Re: [PATCH] fs/bcachefs: Fix __wait_on_freeing_inode() definition of waitqueue entry

[NeilBrown]

On Wed, 09 Oct 2024, Ingo Molnar wrote:
> * NeilBrown <neilb@xxxxxxx> wrote:
> 
> > On Wed, 09 Oct 2024, Stephen Rothwell wrote:
> > > Hi all,
> > > 
> > > After merging the tip tree, today's linux-next build (x86_64 allmodconfig)
> > > failed like this:
> > > 
> > > In file included from include/linux/fs.h:6,
> > >                  from include/linux/highmem.h:5,
> > >                  from include/linux/bvec.h:10,
> > >                  from include/linux/blk_types.h:10,
> > >                  from include/linux/bio.h:10,
> > >                  from fs/bcachefs/bcachefs.h:188,
> > >                  from fs/bcachefs/fs.c:4:
> > > fs/bcachefs/fs.c: In function '__wait_on_freeing_inode':
> > > fs/bcachefs/fs.c:281:31: error: initialization of 'long unsigned int *' from incompatible pointer type 'u32 *' {aka 'unsigned int *'} [-Wincompatible-pointer-types]
> > >   281 |         DEFINE_WAIT_BIT(wait, &inode->v.i_state, __I_NEW);
> > 
> > The fix we want is to replace that line with
> >    struct wait_bit_queue_entry wait;
> > I should have checked more carefully - sorry.
> > 
> > I guess we wait for rc3?
> > 
> > Kent: could you please make that change?  The inode_bit_waitqueue() does
> > initialisation equivalent of DEFINE_WAIT_BIT() so you only need the declaration.
> 
> Since the breakage was introduced via tip:sched/core, I've applied the fix 
> below.
> 
> Does this look good to you?

Perfect, thanks.

NeilBrown

> 
> Thanks,
> 
> 	Ingo
> 
> ================>
> From: Ingo Molnar <mingo@xxxxxxxxxx>
> Date: Wed, 9 Oct 2024 10:00:09 +0200
> Subject: [PATCH] fs/bcachefs: Fix __wait_on_freeing_inode() definition of waitqueue entry
> 
> The following recent commit made DEFINE_WAIT_BIT() type requirements stricter:
> 
>   2382d68d7d43 ("sched: change wake_up_bit() and related function to expect unsigned long *")
> 
> .. which results in a build failure:
> 
>   > fs/bcachefs/fs.c: In function '__wait_on_freeing_inode':
>   > fs/bcachefs/fs.c:281:31: error: initialization of 'long unsigned int *' from incompatible pointer type 'u32 *' {aka 'unsigned int *'} [-Wincompatible-pointer-types]
>   >   281 |         DEFINE_WAIT_BIT(wait, &inode->v.i_state, __I_NEW);
> 
> Since this code relies on the waitqueue initialization within
> inode_bit_waitqueue() anyway, the DEFINE_WAIT_BIT() initialization
> is unnecessary - we can just declare a waitqueue entry.
> 
> Reported-by: Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx>
> Suggested-by: NeilBrown <neilb@xxxxxxx>
> Signed-off-by: Ingo Molnar <mingo@xxxxxxxxxx>
> ---
>  fs/bcachefs/fs.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/bcachefs/fs.c b/fs/bcachefs/fs.c
> index 5bfc26d58270..c410133541ba 100644
> --- a/fs/bcachefs/fs.c
> +++ b/fs/bcachefs/fs.c
> @@ -183,8 +183,9 @@ static void __wait_on_freeing_inode(struct bch_fs *c,
>  				    struct bch_inode_info *inode,
>  				    subvol_inum inum)
>  {
> +	struct wait_bit_queue_entry wait;
>  	wait_queue_head_t *wq;
> -	DEFINE_WAIT_BIT(wait, &inode->v.i_state, __I_NEW);
> +
>  	wq = inode_bit_waitqueue(&wait, &inode->v, __I_NEW);
>  	prepare_to_wait(wq, &wait.wq_entry, TASK_UNINTERRUPTIBLE);
>  	spin_unlock(&inode->v.i_lock);
> 
>