linux-kernel - Re: [Suggestion] about calling debug_hotplug_cpu() which enabled by 'allmodconfig' for a x86

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 03 Oct 2013 22:21:53 +0800
From:	Chen Gang <gang.chen@...anux.com>
To:	Toshi Kani <toshi.kani@...com>
CC:	Thomas Gleixner <tglx@...utronix.de>,
	"mingo@...hat.com" <mingo@...hat.com>,
	"H. Peter Anvin" <hpa@...or.com>,
	the arch/x86 maintainers <x86@...nel.org>,
	"Yu, Fenghua" <fenghua.yu@...el.com>,
	"rafael.j.wysocki@...el.com" <rafael.j.wysocki@...el.com>,
	"isimatu.yasuaki@...fujitsu.com" <isimatu.yasuaki@...fujitsu.com>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [Suggestion] about calling debug_hotplug_cpu() which enabled
 by 'allmodconfig' for a x86_64 dual core laptop.


One root cause found (but not only one), allmodconfig will enable
CONFIG_SYSFS_DEPRECATED which for new kernel on old distribution.


I am comparing from allmodconfig config file to defconfig config file,
the attachments are middle files (one enable CONFIG_SYSFS_DEPRECATED
which can not boot up, the other disable it which can boot up).

The related information in "init/Kconfig"

  config SYSFS_DEPRECATED
    bool "Enable deprecated sysfs features to support old userspace tools"
    depends on SYSFS
    default n
    help
      This option adds code that switches the layout of the "block" class
      devices, to not show up in /sys/class/block/, but only in
      /sys/block/.

      This switch is only active when the sysfs.deprecated=1 boot option is
      passed or the SYSFS_DEPRECATED_V2 option is set.

      This option allows new kernels to run on old distributions and tools,
      which might get confused by /sys/class/block/. Since 2007/2008 all
      major distributions and tools handle this just fine.

      Recent distributions and userspace tools after 2009/2010 depend on
      the existence of /sys/class/block/, and will not work with this
      option enabled.

      Only if you are using a new kernel on an old distribution, you might
      need to say Y here.


I will continue to analyze another related issues for allmodconfig on
my laptop (after disable SYSFS_DEPRECATED, for allmodconfig, it still
can not boot up successfully, although my 'middle' config file can).


BTW: for defconfig on my PC (x86_64 dual core, fedora 17), need enable
"Device Drivers" -> "Generic Driver Options" -> "Maintain a devtmpfs
...", or can not start up. is it suitable to send related patch for it?


Thanks.

On 10/03/2013 08:49 AM, Chen Gang wrote:
> On 10/03/2013 03:33 AM, Toshi Kani wrote:
>> On Thu, 2013-10-03 at 00:41 +0800, Chen Gang F T wrote:
>>> On 10/02/2013 11:28 PM, Toshi Kani wrote:
>>>> On Wed, 2013-10-02 at 03:10 +0000, Chen Gang wrote:
>>>>> Hello Maintainers:
>>>>>
>>>>> Under my x86_64 dual core laptop, I build kernel next-20130927 with
>>>>> 'allmodconfig', and install it, the machine can not start. Related
>>>>> information is:
>>>>>
>>>>>   after call debug_hotplug_cpu(), output "cpu 0 is offline" .... and then "Failed to execute /init".
>>>>>
>>>>> After remove "_debug_hotplug_cpu(0, 0);", can pass the issue (but will
>>>>> fail in another place). I guess, the reason is my laptop cpu is not
>>>>> 'hotplug', but have to call debug_hotplug_cpu() with allmodconfig.
>>>>>
>>>>>
>>>>> I will continue analyzing, welcome any additional suggestions or
>>>>> completions.
>>>>
>>>> allmodconfig sets CONFIG_DEBUG_HOTPLUG_CPU0 to y (which is defined as
>>>> "def_bool n"), which puts CPU0 offline during boot for testing.  This
>>>> debug feature is causing the problem on your laptop.  I am not familiar
>>>> with allmodconfig, but it seems that it enables all the config options,
>>>> preferably with 'm'.
>>>>
>>>
>>> Hmm... excuse me, I don't know: for laptop, if cpu0 is offline, whether
>>> it still can work or not. If any members know about it, please tell me,
>>> thanks.
>>
>> CPU online/offline works on laptops as long as they have more than one
>> CPU.  You can boot with other kernel and test this feature
>> with /sys/devices/system/cpu/cpuX/online on your laptop.  cpu0 is a
>> special case and you may need to specify "cpu0_hotplug" boot option.
>>
> 
> Thank you for your confirmation, "for dual core (although laptop), after
> system boot up, can let cpu0 offline". :-)
> 
>>> In my opinion, if enable DEBUG_HOTPLUG_CPU0, but "should not let cpu0
>>> offline for laptop", we need let 'offline' operation fail.
>>
>> Whether it is a laptop or not should not be a matter here.  But I agree
>> with you that enabling CONFIG_DEBUG_HOTPLUG_CPU0 is not useful (and can
>> be harmful) unless a user really intends to test this feature.
>>
> 
> Hmm... I will continue analyzing. After finish analyzing (let kernel
> boot up OK), I should give a confirmation: "during boot up, can we still
> let cpu0 offline for dual core?".
> 
> In fact, 'allmodconfig' is just a mad user which often does useless
> things to intend to find issues.
> 
> Thanks.
> 
>> Thanks,
>> -Toshi
>>
>>> And I am analyzing (just constructing environments: KVM, kgdb ...),
>>> before let kernel start successfully, only according to my current
>>> proofs, we can not say "let laptop cpu0 offline" must be an issue.
>>
>>
>>
>>
>>
> 
> 


-- 
Chen Gang

View attachment "config_554_ok" of type "text/plain" (98961 bytes)

View attachment "config_555_fail_dev" of type "text/plain" (98979 bytes)