[f2fs-dev] f2fs async buffered write patch

[f2fs-dev] f2fs async buffered write patch

[Jens Axboe]

Hi,

I came across this patch in a news posting:

https://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs.git/commit/?h=dev&id=d618126911829523e35a61f4a5a4ad159b1b2c8d

which has me a bit worried. As far as I can tell, all that patch does is
set FMODE_BUF_WASYNC, and then just hope that the lower layers handle
the rest?

What happens if iocb->ki_flags & IOCB_NOWAIT is true, and now we do:

generic_perform_write(iocb, from)
	...
	->write_begin() <- does this block?
	...
	->write_end() <- or this one?
	...
	balance_dirty_pages_ratelimited() <- this one surely does...

If you look just one level down the latter to
balance_dirty_pages_ratelimited_flags(), you'll even see the 'flags'
argument documented there.

This looks pretty haphazard and cannot possibly work as-is, so please
get this reverted until f2fs is converted to iomap, or IOCB_NOWAIT is
handled by generic_perform_write() and below.

-- 
Jens Axboe



_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

Re: [f2fs-dev] f2fs async buffered write patch

[Jaegeuk Kim]

On 06/19, Jens Axboe wrote:
> Hi,
> 
> I came across this patch in a news posting:
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs.git/commit/?h=dev&id=d618126911829523e35a61f4a5a4ad159b1b2c8d
> 
> which has me a bit worried. As far as I can tell, all that patch does is
> set FMODE_BUF_WASYNC, and then just hope that the lower layers handle
> the rest?
> 
> What happens if iocb->ki_flags & IOCB_NOWAIT is true, and now we do:
> 
> generic_perform_write(iocb, from)
> 	...
> 	->write_begin() <- does this block?
> 	...
> 	->write_end() <- or this one?
> 	...
> 	balance_dirty_pages_ratelimited() <- this one surely does...
> 
> If you look just one level down the latter to
> balance_dirty_pages_ratelimited_flags(), you'll even see the 'flags'
> argument documented there.
> 
> This looks pretty haphazard and cannot possibly work as-is, so please
> get this reverted until f2fs is converted to iomap, or IOCB_NOWAIT is
> handled by generic_perform_write() and below.

Thank you for pointing that out. It seems I haven't reviewed it carefully.
Hence I removed it from -next, and hope to have some time to convert iomap
soon.

Thanks,

> 
> -- 
> Jens Axboe


_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

Re: [f2fs-dev] f2fs async buffered write patch

[Jens Axboe]

On 6/20/23 12:16?AM, Jaegeuk Kim wrote:
> On 06/19, Jens Axboe wrote:
>> Hi,
>>
>> I came across this patch in a news posting:
>>
>> https://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs.git/commit/?h=dev&id=d618126911829523e35a61f4a5a4ad159b1b2c8d
>>
>> which has me a bit worried. As far as I can tell, all that patch does is
>> set FMODE_BUF_WASYNC, and then just hope that the lower layers handle
>> the rest?
>>
>> What happens if iocb->ki_flags & IOCB_NOWAIT is true, and now we do:
>>
>> generic_perform_write(iocb, from)
>> 	...
>> 	->write_begin() <- does this block?
>> 	...
>> 	->write_end() <- or this one?
>> 	...
>> 	balance_dirty_pages_ratelimited() <- this one surely does...
>>
>> If you look just one level down the latter to
>> balance_dirty_pages_ratelimited_flags(), you'll even see the 'flags'
>> argument documented there.
>>
>> This looks pretty haphazard and cannot possibly work as-is, so please
>> get this reverted until f2fs is converted to iomap, or IOCB_NOWAIT is
>> handled by generic_perform_write() and below.
> 
> Thank you for pointing that out. It seems I haven't reviewed it carefully.
> Hence I removed it from -next, and hope to have some time to convert iomap
> soon.

Thanks - would be great to get FMODE_BUF_WASYNC enabled obviously, just
needs a bit more work to get there.

-- 
Jens Axboe



_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

Re: [f2fs-dev] f2fs async buffered write patch

[Yangtao Li via Linux-f2fs-devel]

+cc willy@infradead.org


On 2023/6/20 4:43, Jens Axboe wrote:

> Hi,
>
> I came across this patch in a news posting:
>
> https://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs.git/commit/?h=dev&id=d618126911829523e35a61f4a5a4ad159b1b2c8d
>
> which has me a bit worried. As far as I can tell, all that patch does is
> set FMODE_BUF_WASYNC, and then just hope that the lower layers handle
> the rest?
>
> What happens if iocb->ki_flags & IOCB_NOWAIT is true, and now we do:
>
> generic_perform_write(iocb, from)
> 	...
> 	->write_begin() <- does this block?

Most file systems have not been changed to the iomap method,

and the generic_perform_write method is still used. So it seems

  reasonable to make IOCB_NOWAIT be handled correctly by the

  generic_perform_write function. This means that we need to modify

  several places mentioned by Jens in the generic_perform_write function,

  and need to pass AOP_FLAG_xxx or iocb into write_begin.


I noticed that Matthew Wilcox removed the flags parameter in write_begin

in a previous commit, maybe we can add it back?


Thx

> 	...
> 	->write_end() <- or this one?
> 	...
> 	balance_dirty_pages_ratelimited() <- this one surely does...
>
> If you look just one level down the latter to
> balance_dirty_pages_ratelimited_flags(), you'll even see the 'flags'
> argument documented there.
>
> This looks pretty haphazard and cannot possibly work as-is, so please
> get this reverted until f2fs is converted to iomap, or IOCB_NOWAIT is
> handled by generic_perform_write() and below.
>


_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

Re: [f2fs-dev] f2fs async buffered write patch

[Matthew Wilcox]

On Mon, Jun 26, 2023 at 03:31:57PM +0800, Yangtao Li wrote:
> +cc willy@infradead.org
> 
> 
> On 2023/6/20 4:43, Jens Axboe wrote:
> 
> > Hi,
> > 
> > I came across this patch in a news posting:
> > 
> > https://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs.git/commit/?h=dev&id=d618126911829523e35a61f4a5a4ad159b1b2c8d
> > 
> > which has me a bit worried. As far as I can tell, all that patch does is
> > set FMODE_BUF_WASYNC, and then just hope that the lower layers handle
> > the rest?
> > 
> > What happens if iocb->ki_flags & IOCB_NOWAIT is true, and now we do:
> > 
> > generic_perform_write(iocb, from)
> > 	...
> > 	->write_begin() <- does this block?
> 
> Most file systems have not been changed to the iomap method,

That's what needs to be fixed.

> and the generic_perform_write method is still used. So it seems
> 
>  reasonable to make IOCB_NOWAIT be handled correctly by the
> 
>  generic_perform_write function. This means that we need to modify

No.  Use the modern infrastructure.

>  several places mentioned by Jens in the generic_perform_write function,
> 
>  and need to pass AOP_FLAG_xxx or iocb into write_begin.

Definitely not.

> 
> I noticed that Matthew Wilcox removed the flags parameter in write_begin
> 
> in a previous commit, maybe we can add it back?

No.


_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel