[PATCH] f2fs: skip adding a discard command if exists

[PATCH] f2fs: skip adding a discard command if exists

[Jaegeuk Kim]

When recovering zoned UFS, sometimes we add the same zone to discard multiple
times. Simple workaround is to bypass adding it.

Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
---
 fs/f2fs/segment.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index 727d016318f9..f4ffd64b44b2 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -1380,7 +1380,8 @@ static void __insert_discard_cmd(struct f2fs_sb_info *sbi,
 			p = &(*p)->rb_right;
 			leftmost = false;
 		} else {
-			f2fs_bug_on(sbi, 1);
+			/* Let's skip to add, if exists */
+			return;
 		}
 	}
 
-- 
2.43.0.rc0.421.g78406f8d94-goog

Re: [f2fs-dev] [PATCH] f2fs: skip adding a discard command if exists

[Chao Yu]

On 2023/11/15 5:24, Jaegeuk Kim wrote:
> When recovering zoned UFS, sometimes we add the same zone to discard multiple
> times. Simple workaround is to bypass adding it.

What about skipping f2fs_bug_on() just for zoned UFS case? so that the check
condition can still be used for non-zoned UFS case.

Thanks,

> 
> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
> ---
>   fs/f2fs/segment.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
> index 727d016318f9..f4ffd64b44b2 100644
> --- a/fs/f2fs/segment.c
> +++ b/fs/f2fs/segment.c
> @@ -1380,7 +1380,8 @@ static void __insert_discard_cmd(struct f2fs_sb_info *sbi,
>   			p = &(*p)->rb_right;
>   			leftmost = false;
>   		} else {
> -			f2fs_bug_on(sbi, 1);
> +			/* Let's skip to add, if exists */
> +			return;
>   		}
>   	}
>   

Re: [f2fs-dev] [PATCH] f2fs: skip adding a discard command if exists

[Jaegeuk Kim]

On 11/15, Chao Yu wrote:
> On 2023/11/15 5:24, Jaegeuk Kim wrote:
> > When recovering zoned UFS, sometimes we add the same zone to discard multiple
> > times. Simple workaround is to bypass adding it.
> 
> What about skipping f2fs_bug_on() just for zoned UFS case? so that the check
> condition can still be used for non-zoned UFS case.

Hmm, I've never seen this bug_on before, but even this really happens, it does
not make sense to move forward to create duplicate commands resulting in a loop.

So, the question is, do we really need to check this? Have we hit this before?

> 
> Thanks,
> 
> > 
> > Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
> > ---
> >   fs/f2fs/segment.c | 3 ++-
> >   1 file changed, 2 insertions(+), 1 deletion(-)
> > 
> > diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
> > index 727d016318f9..f4ffd64b44b2 100644
> > --- a/fs/f2fs/segment.c
> > +++ b/fs/f2fs/segment.c
> > @@ -1380,7 +1380,8 @@ static void __insert_discard_cmd(struct f2fs_sb_info *sbi,
> >   			p = &(*p)->rb_right;
> >   			leftmost = false;
> >   		} else {
> > -			f2fs_bug_on(sbi, 1);
> > +			/* Let's skip to add, if exists */
> > +			return;
> >   		}
> >   	}

Re: [f2fs-dev] [PATCH] f2fs: skip adding a discard command if exists

[Chao Yu]

On 2023/11/15 10:45, Jaegeuk Kim wrote:
> On 11/15, Chao Yu wrote:
>> On 2023/11/15 5:24, Jaegeuk Kim wrote:
>>> When recovering zoned UFS, sometimes we add the same zone to discard multiple
>>> times. Simple workaround is to bypass adding it.
>>
>> What about skipping f2fs_bug_on() just for zoned UFS case? so that the check
>> condition can still be used for non-zoned UFS case.
> 
> Hmm, I've never seen this bug_on before, but even this really happens, it does

I've never seen it was been triggered as well.

> not make sense to move forward to create duplicate commands resulting in a loop.

Agreed.

It looks those codes were copied from extent_cache code base, do we need to fix
all cases to avoid loop?

> 
> So, the question is, do we really need to check this? Have we hit this before?
Not sure, just be worry about that flaw of newly developed feature can make
code run into that branch.

Thanks,

> 
>>
>> Thanks,
>>
>>>
>>> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
>>> ---
>>>    fs/f2fs/segment.c | 3 ++-
>>>    1 file changed, 2 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
>>> index 727d016318f9..f4ffd64b44b2 100644
>>> --- a/fs/f2fs/segment.c
>>> +++ b/fs/f2fs/segment.c
>>> @@ -1380,7 +1380,8 @@ static void __insert_discard_cmd(struct f2fs_sb_info *sbi,
>>>    			p = &(*p)->rb_right;
>>>    			leftmost = false;
>>>    		} else {
>>> -			f2fs_bug_on(sbi, 1);
>>> +			/* Let's skip to add, if exists */
>>> +			return;
>>>    		}
>>>    	}

Re: [f2fs-dev] [PATCH] f2fs: skip adding a discard command if exists

[Jaegeuk Kim]

On 11/15, Chao Yu wrote:
> On 2023/11/15 10:45, Jaegeuk Kim wrote:
> > On 11/15, Chao Yu wrote:
> > > On 2023/11/15 5:24, Jaegeuk Kim wrote:
> > > > When recovering zoned UFS, sometimes we add the same zone to discard multiple
> > > > times. Simple workaround is to bypass adding it.
> > > 
> > > What about skipping f2fs_bug_on() just for zoned UFS case? so that the check
> > > condition can still be used for non-zoned UFS case.
> > 
> > Hmm, I've never seen this bug_on before, but even this really happens, it does
> 
> I've never seen it was been triggered as well.
> 
> > not make sense to move forward to create duplicate commands resulting in a loop.
> 
> Agreed.
> 
> It looks those codes were copied from extent_cache code base, do we need to fix
> all cases to avoid loop?

Not sure other cases yet.. let's do one by one, since I hit this in real.

> 
> > 
> > So, the question is, do we really need to check this? Have we hit this before?
> Not sure, just be worry about that flaw of newly developed feature can make
> code run into that branch.
> 
> Thanks,
> 
> > 
> > > 
> > > Thanks,
> > > 
> > > > 
> > > > Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
> > > > ---
> > > >    fs/f2fs/segment.c | 3 ++-
> > > >    1 file changed, 2 insertions(+), 1 deletion(-)
> > > > 
> > > > diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
> > > > index 727d016318f9..f4ffd64b44b2 100644
> > > > --- a/fs/f2fs/segment.c
> > > > +++ b/fs/f2fs/segment.c
> > > > @@ -1380,7 +1380,8 @@ static void __insert_discard_cmd(struct f2fs_sb_info *sbi,
> > > >    			p = &(*p)->rb_right;
> > > >    			leftmost = false;
> > > >    		} else {
> > > > -			f2fs_bug_on(sbi, 1);
> > > > +			/* Let's skip to add, if exists */
> > > > +			return;
> > > >    		}
> > > >    	}

Re: [f2fs-dev] [PATCH] f2fs: skip adding a discard command if exists

[Chao Yu]

On 2023/11/18 1:41, Jaegeuk Kim wrote:
> Not sure other cases yet.. let's do one by one, since I hit this in real.

Sure.

>>>>> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>

Reviewed-by: Chao Yu <chao@kernel.org>

Thansk,

Re: [f2fs-dev] [PATCH] f2fs: skip adding a discard command if exists

[patchwork-bot+f2fs]

Hello:

This patch was applied to jaegeuk/f2fs.git (dev)
by Jaegeuk Kim <jaegeuk@kernel.org>:

On Tue, 14 Nov 2023 13:24:14 -0800 you wrote:
> When recovering zoned UFS, sometimes we add the same zone to discard multiple
> times. Simple workaround is to bypass adding it.
> 
> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
> ---
>  fs/f2fs/segment.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)

Here is the summary with links:
  - [f2fs-dev] f2fs: skip adding a discard command if exists
    https://git.kernel.org/jaegeuk/f2fs/c/bbd3efed3383

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html