[PATCH v2 0/2] Fix issue with alternatives/paravirt patches

[PATCH v2 0/2] Fix issue with alternatives/paravirt patches

[Jessica Yu]

Hi,

A few months ago, Chris Arges reported a bug involving alternatives/paravirt
patching that was discussed here [1] and here [2]. To briefly summarize the
bug, patch modules that contained .altinstructions or .parainstructions
sections would break because these alternative/paravirt patches would be
applied first by the module loader (see x86 module_finalize()), then
livepatch would later clobber these patches when applying per-object
relocations. This lead to crashes and unpredictable behavior.

One conclusion we reached from our last discussion was that we will
need to introduce some arch-specific code to address this problem.
This patchset presents a possible fix for the bug by adding a new
arch-specific arch_klp_init_object_loaded() function that by default
does nothing but can be overridden by different arches.

To fix this issue for x86, since we can access a patch module's Elf
sections through mod->klp_info, we can simply delay the calls to
apply_paravirt() and apply_alternatives() to arch_klp_init_object_loaded(),
which is called after relocations have been written for an object.
In addition, for patch modules, .parainstructions and .altinstructions are
prefixed by ".klp.arch.${objname}" so that the module loader ignores them
and livepatch can apply them manually.

Currently for kpatch, we don't support including jump table sections in
the patch module, and supporting .smp_locks is currently broken, so we
don't consider those sections (for now).

I did some light testing with some patches to kvm and verified that the
original issue reported in [2] was fixed.

Based on linux-next.

v1 here:
http://lkml.kernel.org/g/1467772500-26092-1-git-send-email-jeyu@redhat.com

v2:
 - add BUILD_BUG_ON() check in arch_klp_init_object_loaded (x86)

[1] http://thread.gmane.org/gmane.linux.kernel/2185604/
[2] https://github.com/dynup/kpatch/issues/580

Jessica Yu (2):
  livepatch: use arch_klp_init_object_loaded() to finish arch-specific tasks
  livepatch/x86: apply alternatives and paravirt patches after relocations

 arch/x86/kernel/Makefile    |  1 +
 arch/x86/kernel/livepatch.c | 65 +++++++++++++++++++++++++++++++++++++++++++++
 include/linux/livepatch.h   |  3 +++
 kernel/livepatch/core.c     | 12 +++++++--
 4 files changed, 79 insertions(+), 2 deletions(-)
 create mode 100644 arch/x86/kernel/livepatch.c

-- 
2.5.5

[PATCH v2 1/2] livepatch: use arch_klp_init_object_loaded() to finish arch-specific tasks

[Jessica Yu]

Introduce arch_klp_init_object_loaded() to complete any additional
arch-specific tasks during patching. Architecture code may override this
function.

Signed-off-by: Jessica Yu <jeyu@redhat.com>
---
 include/linux/livepatch.h |  3 +++
 kernel/livepatch/core.c   | 12 ++++++++++--
 2 files changed, 13 insertions(+), 2 deletions(-)

diff --git a/include/linux/livepatch.h b/include/linux/livepatch.h
index a93a0b2..9072f04 100644
--- a/include/linux/livepatch.h
+++ b/include/linux/livepatch.h
@@ -116,6 +116,9 @@ int klp_unregister_patch(struct klp_patch *);
 int klp_enable_patch(struct klp_patch *);
 int klp_disable_patch(struct klp_patch *);
 
+void arch_klp_init_object_loaded(struct klp_patch *patch,
+				 struct klp_object *obj);
+
 /* Called from the module loader during module coming/going states */
 int klp_module_coming(struct module *mod);
 void klp_module_going(struct module *mod);
diff --git a/kernel/livepatch/core.c b/kernel/livepatch/core.c
index 5c2bc10..164eff6 100644
--- a/kernel/livepatch/core.c
+++ b/kernel/livepatch/core.c
@@ -274,7 +274,6 @@ static int klp_write_object_relocations(struct module *pmod,
 
 	objname = klp_is_module(obj) ? obj->name : "vmlinux";
 
-	module_disable_ro(pmod);
 	/* For each klp relocation section */
 	for (i = 1; i < pmod->klp_info->hdr.e_shnum; i++) {
 		sec = pmod->klp_info->sechdrs + i;
@@ -309,7 +308,6 @@ static int klp_write_object_relocations(struct module *pmod,
 			break;
 	}
 
-	module_enable_ro(pmod);
 	return ret;
 }
 
@@ -763,6 +761,12 @@ static int klp_init_func(struct klp_object *obj, struct klp_func *func)
 				    func->old_sympos ? func->old_sympos : 1);
 }
 
+/* Arches may override this to finish any remaining arch-specific tasks */
+void __weak arch_klp_init_object_loaded(struct klp_patch *patch,
+					struct klp_object *obj)
+{
+}
+
 /* parts of the initialization that is done only when the object is loaded */
 static int klp_init_object_loaded(struct klp_patch *patch,
 				  struct klp_object *obj)
@@ -770,10 +774,14 @@ static int klp_init_object_loaded(struct klp_patch *patch,
 	struct klp_func *func;
 	int ret;
 
+	module_disable_ro(patch->mod);
 	ret = klp_write_object_relocations(patch->mod, obj);
 	if (ret)
 		return ret;
 
+	arch_klp_init_object_loaded(patch, obj);
+	module_enable_ro(patch->mod);
+
 	klp_for_each_func(obj, func) {
 		ret = klp_find_object_symbol(obj->name, func->old_name,
 					     func->old_sympos,
-- 
2.5.5

[PATCH v2 2/2] livepatch/x86: apply alternatives and paravirt patches after relocations

[Jessica Yu]

Implement arch_klp_init_object_loaded() for x86, which applies
alternatives/paravirt patches. This fixes the order in which relocations
and alternatives/paravirt patches are applied.

Previously, if a patch module had alternatives or paravirt patches,
these were applied first by the module loader before livepatch can apply
per-object relocations. The (buggy) sequence of events was:

(1) Load patch module
(2) Apply alternatives and paravirt patches to patch module
    * Note that these are applied to the new functions in the patch module
(3) Apply per-object relocations to patch module when target module loads.
    * This clobbers what was written in step 2

This lead to crashes and corruption in general, since livepatch would
overwrite or step on previously applied alternative/paravirt patches.
The correct sequence of events should be:

(1) Load patch module
(2) Apply per-object relocations to patch module
(3) Apply alternatives and paravirt patches to patch module

This is fixed by delaying paravirt/alternatives patching until after
relocations are applied. Any .altinstructions or .parainstructions sections in
a patch module are prefixed with ".klp.arch.${objname}" and applied in
arch_klp_init_object_loaded().

Signed-off-by: Jessica Yu <jeyu@redhat.com>
---
 arch/x86/kernel/Makefile    |  1 +
 arch/x86/kernel/livepatch.c | 65 +++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 66 insertions(+)
 create mode 100644 arch/x86/kernel/livepatch.c

diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
index 0503f5b..4f656fe 100644
--- a/arch/x86/kernel/Makefile
+++ b/arch/x86/kernel/Makefile
@@ -83,6 +83,7 @@ obj-$(CONFIG_X86_MPPARSE)	+= mpparse.o
 obj-y				+= apic/
 obj-$(CONFIG_X86_REBOOTFIXUPS)	+= reboot_fixups_32.o
 obj-$(CONFIG_DYNAMIC_FTRACE)	+= ftrace.o
+obj-$(CONFIG_LIVEPATCH)	+= livepatch.o
 obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o
 obj-$(CONFIG_FTRACE_SYSCALLS)	+= ftrace.o
 obj-$(CONFIG_X86_TSC)		+= trace_clock.o
diff --git a/arch/x86/kernel/livepatch.c b/arch/x86/kernel/livepatch.c
new file mode 100644
index 0000000..78209f1
--- /dev/null
+++ b/arch/x86/kernel/livepatch.c
@@ -0,0 +1,65 @@
+/*
+ * livepatch.c - x86-specific Kernel Live Patching Core
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version 2
+ * of the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/module.h>
+#include <linux/kallsyms.h>
+#include <linux/livepatch.h>
+#include <asm/text-patching.h>
+
+/* Apply per-object alternatives. Based on x86 module_finalize() */
+void arch_klp_init_object_loaded(struct klp_patch *patch,
+				 struct klp_object *obj)
+{
+	int cnt;
+	struct klp_modinfo *info;
+	Elf_Shdr *s, *alt = NULL, *para = NULL;
+	void *aseg, *pseg;
+	const char *objname;
+	char sec_objname[MODULE_NAME_LEN];
+	char secname[KSYM_NAME_LEN];
+
+	info = patch->mod->klp_info;
+	objname = obj->name ? obj->name : "vmlinux";
+
+	/* See livepatch core code for BUILD_BUG_ON() explanation */
+	BUILD_BUG_ON(MODULE_NAME_LEN < 56 || KSYM_NAME_LEN != 128);
+
+	for (s = info->sechdrs; s < info->sechdrs + info->hdr.e_shnum; s++) {
+		/* Apply per-object .klp.arch sections */
+		cnt = sscanf(info->secstrings + s->sh_name,
+			     ".klp.arch.%55[^.].%127s",
+			     sec_objname, secname);
+		if (cnt != 2)
+			continue;
+		if (strcmp(sec_objname, objname))
+			continue;
+		if (!strcmp("altinstructions", secname))
+			alt = s;
+		if (!strcmp("parainstructions", secname))
+			para = s;
+	}
+
+	if (alt) {
+		aseg = (void *) alt->sh_addr;
+		apply_alternatives(aseg, aseg + alt->sh_size);
+	}
+
+	if (para) {
+		pseg = (void *) para->sh_addr;
+		apply_paravirt(pseg, pseg + para->sh_size);
+	}
+}
-- 
2.5.5

Re: [PATCH v2 0/2] Fix issue with alternatives/paravirt patches

[Josh Poimboeuf]

On Thu, Jul 21, 2016 at 01:23:58AM -0400, Jessica Yu wrote:
> Hi,
> 
> A few months ago, Chris Arges reported a bug involving alternatives/paravirt
> patching that was discussed here [1] and here [2]. To briefly summarize the
> bug, patch modules that contained .altinstructions or .parainstructions
> sections would break because these alternative/paravirt patches would be
> applied first by the module loader (see x86 module_finalize()), then
> livepatch would later clobber these patches when applying per-object
> relocations. This lead to crashes and unpredictable behavior.
> 
> One conclusion we reached from our last discussion was that we will
> need to introduce some arch-specific code to address this problem.
> This patchset presents a possible fix for the bug by adding a new
> arch-specific arch_klp_init_object_loaded() function that by default
> does nothing but can be overridden by different arches.
> 
> To fix this issue for x86, since we can access a patch module's Elf
> sections through mod->klp_info, we can simply delay the calls to
> apply_paravirt() and apply_alternatives() to arch_klp_init_object_loaded(),
> which is called after relocations have been written for an object.
> In addition, for patch modules, .parainstructions and .altinstructions are
> prefixed by ".klp.arch.${objname}" so that the module loader ignores them
> and livepatch can apply them manually.
> 
> Currently for kpatch, we don't support including jump table sections in
> the patch module, and supporting .smp_locks is currently broken, so we
> don't consider those sections (for now).
> 
> I did some light testing with some patches to kvm and verified that the
> original issue reported in [2] was fixed.
> 
> Based on linux-next.

For the series:

Acked-by: Josh Poimboeuf <jpoimboe@redhat.com>

-- 
Josh

Re: [PATCH v2 1/2] livepatch: use arch_klp_init_object_loaded() to finish arch-specific tasks

[Miroslav Benes]

On Thu, 21 Jul 2016, Jessica Yu wrote:

> Introduce arch_klp_init_object_loaded() to complete any additional
> arch-specific tasks during patching. Architecture code may override this
> function.
> 
> Signed-off-by: Jessica Yu <jeyu@redhat.com>
> ---
>  include/linux/livepatch.h |  3 +++
>  kernel/livepatch/core.c   | 12 ++++++++++--
>  2 files changed, 13 insertions(+), 2 deletions(-)
> 
> diff --git a/include/linux/livepatch.h b/include/linux/livepatch.h
> index a93a0b2..9072f04 100644
> --- a/include/linux/livepatch.h
> +++ b/include/linux/livepatch.h
> @@ -116,6 +116,9 @@ int klp_unregister_patch(struct klp_patch *);
>  int klp_enable_patch(struct klp_patch *);
>  int klp_disable_patch(struct klp_patch *);
>  
> +void arch_klp_init_object_loaded(struct klp_patch *patch,
> +				 struct klp_object *obj);
> +
>  /* Called from the module loader during module coming/going states */
>  int klp_module_coming(struct module *mod);
>  void klp_module_going(struct module *mod);
> diff --git a/kernel/livepatch/core.c b/kernel/livepatch/core.c
> index 5c2bc10..164eff6 100644
> --- a/kernel/livepatch/core.c
> +++ b/kernel/livepatch/core.c
> @@ -274,7 +274,6 @@ static int klp_write_object_relocations(struct module *pmod,
>  
>  	objname = klp_is_module(obj) ? obj->name : "vmlinux";
>  
> -	module_disable_ro(pmod);
>  	/* For each klp relocation section */
>  	for (i = 1; i < pmod->klp_info->hdr.e_shnum; i++) {
>  		sec = pmod->klp_info->sechdrs + i;
> @@ -309,7 +308,6 @@ static int klp_write_object_relocations(struct module *pmod,
>  			break;
>  	}
>  
> -	module_enable_ro(pmod);
>  	return ret;
>  }
>  
> @@ -763,6 +761,12 @@ static int klp_init_func(struct klp_object *obj, struct klp_func *func)
>  				    func->old_sympos ? func->old_sympos : 1);
>  }
>  
> +/* Arches may override this to finish any remaining arch-specific tasks */
> +void __weak arch_klp_init_object_loaded(struct klp_patch *patch,
> +					struct klp_object *obj)
> +{
> +}
> +
>  /* parts of the initialization that is done only when the object is loaded */
>  static int klp_init_object_loaded(struct klp_patch *patch,
>  				  struct klp_object *obj)
> @@ -770,10 +774,14 @@ static int klp_init_object_loaded(struct klp_patch *patch,
>  	struct klp_func *func;
>  	int ret;
>  
> +	module_disable_ro(patch->mod);
>  	ret = klp_write_object_relocations(patch->mod, obj);
>  	if (ret)
>  		return ret;

We need to call module_enable_ro(patch->mod) here before returning back.

Miroslav

>  
> +	arch_klp_init_object_loaded(patch, obj);
> +	module_enable_ro(patch->mod);
> +
>  	klp_for_each_func(obj, func) {
>  		ret = klp_find_object_symbol(obj->name, func->old_name,
>  					     func->old_sympos,
> -- 
> 2.5.5
> 

Re: [PATCH v2 2/2] livepatch/x86: apply alternatives and paravirt patches after relocations

[Miroslav Benes]

On Thu, 21 Jul 2016, Jessica Yu wrote:

> Implement arch_klp_init_object_loaded() for x86, which applies
> alternatives/paravirt patches. This fixes the order in which relocations
> and alternatives/paravirt patches are applied.
> 
> Previously, if a patch module had alternatives or paravirt patches,
> these were applied first by the module loader before livepatch can apply
> per-object relocations. The (buggy) sequence of events was:
> 
> (1) Load patch module
> (2) Apply alternatives and paravirt patches to patch module
>     * Note that these are applied to the new functions in the patch module
> (3) Apply per-object relocations to patch module when target module loads.
>     * This clobbers what was written in step 2
> 
> This lead to crashes and corruption in general, since livepatch would
> overwrite or step on previously applied alternative/paravirt patches.
> The correct sequence of events should be:
> 
> (1) Load patch module
> (2) Apply per-object relocations to patch module
> (3) Apply alternatives and paravirt patches to patch module
> 
> This is fixed by delaying paravirt/alternatives patching until after
> relocations are applied. Any .altinstructions or .parainstructions sections in
> a patch module are prefixed with ".klp.arch.${objname}" and applied in
> arch_klp_init_object_loaded().

Would it make sense to update our documentation and include this info?

> Signed-off-by: Jessica Yu <jeyu@redhat.com>
> ---
>  arch/x86/kernel/Makefile    |  1 +
>  arch/x86/kernel/livepatch.c | 65 +++++++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 66 insertions(+)
>  create mode 100644 arch/x86/kernel/livepatch.c
> 
> diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
> index 0503f5b..4f656fe 100644
> --- a/arch/x86/kernel/Makefile
> +++ b/arch/x86/kernel/Makefile
> @@ -83,6 +83,7 @@ obj-$(CONFIG_X86_MPPARSE)	+= mpparse.o
>  obj-y				+= apic/
>  obj-$(CONFIG_X86_REBOOTFIXUPS)	+= reboot_fixups_32.o
>  obj-$(CONFIG_DYNAMIC_FTRACE)	+= ftrace.o
> +obj-$(CONFIG_LIVEPATCH)	+= livepatch.o
>  obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o
>  obj-$(CONFIG_FTRACE_SYSCALLS)	+= ftrace.o
>  obj-$(CONFIG_X86_TSC)		+= trace_clock.o
> diff --git a/arch/x86/kernel/livepatch.c b/arch/x86/kernel/livepatch.c
> new file mode 100644
> index 0000000..78209f1
> --- /dev/null
> +++ b/arch/x86/kernel/livepatch.c
> @@ -0,0 +1,65 @@
> +/*
> + * livepatch.c - x86-specific Kernel Live Patching Core
> + *
> + * This program is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU General Public License
> + * as published by the Free Software Foundation; either version 2
> + * of the License, or (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program; if not, see <http://www.gnu.org/licenses/>.
> + */
> +
> +#include <linux/module.h>
> +#include <linux/kallsyms.h>
> +#include <linux/livepatch.h>
> +#include <asm/text-patching.h>

It seems I don't have such header file here and it causes a build error. 
Shouldn't it be <asm/alternative.h>?

> +
> +/* Apply per-object alternatives. Based on x86 module_finalize() */
> +void arch_klp_init_object_loaded(struct klp_patch *patch,
> +				 struct klp_object *obj)
> +{
> +	int cnt;
> +	struct klp_modinfo *info;
> +	Elf_Shdr *s, *alt = NULL, *para = NULL;
> +	void *aseg, *pseg;
> +	const char *objname;
> +	char sec_objname[MODULE_NAME_LEN];
> +	char secname[KSYM_NAME_LEN];
> +
> +	info = patch->mod->klp_info;
> +	objname = obj->name ? obj->name : "vmlinux";
> +
> +	/* See livepatch core code for BUILD_BUG_ON() explanation */
> +	BUILD_BUG_ON(MODULE_NAME_LEN < 56 || KSYM_NAME_LEN != 128);
> +
> +	for (s = info->sechdrs; s < info->sechdrs + info->hdr.e_shnum; s++) {
> +		/* Apply per-object .klp.arch sections */
> +		cnt = sscanf(info->secstrings + s->sh_name,
> +			     ".klp.arch.%55[^.].%127s",
> +			     sec_objname, secname);
> +		if (cnt != 2)
> +			continue;
> +		if (strcmp(sec_objname, objname))
> +			continue;
> +		if (!strcmp("altinstructions", secname))
> +			alt = s;
> +		if (!strcmp("parainstructions", secname))
> +			para = s;
> +	}
> +
> +	if (alt) {
> +		aseg = (void *) alt->sh_addr;
> +		apply_alternatives(aseg, aseg + alt->sh_size);
> +	}
> +
> +	if (para) {
> +		pseg = (void *) para->sh_addr;
> +		apply_paravirt(pseg, pseg + para->sh_size);
> +	}
> +}

Otherwise it looks good.

Regards,
Miroslav

Re: [PATCH v2 2/2] livepatch/x86: apply alternatives and paravirt patches after relocations

[Petr Mladek]

On Tue 2016-08-02 10:59:07, Miroslav Benes wrote:
> On Thu, 21 Jul 2016, Jessica Yu wrote:
> 
> > Implement arch_klp_init_object_loaded() for x86, which applies
> > alternatives/paravirt patches. This fixes the order in which relocations
> > and alternatives/paravirt patches are applied.
> > 
> > diff --git a/arch/x86/kernel/livepatch.c b/arch/x86/kernel/livepatch.c
> > new file mode 100644
> > index 0000000..78209f1
> > --- /dev/null
> > +++ b/arch/x86/kernel/livepatch.c
> > +#include <linux/module.h>
> > +#include <linux/kallsyms.h>
> > +#include <linux/livepatch.h>
> > +#include <asm/text-patching.h>
> 
> It seems I don't have such header file here and it causes a build error. 
> Shouldn't it be <asm/alternative.h>?

It is in linux-next, see 35de5b0692aaa ("x86/asm: Stop depending on
ptrace.h in alternative.h").

I agree with the other Mirek's comments. Otherwise, it looks good.

Thanks for working on it,
Petr

Re: [PATCH v2 2/2] livepatch/x86: apply alternatives and paravirt patches after relocations

[Jiri Kosina]

On Tue, 2 Aug 2016, Miroslav Benes wrote:

> > +#include <linux/module.h>
> > +#include <linux/kallsyms.h>
> > +#include <linux/livepatch.h>
> > +#include <asm/text-patching.h>
> 
> It seems I don't have such header file here and it causes a build error. 
> Shouldn't it be <asm/alternative.h>?

I've now refreshed our for-next branch so that it's based on v4.7 to allow 
for it being used as a up-to-date basis for development.

(there have been no KLP patches this merge window, which made for-next 
also quite silent).

Thanks,

-- 
Jiri Kosina
SUSE Labs