All of lore.kernel.org
 help / color / mirror / Atom feed
From: Petr Mladek <pmladek-IBi9RG/b67k@public.gmane.org>
To: Jessica Yu <jeyu-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
Cc: Rusty Russell <rusty-8n+1lVoiYb80n/F98K4Iww@public.gmane.org>,
	Josh Poimboeuf <jpoimboe-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>,
	Seth Jennings <sjenning-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>,
	Jiri Kosina <jikos-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>,
	Vojtech Pavlik <vojtech-IBi9RG/b67k@public.gmane.org>,
	Jonathan Corbet <corbet-T1hC0tSOHrs@public.gmane.org>,
	Miroslav Benes <mbenes-AlSwsSmVLrQ@public.gmane.org>,
	linux-api-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
	live-patching-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
	x86-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org,
	linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
	linux-s390-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
	linux-doc-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
Subject: Re: [RFC PATCH v4 2/6] module: preserve Elf information for livepatch modules
Date: Tue, 9 Feb 2016 09:44:43 +0100	[thread overview]
Message-ID: <20160209084443.GA12548@pathway.suse.cz> (raw)
In-Reply-To: <1454548271-24923-3-git-send-email-jeyu-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>

On Wed 2016-02-03 20:11:07, Jessica Yu wrote:
> For livepatch modules, copy Elf section, symbol, and string information
> from the load_info struct in the module loader. Persist copies of the
> original symbol table and string table.
> 
> Livepatch manages its own relocation sections in order to reuse module
> loader code to write relocations. Livepatch modules must preserve Elf
> information such as section indices in order to apply livepatch relocation
> sections using the module loader's apply_relocate_add() function.
> 
> In order to apply livepatch relocation sections, livepatch modules must
> keep a complete copy of their original symbol table in memory. Normally, a
> stripped down copy of a module's symbol table (containing only "core"
> symbols) is made available through module->core_symtab. But for livepatch
> modules, the symbol table copied into memory on module load must be exactly
> the same as the symbol table produced when the patch module was compiled.
> This is because the relocations in each livepatch relocation section refer
> to their respective symbols with their symbol indices, and the original
> symbol indices (and thus the symtab ordering) must be preserved in order
> for apply_relocate_add() to find the right symbol.
> 
> diff --git a/kernel/module.c b/kernel/module.c
> index 71c77ed..9c16eb2 100644
> --- a/kernel/module.c
> +++ b/kernel/module.c
> @@ -3222,6 +3331,12 @@ static noinline int do_init_module(struct module *mod)
>  	 */
>  	current->flags &= ~PF_USED_ASYNC;
>  
> +#ifdef CONFIG_KALLSYMS
> +	/* Make symtab and strtab available prior to module init call */
> +	mod->num_symtab = mod->core_num_syms;
> +	mod->symtab = mod->core_symtab;
> +	mod->strtab = mod->core_strtab;
> +#endif

This should be done with module_mutex. Otherwise, it looks racy
at least against module_kallsyms_on_each_symbol().

BTW: I wonder why even the original code is not racy
for example against module_get_kallsym. It is called
without the mutex. This code sets the number of entries
before the pointer to the entries.

Note that the module is in the list even in the UNFORMED state.


>  	do_mod_ctors(mod);
>  	/* Start the module */
>  	if (mod->init != NULL)
> @@ -3266,11 +3381,6 @@ static noinline int do_init_module(struct module *mod)
>  	/* Drop initial reference. */
>  	module_put(mod);
>  	trim_init_extable(mod);
> -#ifdef CONFIG_KALLSYMS
> -	mod->num_symtab = mod->core_num_syms;
> -	mod->symtab = mod->core_symtab;
> -	mod->strtab = mod->core_strtab;
> -#endif
>  	mod_tree_remove_init(mod);
>  	disable_ro_nx(&mod->init_layout);
>  	module_arch_freeing_init(mod);

In each case, it was called with the mutex here.

Best Regards,
Petr

WARNING: multiple messages have this Message-ID (diff)
From: Petr Mladek <pmladek@suse.com>
To: Jessica Yu <jeyu@redhat.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>,
	Josh Poimboeuf <jpoimboe@redhat.com>,
	Seth Jennings <sjenning@redhat.com>,
	Jiri Kosina <jikos@kernel.org>, Vojtech Pavlik <vojtech@suse.com>,
	Jonathan Corbet <corbet@lwn.net>, Miroslav Benes <mbenes@suse.cz>,
	linux-api@vger.kernel.org, live-patching@vger.kernel.org,
	x86@kernel.org, linux-kernel@vger.kernel.org,
	linux-s390@vger.kernel.org, linux-doc@vger.kernel.org
Subject: Re: [RFC PATCH v4 2/6] module: preserve Elf information for livepatch modules
Date: Tue, 9 Feb 2016 09:44:43 +0100	[thread overview]
Message-ID: <20160209084443.GA12548@pathway.suse.cz> (raw)
In-Reply-To: <1454548271-24923-3-git-send-email-jeyu@redhat.com>

On Wed 2016-02-03 20:11:07, Jessica Yu wrote:
> For livepatch modules, copy Elf section, symbol, and string information
> from the load_info struct in the module loader. Persist copies of the
> original symbol table and string table.
> 
> Livepatch manages its own relocation sections in order to reuse module
> loader code to write relocations. Livepatch modules must preserve Elf
> information such as section indices in order to apply livepatch relocation
> sections using the module loader's apply_relocate_add() function.
> 
> In order to apply livepatch relocation sections, livepatch modules must
> keep a complete copy of their original symbol table in memory. Normally, a
> stripped down copy of a module's symbol table (containing only "core"
> symbols) is made available through module->core_symtab. But for livepatch
> modules, the symbol table copied into memory on module load must be exactly
> the same as the symbol table produced when the patch module was compiled.
> This is because the relocations in each livepatch relocation section refer
> to their respective symbols with their symbol indices, and the original
> symbol indices (and thus the symtab ordering) must be preserved in order
> for apply_relocate_add() to find the right symbol.
> 
> diff --git a/kernel/module.c b/kernel/module.c
> index 71c77ed..9c16eb2 100644
> --- a/kernel/module.c
> +++ b/kernel/module.c
> @@ -3222,6 +3331,12 @@ static noinline int do_init_module(struct module *mod)
>  	 */
>  	current->flags &= ~PF_USED_ASYNC;
>  
> +#ifdef CONFIG_KALLSYMS
> +	/* Make symtab and strtab available prior to module init call */
> +	mod->num_symtab = mod->core_num_syms;
> +	mod->symtab = mod->core_symtab;
> +	mod->strtab = mod->core_strtab;
> +#endif

This should be done with module_mutex. Otherwise, it looks racy
at least against module_kallsyms_on_each_symbol().

BTW: I wonder why even the original code is not racy
for example against module_get_kallsym. It is called
without the mutex. This code sets the number of entries
before the pointer to the entries.

Note that the module is in the list even in the UNFORMED state.


>  	do_mod_ctors(mod);
>  	/* Start the module */
>  	if (mod->init != NULL)
> @@ -3266,11 +3381,6 @@ static noinline int do_init_module(struct module *mod)
>  	/* Drop initial reference. */
>  	module_put(mod);
>  	trim_init_extable(mod);
> -#ifdef CONFIG_KALLSYMS
> -	mod->num_symtab = mod->core_num_syms;
> -	mod->symtab = mod->core_symtab;
> -	mod->strtab = mod->core_strtab;
> -#endif
>  	mod_tree_remove_init(mod);
>  	disable_ro_nx(&mod->init_layout);
>  	module_arch_freeing_init(mod);

In each case, it was called with the mutex here.

Best Regards,
Petr

  parent reply	other threads:[~2016-02-09  8:44 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-02-04  1:11 [RFC PATCH v4 0/6] (mostly) Arch-independent livepatch Jessica Yu
2016-02-04  1:11 ` Jessica Yu
     [not found] ` <1454548271-24923-1-git-send-email-jeyu-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2016-02-04  1:11   ` [RFC PATCH v4 1/6] Elf: add livepatch-specific Elf constants Jessica Yu
2016-02-04  1:11     ` Jessica Yu
2016-02-04  1:11 ` [RFC PATCH v4 2/6] module: preserve Elf information for livepatch modules Jessica Yu
2016-02-08 20:10   ` Josh Poimboeuf
     [not found]     ` <20160208201039.GC23106-8wJ5/zUtDR0XGNroddHbYwC/G2K4zDHf@public.gmane.org>
2016-02-08 20:34       ` Jessica Yu
2016-02-08 20:34         ` Jessica Yu
     [not found]   ` <1454548271-24923-3-git-send-email-jeyu-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2016-02-09  8:44     ` Petr Mladek [this message]
2016-02-09  8:44       ` [RFC PATCH v4 2/6] " Petr Mladek
2016-02-09 10:33       ` Jiri Kosina
     [not found]         ` <alpine.LNX.2.00.1602091131510.22727-YHPUNQjx9ReKbouaWp301Q@public.gmane.org>
2016-02-09 12:31           ` Petr Mladek
2016-02-09 12:31             ` Petr Mladek
2016-02-10  0:18             ` Rusty Russell
2016-02-10 15:53   ` Petr Mladek
2016-02-04  1:11 ` [RFC PATCH v4 3/6] module: s390: keep mod_arch_specific " Jessica Yu
2016-02-04  1:37   ` Jessica Yu
2016-02-04 21:03     ` Josh Poimboeuf
2016-02-05 15:32       ` Miroslav Benes
2016-02-04  1:11 ` [RFC PATCH v4 4/6] livepatch: reuse module loader code to write relocations Jessica Yu
2016-02-08 15:05   ` Miroslav Benes
2016-02-09 13:32     ` Miroslav Benes
     [not found]   ` <1454548271-24923-5-git-send-email-jeyu-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2016-02-08 20:26     ` Josh Poimboeuf
2016-02-08 20:26       ` Josh Poimboeuf
     [not found]       ` <20160208202606.GD23106-8wJ5/zUtDR0XGNroddHbYwC/G2K4zDHf@public.gmane.org>
2016-02-10  0:56         ` Jessica Yu
2016-02-10  0:56           ` Jessica Yu
2016-02-09 14:01     ` [RFC PATCH v4 4/6] " Petr Mladek
2016-02-09 14:01       ` Petr Mladek
2016-02-09 15:57       ` Miroslav Benes
2016-02-10  1:21       ` Jessica Yu
2016-02-04  1:11 ` [RFC PATCH v4 5/6] samples: livepatch: mark as livepatch module Jessica Yu
2016-02-04  1:11 ` [RFC PATCH v4 6/6] Documentation: livepatch: outline Elf format and requirements for patch modules Jessica Yu
2016-02-08 14:54 ` [RFC PATCH v4 0/6] (mostly) Arch-independent livepatch Miroslav Benes
     [not found]   ` <alpine.LNX.2.00.1602081548070.12964-ztGlSCb7Y1iN3ZZ/Hiejyg@public.gmane.org>
2016-02-08 20:28     ` Josh Poimboeuf
2016-02-08 20:28       ` Josh Poimboeuf
2016-02-09 15:56 ` Petr Mladek

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20160209084443.GA12548@pathway.suse.cz \
    --to=pmladek-ibi9rg/b67k@public.gmane.org \
    --cc=corbet-T1hC0tSOHrs@public.gmane.org \
    --cc=jeyu-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org \
    --cc=jikos-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org \
    --cc=jpoimboe-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org \
    --cc=linux-api-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
    --cc=linux-doc-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
    --cc=linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
    --cc=linux-s390-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
    --cc=live-patching-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
    --cc=mbenes-AlSwsSmVLrQ@public.gmane.org \
    --cc=rusty-8n+1lVoiYb80n/F98K4Iww@public.gmane.org \
    --cc=sjenning-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org \
    --cc=vojtech-IBi9RG/b67k@public.gmane.org \
    --cc=x86-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.