expansion *type)
+{
+	type->name = NULL;
+	type->len = 0;
+	INIT_LIST_HEAD(&type->expanded);
+}
+
+static inline void type_expansion_free(struct type_expansion *type)
+{
+	free(type->name);
+	type->name = NULL;
+	type->len = 0;
+	type_list_free(&type->expanded);
+}
+
+static void type_expansion_append(struct type_expansion *type, const char *s,
+				  void *owned)
+{
+	type->len += type_list_append(&type->expanded, s, owned);
+}
+
+/*
+ * type_map -- the longest expansions for each type.
+ *
+ * const char *name -> struct type_expansion *
+ */
+#define TYPE_HASH_BITS 12
+static HASHTABLE_DEFINE(type_map, 1 << TYPE_HASH_BITS);
+
+static int type_map_get(const char *name, struct type_expansion **res)
+{
+	struct type_expansion *e;
+
+	hash_for_each_possible(type_map, e, hash, hash_str(name)) {
+		if (!strcmp(name, e->name)) {
+			*res = e;
+			return 0;
+		}
+	}
+
+	return -1;
+}
+
+static void type_map_add(const char *name, struct type_expansion *type)
+{
+	struct type_expansion *e;
+
+	if (type_map_get(name, &e)) {
+		e = xmalloc(sizeof(struct type_expansion));
+		type_expansion_init(e);
+		e->name = xstrdup(name);
+
+		hash_add(type_map, &e->hash, hash_str(e->name));
+
+		if (dump_types)
+			debug("adding %s", e->name);
+	} else {
+		/* Use the longest available expansion */
+		if (type->len <= e->len)
+			return;
+
+		type_list_free(&e->expanded);
+
+		if (dump_types)
+			debug("replacing %s", e->name);
+	}
+
+	/* Take ownership of type->expanded */
+	list_replace_init(&type->expanded, &e->expanded);
+	e->len = type->len;
+
+	if (dump_types) {
+		checkp(fputs(e->name, stderr));
+		checkp(fputs(" ", stderr));
+		type_list_write(&e->expanded, stderr);
+		checkp(fputs("\n", stderr));
+	}
+}
+
+static void type_map_write(FILE *file)
+{
+	struct type_expansion *e;
+	struct hlist_node *tmp;
+
+	if (!file)
+		return;
+
+	hash_for_each_safe(type_map, e, tmp, hash) {
+		checkp(fputs(e->name, file));
+		checkp(fputs(" ", file));
+		type_list_write(&e->expanded, file);
+		checkp(fputs("\n", file));
+	}
+}
+
+static void type_map_free(void)
+{
+	struct type_expansion *e;
+	struct hlist_node *tmp;
+
+	hash_for_each_safe(type_map, e, tmp, hash) {
+		type_expansion_free(e);
+		free(e);
+	}
+
+	hash_init(type_map);
+}
+
+/*
+ * CRC for a type, with an optional fully expanded type string for
+ * debugging.
+ */
+struct version {
+	struct type_expansion type;
+	unsigned long crc;
+};
+
+static void version_init(struct version *version)
+{
+	version->crc = crc32(0, NULL, 0);
+	type_expansion_init(&version->type);
+}
+
+static void version_free(struct version *version)
+{
+	type_expansion_free(&version->type);
+}
+
+static void version_add(struct version *version, const char *s)
+{
+	version->crc = crc32(version->crc, (void *)s, strlen(s));
+	if (dump_versions)
+		type_expansion_append(&version->type, s, NULL);
+}
+
+/*
+ * Type reference format: <prefix>#<name>, where prefix:
+ * 	s -> structure
+ * 	u -> union
+ * 	e -> enum
+ * 	t -> typedef
+ *
+ * Names with spaces are additionally wrapped in single quotes.
+ */
+static inline bool is_type_prefix(const char *s)
+{
+	return (s[0] == 's' || s[0] == 'u' || s[0] == 'e' || s[0] == 't') &&
+	       s[1] == '#';
+}
+
+static char get_type_prefix(int tag)
+{
+	switch (tag) {
+	case DW_TAG_class_type:
+	case DW_TAG_structure_type:
+		return 's';
+	case DW_TAG_union_type:
+		return 'u';
+	case DW_TAG_enumeration_type:
+		return 'e';
+	case DW_TAG_typedef_type:
+		return 't';
+	default:
+		return 0;
+	}
+}
+
+static char *get_type_name(struct die *cache)
+{
+	const char *quote;
+	char prefix;
+	char *name;
+
+	if (cache->state == DIE_INCOMPLETE) {
+		warn("found incomplete cache entry: %p", cache);
+		return NULL;
+	}
+	if (cache->state == DIE_SYMBOL)
+		return NULL;
+	if (!cache->fqn || !*cache->fqn)
+		return NULL;
+
+	prefix = get_type_prefix(cache->tag);
+	if (!prefix)
+		return NULL;
+
+	/* Wrap names with spaces in single quotes */
+	quote = strstr(cache->fqn, " ") ? "'" : "";
+
+	/* <prefix>#<type_name>\0 */
+	if (asprintf(&name, "%c#%s%s%s", prefix, quote, cache->fqn, quote) < 0)
+		error("asprintf failed for '%s'", cache->fqn);
+
+	return name;
+}
+
+static void __calculate_version(struct version *version, struct list_head *list)
+{
+	struct type_list_entry *entry;
+	struct type_expansion *e;
+
+	/* Calculate a CRC over an expanded type string */
+	list_for_each_entry(entry, list, list) {
+		if (is_type_prefix(entry->str)) {
+			check(type_map_get(entry->str, &e));
+
+			/*
+			 * It's sufficient to expand each type reference just
+			 * once to detect changes.
+			 */
+			if (cache_was_expanded(&expansion_cache, e)) {
+				version_add(version, entry->str);
+			} else {
+				cache_mark_expanded(&expansion_cache, e);
+				__calculate_version(version, &e->expanded);
+			}
+		} else {
+			version_add(version, entry->str);
+		}
+	}
+}
+
+static void calculate_version(struct version *version, struct list_head *list)
+{
+	version_init(version);
+	__calculate_version(version, list);
+	cache_free(&expansion_cache);
+}
+
+static void __type_expand(struct die *cache, struct type_expansion *type,
+			  bool recursive);
+
+static void type_expand_child(struct die *cache, struct type_expansion *type,
+			      bool recursive)
+{
+	struct type_expansion child;
+	char *name;
+
+	name = get_type_name(cache);
+	if (!name) {
+		__type_expand(cache, type, recursive);
+		return;
+	}
+
+	if (recursive && !__cache_was_expanded(&expansion_cache, cache->addr)) {
+		__cache_mark_expanded(&expansion_cache, cache->addr);
+		type_expansion_init(&child);
+		__type_expand(cache, &child, true);
+		type_map_add(name, &child);
+		type_expansion_free(&child);
+	}
+
+	type_expansion_append(type, name, name);
+}
+
+static void __type_expand(struct die *cache, struct type_expansion *type,
+			  bool recursive)
+{
+	struct die_fragment *df;
+	struct die *child;
+
+	list_for_each_entry(df, &cache->fragments, list) {
+		switch (df->type) {
+		case FRAGMENT_STRING:
+			type_expansion_append(type, df->data.str, NULL);
+			break;
+		case FRAGMENT_DIE:
+			/* Use a complete die_map expansion if available */
+			if (__die_map_get(df->data.addr, DIE_COMPLETE,
+					  &child) &&
+			    __die_map_get(df->data.addr, DIE_UNEXPANDED,
+					  &child))
+				error("unknown child: %" PRIxPTR,
+				      df->data.addr);
+
+			type_expand_child(child, type, recursive);
+			break;
+		case FRAGMENT_LINEBREAK:
+			/*
+			 * Keep whitespace in the symtypes format, but avoid
+			 * repeated spaces.
+			 */
+			if (list_is_last(&df->list, &cache->fragments) ||
+			    list_next_entry(df, list)->type !=
+				    FRAGMENT_LINEBREAK)
+				type_expansion_append(type, " ", NULL);
+			break;
+		default:
+			error("empty die_fragment in %p", cache);
+		}
+	}
+}
+
+static void type_expand(struct die *cache, struct type_expansion *type,
+			bool recursive)
+{
+	type_expansion_init(type);
+	__type_expand(cache, type, recursive);
+	cache_free(&expansion_cache);
+}
+
+static void expand_type(struct die *cache, void *arg)
+{
+	struct type_expansion type;
+	char *name;
+
+	if (cache->mapped)
+		return;
+
+	cache->mapped = true;
+
+	/*
+	 * Skip unexpanded die_map entries if there's a complete
+	 * expansion available for this DIE.
+	 */
+	if (cache->state == DIE_UNEXPANDED &&
+	    !__die_map_get(cache->addr, DIE_COMPLETE, &cache)) {
+		if (cache->mapped)
+			return;
+
+		cache->mapped = true;
+	}
+
+	name = get_type_name(cache);
+	if (!name)
+		return;
+
+	debug("%s", name);
+	type_expand(cache, &type, true);
+	type_map_add(name, &type);
+
+	type_expansion_free(&type);
+	free(name);
+}
+
+static void expand_symbol(struct symbol *sym, void *arg)
+{
+	struct type_expansion type;
+	struct version version;
+	struct die *cache;
+
+	/*
+	 * No need to expand again unless we want a symtypes file entry
+	 * for the symbol. Note that this means `sym` has the same address
+	 * as another symbol that was already processed.
+	 */
+	if (!symtypes && sym->state == SYMBOL_PROCESSED)
+		return;
+
+	if (__die_map_get(sym->die_addr, DIE_SYMBOL, &cache))
+		return; /* We'll warn about missing CRCs later. */
+
+	type_expand(cache, &type, false);
+
+	/* If the symbol already has a version, don't calculate it again. */
+	if (sym->state != SYMBOL_PROCESSED) {
+		calculate_version(&version, &type.expanded);
+		symbol_set_crc(sym, version.crc);
+		debug("%s = %lx", sym->name, version.crc);
+
+		if (dump_versions) {
+			checkp(fputs(sym->name, stderr));
+			checkp(fputs(" ", stderr));
+			type_list_write(&version.type.expanded, stderr);
+			checkp(fputs("\n", stderr));
+		}
+
+		version_free(&version);
+	}
+
+	/* These aren't needed in type_map unless we want a symtypes file. */
+	if (symtypes)
+		type_map_add(sym->name, &type);
+
+	type_expansion_free(&type);
+}
+
+void generate_symtypes_and_versions(FILE *file)
+{
+	cache_init(&expansion_cache);
+
+	/*
+	 * die_map processing:
+	 *
+	 *   1. die_map contains all types referenced in exported symbol
+	 *      signatures, but can contain duplicates just like the original
+	 *      DWARF, and some references may not be fully expanded depending
+	 *      on how far we processed the DIE tree for that specific symbol.
+	 *
+	 *      For each die_map entry, find the longest available expansion,
+	 *      and add it to type_map.
+	 */
+	die_map_for_each(expand_type, NULL);
+
+	/*
+	 *   2. For each exported symbol, expand the die_map type, and use
+	 *      type_map expansions to calculate a symbol version from the
+	 *      fully expanded type string.
+	 */
+	symbol_for_each(expand_symbol, NULL);
+
+	/*
+	 *   3. If a symtypes file is requested, write type_map contents to
+	 *      the file.
+	 */
+	type_map_write(file);
+	type_map_free();
+}

scripts/genksyms/genksyms.c

@@ -12,18 +12,19 @@
 
 #include <stdio.h>
 #include <string.h>
+#include <stdint.h>
 #include <stdlib.h>
 #include <unistd.h>
 #include <assert.h>
 #include <stdarg.h>
 #include <getopt.h>
 
+#include <hashtable.h>
+
 #include "genksyms.h"
 /*----------------------------------------------------------------------*/
 
-#define HASH_BUCKETS  4096
-
-static struct symbol *symtab[HASH_BUCKETS];
+static HASHTABLE_DEFINE(symbol_hashtable, 1U << 12);
 static FILE *debugfile;
 
 int cur_line = 1;
@@ -60,7 +61,7 @@ static void print_type_name(enum symbol_type type, const char *name);
 
 /*----------------------------------------------------------------------*/
 
-static const unsigned int crctab32[] = {
+static const uint32_t crctab32[] = {
 	0x00000000U, 0x77073096U, 0xee0e612cU, 0x990951baU, 0x076dc419U,
 	0x706af48fU, 0xe963a535U, 0x9e6495a3U, 0x0edb8832U, 0x79dcb8a4U,
 	0xe0d5e91eU, 0x97d2d988U, 0x09b64c2bU, 0x7eb17cbdU, 0xe7b82d07U,
@@ -115,19 +116,19 @@ static const unsigned int crctab32[] = {
 	0x2d02ef8dU
 };
 
-static unsigned long partial_crc32_one(unsigned char c, unsigned long crc)
+static uint32_t partial_crc32_one(uint8_t c, uint32_t crc)
 {
 	return crctab32[(crc ^ c) & 0xff] ^ (crc >> 8);
 }
 
-static unsigned long partial_crc32(const char *s, unsigned long crc)
+static uint32_t partial_crc32(const char *s, uint32_t crc)
 {
 	while (*s)
 		crc = partial_crc32_one(*s++, crc);
 	return crc;
 }
 
-static unsigned long crc32(const char *s)
+static uint32_t crc32(const char *s)
 {
 	return partial_crc32(s, 0xffffffff) ^ 0xffffffff;
 }
@@ -151,14 +152,14 @@ static enum symbol_type map_to_ns(enum symbol_type t)
 
 struct symbol *find_symbol(const char *name, enum symbol_type ns, int exact)
 {
-	unsigned long h = crc32(name) % HASH_BUCKETS;
 	struct symbol *sym;
 
-	for (sym = symtab[h]; sym; sym = sym->hash_next)
+	hash_for_each_possible(symbol_hashtable, sym, hnode, crc32(name)) {
 		if (map_to_ns(sym->type) == map_to_ns(ns) &&
 		    strcmp(name, sym->name) == 0 &&
 		    sym->is_declared)
 			break;
+	}
 
 	if (exact && sym && sym->type != ns)
 		return NULL;
@@ -224,64 +225,56 @@ static struct symbol *__add_symbol(const char *name, enum symbol_type type,
 			return NULL;
 	}
 
-	h = crc32(name) % HASH_BUCKETS;
-	for (sym = symtab[h]; sym; sym = sym->hash_next) {
-		if (map_to_ns(sym->type) == map_to_ns(type) &&
-		    strcmp(name, sym->name) == 0) {
-			if (is_reference)
-				/* fall through */ ;
-			else if (sym->type == type &&
-				 equal_list(sym->defn, defn)) {
-				if (!sym->is_declared && sym->is_override) {
-					print_location();
-					print_type_name(type, name);
-					fprintf(stderr, " modversion is "
-						"unchanged\n");
-				}
-				sym->is_declared = 1;
-				return sym;
-			} else if (!sym->is_declared) {
-				if (sym->is_override && flag_preserve) {
-					print_location();
-					fprintf(stderr, "ignoring ");
-					print_type_name(type, name);
-					fprintf(stderr, " modversion change\n");
-					sym->is_declared = 1;
-					return sym;
-				} else {
-					status = is_unknown_symbol(sym) ?
-						STATUS_DEFINED : STATUS_MODIFIED;
-				}
-			} else {
-				error_with_pos("redefinition of %s", name);
-				return sym;
+	h = crc32(name);
+	hash_for_each_possible(symbol_hashtable, sym, hnode, h) {
+		if (map_to_ns(sym->type) != map_to_ns(type) ||
+		    strcmp(name, sym->name))
+			continue;
+
+		if (is_reference) {
+			break;
+		} else if (sym->type == type && equal_list(sym->defn, defn)) {
+			if (!sym->is_declared && sym->is_override) {
+				print_location();
+				print_type_name(type, name);
+				fprintf(stderr, " modversion is unchanged\n");
 			}
+			sym->is_declared = 1;
+		} else if (sym->is_declared) {
+			error_with_pos("redefinition of %s", name);
+		} else if (sym->is_override && flag_preserve) {
+			print_location();
+			fprintf(stderr, "ignoring ");
+			print_type_name(type, name);
+			fprintf(stderr, " modversion change\n");
+			sym->is_declared = 1;
+		} else {
+			status = is_unknown_symbol(sym) ?
+					STATUS_DEFINED : STATUS_MODIFIED;
 			break;
 		}
+		free_list(defn, NULL);
+		return sym;
 	}
 
 	if (sym) {
-		struct symbol **psym;
+		hash_del(&sym->hnode);
 
-		for (psym = &symtab[h]; *psym; psym = &(*psym)->hash_next) {
-			if (*psym == sym) {
-				*psym = sym->hash_next;
-				break;
-			}
-		}
+		free_list(sym->defn, NULL);
+		free(sym->name);
+		free(sym);
 		--nsyms;
 	}
 
 	sym = xmalloc(sizeof(*sym));
-	sym->name = name;
+	sym->name = xstrdup(name);
 	sym->type = type;
 	sym->defn = defn;
 	sym->expansion_trail = NULL;
 	sym->visited = NULL;
 	sym->is_extern = is_extern;
 
-	sym->hash_next = symtab[h];
-	symtab[h] = sym;
+	hash_add(symbol_hashtable, &sym->hnode, h);
 
 	sym->is_declared = !is_reference;
 	sym->status = status;
@@ -480,7 +473,7 @@ static void read_reference(FILE *f)
 			defn = def;
 			def = read_node(f);
 		}
-		subsym = add_reference_symbol(xstrdup(sym->string), sym->tag,
+		subsym = add_reference_symbol(sym->string, sym->tag,
 					      defn, is_extern);
 		subsym->is_override = is_override;
 		free_node(sym);
@@ -525,7 +518,7 @@ static void print_list(FILE * f, struct string_list *list)
 	}
 }
 
-static unsigned long expand_and_crc_sym(struct symbol *sym, unsigned long crc)
+static uint32_t expand_and_crc_sym(struct symbol *sym, uint32_t crc)
 {
 	struct string_list *list = sym->defn;
 	struct string_list **e, **b;
@@ -632,7 +625,7 @@ static unsigned long expand_and_crc_sym(struct symbol *sym, unsigned long crc)
 void export_symbol(const char *name)
 {
 	struct symbol *sym;
-	unsigned long crc;
+	uint32_t crc;
 	int has_changed = 0;
 
 	sym = find_symbol(name, SYM_NORMAL, 0);
@@ -680,7 +673,7 @@ void export_symbol(const char *name)
 	if (flag_dump_defs)
 		fputs(">\n", debugfile);
 
-	printf("#SYMVER %s 0x%08lx\n", name, crc);
+	printf("#SYMVER %s 0x%08lx\n", name, (unsigned long)crc);
 }
 
 /*----------------------------------------------------------------------*/
@@ -832,9 +825,9 @@ int main(int argc, char **argv)
 	}
 
 	if (flag_debug) {
-		fprintf(debugfile, "Hash table occupancy %d/%d = %g\n",
-			nsyms, HASH_BUCKETS,
-			(double)nsyms / (double)HASH_BUCKETS);
+		fprintf(debugfile, "Hash table occupancy %d/%zd = %g\n",
+			nsyms, HASH_SIZE(symbol_hashtable),
+			(double)nsyms / HASH_SIZE(symbol_hashtable));
 	}
 
 	if (dumpfile)

scripts/genksyms/genksyms.h

@@ -14,6 +14,8 @@
 
 #include <stdio.h>
 
+#include <list_types.h>
+
 enum symbol_type {
 	SYM_NORMAL, SYM_TYPEDEF, SYM_ENUM, SYM_STRUCT, SYM_UNION,
 	SYM_ENUM_CONST
@@ -31,8 +33,8 @@ struct string_list {
 };
 
 struct symbol {
-	struct symbol *hash_next;
-	const char *name;
+	struct hlist_node hnode;
+	char *name;
 	enum symbol_type type;
 	struct string_list *defn;
 	struct symbol *expansion_trail;

scripts/genksyms/parse.y

@@ -152,14 +152,19 @@ simple_declaration:
 	;
 
 init_declarator_list_opt:
-	/* empty */				{ $$ = NULL; }
-	| init_declarator_list
+	/* empty */			{ $$ = NULL; }
+	| init_declarator_list		{ free_list(decl_spec, NULL); $$ = $1; }
 	;
 
 init_declarator_list:
 	init_declarator
 		{ struct string_list *decl = *$1;
 		  *$1 = NULL;
+
+		  /* avoid sharing among multiple init_declarators */
+		  if (decl_spec)
+		    decl_spec = copy_list_range(decl_spec, NULL);
+
 		  add_symbol(current_name,
 			     is_typedef ? SYM_TYPEDEF : SYM_NORMAL, decl, is_extern);
 		  current_name = NULL;
@@ -170,6 +175,11 @@ init_declarator_list:
 		  *$3 = NULL;
 		  free_list(*$2, NULL);
 		  *$2 = decl_spec;
+
+		  /* avoid sharing among multiple init_declarators */
+		  if (decl_spec)
+		    decl_spec = copy_list_range(decl_spec, NULL);
+
 		  add_symbol(current_name,
 			     is_typedef ? SYM_TYPEDEF : SYM_NORMAL, decl, is_extern);
 		  current_name = NULL;
@@ -472,12 +482,12 @@ enumerator_list:
 enumerator:
 	IDENT
 		{
-			const char *name = strdup((*$1)->string);
+			const char *name = (*$1)->string;
 			add_symbol(name, SYM_ENUM_CONST, NULL, 0);
 		}
 	| IDENT '=' EXPRESSION_PHRASE
 		{
-			const char *name = strdup((*$1)->string);
+			const char *name = (*$1)->string;
 			struct string_list *expr = copy_list_range(*$3, *$2);
 			add_symbol(name, SYM_ENUM_CONST, expr, 0);
 		}

scripts/kconfig/Makefile

@@ -105,9 +105,11 @@ configfiles = $(wildcard $(srctree)/kernel/configs/$(1) $(srctree)/arch/$(SRCARC
 all-config-fragments = $(call configfiles,*.config)
 config-fragments = $(call configfiles,$@)
 
+cmd_merge_fragments = $(srctree)/scripts/kconfig/merge_config.sh -m $(KCONFIG_CONFIG) $(config-fragments)
+
 %.config: $(obj)/conf
 	$(if $(config-fragments),, $(error $@ fragment does not exists on this architecture))
-	$(Q)$(CONFIG_SHELL) $(srctree)/scripts/kconfig/merge_config.sh -m $(KCONFIG_CONFIG) $(config-fragments)
+	$(call cmd,merge_fragments)
 	$(Q)$(MAKE) -f $(srctree)/Makefile olddefconfig
 
 PHONY += tinyconfig

scripts/kconfig/qconf.cc

@@ -1464,8 +1464,8 @@ void ConfigMainWindow::loadConfig(void)
 {
 	QString str;
 
-	str = QFileDialog::getOpenFileName(this, "", configname);
-	if (str.isNull())
+	str = QFileDialog::getOpenFileName(this, QString(), configname);
+	if (str.isEmpty())
 		return;
 
 	if (conf_read(str.toLocal8Bit().constData()))
@@ -1491,8 +1491,8 @@ void ConfigMainWindow::saveConfigAs(void)
 {
 	QString str;
 
-	str = QFileDialog::getSaveFileName(this, "", configname);
-	if (str.isNull())
+	str = QFileDialog::getSaveFileName(this, QString(), configname);
+	if (str.isEmpty())
 		return;
 
 	if (conf_write(str.toLocal8Bit().constData())) {

scripts/mod/modpost.c

@@ -33,6 +33,10 @@ static bool module_enabled;
 static bool modversions;
 /* Is CONFIG_MODULE_SRCVERSION_ALL set? */
 static bool all_versions;
+/* Is CONFIG_BASIC_MODVERSIONS set? */
+static bool basic_modversions;
+/* Is CONFIG_EXTENDED_MODVERSIONS set? */
+static bool extended_modversions;
 /* If we are modposting external module set to 1 */
 static bool external_module;
 /* Only warn about unresolved symbols */
@@ -1805,6 +1809,49 @@ static void add_exported_symbols(struct buffer *buf, struct module *mod)
 	}
 }
 
+/**
+ * Record CRCs for unresolved symbols, supporting long names
+ */
+static void add_extended_versions(struct buffer *b, struct module *mod)
+{
+	struct symbol *s;
+
+	if (!extended_modversions)
+		return;
+
+	buf_printf(b, "\n");
+	buf_printf(b, "static const u32 ____version_ext_crcs[]\n");
+	buf_printf(b, "__used __section(\"__version_ext_crcs\") = {\n");
+	list_for_each_entry(s, &mod->unresolved_symbols, list) {
+		if (!s->module)
+			continue;
+		if (!s->crc_valid) {
+			warn("\"%s\" [%s.ko] has no CRC!\n",
+				s->name, mod->name);
+			continue;
+		}
+		buf_printf(b, "\t0x%08x,\n", s->crc);
+	}
+	buf_printf(b, "};\n");
+
+	buf_printf(b, "static const char ____version_ext_names[]\n");
+	buf_printf(b, "__used __section(\"__version_ext_names\") =\n");
+	list_for_each_entry(s, &mod->unresolved_symbols, list) {
+		if (!s->module)
+			continue;
+		if (!s->crc_valid)
+			/*
+			 * We already warned on this when producing the crc
+			 * table.
+			 * We need to skip its name too, as the indexes in
+			 * both tables need to align.
+			 */
+			continue;
+		buf_printf(b, "\t\"%s\\0\"\n", s->name);
+	}
+	buf_printf(b, ";\n");
+}
+
 /**
  * Record CRCs for unresolved symbols
  **/
@@ -1812,7 +1859,7 @@ static void add_versions(struct buffer *b, struct module *mod)
 {
 	struct symbol *s;
 
-	if (!modversions)
+	if (!basic_modversions)
 		return;
 
 	buf_printf(b, "\n");
@@ -1828,11 +1875,16 @@ static void add_versions(struct buffer *b, struct module *mod)
 			continue;
 		}
 		if (strlen(s->name) >= MODULE_NAME_LEN) {
-			error("too long symbol \"%s\" [%s.ko]\n",
-			      s->name, mod->name);
-			break;
+			if (extended_modversions) {
+				/* this symbol will only be in the extended info */
+				continue;
+			} else {
+				error("too long symbol \"%s\" [%s.ko]\n",
+				      s->name, mod->name);
+				break;
+			}
 		}
-		buf_printf(b, "\t{ %#8x, \"%s\" },\n",
+		buf_printf(b, "\t{ 0x%08x, \"%s\" },\n",
 			   s->crc, s->name);
 	}
 
@@ -1961,6 +2013,7 @@ static void write_mod_c_file(struct module *mod)
 	add_header(&buf, mod);
 	add_exported_symbols(&buf, mod);
 	add_versions(&buf, mod);
+	add_extended_versions(&buf, mod);
 	add_depends(&buf, mod);
 
 	buf_printf(&buf, "\n");
@@ -2126,7 +2179,7 @@ int main(int argc, char **argv)
 	LIST_HEAD(dump_lists);
 	struct dump_list *dl, *dl2;
 
-	while ((opt = getopt(argc, argv, "ei:MmnT:to:au:WwENd:")) != -1) {
+	while ((opt = getopt(argc, argv, "ei:MmnT:to:au:WwENd:xb")) != -1) {
 		switch (opt) {
 		case 'e':
 			external_module = true;
@@ -2175,6 +2228,12 @@ int main(int argc, char **argv)
 		case 'd':
 			missing_namespace_deps = optarg;
 			break;
+		case 'b':
+			basic_modversions = true;
+			break;
+		case 'x':
+			extended_modversions = true;
+			break;
 		default:
 			exit(1);
 		}

scripts/package/PKGBUILD

@@ -22,7 +22,6 @@ license=(GPL-2.0-only)
 makedepends=(
 	bc
 	bison
-	cpio
 	flex
 	gettext
 	kmod

scripts/package/builddeb

@@ -5,10 +5,12 @@
 #
 # Simple script to generate a deb package for a Linux kernel. All the
 # complexity of what to do with a kernel after it is installed or removed
-# is left to other scripts and packages: they can install scripts in the
-# /etc/kernel/{pre,post}{inst,rm}.d/ directories (or an alternative location
-# specified in KDEB_HOOKDIR) that will be called on package install and
-# removal.
+# is left to other scripts and packages. Scripts can be placed into the
+# preinst, postinst, prerm and postrm directories in /etc/kernel or
+# /usr/share/kernel. A different list of search directories can be given
+# via KDEB_HOOKDIR. Scripts in directories earlier in the list will
+# override scripts of the same name in later directories.  The script will
+# be called on package installation and removal.
 
 set -eu
 
@@ -74,10 +76,8 @@ install_maint_scripts () {
 	# kernel packages, as well as kernel packages built using make-kpkg.
 	# make-kpkg sets $INITRD to indicate whether an initramfs is wanted, and
 	# so do we; recent versions of dracut and initramfs-tools will obey this.
-	debhookdir=${KDEB_HOOKDIR:-/etc/kernel}
+	debhookdir=${KDEB_HOOKDIR:-/etc/kernel /usr/share/kernel}
 	for script in postinst postrm preinst prerm; do
-		mkdir -p "${pdir}${debhookdir}/${script}.d"
-
 		mkdir -p "${pdir}/DEBIAN"
 		cat <<-EOF > "${pdir}/DEBIAN/${script}"
 		#!/bin/sh
@@ -90,7 +90,15 @@ install_maint_scripts () {
 		# Tell initramfs builder whether it's wanted
 		export INITRD=$(if_enabled_echo CONFIG_BLK_DEV_INITRD Yes No)
 
-		test -d ${debhookdir}/${script}.d && run-parts --arg="${KERNELRELEASE}" --arg="/${installed_image_path}" ${debhookdir}/${script}.d
+		# run-parts will error out if one of its directory arguments does not
+		# exist, so filter the list of hook directories accordingly.
+		hookdirs=
+		for dir in ${debhookdir}; do
+			test -d "\$dir/${script}.d" || continue
+			hookdirs="\$hookdirs \$dir/${script}.d"
+		done
+		hookdirs="\${hookdirs# }"
+		test -n "\$hookdirs" && run-parts --arg="${KERNELRELEASE}" --arg="/${installed_image_path}" \$hookdirs
 		exit 0
 		EOF
 		chmod 755 "${pdir}/DEBIAN/${script}"

scripts/package/install-extmod-build

@@ -49,17 +49,10 @@ mkdir -p "${destdir}"
 # This caters to host programs that participate in Kbuild. objtool and
 # resolve_btfids are out of scope.
 if [ "${CC}" != "${HOSTCC}" ]; then
-	echo "Rebuilding host programs with ${CC}..."
-
-	# This leverages external module building.
-	# - Clear sub_make_done to allow the top-level Makefile to redo sub-make.
-	# - Filter out --no-print-directory to print "Entering directory" logs
-	#   when Make changes the working directory.
-	unset sub_make_done
-	MAKEFLAGS=$(echo "${MAKEFLAGS}" | sed s/--no-print-directory//)
-
-	cat <<-'EOF' >  "${destdir}/Kbuild"
-	subdir-y := scripts
+	cat "${destdir}/scripts/Makefile" - <<-'EOF' > "${destdir}/scripts/Kbuild"
+	subdir-y += basic
+	hostprogs-always-y += mod/modpost
+	mod/modpost-objs := $(addprefix mod/, modpost.o file2alias.o sumversion.o symsearch.o)
 	EOF
 
 	# HOSTCXX is not overridden. The C++ compiler is used to build:
@@ -67,20 +60,12 @@ if [ "${CC}" != "${HOSTCC}" ]; then
 	# - GCC plugins, which will not work on the installed system even after
 	#   being rebuilt.
 	#
-	# Use the single-target build to avoid the modpost invocation, which
-	# would overwrite Module.symvers.
-	"${MAKE}" HOSTCC="${CC}" KBUILD_OUTPUT=. KBUILD_EXTMOD="${destdir}" scripts/
+	# Clear VPATH and srcroot because the source files reside in the output
+	# directory.
+	# shellcheck disable=SC2016 # $(MAKE), $(CC), and $(build) will be expanded by Make
+	"${MAKE}" run-command KBUILD_RUN_COMMAND='+$(MAKE) HOSTCC=$(CC) VPATH= srcroot=. $(build)='"${destdir}"/scripts
 
-	cat <<-'EOF' >  "${destdir}/scripts/Kbuild"
-	subdir-y := basic
-	hostprogs-always-y := mod/modpost
-	mod/modpost-objs := $(addprefix mod/, modpost.o file2alias.o sumversion.o symsearch.o)
-	EOF
-
-	# Run once again to rebuild scripts/basic/ and scripts/mod/modpost.
-	"${MAKE}" HOSTCC="${CC}" KBUILD_OUTPUT=. KBUILD_EXTMOD="${destdir}" scripts/
-
-	rm -f "${destdir}/Kbuild" "${destdir}/scripts/Kbuild"
+	rm -f "${destdir}/scripts/Kbuild"
 fi
 
 find "${destdir}" \( -name '.*.cmd' -o -name '*.o' \) -delete

scripts/package/mkdebian

@@ -205,7 +205,7 @@ Priority: optional
 Maintainer: $maintainer
 Rules-Requires-Root: no
 Build-Depends: debhelper-compat (= 12)
-Build-Depends-Arch: bc, bison, cpio, flex,
+Build-Depends-Arch: bc, bison, flex,
  gcc-${host_gnu} <!pkg.${sourcename}.nokernelheaders>,
  kmod, libelf-dev:native,
  libssl-dev:native, libssl-dev <!pkg.${sourcename}.nokernelheaders>,