Live Patching
 help / color / mirror / Atom feed
From: Petr Mladek <pmladek@suse.com>
To: Zhen Lei <thunder.leizhen@huawei.com>
Cc: Josh Poimboeuf <jpoimboe@kernel.org>,
	Jiri Kosina <jikos@kernel.org>, Miroslav Benes <mbenes@suse.cz>,
	Joe Lawrence <joe.lawrence@redhat.com>,
	live-patching@vger.kernel.org, linux-kernel@vger.kernel.org,
	Masahiro Yamada <masahiroy@kernel.org>,
	Alexei Starovoitov <ast@kernel.org>, Jiri Olsa <jolsa@kernel.org>,
	Kees Cook <keescook@chromium.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	Luis Chamberlain <mcgrof@kernel.org>,
	linux-modules@vger.kernel.org
Subject: Re: [PATCH v2 7/8] livepatch: Improve the search performance of module_kallsyms_on_each_symbol()
Date: Tue, 20 Sep 2022 14:08:07 +0200	[thread overview]
Message-ID: <YymtJwYB7Q9mTPgS@alley> (raw)
In-Reply-To: <20220909130016.727-8-thunder.leizhen@huawei.com>

On Fri 2022-09-09 21:00:15, Zhen Lei wrote:
> Currently we traverse all symbols of all modules to find the specified
> function for the specified module. But in reality, we just need to find
> the given module and then traverse all the symbols in it.

I agree that it might be noticeable speedup.

> In order to achieve this purpose, split the call to hook 'fn' into two
> phases:
> 1. Finds the given module. Pass pointer 'mod'. Hook 'fn' directly returns
>    the comparison result of the module name without comparing the function
>    name.
> 2. Finds the given function in that module. Pass pointer 'mod = NULL'.
>    Hook 'fn' skip the comparison of module name and directly compare
>    function names.
> 
> Phase1: mod1-->mod2..(subsequent modules do not need to be compared)
>                 |
> Phase2:          -->f1-->f2-->f3
> 
> Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com>
> ---
>  kernel/livepatch/core.c  |  7 ++-----
>  kernel/module/kallsyms.c | 13 ++++++++++++-
>  2 files changed, 14 insertions(+), 6 deletions(-)
> 
> diff --git a/kernel/livepatch/core.c b/kernel/livepatch/core.c
> index 31b57ccf908017e..98e23137e4133bc 100644
> --- a/kernel/livepatch/core.c
> +++ b/kernel/livepatch/core.c
> @@ -130,15 +130,12 @@ static int klp_find_callback(void *data, const char *name,
>  {
>  	struct klp_find_arg *args = data;
>  
> -	if ((mod && !args->objname) || (!mod && args->objname))
> -		return 0;
> +	if (mod)
> +		return strcmp(args->objname, mod->name);
>  
>  	if (strcmp(args->name, name))
>  		return 0;
>  
> -	if (args->objname && strcmp(args->objname, mod->name))
> -		return 0;
> -
>  	args->addr = addr;
>  	args->count++;
>  
> diff --git a/kernel/module/kallsyms.c b/kernel/module/kallsyms.c
> index f5c5c9175333df7..b033613e6c7e3bb 100644
> --- a/kernel/module/kallsyms.c
> +++ b/kernel/module/kallsyms.c
> @@ -510,6 +510,11 @@ int module_kallsyms_on_each_symbol(int (*fn)(void *, const char *,
>  		if (mod->state == MODULE_STATE_UNFORMED)
>  			continue;
>  
> +		/* check mod->name first */
> +		ret = fn(data, NULL, mod, 0);
> +		if (ret)
> +			continue;

Hmm, it somehow gets too complicated. The same fn() callback has to
behave correctly in 3 different situations. I would suggest to
simplify everything:


1. Pass the requested modname as a parameter to module_kallsyms_on_each_symbol()

/*
 * Iterate over all symbols in the given @modname. For symbols from
 * vmlinux use kallsyms_on_each_symbol() instead.
 */
int module_kallsyms_on_each_symbol(const char *modname,
				   int (*fn)(void *, const char *,
					     struct module *, unsigned long),
				   void *data)

and do here:

		if (strcmp(modname, mod->name))
			continue;


2. We do not even need to pass .objname in struct klp_find_arg
   could simplify the callback:

static int klp_find_callback(void *data, const char *name,
			     struct module *mod, unsigned long addr)
{
	struct klp_find_arg *args = data;

	if (strcmp(args->name, name))
		return 0;

	args->addr = addr;
	args->count++;

	/*
	 * Finish the search when the symbol is found for the desired position
	 * or the position is not defined for a non-unique symbol.
	 */
	if ((args->pos && (args->count == args->pos)) ||
	    (!args->pos && (args->count > 1)))
		return 1;

	return 0;
}

3. As a result the *mod parameter won't be used by any existing
   fn() callback and could be removed. This should be done as
   a separate patch. It touches also ftrace_lookup_symbols().

Best Regards,
Petr

  reply	other threads:[~2022-09-20 12:08 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-09-09 13:00 [PATCH v2 0/8] kallsyms: Optimizes the performance of lookup symbols Zhen Lei
2022-09-09 13:00 ` [PATCH v2 1/8] scripts/kallsyms: don't compress symbol type when CONFIG_KALLSYMS_ALL=y Zhen Lei
2022-09-20 17:26   ` Petr Mladek
2022-09-21  2:42     ` Leizhen (ThunderTown)
2022-09-21  6:51       ` Leizhen (ThunderTown)
2022-09-21  7:13       ` Petr Mladek
2022-09-09 13:00 ` [PATCH v2 2/8] scripts/kallsyms: rename build_initial_tok_table() Zhen Lei
2022-09-09 13:00 ` [PATCH v2 3/8] kallsyms: Adjust the types of some local variables Zhen Lei
2022-09-09 13:00 ` [PATCH v2 4/8] kallsyms: Improve the performance of kallsyms_lookup_name() Zhen Lei
2022-09-09 13:00 ` [PATCH v2 5/8] kallsyms: Add helper kallsyms_on_each_match_symbol() Zhen Lei
2022-09-09 13:00 ` [PATCH v2 6/8] livepatch: Use kallsyms_on_each_match_symbol() to improve performance Zhen Lei
2022-09-09 13:00 ` [PATCH v2 7/8] livepatch: Improve the search performance of module_kallsyms_on_each_symbol() Zhen Lei
2022-09-20 12:08   ` Petr Mladek [this message]
2022-09-20 14:01     ` Leizhen (ThunderTown)
2022-09-21  6:56       ` Petr Mladek
2022-09-09 13:00 ` [PATCH v2 8/8] kallsyms: Add self-test facility Zhen Lei
2022-09-17  8:07   ` Kees Cook
2022-09-17 12:40     ` Leizhen (ThunderTown)
2022-09-16  1:27 ` [PATCH v2 0/8] kallsyms: Optimizes the performance of lookup symbols Leizhen (ThunderTown)
2022-09-16  3:17   ` Leizhen (ThunderTown)

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=YymtJwYB7Q9mTPgS@alley \
    --to=pmladek@suse.com \
    --cc=akpm@linux-foundation.org \
    --cc=ast@kernel.org \
    --cc=jikos@kernel.org \
    --cc=joe.lawrence@redhat.com \
    --cc=jolsa@kernel.org \
    --cc=jpoimboe@kernel.org \
    --cc=keescook@chromium.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-modules@vger.kernel.org \
    --cc=live-patching@vger.kernel.org \
    --cc=masahiroy@kernel.org \
    --cc=mbenes@suse.cz \
    --cc=mcgrof@kernel.org \
    --cc=thunder.leizhen@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