Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1440630 > unrolled thread

Re: [PATCH v2 2/2] arm64: implement live patching

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH v2 2/2] arm64: implement live patching Miroslav Benes <mbenes@suse.cz> - 2016-07-11 16:10 +0200
    Re: arm64: implement live patching Jessica Yu <jeyu@redhat.com> - 2016-07-12 00:00 +0200
      Re: arm64: implement live patching Miroslav Benes <mbenes@suse.cz> - 2016-07-12 11:50 +0200
        [PATCH] arm64: take SHN_LIVEPATCH syms into account when calculating  plt_max_entries Jessica Yu <jeyu@redhat.com> - 2016-07-13 02:20 +0200

#1440630 — Re: [PATCH v2 2/2] arm64: implement live patching

On Mon, 27 Jun 2016, Torsten Duwe wrote:

> diff --git a/arch/arm64/include/asm/livepatch.h b/arch/arm64/include/asm/livepatch.h
> new file mode 100644
> index 0000000..6b9a3d1
> --- /dev/null
> +++ b/arch/arm64/include/asm/livepatch.h
> @@ -0,0 +1,37 @@
> +/*
> + * livepatch.h - arm64-specific Kernel Live Patching Core
> + *
> + * Copyright (C) 2016 SUSE
> + *
> + * 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/>.
> + */
> +#ifndef _ASM_ARM64_LIVEPATCH_H
> +#define _ASM_ARM64_LIVEPATCH_H
> +
> +#include <linux/module.h>
> +#include <linux/ftrace.h>
> +
> +#ifdef CONFIG_LIVEPATCH

A nit but we removed such guards in the other header files.

> +static inline int klp_check_compiler_support(void)
> +{
> +	return 0;
> +}
> +
> +static inline void klp_arch_set_pc(struct pt_regs *regs, unsigned long ip)
> +{
> +	regs->pc = ip;
> +}
> +#endif /* CONFIG_LIVEPATCH */

I also checked mod_arch_specific structure because of the way we deal 
with relocations. It is defined only if CONFIG_ARM64_MODULE_PLTS is 
enabled and there is a pointer to 'struct elf64_shdr' called plt. It is 
used indirectly in apply_relocate_add() so we need it to stay. However it 
points to an existing Elf section and SHF_ALLOC is added to its sh_flags 
in module_frob_arch_sections() (arch/arm64/kernel/module-plts.c). 
Therefore we should be ok.

Jessica, could you check it as well, please?

Thanks,
Miroslav

[toc] | [next] | [standalone]

#1440926 — Re: arm64: implement live patching

+++ Miroslav Benes [11/07/16 16:03 +0200]:
>On Mon, 27 Jun 2016, Torsten Duwe wrote:
>
>> diff --git a/arch/arm64/include/asm/livepatch.h b/arch/arm64/include/asm/livepatch.h
>> new file mode 100644
>> index 0000000..6b9a3d1
>> --- /dev/null
>> +++ b/arch/arm64/include/asm/livepatch.h
>> @@ -0,0 +1,37 @@
>> +/*
>> + * livepatch.h - arm64-specific Kernel Live Patching Core
>> + *
>> + * Copyright (C) 2016 SUSE
>> + *
>> + * 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/>.
>> + */
>> +#ifndef _ASM_ARM64_LIVEPATCH_H
>> +#define _ASM_ARM64_LIVEPATCH_H
>> +
>> +#include <linux/module.h>
>> +#include <linux/ftrace.h>
>> +
>> +#ifdef CONFIG_LIVEPATCH
>
>A nit but we removed such guards in the other header files.
>
>> +static inline int klp_check_compiler_support(void)
>> +{
>> +	return 0;
>> +}
>> +
>> +static inline void klp_arch_set_pc(struct pt_regs *regs, unsigned long ip)
>> +{
>> +	regs->pc = ip;
>> +}
>> +#endif /* CONFIG_LIVEPATCH */
>
>I also checked mod_arch_specific structure because of the way we deal
>with relocations. It is defined only if CONFIG_ARM64_MODULE_PLTS is
>enabled and there is a pointer to 'struct elf64_shdr' called plt. It is
>used indirectly in apply_relocate_add() so we need it to stay. However it
>points to an existing Elf section and SHF_ALLOC is added to its sh_flags
>in module_frob_arch_sections() (arch/arm64/kernel/module-plts.c).
>Therefore we should be ok.
>
>Jessica, could you check it as well, please?

That sounds right, the plt will remain in module core memory, so we
are fine there.

However I think the plt->sh_size calculation will be incorrect for
livepatch modules. In calculating mod->arch.plt_max_entries (see:
module-plts.c), count_plts() is called for every rela section.
For livepatch modules, this means count_plts() will also be called for
our .klp.rela sections, which is correct behavior. However,
count_plts() only considers relas referring to SHN_UNDEF symbols, and
since every rela in a klp rela section refers to a SHN_LIVEPATCH
symbol, these are all ignored. So count_plts() may return an incorrect
value for a klp rela section.

Miroslav, can you confirm the issue? I think the fix would be easy
though; we can just add an additional check for SHN_LIVEPATCH in
count_plts(). 

Jessica

[toc] | [prev] | [next] | [standalone]

#1441185 — Re: arm64: implement live patching

On Mon, 11 Jul 2016, Jessica Yu wrote:

> +++ Miroslav Benes [11/07/16 16:03 +0200]:
> > On Mon, 27 Jun 2016, Torsten Duwe wrote:
> > 
> > > diff --git a/arch/arm64/include/asm/livepatch.h
> > > b/arch/arm64/include/asm/livepatch.h
> > > new file mode 100644
> > > index 0000000..6b9a3d1
> > > --- /dev/null
> > > +++ b/arch/arm64/include/asm/livepatch.h
> > > @@ -0,0 +1,37 @@
> > > +/*
> > > + * livepatch.h - arm64-specific Kernel Live Patching Core
> > > + *
> > > + * Copyright (C) 2016 SUSE
> > > + *
> > > + * 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/>.
> > > + */
> > > +#ifndef _ASM_ARM64_LIVEPATCH_H
> > > +#define _ASM_ARM64_LIVEPATCH_H
> > > +
> > > +#include <linux/module.h>
> > > +#include <linux/ftrace.h>
> > > +
> > > +#ifdef CONFIG_LIVEPATCH
> > 
> > A nit but we removed such guards in the other header files.
> > 
> > > +static inline int klp_check_compiler_support(void)
> > > +{
> > > +	return 0;
> > > +}
> > > +
> > > +static inline void klp_arch_set_pc(struct pt_regs *regs, unsigned long
> > > ip)
> > > +{
> > > +	regs->pc = ip;
> > > +}
> > > +#endif /* CONFIG_LIVEPATCH */
> > 
> > I also checked mod_arch_specific structure because of the way we deal
> > with relocations. It is defined only if CONFIG_ARM64_MODULE_PLTS is
> > enabled and there is a pointer to 'struct elf64_shdr' called plt. It is
> > used indirectly in apply_relocate_add() so we need it to stay. However it
> > points to an existing Elf section and SHF_ALLOC is added to its sh_flags
> > in module_frob_arch_sections() (arch/arm64/kernel/module-plts.c).
> > Therefore we should be ok.
> > 
> > Jessica, could you check it as well, please?
> 
> That sounds right, the plt will remain in module core memory, so we
> are fine there.
> 
> However I think the plt->sh_size calculation will be incorrect for
> livepatch modules. In calculating mod->arch.plt_max_entries (see:
> module-plts.c), count_plts() is called for every rela section.
> For livepatch modules, this means count_plts() will also be called for
> our .klp.rela sections, which is correct behavior. However,
> count_plts() only considers relas referring to SHN_UNDEF symbols, and
> since every rela in a klp rela section refers to a SHN_LIVEPATCH
> symbol, these are all ignored. So count_plts() may return an incorrect
> value for a klp rela section.

You're right. During the patch module creation we basically transform all 
SHN_UNDEF relas to SHN_LIVEPATCH, right? We must take it into account 
here.
 
> Miroslav, can you confirm the issue? I think the fix would be easy
> though; we can just add an additional check for SHN_LIVEPATCH in
> count_plts(). 

Yes, such a check should be sufficient.

Thanks for looking into it.
Miroslav

[toc] | [prev] | [next] | [standalone]

#1441929 — [PATCH] arm64: take SHN_LIVEPATCH syms into account when calculating plt_max_entries

SHN_LIVEPATCH symbols are technically a subset of SHN_UNDEF/undefined
symbols, except that their addresses are resolved by livepatch at runtime.
Therefore, when calculating the upper-bound for the number of plt entries
to allocate, make sure to take livepatch symbols into account as well.

Signed-off-by: Jessica Yu <jeyu@redhat.com>
---
 arch/arm64/kernel/module-plts.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/kernel/module-plts.c b/arch/arm64/kernel/module-plts.c
index 1ce90d8..1e95dc1 100644
--- a/arch/arm64/kernel/module-plts.c
+++ b/arch/arm64/kernel/module-plts.c
@@ -122,7 +122,8 @@ static unsigned int count_plts(Elf64_Sym *syms, Elf64_Rela *rela, int num)
                         * as well, so modules can never grow beyond that limit.
                         */
                        s = syms + ELF64_R_SYM(rela[i].r_info);
-                       if (s->st_shndx != SHN_UNDEF)
+                       if (s->st_shndx != SHN_UNDEF &&
+                           s->st_shndx != SHN_LIVEPATCH)
                                break;

                        /*
-- 
2.5.5

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web