mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Qi Xi <xiqi2@huawei.com>
To: <bobo.shaobowang@huawei.com>, <xiexiuqi@huawei.com>,
	<arnd@arndb.de>, <masahiroy@kernel.org>, <kuba@kernel.org>,
	<edumazet@google.com>, <linux-arch@vger.kernel.org>
Cc: <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH v3] once: fix race by moving DO_ONCE to separate section
Date: Thu, 25 Sep 2025 09:49:44 +0800	[thread overview]
Message-ID: <2fc167c7-9d32-4869-a0be-cd17e6701248@huawei.com> (raw)
In-Reply-To: <5c88f8c4-c06b-4eb9-8883-202d15130938@huawei.com>

kindly ping

On 09/09/2025 20:01, Qi Xi wrote:
>
> On 09/09/2025 19:29, Qi Xi wrote:
>> The commit c2c60ea37e5b ("once: use __section(".data.once")") moved
>> DO_ONCE's ___done variable to .data.once section, which conflicts with
>> DO_ONCE_LITE() that also uses the same section.
>>
>> This creates a race condition when clear_warn_once is used:
>>
>> Thread 1 (DO_ONCE)             Thread 2 (DO_ONCE)
>> __do_once_start
>>      read ___done (false)
>>      acquire once_lock
>> execute func
>> __do_once_done
>>      write ___done (true)      __do_once_start
>>      release once_lock             // Thread 3 clear_warn_once reset 
>> ___done
>>                                    read ___done (false)
>>                                    acquire once_lock
>>                                execute func
>> schedule once_work            __do_once_done
>> once_deferred: OK             write ___done (true)
>> static_branch_disable         release once_lock
>>                                schedule once_work
>>                                once_deferred:
>>                                    BUG_ON(!static_key_enabled)
> When simulating concurrent execution between DO_ONCE() and
> clear_warn_once (clears the entire .data..once section in 
> __do_once_done()),
> BUG_ON() can be easily reproduced.
>
> +#include <asm/sections.h>
>  void __do_once_done(bool *done, struct static_key_true *once_key,
>                     unsigned long *flags, struct module *mod)
>         __releases(once_lock)
>  {
>         *done = true;
>         spin_unlock_irqrestore(&once_lock, *flags);
> +       memset(__start_once, 0, __end_once - __start_once);
> +       pr_info("__do_once_done done: %d\n", *done); // *done equals 0
>         once_disable_jump(once_key, mod);
>  }
>
>>
>> DO_ONCE_LITE() in once_lite.h is used by WARN_ON_ONCE() and other 
>> warning
>> macros. Keep its ___done flag in the .data..once section and allow 
>> resetting
>> by clear_warn_once, as originally intended.
>>
>> In contrast, DO_ONCE() is used for functions like get_random_once() and
>> relies on its ___done flag for internal synchronization. We should 
>> not reset
>> DO_ONCE() by clear_warn_once.
>>
>> Fix it by isolating DO_ONCE's ___done into a separate .data..do_once 
>> section,
>> shielding it from clear_warn_once.
>>
>> Fixes: c2c60ea37e5b ("once: use __section(".data.once")")
>> Reported-by: Hulk Robot <hulkci@huawei.com>
>> Signed-off-by: Qi Xi <xiqi2@huawei.com>
>> ---
>> v3 -> v2: apply the same section change to DO_ONCE_SLEEPABLE().
>> v2 -> v1: add comments for DO_ONCE_LITE() and DO_ONCE().
>> ---
>>   include/asm-generic/vmlinux.lds.h | 1 +
>>   include/linux/once.h              | 4 ++--
>>   2 files changed, 3 insertions(+), 2 deletions(-)
>>
>> diff --git a/include/asm-generic/vmlinux.lds.h 
>> b/include/asm-generic/vmlinux.lds.h
>> index 883dbac79da9..94850b52e5cc 100644
>> --- a/include/asm-generic/vmlinux.lds.h
>> +++ b/include/asm-generic/vmlinux.lds.h
>> @@ -384,6 +384,7 @@
>>       __start_once = .;                        \
>>       *(.data..once)                            \
>>       __end_once = .;                            \
>> +    *(.data..do_once)                        \
>>       STRUCT_ALIGN();                            \
>>       *(__tracepoints)                        \
>>       /* implement dynamic printk debug */                \
>> diff --git a/include/linux/once.h b/include/linux/once.h
>> index 30346fcdc799..449a0e34ad5a 100644
>> --- a/include/linux/once.h
>> +++ b/include/linux/once.h
>> @@ -46,7 +46,7 @@ void __do_once_sleepable_done(bool *done, struct 
>> static_key_true *once_key,
>>   #define DO_ONCE(func, ...)                             \
>>       ({                                     \
>>           bool ___ret = false;                         \
>> -        static bool __section(".data..once") ___done = false;         \
>> +        static bool __section(".data..do_once") ___done = false;     \
>>           static DEFINE_STATIC_KEY_TRUE(___once_key);             \
>>           if (static_branch_unlikely(&___once_key)) {             \
>>               unsigned long ___flags;                     \
>> @@ -64,7 +64,7 @@ void __do_once_sleepable_done(bool *done, struct 
>> static_key_true *once_key,
>>   #define DO_ONCE_SLEEPABLE(func, ...)                        \
>>       ({                                    \
>>           bool ___ret = false;                        \
>> -        static bool __section(".data..once") ___done = false;        \
>> +        static bool __section(".data..do_once") ___done = false;    \
>>           static DEFINE_STATIC_KEY_TRUE(___once_key); \
>>           if (static_branch_unlikely(&___once_key)) {            \
>>               ___ret = __do_once_sleepable_start(&___done);        \

  reply	other threads:[~2025-09-25  1:49 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-09-09 11:29 Qi Xi
2025-09-09 12:01 ` Qi Xi
2025-09-25  1:49   ` Qi Xi [this message]
2025-09-25  6:14 ` Arnd Bergmann
2025-09-26  0:19   ` Nathan Chancellor

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=2fc167c7-9d32-4869-a0be-cd17e6701248@huawei.com \
    --to=xiqi2@huawei.com \
    --cc=arnd@arndb.de \
    --cc=bobo.shaobowang@huawei.com \
    --cc=edumazet@google.com \
    --cc=kuba@kernel.org \
    --cc=linux-arch@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=masahiroy@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

all inboxes | Powered by JetHome®