From: Miroslav Benes <mbenes@suse.cz>
To: Cheng Jian <cj.chengjian@huawei.com>
Cc: linux-kernel@vger.kernel.org, live-patching@vger.kernel.org,
chenwandun@huawei.com, xiexiuqi@huawei.com,
bobo.shaobowang@huawei.com, huawei.libin@huawei.com,
jeyu@kernel.org, jikos@kernel.org
Subject: Re: [PATCH] module: make module symbols visible after init
Date: Wed, 3 Jun 2020 19:00:10 +0200 (CEST) [thread overview]
Message-ID: <alpine.LSU.2.21.2006031848020.26737@pobox.suse.cz> (raw)
In-Reply-To: <20200603141200.17745-1-cj.chengjian@huawei.com>
Hi,
I'm confused...
On Wed, 3 Jun 2020, Cheng Jian wrote:
> When lookup the symbols of module by module_kallsyms_lookup_name(),
> the symbols address is visible only if the module's status isn't
> MODULE_STATE_UNFORMED, This is problematic.
>
> When complete_formation is done, the state of the module is modified
> to MODULE_STATE_COMING, and the symbol of module is visible to the
> outside.
>
> At this time, the init function of the module has not been called,
> so if the address of the function symbol has been found and called,
> it may cause some exceptions.
>
> For livepatch module, the relocation information of the livepatch
> module is completed in init by klp_write_object_relocations(), and
> the symbol name of the old and new functions are the same. Therefore,
> when we lookup the symbol, we may get the function address of the
> livepatch module. a crash can occurs when we call this function.
>
> CPU 0 CPU 1
> ==================================================
> load_module
> add_unformed_module # MODULE_STATE_UNFORMED;
> post_relocation
> complete_formation # MODULE_STATE_COMING;
> ------------------
> module_kallsymc_lookup_name("A")
> call A() # CRASH
> ------------------
> do_init_module
> klp_write_object_relocations
> mod->state = MODULE_STATE_LIVE;
We don't call module_kallsymc_lookup_name() anywhere in livepatch if I am
not missing something. So is this your code? Then I could see the problem.
You get the address of a function from a livepatch module and call it,
which is not correct.
I see two options...
1. don't use the same name for the new function. Use some kind of prefix.
It is more bulletproof anyway.
2. module_kallsyms_lookup_name() accepts a module name as a prefix of a
symbol. So you can use module_kallsyms_lookup_name("module:A") and it
should return A from that particular module only (if it exists).
> In commit 0bd476e6c671 ("kallsyms: unexport kallsyms_lookup_name() and
> kallsyms_on_each_symbol()") restricts the invocation for kernel unexported
> symbols, but it is still incorrect to make the symbols of non-LIVE modules
> visible to the outside.
Why? It could easily break something somewhere. I didn't check properly,
but module states are not safe to play with, so I'd be conservative here.
> Signed-off-by: Cheng Jian <cj.chengjian@huawei.com>
> ---
> kernel/module.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/kernel/module.c b/kernel/module.c
> index 64a2b4daaaa5..96c9cb64de57 100644
> --- a/kernel/module.c
> +++ b/kernel/module.c
> @@ -4220,7 +4220,7 @@ unsigned long module_kallsyms_lookup_name(const char *name)
> ret = find_kallsyms_symbol_value(mod, colon+1);
> } else {
> list_for_each_entry_rcu(mod, &modules, list) {
> - if (mod->state == MODULE_STATE_UNFORMED)
> + if (mod->state != MODULE_STATE_LIVE)
> continue;
> if ((ret = find_kallsyms_symbol_value(mod, name)) != 0)
> break;
Thanks,
Miroslav
next prev parent reply other threads:[~2020-06-03 17:00 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-06-03 14:12 [PATCH] module: make module symbols visible after init Cheng Jian
2020-06-03 17:00 ` Miroslav Benes [this message]
2020-06-04 9:49 ` Jessica Yu
2020-06-04 8:57 ` Petr Mladek
2020-06-04 12:55 ` chengjian (D)
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=alpine.LSU.2.21.2006031848020.26737@pobox.suse.cz \
--to=mbenes@suse.cz \
--cc=bobo.shaobowang@huawei.com \
--cc=chenwandun@huawei.com \
--cc=cj.chengjian@huawei.com \
--cc=huawei.libin@huawei.com \
--cc=jeyu@kernel.org \
--cc=jikos@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=live-patching@vger.kernel.org \
--cc=xiexiuqi@huawei.com \
/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 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).