Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Mon, 19 Dec 2022 16:01:00 +0800
From:   Baoquan He <bhe@redhat.com>
To:     Lorenzo Stoakes <lstoakes@gmail.com>
Cc:     linux-kernel@vger.kernel.org, linux-mm@kvack.org, urezki@gmail.com,
        stephen.s.brennan@oracle.com, willy@infradead.org,
        akpm@linux-foundation.org, hch@infradead.org
Subject: Re: [PATCH v2 2/7] mm/vmalloc.c: add flags to mark vm_map_ram area
Message-ID: <Y6AaPKT7mdVxdHRl@MiWiFi-R3L-srv>
References: <20221217015435.73889-1-bhe@redhat.com>
 <20221217015435.73889-3-bhe@redhat.com>
 <Y52rllbXHMwbpJ8K@lucifer>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <Y52rllbXHMwbpJ8K@lucifer>
Precedence: bulk

On 12/17/22 at 11:44am, Lorenzo Stoakes wrote:
> On Sat, Dec 17, 2022 at 09:54:30AM +0800, Baoquan He wrote:
> > @@ -2229,8 +2236,12 @@ void vm_unmap_ram(const void *mem, unsigned int count)
> >  		return;
> >  	}
> >
> > -	va = find_vmap_area(addr);
> > +	spin_lock(&vmap_area_lock);
> > +	va = __find_vmap_area((unsigned long)addr, &vmap_area_root);
> >  	BUG_ON(!va);
> > +	if (va)
> > +		va->flags &= ~VMAP_RAM;
> > +	spin_unlock(&vmap_area_lock);
> >  	debug_check_no_locks_freed((void *)va->va_start,
> >  				    (va->va_end - va->va_start));
> >  	free_unmap_vmap_area(va);
> 
> Would it be better to perform the BUG_ON() after the lock is released? You
> already check if va exists before unmasking so it's safe.

It's a little unclear to me why we care BUG_ON() is performed before or
after the lock released. We won't have a stable kernel after BUG_ON()(),
right?
> 
> Also, do we want to clear VMAP_BLOCK here?

I do, but I don't find a good place to clear VMAP_BLOCK.

In v1, I tried to clear it in free_vmap_area_noflush() as below,
Uladzislau dislikes it. So I remove it. My thinking is when we unmap and
free the vmap area, the vmap_area is moved from vmap_area_root into
&free_vmap_area_root. When we allocate a new vmap_area via
alloc_vmap_area(), we will allocate a new va by kmem_cache_alloc_node(),
the va->flags must be 0. Seems not initializing it to 0 won't impact
thing.

diff --git a/mm/vmalloc.c b/mm/vmalloc.c
index 5d3fd3e6fe09..d6f376060d83 100644
--- a/mm/vmalloc.c
+++ b/mm/vmalloc.c
@@ -1815,6 +1815,7 @@ static void free_vmap_area_noflush(struct vmap_area *va)

        spin_lock(&vmap_area_lock);
        unlink_va(va, &vmap_area_root);
+       va->flags = 0;
        spin_unlock(&vmap_area_lock);

        nr_lazy = atomic_long_add_return((va->va_end - va->va_start) >>

>