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;
Date:   Tue, 19 Oct 2021 08:59:30 +0200
From:   Michal Hocko <mhocko@suse.com>
To:     NeilBrown <neilb@suse.de>
Cc:     linux-mm@kvack.org, Dave Chinner <david@fromorbit.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Christoph Hellwig <hch@infradead.org>,
        Uladzislau Rezki <urezki@gmail.com>,
        linux-fsdevel@vger.kernel.org, LKML <linux-kernel@vger.kernel.org>,
        Ilya Dryomov <idryomov@gmail.com>,
        Jeff Layton <jlayton@kernel.org>
Subject: Re: [RFC 1/3] mm/vmalloc: alloc GFP_NO{FS,IO} for vmalloc
Message-ID: <YW5s0qo64mFaQMQj@dhcp22.suse.cz>
References: <20211018114712.9802-1-mhocko@kernel.org>
 <20211018114712.9802-2-mhocko@kernel.org>
 <163460424165.17149.585825289709126969@noble.neil.brown.name>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <163460424165.17149.585825289709126969@noble.neil.brown.name>
Precedence: bulk

On Tue 19-10-21 11:44:01, Neil Brown wrote:
> On Mon, 18 Oct 2021, Michal Hocko wrote:
[...]
> > @@ -2930,8 +2932,24 @@ static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask,
> >  		goto fail;
> >  	}
> >  
> > -	if (vmap_pages_range(addr, addr + size, prot, area->pages,
> > -			page_shift) < 0) {
> > +	/*
> > +	 * page tables allocations ignore external gfp mask, enforce it
> > +	 * by the scope API
> > +	 */
> > +	if ((gfp_mask & (__GFP_FS | __GFP_IO)) == __GFP_IO)
> > +		flags = memalloc_nofs_save();
> > +	else if (!(gfp_mask & (__GFP_FS | __GFP_IO)))
> 
> I would *much* rather this were written
> 
>         else if ((gfp_mask & (__GFP_FS | __GFP_IO)) == 0)

Sure, this looks better indeed.

> so that the comparison with the previous test is more obvious.  Ditto
> for similar code below.
> It could even be
> 
>    switch (gfp_mask & (__GFP_FS | __GFP_IO)) {
>    case __GFP__IO: flags = memalloc_nofs_save(); break;
>    case 0:         flags = memalloc_noio_save(); break;
>    }
> 
> But I'm not completely convinced that is an improvement.

I am not a great fan of this though.

> In terms of functionality this looks good.

Thanks for the review!

-- 
Michal Hocko
SUSE Labs