Date: Sun, 27 Aug 2017 21:52:38 -0700
From: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
To: Michael Ellerman <mpe@ellerman.id.au>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>, mikey@neuling.org,
        stewart@linux.vnet.ibm.com, apopple@au1.ibm.com, hbabu@us.ibm.com,
        oohall@gmail.com, linuxppc-dev@ozlabs.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH v7 06/12] powerpc/vas: Define helpers to alloc/free
 windows
References: <1503556688-15412-1-git-send-email-sukadev@linux.vnet.ibm.com>
 <1503556688-15412-7-git-send-email-sukadev@linux.vnet.ibm.com>
 <878ti8uh0r.fsf@concordia.ellerman.id.au>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <878ti8uh0r.fsf@concordia.ellerman.id.au>
User-Agent: Mutt/1.7.1 (2016-10-04)
Message-Id: <20170828045238.GD12907@us.ibm.com>
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 2028
Lines: 86

Michael Ellerman [mpe@ellerman.id.au] wrote:
> Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com> writes:
> > diff --git a/arch/powerpc/platforms/powernv/vas-window.c b/arch/powerpc/platforms/powernv/vas-window.c

> > +	rc = ida_pre_get(ida, GFP_KERNEL);
> > +	if (!rc)
> > +		return -EAGAIN;
> > +
> > +	spin_lock(&vas_ida_lock);
> > +	rc = ida_get_new_above(ida, 0, &winid);
> 
> If you're passing 0 you can just use ida_get_new().

Ok.

> 
> Or did you actually want to exclude 0? In which case you should pass 1.
> 
> > +	spin_unlock(&vas_ida_lock);
> > +
> > +	if (rc)
> > +		return rc;
> 
> You're supposed to handle EAGAIN I thought.

Yes, I will retry the pre_get()
> 
> > +
> > +	if (winid > VAS_WINDOWS_PER_CHIP) {
> > +		pr_err("VAS: Too many (%d) open windows\n", winid);
> > +		vas_release_window_id(ida, winid);
> > +		return -EAGAIN;
> > +	}
> > +
> > +	return winid;
> > +}
> > +
> > +void vas_window_free(struct vas_window *window)
> 
> static.

Ok

> 
> > +{
> > +	int winid = window->winid;
> > +	struct vas_instance *vinst = window->vinst;
> > +
> > +	unmap_winctx_mmio_bars(window);
> > +	kfree(window);
> > +
> > +	vas_release_window_id(&vinst->ida, winid);
> > +}
> > +
> > +struct vas_window *vas_window_alloc(struct vas_instance *vinst)
> > +{
> > +	int winid;
> > +	struct vas_window *window;
> > +
> > +	winid = vas_assign_window_id(&vinst->ida);
> > +	if (winid < 0)
> > +		return ERR_PTR(winid);
> > +
> > +	window = kzalloc(sizeof(*window), GFP_KERNEL);
> > +	if (!window)
> > +		return ERR_PTR(-ENOMEM);
> 
> You leak an id here.

Argh. Yes.

> 
> The error handling would be easier in here if the caller did the alloc,
> or if you split alloc and init, and alloc just did the kzalloc().

I was trying to simplify error handling in the callers where they have
to only deal with one failure now.
> 
> One of the callers even prints "unable to allocate memory" if this
> function fails, but that's not accurate, there's several failure modes.

Yes, will fix that message and the leaks.

Thanks,

Suka