Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

[PATCH v3 0/3] modversions: Fix CRC mangling under CONFIG_RELOCATABLE=y

88 views
Skip to first unread message

Ard Biesheuvel

unread,
Oct 27, 2016, 12:30:06 PM10/27/16
to
This series is a followup to the single patch 'modversions: treat symbol
CRCs as 32 bit quantities on 64 bit archs', of which two versions have
been sent out so far [0][1]

As pointed out by Michael, GNU ld behaves a bit differently between arm64
and PowerPC64, and where the former gets rid of all runtime relocations
related to CRCs, the latter is not as easily convinced.

Patch #1 fixes the issue where CRCs are corrupted by the runtime relocation
routines for 32-bit PowerPC, for which the original fix was effectively
reverted by commit 0e0ed6406e61 ("powerpc/modules: Module CRC relocation fix
causes perf issues")

Patch #2 adds handling of R_PPC64_ADDR32 relocations against the NULL .dynsym
symbol entry to the PPC64 runtime relocation routines, so it is prepared to
deal with CRCs being emitted as 32-bit quantities.

Patch #3 is the original patch from the v1 and v2 submissions.

Changes since v2:
- added #1 and #2
- updated #3 to deal with CRC entries being emitted from assembler
- added Rusty's ack (#3)

Branch can be found here:
https://git.kernel.org/cgit/linux/kernel/git/ardb/linux.git/log/?h=kcrctab-reloc

[0] http://marc.info/?l=linux-kernel&m=147652300207369&w=2
[1] http://marc.info/?l=linux-kernel&m=147695629614409&w=2

Ard Biesheuvel (3):
powerpc/reloc32: fix corrupted modversion CRCs
powerpc/reloc64: add support for 32-bit CRC pseudo-symbols
modversions: treat symbol CRCs as 32 bit quantities on 64 bit archs

arch/powerpc/include/asm/module.h | 4 --
arch/powerpc/kernel/module_64.c | 8 ----
arch/powerpc/kernel/reloc_32.S | 36 +++++++++++++--
arch/powerpc/kernel/reloc_64.S | 22 +++++++--
arch/powerpc/relocs_check.sh | 3 +-
include/asm-generic/export.h | 7 +--
include/linux/export.h | 8 ++++
include/linux/module.h | 14 +++---
kernel/module.c | 47 +++++++-------------
9 files changed, 85 insertions(+), 64 deletions(-)

--
2.7.4

Ard Biesheuvel

unread,
Oct 27, 2016, 12:30:06 PM10/27/16
to
Commit 0e0ed6406e61 ("powerpc/modules: Module CRC relocation fix causes
perf issues") fixed an issue with relocatable PIE kernels in a way that
essentially reintroduced the issue again for 32-bit builds.

Since the chosen approach does is not applicable to 32-bit, fix the
issue by updating the runtime relocation routine to ignore the load
offset for the interval [__start___kcrctab, __stop___kcrctab_gpl_future),
which is where the CRCs reside. This ensures that the values of the CRC
pseudo-symbols are no longer made dependent on the runtime load offset.

Signed-off-by: Ard Biesheuvel <ard.bie...@linaro.org>
---
arch/powerpc/kernel/reloc_32.S | 36 +++++++++++++++++---
1 file changed, 32 insertions(+), 4 deletions(-)

diff --git a/arch/powerpc/kernel/reloc_32.S b/arch/powerpc/kernel/reloc_32.S
index f366fedb0872..150686b9febb 100644
--- a/arch/powerpc/kernel/reloc_32.S
+++ b/arch/powerpc/kernel/reloc_32.S
@@ -87,12 +87,12 @@ eodyn: /* End of Dyn Table scan */
* Work out the current offset from the link time address of .rela
* section.
* cur_offset[r7] = rela.run[r9] - rela.link [r7]
- * _stext.link[r12] = _stext.run[r10] - cur_offset[r7]
- * final_offset[r3] = _stext.final[r3] - _stext.link[r12]
+ * _stext.link[r11] = _stext.run[r10] - cur_offset[r7]
+ * final_offset[r3] = _stext.final[r3] - _stext.link[r11]
*/
subf r7, r7, r9 /* cur_offset */
- subf r12, r7, r10
- subf r3, r12, r3 /* final_offset */
+ subf r11, r7, r10
+ subf r3, r11, r3 /* final_offset */

subf r8, r6, r8 /* relaz -= relaent */
/*
@@ -101,6 +101,21 @@ eodyn: /* End of Dyn Table scan */
* r13 - points to the symbol table
*/

+#ifdef CONFIG_MODVERSIONS
+ /*
+ * Treat R_PPC_RELATIVE relocations differently when they target the
+ * interval [__start___kcrctab, __stop___kcrctab_gpl_future): in this
+ * case, the relocated quantities are CRC pseudo-symbols, which should
+ * be preserved as-is, rather than be modified to take the runtime
+ * offset into account.
+ */
+ lwz r10, (p_kcrc_start - 0b)(r12)
+ lwz r11, (p_kcrc_stop - 0b)(r12)
+ subf r12, r7, r12 /* link time addr of 0b */
+ add r10, r10, r12
+ add r11, r11, r12
+#endif
+
/*
* Check if we have a relocation based on symbol
* r5 will hold the value of the symbol.
@@ -135,7 +150,15 @@ get_type:
bne hi16
lwz r4, 0(r9) /* r_offset */
lwz r0, 8(r9) /* r_addend */
+#ifdef CONFIG_MODVERSIONS
+ cmplw r4, r10
+ blt do_add
+ cmplw r4, r11
+ blt skip_add
+do_add:
+#endif
add r0, r0, r3 /* final addend */
+skip_add:
stwx r0, r4, r7 /* memory[r4+r7]) = (u32)r0 */
b nxtrela /* continue */

@@ -207,3 +230,8 @@ p_dyn: .long __dynamic_start - 0b
p_rela: .long __rela_dyn_start - 0b
p_sym: .long __dynamic_symtab - 0b
p_st: .long _stext - 0b
+
+#ifdef CONFIG_MODVERSIONS
+p_kcrc_start: .long __start___kcrctab - 0b
+p_kcrc_stop: .long __stop___kcrctab_gpl_future - 0b
+#endif
--
2.7.4

Ard Biesheuvel

unread,
Oct 27, 2016, 12:30:06 PM10/27/16
to
The modversion 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 a couple of downsides:
- Given that the CRCs are treated as memory addresses, we waste 4 bytes
for each CRC on 64 bit architectures,
- On architectures that support runtime relocation, a R_<arch>_RELATIVE
relocation entry is emitted for each CRC value, which identifies it as
a quantity that requires fixing up based on the actual runtime load
offset of the kernel. This results in corrupted CRCs unless we
explicitly undo the fixup (and this is currently being handled in the
core module code)
- Such runtime relocation entries take up 24 bytes of __init space each,
resulting in a x8 overhead in [uncompressed] kernel size for CRCs.

Switching to explicit 32 bit values on 64 bit architectures fixes most
of these issues, given that 32 bit values are not treated as quantities
that require fixing up based on the actual runtime load offset. Note that
on some ELF64 architectures [such as PPC64], these 32-bit values are still
emitted as runtime relocatable quantities, even if the value resolves to
a build time constant. However, these can easily be distinguished from
variables that do need fixing up, and the CRCs can be emitted correctly
in the arch specific runtime relocation routines right off the bat.

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 not exceed 0xffffffff.

Note that this mostly reverts commit d4703aefdbc8 ("module: handle ppc64
relocating kcrctabs when CONFIG_RELOCATABLE=y")

Acked-by: Rusty Russell <ru...@rustcorp.com.au>
Signed-off-by: Ard Biesheuvel <ard.bie...@linaro.org>
---
arch/powerpc/include/asm/module.h | 4 --
arch/powerpc/kernel/module_64.c | 8 ----
include/asm-generic/export.h | 7 +--
include/linux/export.h | 8 ++++
include/linux/module.h | 14 +++---
kernel/module.c | 47 +++++++-------------
6 files changed, 33 insertions(+), 55 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/asm-generic/export.h b/include/asm-generic/export.h
index 43199a049da5..d245910d026b 100644
--- a/include/asm-generic/export.h
+++ b/include/asm-generic/export.h
@@ -9,18 +9,15 @@
#ifndef KSYM_ALIGN
#define KSYM_ALIGN 8
#endif
-#ifndef KCRC_ALIGN
-#define KCRC_ALIGN 8
-#endif
#else
#define __put .long
#ifndef KSYM_ALIGN
#define KSYM_ALIGN 4
#endif
+#endif
#ifndef KCRC_ALIGN
#define KCRC_ALIGN 4
#endif
-#endif

#ifdef CONFIG_HAVE_UNDERSCORE_SYMBOL_PREFIX
#define KSYM(name) _##name
@@ -52,7 +49,7 @@ KSYM(__kstrtab_\name):
.section ___kcrctab\sec+\name,"a"
.balign KCRC_ALIGN
KSYM(__kcrctab_\name):
- __put KSYM(__crc_\name)
+ .long KSYM(__crc_\name)
.weak KSYM(__crc_\name)
.previous
#endif
diff --git a/include/linux/export.h b/include/linux/export.h
index 2a0f61fbc731..a000d421526d 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" \
+ " .long " 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..18d92e710a13 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)
{
@@ -1257,23 +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 struct module *crc_owner)
+ const u32 *crc)
{
unsigned int i, num_versions;
struct modversion_info *versions;
@@ -1294,10 +1282,10 @@ 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 +1302,7 @@ 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
@@ -1328,8 +1316,7 @@ static inline int check_modstruct_version(Elf_Shdr *sechdrs,
}
preempt_enable();
return check_version(sechdrs, versindex,
- VMLINUX_SYMBOL_STR(module_layout), mod, crc,
- NULL);
+ VMLINUX_SYMBOL_STR(module_layout), mod, crc);
}

/* First part is kernel version, which we ignore if module has crcs. */
@@ -1347,8 +1334,7 @@ static inline int check_version(Elf_Shdr *sechdrs,
unsigned int versindex,
const char *symname,
struct module *mod,
- const unsigned long *crc,
- const struct module *crc_owner)
+ const u32 *crc)
{
return 1;
}
@@ -1375,7 +1361,7 @@ 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;

/*
@@ -1390,8 +1376,7 @@ static const struct kernel_symbol *resolve_symbol(struct module *mod,
if (!sym)
goto unlock;

- if (!check_version(info->sechdrs, info->index.vers, name, mod, crc,
- owner)) {
+ if (!check_version(info->sechdrs, info->index.vers, name, mod, crc)) {
sym = ERR_PTR(-EINVAL);
goto getname;
}
--
2.7.4

Ard Biesheuvel

unread,
Oct 27, 2016, 12:30:07 PM10/27/16
to
This series is a followup to the single patch 'modversions: treat symbol
CRCs as 32 bit quantities on 64 bit archs', of which two versions have
been sent out so far [0][1]

As pointed out by Michael, GNU ld behaves a bit differently between arm64
and PowerPC64, and where the former gets rid of all runtime relocations
related to CRCs, the latter is not as easily convinced.

Patch #1 fixes the issue where CRCs are corrupted by the runtime relocation
routines for 32-bit PowerPC, for which the original fix was effectively
reverted by commit 0e0ed6406e61 ("powerpc/modules: Module CRC relocation fix
causes perf issues")

Patch #2 adds handling of R_PPC64_ADDR32 relocations against the NULL .dynsym
symbol entry to the PPC64 runtime relocation routines, so it is prepared to
deal with CRCs being emitted as 32-bit quantities.

Patch #3 is the original patch from the v1 and v2 submissions.

Changes since v2:
- added #1 and #2
- updated #3 to deal with CRC entries being emitted from assembler
- added Rusty's ack (#3)

[0] http://marc.info/?l=linux-kernel&m=147652300207369&w=2
[1] http://marc.info/?l=linux-kernel&m=147695629614409&w=2

Ard Biesheuvel (3):
powerpc/reloc32: fix corrupted modversion CRCs
powerpc/reloc64: add support for 32-bit CRC pseudo-symbols
modversions: treat symbol CRCs as 32 bit quantities on 64 bit archs

arch/powerpc/include/asm/module.h | 4 --
arch/powerpc/kernel/module_64.c | 8 ----
arch/powerpc/kernel/reloc_32.S | 36 +++++++++++++--
arch/powerpc/kernel/reloc_64.S | 22 +++++++--
arch/powerpc/relocs_check.sh | 3 +-
include/asm-generic/export.h | 7 +--
include/linux/export.h | 8 ++++
include/linux/module.h | 14 +++---
kernel/module.c | 47 +++++++-------------

Suzuki K Poulose

unread,
Oct 28, 2016, 6:30:06 AM10/28/16
to
On 27/10/16 17:27, Ard Biesheuvel wrote:
> Commit 0e0ed6406e61 ("powerpc/modules: Module CRC relocation fix causes
> perf issues") fixed an issue with relocatable PIE kernels in a way that
> essentially reintroduced the issue again for 32-bit builds.
>
> Since the chosen approach does is not applicable to 32-bit, fix the
> issue by updating the runtime relocation routine to ignore the load
> offset for the interval [__start___kcrctab, __stop___kcrctab_gpl_future),
> which is where the CRCs reside. This ensures that the values of the CRC
> pseudo-symbols are no longer made dependent on the runtime load offset.
>
> Signed-off-by: Ard Biesheuvel <ard.bie...@linaro.org>

Ard,

These changes look good to me (having originally written the code).

Reviewed-by : Suzuki K Poulose <suzuki....@arm.com>

Cheers
Suzuki

Ard Biesheuvel

unread,
Nov 4, 2016, 6:00:06 AM11/4/16
to
On 27 October 2016 at 17:27, Ard Biesheuvel <ard.bie...@linaro.org> wrote:
> This series is a followup to the single patch 'modversions: treat symbol
> CRCs as 32 bit quantities on 64 bit archs', of which two versions have
> been sent out so far [0][1]
>
> As pointed out by Michael, GNU ld behaves a bit differently between arm64
> and PowerPC64, and where the former gets rid of all runtime relocations
> related to CRCs, the latter is not as easily convinced.
>
> Patch #1 fixes the issue where CRCs are corrupted by the runtime relocation
> routines for 32-bit PowerPC, for which the original fix was effectively
> reverted by commit 0e0ed6406e61 ("powerpc/modules: Module CRC relocation fix
> causes perf issues")
>
> Patch #2 adds handling of R_PPC64_ADDR32 relocations against the NULL .dynsym
> symbol entry to the PPC64 runtime relocation routines, so it is prepared to
> deal with CRCs being emitted as 32-bit quantities.
>
> Patch #3 is the original patch from the v1 and v2 submissions.
>
> Changes since v2:
> - added #1 and #2
> - updated #3 to deal with CRC entries being emitted from assembler
> - added Rusty's ack (#3)
>
> Branch can be found here:
> https://git.kernel.org/cgit/linux/kernel/git/ardb/linux.git/log/?h=kcrctab-reloc
>
> [0] http://marc.info/?l=linux-kernel&m=147652300207369&w=2
> [1] http://marc.info/?l=linux-kernel&m=147695629614409&w=2
>

Ping?

Michael Ellerman

unread,
Nov 9, 2016, 11:30:05 PM11/9/16
to
Sorry, you didn't cc linuxppc-dev, so it's not in my patchwork list
which tends to mean I miss it.

Will try and test and get back to you.

cheers

Ard Biesheuvel

unread,
Nov 15, 2016, 4:20:05 AM11/15/16
to
Ah, my mistake. Apologies.

> Will try and test and get back to you.
>

Thanks!

Uwe Kleine-König

unread,
Nov 16, 2016, 2:30:05 PM11/16/16
to
On Thu, Oct 27, 2016 at 05:27:08PM +0100, Ard Biesheuvel wrote:
> This series is a followup to the single patch 'modversions: treat symbol
> CRCs as 32 bit quantities on 64 bit archs', of which two versions have
> been sent out so far [0][1]
>
> As pointed out by Michael, GNU ld behaves a bit differently between arm64
> and PowerPC64, and where the former gets rid of all runtime relocations
> related to CRCs, the latter is not as easily convinced.
>
> Patch #1 fixes the issue where CRCs are corrupted by the runtime relocation
> routines for 32-bit PowerPC, for which the original fix was effectively
> reverted by commit 0e0ed6406e61 ("powerpc/modules: Module CRC relocation fix
> causes perf issues")
>
> Patch #2 adds handling of R_PPC64_ADDR32 relocations against the NULL .dynsym
> symbol entry to the PPC64 runtime relocation routines, so it is prepared to
> deal with CRCs being emitted as 32-bit quantities.
>
> Patch #3 is the original patch from the v1 and v2 submissions.

Is this related to me seeing

[ 2.111424] mvneta: module verification failed: signature and/or required key missing - tainting kernel
[ 2.126061] scsi_mod: no symbol version for _clear_bit
[ 2.131257] scsi_mod: Unknown symbol _clear_bit (err -22)
[ 2.138093] mvneta: no symbol version for _clear_bit
[ 2.143117] mvneta: Unknown symbol _clear_bit (err -22)
[ 2.144135] mvmdio: no symbol version for __gnu_mcount_nc
[ 2.144138] mvmdio: Unknown symbol __gnu_mcount_nc (err -22)
...

? If so, this would be great to mention it in the commit log to make
people searching for this issue actually find this patch set.

Best regards
Uwe

--
Pengutronix e.K. | Uwe Kleine-König |
Industrial Linux Solutions | http://www.pengutronix.de/ |

Ard Biesheuvel

unread,
Nov 16, 2016, 3:30:05 PM11/16/16
to
On 16 November 2016 at 19:23, Uwe Kleine-König
No, I don't think it is. My guess would be that it is caused by

https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=4dd1837d75

Ard Biesheuvel

unread,
Nov 25, 2016, 3:50:06 AM11/25/16
to
Ping?

Michael Ellerman

unread,
Nov 25, 2016, 6:30:05 AM11/25/16
to
Sorry :/

I tried testing it last week or so but it was interacting badly with the
other modversion CRC problems we were having (fixed recently in my fixes
branch).

I'll rebase on top of those fixes and try again.

cheers
0 new messages