[PATCH RESEND] tty/sysrq: Do not call sync directly from sysrq_do_reset()

[PATCH RESEND] tty/sysrq: Do not call sync directly from sysrq_do_reset()

[Mark Tomlinson]

sysrq_do_reset() is called in softirq context, so it cannot call
sync() directly. Instead, call orderly_reboot(), which creates a work
item to run /sbin/reboot, or do emergency_sync and restart if the
command fails.

Signed-off-by: Mark Tomlinson <mark.tomlinson@alliedtelesis.co.nz>
---
 drivers/tty/sysrq.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/tty/sysrq.c b/drivers/tty/sysrq.c
index ad1ee5d01b53..f2ca32c1ad7c 100644
--- a/drivers/tty/sysrq.c
+++ b/drivers/tty/sysrq.c
@@ -660,8 +660,7 @@ static void sysrq_do_reset(struct timer_list *t)
 
 	state->reset_requested = true;
 
-	ksys_sync();
-	kernel_restart(NULL);
+	orderly_reboot();
 }
 
 static void sysrq_handle_reset_request(struct sysrq_state *state)
-- 
2.19.2

Re: [PATCH RESEND] tty/sysrq: Do not call sync directly from sysrq_do_reset()

[Greg KH]

On Tue, Nov 27, 2018 at 09:57:23AM +1300, Mark Tomlinson wrote:
> sysrq_do_reset() is called in softirq context, so it cannot call
> sync() directly. Instead, call orderly_reboot(), which creates a work
> item to run /sbin/reboot, or do emergency_sync and restart if the
> command fails.
> 
> Signed-off-by: Mark Tomlinson <mark.tomlinson@alliedtelesis.co.nz>
> ---
>  drivers/tty/sysrq.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/drivers/tty/sysrq.c b/drivers/tty/sysrq.c
> index ad1ee5d01b53..f2ca32c1ad7c 100644
> --- a/drivers/tty/sysrq.c
> +++ b/drivers/tty/sysrq.c
> @@ -660,8 +660,7 @@ static void sysrq_do_reset(struct timer_list *t)
>  
>  	state->reset_requested = true;
>  
> -	ksys_sync();
> -	kernel_restart(NULL);
> +	orderly_reboot();

Is this something new?  Why haven't we had reports of this failing in
the past?  Or has something changed recently to cause this to now be
needed?

thanks,

greg k-h

Re: [PATCH RESEND] tty/sysrq: Do not call sync directly from sysrq_do_reset()

[Mark Tomlinson]

On 27/11/18 11:11 PM, Greg KH wrote:
> On Tue, Nov 27, 2018 at 09:57:23AM +1300, Mark Tomlinson wrote:
>> sysrq_do_reset() is called in softirq context, so it cannot call
>> sync() directly. Instead, call orderly_reboot(), which creates a work
>> item to run /sbin/reboot, or do emergency_sync and restart if the
>> command fails.
>>
>> Signed-off-by: Mark Tomlinson <mark.tomlinson@alliedtelesis.co.nz>
>> ---
>>   drivers/tty/sysrq.c | 3 +--
>>   1 file changed, 1 insertion(+), 2 deletions(-)
>>
>> diff --git a/drivers/tty/sysrq.c b/drivers/tty/sysrq.c
>> index ad1ee5d01b53..f2ca32c1ad7c 100644
>> --- a/drivers/tty/sysrq.c
>> +++ b/drivers/tty/sysrq.c
>> @@ -660,8 +660,7 @@ static void sysrq_do_reset(struct timer_list *t)
>>   
>>   	state->reset_requested = true;
>>   
>> -	ksys_sync();
>> -	kernel_restart(NULL);
>> +	orderly_reboot();
> 
> Is this something new?  Why haven't we had reports of this failing in
> the past?  Or has something changed recently to cause this to now be
> needed?
> 
> thanks,
> 
> greg k-h
> 

As far as I can tell, sysrq_do_reset() is only used when setting the 
"keyset" property in "/chosen/linux,sysrq-reset-seq" (which is what I 
want to do). No DTS file in the kernel tree sets this, so I don't know 
if anyone else is using this. I am wondering whether it ever worked. A 
reboot does occur after a kernel panic, so if the console is not 
available, it may appear to have worked.

Pressing the standard 'sysrq' key followed by 'b'(reboot) instead calls 
sysrq_handle_reboot(), which does not call sync at all, but 
emergency_restart(). Also, the 's'(sync) key calls emergency_sync(), 
which performs the sync via a work queue.