* [PATCH] x86/retpoline: change RETPOLINE into CONFIG_RETPOLINE
@ 2018-12-31 23:42 Nadav Amit
2019-01-02 1:17 ` Zhenzhong Duan
0 siblings, 1 reply; 4+ messages in thread
From: Nadav Amit @ 2018-12-31 23:42 UTC (permalink / raw)
To: Thomas Gleixner
Cc: linux-kernel, x86, Borislav Petkov, Ingo Molnar, Nadav Amit,
Peter Zijlstra, Zhenzhong Duan, David Woodhouse, Andy Lutomirski,
Masahiro Yamada, stable
A recent enhancement intentionally fails the kernel build if the
compiler does not support retpolines and CONFIG_RETPOLINE is set.
However, the patch that introduced it did not change RETPOLINE macro
references into CONFIG_RETPOLINE ones. As a result, indirect branches
that are used by init functions are not kept (i.e., they use
retpolines), and modules that do not use retpolines are marked as
retpoline-safe.
Fix it be changing RETPOLINE into CONFIG_RETPOLINE.
Fixes: 4cd24de3a098 ("x86/retpoline: Make CONFIG_RETPOLINE depend on compiler support")
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Zhenzhong Duan <zhenzhong.duan@oracle.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: David Woodhouse <dwmw@amazon.co.uk>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Cc: stable@vger.kernel.org
Signed-off-by: Nadav Amit <namit@vmware.com>
---
arch/x86/kernel/cpu/bugs.c | 2 +-
include/linux/compiler-gcc.h | 2 +-
include/linux/module.h | 2 +-
3 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c
index 8654b8b0c848..1de0f4170178 100644
--- a/arch/x86/kernel/cpu/bugs.c
+++ b/arch/x86/kernel/cpu/bugs.c
@@ -215,7 +215,7 @@ static enum spectre_v2_mitigation spectre_v2_enabled __ro_after_init =
static enum spectre_v2_user_mitigation spectre_v2_user __ro_after_init =
SPECTRE_V2_USER_NONE;
-#ifdef RETPOLINE
+#ifdef CONFIG_RETPOLINE
static bool spectre_v2_bad_module;
bool retpoline_module_ok(bool has_retpoline)
diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
index 2010493e1040..977ddf2774f9 100644
--- a/include/linux/compiler-gcc.h
+++ b/include/linux/compiler-gcc.h
@@ -68,7 +68,7 @@
*/
#define uninitialized_var(x) x = x
-#ifdef RETPOLINE
+#ifdef CONFIG_RETPOLINE
#define __noretpoline __attribute__((__indirect_branch__("keep")))
#endif
diff --git a/include/linux/module.h b/include/linux/module.h
index fce6b4335e36..0c575f51fe57 100644
--- a/include/linux/module.h
+++ b/include/linux/module.h
@@ -817,7 +817,7 @@ static inline void module_bug_finalize(const Elf_Ehdr *hdr,
static inline void module_bug_cleanup(struct module *mod) {}
#endif /* CONFIG_GENERIC_BUG */
-#ifdef RETPOLINE
+#ifdef CONFIG_RETPOLINE
extern bool retpoline_module_ok(bool has_retpoline);
#else
static inline bool retpoline_module_ok(bool has_retpoline)
--
2.17.1
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH] x86/retpoline: change RETPOLINE into CONFIG_RETPOLINE
2018-12-31 23:42 [PATCH] x86/retpoline: change RETPOLINE into CONFIG_RETPOLINE Nadav Amit
@ 2019-01-02 1:17 ` Zhenzhong Duan
2019-01-02 17:23 ` Nadav Amit
0 siblings, 1 reply; 4+ messages in thread
From: Zhenzhong Duan @ 2019-01-02 1:17 UTC (permalink / raw)
To: Nadav Amit, Thomas Gleixner
Cc: linux-kernel, x86, Borislav Petkov, Ingo Molnar, Peter Zijlstra,
David Woodhouse, Andy Lutomirski, Masahiro Yamada, stable
On 2019/1/1 7:42, Nadav Amit wrote:
> A recent enhancement intentionally fails the kernel build if the
> compiler does not support retpolines and CONFIG_RETPOLINE is set.
>
> However, the patch that introduced it did not change RETPOLINE macro
> references into CONFIG_RETPOLINE ones. As a result, indirect branches
> that are used by init functions are not kept (i.e., they use
> retpolines), and modules that do not use retpolines are marked as
> retpoline-safe.
>
> Fix it be changing RETPOLINE into CONFIG_RETPOLINE.
>
> Fixes: 4cd24de3a098 ("x86/retpoline: Make CONFIG_RETPOLINE depend on compiler support")
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Zhenzhong Duan <zhenzhong.duan@oracle.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: David Woodhouse <dwmw@amazon.co.uk>
> Cc: Andy Lutomirski <luto@kernel.org>
> Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
> Cc: stable@vger.kernel.org
> Signed-off-by: Nadav Amit <namit@vmware.com>
> ---
> arch/x86/kernel/cpu/bugs.c | 2 +-
> include/linux/compiler-gcc.h | 2 +-
> include/linux/module.h | 2 +-
> 3 files changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c
> index 8654b8b0c848..1de0f4170178 100644
> --- a/arch/x86/kernel/cpu/bugs.c
> +++ b/arch/x86/kernel/cpu/bugs.c
> @@ -215,7 +215,7 @@ static enum spectre_v2_mitigation spectre_v2_enabled __ro_after_init =
> static enum spectre_v2_user_mitigation spectre_v2_user __ro_after_init =
> SPECTRE_V2_USER_NONE;
>
> -#ifdef RETPOLINE
> +#ifdef CONFIG_RETPOLINE
> static bool spectre_v2_bad_module;
>
> bool retpoline_module_ok(bool has_retpoline)
> diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
> index 2010493e1040..977ddf2774f9 100644
> --- a/include/linux/compiler-gcc.h
> +++ b/include/linux/compiler-gcc.h
> @@ -68,7 +68,7 @@
> */
> #define uninitialized_var(x) x = x
>
> -#ifdef RETPOLINE
> +#ifdef CONFIG_RETPOLINE
> #define __noretpoline __attribute__((__indirect_branch__("keep")))
> #endif
>
> diff --git a/include/linux/module.h b/include/linux/module.h
> index fce6b4335e36..0c575f51fe57 100644
> --- a/include/linux/module.h
> +++ b/include/linux/module.h
> @@ -817,7 +817,7 @@ static inline void module_bug_finalize(const Elf_Ehdr *hdr,
> static inline void module_bug_cleanup(struct module *mod) {}
> #endif /* CONFIG_GENERIC_BUG */
>
> -#ifdef RETPOLINE
> +#ifdef CONFIG_RETPOLINE
> extern bool retpoline_module_ok(bool has_retpoline);
> #else
> static inline bool retpoline_module_ok(bool has_retpoline)
Someone sent a similar patch last month, see below link, you missed the
change in scripts/mod/modpost.c
https://lkml.org/lkml/2018/12/10/540
But you patch description is more clear.
--
Thanks
Zhenzhong
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] x86/retpoline: change RETPOLINE into CONFIG_RETPOLINE
2019-01-02 1:17 ` Zhenzhong Duan
@ 2019-01-02 17:23 ` Nadav Amit
2019-01-03 2:07 ` Zhenzhong Duan
0 siblings, 1 reply; 4+ messages in thread
From: Nadav Amit @ 2019-01-02 17:23 UTC (permalink / raw)
To: Zhenzhong Duan
Cc: Thomas Gleixner, LKML, X86 ML, Borislav Petkov, Ingo Molnar,
Peter Zijlstra, David Woodhouse, Andy Lutomirski,
Masahiro Yamada, stable
> On Jan 1, 2019, at 5:17 PM, Zhenzhong Duan <zhenzhong.duan@oracle.com> wrote:
>
>
> On 2019/1/1 7:42, Nadav Amit wrote:
>> A recent enhancement intentionally fails the kernel build if the
>> compiler does not support retpolines and CONFIG_RETPOLINE is set.
>>
>> However, the patch that introduced it did not change RETPOLINE macro
>> references into CONFIG_RETPOLINE ones. As a result, indirect branches
>> that are used by init functions are not kept (i.e., they use
>> retpolines), and modules that do not use retpolines are marked as
>> retpoline-safe.
>>
>> Fix it be changing RETPOLINE into CONFIG_RETPOLINE.
>>
>> Fixes: 4cd24de3a098 ("x86/retpoline: Make CONFIG_RETPOLINE depend on compiler support")
>> Cc: Peter Zijlstra <peterz@infradead.org>
>> Cc: Zhenzhong Duan <zhenzhong.duan@oracle.com>
>> Cc: Thomas Gleixner <tglx@linutronix.de>
>> Cc: David Woodhouse <dwmw@amazon.co.uk>
>> Cc: Andy Lutomirski <luto@kernel.org>
>> Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
>> Cc: stable@vger.kernel.org
>> Signed-off-by: Nadav Amit <namit@vmware.com>
>> ---
>> arch/x86/kernel/cpu/bugs.c | 2 +-
>> include/linux/compiler-gcc.h | 2 +-
>> include/linux/module.h | 2 +-
>> 3 files changed, 3 insertions(+), 3 deletions(-)
>>
>> diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c
>> index 8654b8b0c848..1de0f4170178 100644
>> --- a/arch/x86/kernel/cpu/bugs.c
>> +++ b/arch/x86/kernel/cpu/bugs.c
>> @@ -215,7 +215,7 @@ static enum spectre_v2_mitigation spectre_v2_enabled __ro_after_init =
>> static enum spectre_v2_user_mitigation spectre_v2_user __ro_after_init =
>> SPECTRE_V2_USER_NONE;
>> -#ifdef RETPOLINE
>> +#ifdef CONFIG_RETPOLINE
>> static bool spectre_v2_bad_module;
>> bool retpoline_module_ok(bool has_retpoline)
>> diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
>> index 2010493e1040..977ddf2774f9 100644
>> --- a/include/linux/compiler-gcc.h
>> +++ b/include/linux/compiler-gcc.h
>> @@ -68,7 +68,7 @@
>> */
>> #define uninitialized_var(x) x = x
>> -#ifdef RETPOLINE
>> +#ifdef CONFIG_RETPOLINE
>> #define __noretpoline __attribute__((__indirect_branch__("keep")))
>> #endif
>> diff --git a/include/linux/module.h b/include/linux/module.h
>> index fce6b4335e36..0c575f51fe57 100644
>> --- a/include/linux/module.h
>> +++ b/include/linux/module.h
>> @@ -817,7 +817,7 @@ static inline void module_bug_finalize(const Elf_Ehdr *hdr,
>> static inline void module_bug_cleanup(struct module *mod) {}
>> #endif /* CONFIG_GENERIC_BUG */
>> -#ifdef RETPOLINE
>> +#ifdef CONFIG_RETPOLINE
>> extern bool retpoline_module_ok(bool has_retpoline);
>> #else
>> static inline bool retpoline_module_ok(bool has_retpoline)
>
> Someone sent a similar patch last month, see below link, you missed the change in scripts/mod/modpost.c
Indeed. I missed this patch since it wasn’t on tip (and Google wasn’t
helpful either).
I wonder why it has still not been applied.
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] x86/retpoline: change RETPOLINE into CONFIG_RETPOLINE
2019-01-02 17:23 ` Nadav Amit
@ 2019-01-03 2:07 ` Zhenzhong Duan
0 siblings, 0 replies; 4+ messages in thread
From: Zhenzhong Duan @ 2019-01-03 2:07 UTC (permalink / raw)
To: Nadav Amit
Cc: Thomas Gleixner, LKML, X86 ML, Borislav Petkov, Ingo Molnar,
Peter Zijlstra, David Woodhouse, Andy Lutomirski,
Masahiro Yamada, stable
On 2019/1/3 1:23, Nadav Amit wrote:
>> On Jan 1, 2019, at 5:17 PM, Zhenzhong Duan <zhenzhong.duan@oracle.com> wrote:
>>
>>
>> On 2019/1/1 7:42, Nadav Amit wrote:
>>> A recent enhancement intentionally fails the kernel build if the
>>> compiler does not support retpolines and CONFIG_RETPOLINE is set.
>>>
>>> However, the patch that introduced it did not change RETPOLINE macro
>>> references into CONFIG_RETPOLINE ones. As a result, indirect branches
>>> that are used by init functions are not kept (i.e., they use
>>> retpolines), and modules that do not use retpolines are marked as
>>> retpoline-safe.
>>>
>>> Fix it be changing RETPOLINE into CONFIG_RETPOLINE.
>>>
>>> Fixes: 4cd24de3a098 ("x86/retpoline: Make CONFIG_RETPOLINE depend on compiler support")
>>> Cc: Peter Zijlstra <peterz@infradead.org>
>>> Cc: Zhenzhong Duan <zhenzhong.duan@oracle.com>
>>> Cc: Thomas Gleixner <tglx@linutronix.de>
>>> Cc: David Woodhouse <dwmw@amazon.co.uk>
>>> Cc: Andy Lutomirski <luto@kernel.org>
>>> Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
>>> Cc: stable@vger.kernel.org
>>> Signed-off-by: Nadav Amit <namit@vmware.com>
>>> ---
>>> arch/x86/kernel/cpu/bugs.c | 2 +-
>>> include/linux/compiler-gcc.h | 2 +-
>>> include/linux/module.h | 2 +-
>>> 3 files changed, 3 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c
>>> index 8654b8b0c848..1de0f4170178 100644
>>> --- a/arch/x86/kernel/cpu/bugs.c
>>> +++ b/arch/x86/kernel/cpu/bugs.c
>>> @@ -215,7 +215,7 @@ static enum spectre_v2_mitigation spectre_v2_enabled __ro_after_init =
>>> static enum spectre_v2_user_mitigation spectre_v2_user __ro_after_init =
>>> SPECTRE_V2_USER_NONE;
>>> -#ifdef RETPOLINE
>>> +#ifdef CONFIG_RETPOLINE
>>> static bool spectre_v2_bad_module;
>>> bool retpoline_module_ok(bool has_retpoline)
>>> diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
>>> index 2010493e1040..977ddf2774f9 100644
>>> --- a/include/linux/compiler-gcc.h
>>> +++ b/include/linux/compiler-gcc.h
>>> @@ -68,7 +68,7 @@
>>> */
>>> #define uninitialized_var(x) x = x
>>> -#ifdef RETPOLINE
>>> +#ifdef CONFIG_RETPOLINE
>>> #define __noretpoline __attribute__((__indirect_branch__("keep")))
>>> #endif
>>> diff --git a/include/linux/module.h b/include/linux/module.h
>>> index fce6b4335e36..0c575f51fe57 100644
>>> --- a/include/linux/module.h
>>> +++ b/include/linux/module.h
>>> @@ -817,7 +817,7 @@ static inline void module_bug_finalize(const Elf_Ehdr *hdr,
>>> static inline void module_bug_cleanup(struct module *mod) {}
>>> #endif /* CONFIG_GENERIC_BUG */
>>> -#ifdef RETPOLINE
>>> +#ifdef CONFIG_RETPOLINE
>>> extern bool retpoline_module_ok(bool has_retpoline);
>>> #else
>>> static inline bool retpoline_module_ok(bool has_retpoline)
>> Someone sent a similar patch last month, see below link, you missed the change in scripts/mod/modpost.c
> Indeed. I missed this patch since it wasn’t on tip (and Google wasn’t
> helpful either).
>
> I wonder why it has still not been applied.
See https://lkml.org/lkml/2018/12/21/244
--
Thanks
Zhenzhong
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2019-01-03 2:08 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-12-31 23:42 [PATCH] x86/retpoline: change RETPOLINE into CONFIG_RETPOLINE Nadav Amit
2019-01-02 1:17 ` Zhenzhong Duan
2019-01-02 17:23 ` Nadav Amit
2019-01-03 2:07 ` Zhenzhong Duan
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).