Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20210511104647.604-1-songmuchun@bytedance.com>
 <20210511104647.604-11-songmuchun@bytedance.com> <20210511234041.GP1872259@dread.disaster.area>
In-Reply-To: <20210511234041.GP1872259@dread.disaster.area>
From:   Muchun Song <songmuchun@bytedance.com>
Date:   Wed, 12 May 2021 11:20:10 +0800
Message-ID: <CAMZfGtV3cOY0JwMp_wr+PcHyYZsL4-2v-9YjXnPgpvJXgMvrTw@mail.gmail.com>
Subject: Re: [External] Re: [PATCH 10/17] fs: introduce alloc_inode_sb() to
 allocate filesystems specific inode
To:     Dave Chinner <david@fromorbit.com>
Cc:     Matthew Wilcox <willy@infradead.org>,
        Andrew Morton <akpm@linux-foundation.org>,
        Johannes Weiner <hannes@cmpxchg.org>,
        Michal Hocko <mhocko@kernel.org>,
        Vladimir Davydov <vdavydov.dev@gmail.com>,
        Shakeel Butt <shakeelb@google.com>,
        Roman Gushchin <guro@fb.com>, Yang Shi <shy828301@gmail.com>,
        alexs@kernel.org, Wei Yang <richard.weiyang@gmail.com>,
        trond.myklebust@hammerspace.com, anna.schumaker@netapp.com,
        linux-fsdevel <linux-fsdevel@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Linux Memory Management List <linux-mm@kvack.org>,
        linux-nfs@vger.kernel.org, zhengqi.arch@bytedance.com,
        Xiongchun duan <duanxiongchun@bytedance.com>,
        fam.zheng@bytedance.com
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, May 12, 2021 at 7:40 AM Dave Chinner <david@fromorbit.com> wrote:
>
> On Tue, May 11, 2021 at 06:46:40PM +0800, Muchun Song wrote:
> > The allocated inode cache will be added into its memcg lru list later,
> > but we do not allocate list_lru in the later patch. So the caller should
> > call kmem_cache_alloc_lru() to allocate inode and related list_lru.
> > Introduce alloc_inode_sb() to do that and convert all inodes allocation
> > to it.
>
> FWIW, this probably needs a documentation update to mention that
> inodes should always be allocated through alloc_inode_sb() rather
> than kmem_cache_alloc(). It's a "** mandatory **" requirement as per
> Documentation/filesytems/porting.rst.

Make sense to me. I'll fix it in the next version.

>
> Also,
>
> > diff --git a/include/linux/fs.h b/include/linux/fs.h
> > index c3c88fdb9b2a..d8d5d4eb68d6 100644
> > --- a/include/linux/fs.h
> > +++ b/include/linux/fs.h
> > @@ -41,6 +41,7 @@
> >  #include <linux/stddef.h>
> >  #include <linux/mount.h>
> >  #include <linux/cred.h>
> > +#include <linux/slab.h>
> >
> >  #include <asm/byteorder.h>
> >  #include <uapi/linux/fs.h>
> > @@ -3200,6 +3201,12 @@ extern void free_inode_nonrcu(struct inode *inode);
> >  extern int should_remove_suid(struct dentry *);
> >  extern int file_remove_privs(struct file *);
> >
> > +static inline void *
> > +alloc_inode_sb(struct super_block *sb, struct kmem_cache *cache, gfp_t gfp)
> > +{
> > +     return kmem_cache_alloc_lru(cache, &sb->s_inode_lru, gfp);
> > +}
> > +
>
> This really needs a kerneldoc comment explaining that it must be
> used for allocating inodes to set up the inode reclaim context
> correctly....

Will do.

>
> /me wonders if we should add a BUG_ON() check in inode_init_always()
> to capture filesystems that don't call through
> kmem_cache_alloc_lru() for inodes?

Good point. IMHO, I think that the BUG_ON check may be useful.
Actually, it can catch such bugs.

Any suggestions from others?

Thanks.

>
> Cheers,
>
> Dave.
> --
> Dave Chinner
> david@fromorbit.com