Backport patch "mm: fix missing handler for __GFP_NOWARN" to linux-5.10.y

Backport patch "mm: fix missing handler for __GFP_NOWARN" to linux-5.10.y

[Ye Weihua]

The following patch is required to be patched in linux-5.10.y:


     3f913fc5f974 mm: fix missing handler for __GFP_NOWARN


Commit 6b9dbedbe349 ("tty: fix deadlock caused by calling printk() under 
tty_port->lock")

was backported to linux-5.10.y. But __GFP_NOWARN flag is still not check 
in fail_dump(), and

deadlock issues still occur.

Re: Backport patch "mm: fix missing handler for __GFP_NOWARN" to linux-5.10.y

[Greg KH]

On Fri, Sep 09, 2022 at 11:40:07AM +0800, Ye Weihua wrote:
> The following patch is required to be patched in linux-5.10.y:
> 
> 
>     3f913fc5f974 mm: fix missing handler for __GFP_NOWARN
> 
> 
> Commit 6b9dbedbe349 ("tty: fix deadlock caused by calling printk() under
> tty_port->lock")
> 
> was backported to linux-5.10.y. But __GFP_NOWARN flag is still not check in
> fail_dump(), and
> 
> deadlock issues still occur.
> 

What about all of the other stable kernel trees that the tty patch was
backported to?  Do they also need the mm change as well?  That would
include 4.9.y, 4.14.y, 4.19.y, 5.4.y, 5.10.y, and 5.15.y.

thanks,

greg k-h

Re: Backport patch "mm: fix missing handler for __GFP_NOWARN" to linux-5.10.y

[Ye Weihua]

On 2022/9/10 14:34, Greg KH wrote:
> On Fri, Sep 09, 2022 at 11:40:07AM +0800, Ye Weihua wrote:
>> The following patch is required to be patched in linux-5.10.y:
>>
>>
>>      3f913fc5f974 mm: fix missing handler for __GFP_NOWARN
>>
>>
>> Commit 6b9dbedbe349 ("tty: fix deadlock caused by calling printk() under
>> tty_port->lock")
>>
>> was backported to linux-5.10.y. But __GFP_NOWARN flag is still not check in
>> fail_dump(), and
>>
>> deadlock issues still occur.
>>
> What about all of the other stable kernel trees that the tty patch was
> backported to?  Do they also need the mm change as well?  That would
> include 4.9.y, 4.14.y, 4.19.y, 5.4.y, 5.10.y, and 5.15.y.

I checked the branches and found that the status of each branch was the 
same. That is, the commit 6b9dbedbe349 ("tty: fix deadlock caused by 
calling printk() under tty_port->lock") was backported but the commit 
3f913fc5f974 ("mm: fix missing handler for __GFP_NOWARN") was not. 
Therefore, the problem occurred in all branches. The commit "mm: fix 
missing handler for __GFP_NOWARN" should be backported to 4.9.y, 4.14.y, 
4.19.y, 5.4.y, 5.10.y, and 5.15.y.

Re: Backport patch "mm: fix missing handler for __GFP_NOWARN" to linux-5.10.y

[Greg KH]

On Tue, Sep 13, 2022 at 10:21:38AM +0800, Ye Weihua wrote:
> 
> On 2022/9/10 14:34, Greg KH wrote:
> > On Fri, Sep 09, 2022 at 11:40:07AM +0800, Ye Weihua wrote:
> > > The following patch is required to be patched in linux-5.10.y:
> > > 
> > > 
> > >      3f913fc5f974 mm: fix missing handler for __GFP_NOWARN
> > > 
> > > 
> > > Commit 6b9dbedbe349 ("tty: fix deadlock caused by calling printk() under
> > > tty_port->lock")
> > > 
> > > was backported to linux-5.10.y. But __GFP_NOWARN flag is still not check in
> > > fail_dump(), and
> > > 
> > > deadlock issues still occur.
> > > 
> > What about all of the other stable kernel trees that the tty patch was
> > backported to?  Do they also need the mm change as well?  That would
> > include 4.9.y, 4.14.y, 4.19.y, 5.4.y, 5.10.y, and 5.15.y.
> 
> I checked the branches and found that the status of each branch was the
> same. That is, the commit 6b9dbedbe349 ("tty: fix deadlock caused by calling
> printk() under tty_port->lock") was backported but the commit 3f913fc5f974
> ("mm: fix missing handler for __GFP_NOWARN") was not. Therefore, the problem
> occurred in all branches. The commit "mm: fix missing handler for
> __GFP_NOWARN" should be backported to 4.9.y, 4.14.y, 4.19.y, 5.4.y, 5.10.y,
> and 5.15.y.

Ok, can you provide a proper backport that has been tested for all of
these branches as it does not apply cleanly as-is.

Or we can revert the tty patch, which do you think is better?

thanks,

greg k-h

Re: Backport patch "mm: fix missing handler for __GFP_NOWARN" to linux-5.10.y

[Ye Weihua]

On 2022/9/13 19:33, Greg KH wrote:
> On Tue, Sep 13, 2022 at 10:21:38AM +0800, Ye Weihua wrote:
>> On 2022/9/10 14:34, Greg KH wrote:
>>> On Fri, Sep 09, 2022 at 11:40:07AM +0800, Ye Weihua wrote:
>>>> The following patch is required to be patched in linux-5.10.y:
>>>>
>>>>
>>>>       3f913fc5f974 mm: fix missing handler for __GFP_NOWARN
>>>>
>>>>
>>>> Commit 6b9dbedbe349 ("tty: fix deadlock caused by calling printk() under
>>>> tty_port->lock")
>>>>
>>>> was backported to linux-5.10.y. But __GFP_NOWARN flag is still not check in
>>>> fail_dump(), and
>>>>
>>>> deadlock issues still occur.
>>>>
>>> What about all of the other stable kernel trees that the tty patch was
>>> backported to?  Do they also need the mm change as well?  That would
>>> include 4.9.y, 4.14.y, 4.19.y, 5.4.y, 5.10.y, and 5.15.y.
>> I checked the branches and found that the status of each branch was the
>> same. That is, the commit 6b9dbedbe349 ("tty: fix deadlock caused by calling
>> printk() under tty_port->lock") was backported but the commit 3f913fc5f974
>> ("mm: fix missing handler for __GFP_NOWARN") was not. Therefore, the problem
>> occurred in all branches. The commit "mm: fix missing handler for
>> __GFP_NOWARN" should be backported to 4.9.y, 4.14.y, 4.19.y, 5.4.y, 5.10.y,
>> and 5.15.y.
> Ok, can you provide a proper backport that has been tested for all of
> these branches as it does not apply cleanly as-is.
>
> Or we can revert the tty patch, which do you think is better?

Ok, I will send a backport for these branches soon. Thank you for the 
reminder.