[PATCH v2] modversions: treat symbol CRCs as 32 bit quantities on 64 bit archs

From: Ard Biesheuvel
Date: Thu Oct 20 2016 - 05:38:11 EST


The symbol CRCs are emitted as ELF symbols, which allows us to easily
populate the kcrctab sections by relying on the linker to associate
each kcrctab slot with the correct value.

This has two downsides:
- given that the CRCs are treated as pointers, we waste 4 bytes for
each CRC on 64 bit architectures,
- on architectures that support runtime relocation, a relocation entry is
emitted for each CRC value, which may take up 24 bytes of __init space
(on ELF64 systems)

This comes down to a x8 overhead in [uncompressed] kernel size. In addition,
each relocation has to be reverted before the CRC value can be used.

Switching to explicit 32 bit values on 64 bit architectures fixes both
issues, since 32 bit values are not treated as relocatable quantities on
ELF64 systems, even if the value ultimately resolves to a linker supplied
value.

So redefine all CRC fields and variables as u32, and redefine the
__CRC_SYMBOL() macro for 64 bit builds to emit the CRC reference using
inline assembler (which is necessary since 64-bit C code cannot use
32-bit types to hold memory addresses, even if they are ultimately
resolved using values that do no exceed 0xffffffff).

Also remove the special handling for PPC64, this should no longer be
required.

Signed-off-by: Ard Biesheuvel <ard.biesheuvel@xxxxxxxxxx>
---
v2: drop the change to struct modversion_info: it affects the layout of the
__versions section, which is consumed by userland tools as well, so it is
effectively ABI

On an arm64 defconfig build with CONFIG_RELOCATABLE=y, this patch reduces
the CRC footprint by 24 KB for .rodata, and by 217 KB for .init

Before:
[ 9] __kcrctab PROGBITS ffff000008b992a8 00b292a8
0000000000009440 0000000000000000 A 0 0 8
[10] __kcrctab_gpl PROGBITS ffff000008ba26e8 00b326e8
0000000000008d40 0000000000000000 A 0 0 8
...
[22] .rela RELA ffff000008c96e20 00c26e20
00000000001cc758 0000000000000018 A 0 0 8

After:
[ 9] __kcrctab PROGBITS ffff000008b728a8 00b028a8
0000000000004a20 0000000000000000 A 0 0 1
[10] __kcrctab_gpl PROGBITS ffff000008b772c8 00b072c8
00000000000046a0 0000000000000000 A 0 0 1
...
[22] .rela RELA ffff000008c66e20 00bf6e20
00000000001962d8 0000000000000018 A 0 0 8

arch/powerpc/include/asm/module.h | 4 --
arch/powerpc/kernel/module_64.c | 8 ----
include/linux/export.h | 8 ++++
include/linux/module.h | 14 +++----
kernel/module.c | 39 +++++++-------------
5 files changed, 29 insertions(+), 44 deletions(-)

diff --git a/arch/powerpc/include/asm/module.h b/arch/powerpc/include/asm/module.h
index cd4ffd86765f..94a7f7aa3ae8 100644
--- a/arch/powerpc/include/asm/module.h
+++ b/arch/powerpc/include/asm/module.h
@@ -94,9 +94,5 @@ struct exception_table_entry;
void sort_ex_table(struct exception_table_entry *start,
struct exception_table_entry *finish);

-#if defined(CONFIG_MODVERSIONS) && defined(CONFIG_PPC64)
-#define ARCH_RELOCATES_KCRCTAB
-#define reloc_start PHYSICAL_START
-#endif
#endif /* __KERNEL__ */
#endif /* _ASM_POWERPC_MODULE_H */
diff --git a/arch/powerpc/kernel/module_64.c b/arch/powerpc/kernel/module_64.c
index 183368e008cf..be9b2d5ff846 100644
--- a/arch/powerpc/kernel/module_64.c
+++ b/arch/powerpc/kernel/module_64.c
@@ -286,14 +286,6 @@ static void dedotify_versions(struct modversion_info *vers,
for (end = (void *)vers + size; vers < end; vers++)
if (vers->name[0] == '.') {
memmove(vers->name, vers->name+1, strlen(vers->name));
-#ifdef ARCH_RELOCATES_KCRCTAB
- /* The TOC symbol has no CRC computed. To avoid CRC
- * check failing, we must force it to the expected
- * value (see CRC check in module.c).
- */
- if (!strcmp(vers->name, "TOC."))
- vers->crc = -(unsigned long)reloc_start;
-#endif
}
}

diff --git a/include/linux/export.h b/include/linux/export.h
index 2a0f61fbc731..fa51ab2ad190 100644
--- a/include/linux/export.h
+++ b/include/linux/export.h
@@ -41,6 +41,7 @@ extern struct module __this_module;

#if defined(__KERNEL__) && !defined(__GENKSYMS__)
#ifdef CONFIG_MODVERSIONS
+#ifndef CONFIG_64BIT
/* Mark the CRC weak since genksyms apparently decides not to
* generate a checksums for some symbols */
#define __CRC_SYMBOL(sym, sec) \
@@ -50,6 +51,13 @@ extern struct module __this_module;
__attribute__((section("___kcrctab" sec "+" #sym), used)) \
= (unsigned long) &__crc_##sym;
#else
+#define __CRC_SYMBOL(sym, sec) \
+ asm(" .section \"___kcrctab" sec "+" #sym "\", \"a\" \n" \
+ " .weak " VMLINUX_SYMBOL_STR(__crc_##sym) " \n" \
+ " .word " VMLINUX_SYMBOL_STR(__crc_##sym) " \n" \
+ " .previous \n");
+#endif
+#else
#define __CRC_SYMBOL(sym, sec)
#endif

diff --git a/include/linux/module.h b/include/linux/module.h
index 0c3207d26ac0..e0067673f5e5 100644
--- a/include/linux/module.h
+++ b/include/linux/module.h
@@ -346,7 +346,7 @@ struct module {

/* Exported symbols */
const struct kernel_symbol *syms;
- const unsigned long *crcs;
+ const u32 *crcs;
unsigned int num_syms;

/* Kernel parameters. */
@@ -359,18 +359,18 @@ struct module {
/* GPL-only exported symbols. */
unsigned int num_gpl_syms;
const struct kernel_symbol *gpl_syms;
- const unsigned long *gpl_crcs;
+ const u32 *gpl_crcs;

#ifdef CONFIG_UNUSED_SYMBOLS
/* unused exported symbols. */
const struct kernel_symbol *unused_syms;
- const unsigned long *unused_crcs;
+ const u32 *unused_crcs;
unsigned int num_unused_syms;

/* GPL-only, unused exported symbols. */
unsigned int num_unused_gpl_syms;
const struct kernel_symbol *unused_gpl_syms;
- const unsigned long *unused_gpl_crcs;
+ const u32 *unused_gpl_crcs;
#endif

#ifdef CONFIG_MODULE_SIG
@@ -382,7 +382,7 @@ struct module {

/* symbols that will be GPL-only in the near future. */
const struct kernel_symbol *gpl_future_syms;
- const unsigned long *gpl_future_crcs;
+ const u32 *gpl_future_crcs;
unsigned int num_gpl_future_syms;

/* Exception table */
@@ -523,7 +523,7 @@ struct module *find_module(const char *name);

struct symsearch {
const struct kernel_symbol *start, *stop;
- const unsigned long *crcs;
+ const u32 *crcs;
enum {
NOT_GPL_ONLY,
GPL_ONLY,
@@ -539,7 +539,7 @@ struct symsearch {
*/
const struct kernel_symbol *find_symbol(const char *name,
struct module **owner,
- const unsigned long **crc,
+ const u32 **crc,
bool gplok,
bool warn);

diff --git a/kernel/module.c b/kernel/module.c
index f57dd63186e6..90ecdad07e1a 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -386,16 +386,16 @@ extern const struct kernel_symbol __start___ksymtab_gpl[];
extern const struct kernel_symbol __stop___ksymtab_gpl[];
extern const struct kernel_symbol __start___ksymtab_gpl_future[];
extern const struct kernel_symbol __stop___ksymtab_gpl_future[];
-extern const unsigned long __start___kcrctab[];
-extern const unsigned long __start___kcrctab_gpl[];
-extern const unsigned long __start___kcrctab_gpl_future[];
+extern const u32 __start___kcrctab[];
+extern const u32 __start___kcrctab_gpl[];
+extern const u32 __start___kcrctab_gpl_future[];
#ifdef CONFIG_UNUSED_SYMBOLS
extern const struct kernel_symbol __start___ksymtab_unused[];
extern const struct kernel_symbol __stop___ksymtab_unused[];
extern const struct kernel_symbol __start___ksymtab_unused_gpl[];
extern const struct kernel_symbol __stop___ksymtab_unused_gpl[];
-extern const unsigned long __start___kcrctab_unused[];
-extern const unsigned long __start___kcrctab_unused_gpl[];
+extern const u32 __start___kcrctab_unused[];
+extern const u32 __start___kcrctab_unused_gpl[];
#endif

#ifndef CONFIG_MODVERSIONS
@@ -494,7 +494,7 @@ struct find_symbol_arg {

/* Output */
struct module *owner;
- const unsigned long *crc;
+ const u32 *crc;
const struct kernel_symbol *sym;
};

@@ -560,7 +560,7 @@ static bool find_symbol_in_section(const struct symsearch *syms,
* (optional) module which owns it. Needs preempt disabled or module_mutex. */
const struct kernel_symbol *find_symbol(const char *name,
struct module **owner,
- const unsigned long **crc,
+ const u32 **crc,
bool gplok,
bool warn)
{
@@ -1257,22 +1257,11 @@ static int try_to_force_load(struct module *mod, const char *reason)
}

#ifdef CONFIG_MODVERSIONS
-/* If the arch applies (non-zero) relocations to kernel kcrctab, unapply it. */
-static unsigned long maybe_relocated(unsigned long crc,
- const struct module *crc_owner)
-{
-#ifdef ARCH_RELOCATES_KCRCTAB
- if (crc_owner == NULL)
- return crc - (unsigned long)reloc_start;
-#endif
- return crc;
-}
-
static int check_version(Elf_Shdr *sechdrs,
unsigned int versindex,
const char *symname,
struct module *mod,
- const unsigned long *crc,
+ const u32 *crc,
const struct module *crc_owner)
{
unsigned int i, num_versions;
@@ -1294,10 +1283,10 @@ static int check_version(Elf_Shdr *sechdrs,
if (strcmp(versions[i].name, symname) != 0)
continue;

- if (versions[i].crc == maybe_relocated(*crc, crc_owner))
+ if (versions[i].crc == *crc)
return 1;
- pr_debug("Found checksum %lX vs module %lX\n",
- maybe_relocated(*crc, crc_owner), versions[i].crc);
+ pr_debug("Found checksum %X vs module %lX\n",
+ *crc, versions[i].crc);
goto bad_version;
}

@@ -1314,7 +1303,7 @@ static inline int check_modstruct_version(Elf_Shdr *sechdrs,
unsigned int versindex,
struct module *mod)
{
- const unsigned long *crc;
+ const u32 *crc;

/*
* Since this should be found in kernel (which can't be removed), no
@@ -1347,7 +1336,7 @@ static inline int check_version(Elf_Shdr *sechdrs,
unsigned int versindex,
const char *symname,
struct module *mod,
- const unsigned long *crc,
+ const u32 *crc,
const struct module *crc_owner)
{
return 1;
@@ -1375,7 +1364,7 @@ static const struct kernel_symbol *resolve_symbol(struct module *mod,
{
struct module *owner;
const struct kernel_symbol *sym;
- const unsigned long *crc;
+ const u32 *crc;
int err;

/*
--
2.7.4