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;
From:   Michael Walle <michael@walle.cc>
To:     andriy.shevchenko@linux.intel.com
Cc:     djrscally@gmail.com, gregkh@linuxfoundation.org,
        heikki.krogerus@linux.intel.com, lenb@kernel.org,
        linux-acpi@vger.kernel.org, linux-kernel@vger.kernel.org,
        nuno.sa@analog.com, rafael@kernel.org,
        sakari.ailus@linux.intel.com, Russell King <linux@armlinux.org.uk>,
        Horatiu Vultur <horatiu.vultur@microchip.com>,
        Michael Walle <michael@walle.cc>
Subject: Re: [PATCH v4 1/1] device property: Allow error pointer to be passed to fwnode APIs
Date:   Mon, 14 Mar 2022 20:51:37 +0100
Message-Id: <20220314195138.20036-1-michael@walle.cc>
In-Reply-To: <20220308123712.18613-1-andriy.shevchenko@linux.intel.com>
References: <20220308123712.18613-1-andriy.shevchenko@linux.intel.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Precedence: bulk

Hi Andy,

> Some of the fwnode APIs might return an error pointer instead of NULL
> or valid fwnode handle. The result of such API call may be considered
> optional and hence the test for it is usually done in a form of
> 
> 	fwnode = fwnode_find_reference(...);
> 	if (IS_ERR(fwnode))
> 		...error handling...
> 
> Nevertheless the resulting fwnode may have bumped the reference count
> and hence caller of the above API is obliged to call fwnode_handle_put().
> Since fwnode may be not valid either as NULL or error pointer the check
> has to be performed there. This approach uglifies the code and adds
> a point of making a mistake, i.e. forgetting about error point case.
> 
> To prevent this, allow an error pointer to be passed to the fwnode APIs.
> 
> Fixes: 83b34afb6b79 ("device property: Introduce fwnode_find_reference()")
> Reported-by: Nuno Sá <nuno.sa@analog.com>
> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> Tested-by: Nuno Sá <nuno.sa@analog.com>
> Reviewed-by: Sakari Ailus <sakari.ailus@linux.intel.com>
> Acked-by: Nuno Sá <nuno.sa@analog.com>

This breaks SFP/phylink (using the lan966x switch) on my board. See below
for more details.

[..]

> @@ -480,15 +485,16 @@ int fwnode_property_get_reference_args(const struct fwnode_handle *fwnode,
>  {
>  	int ret;
>  
> +	if (IS_ERR_OR_NULL(fwnode))
> +		return -ENOENT;
> +
>  	ret = fwnode_call_int_op(fwnode, get_reference_args, prop, nargs_prop,
>  				 nargs, index, args);
> +	if (ret == 0)

Should this be "if (ret == 0 || IS_ERR_OR_NULL(fwnode->secondary))" ?

> +		return ret;
>  
> -	if (ret < 0 && !IS_ERR_OR_NULL(fwnode) &&
> -	    !IS_ERR_OR_NULL(fwnode->secondary))
> -		ret = fwnode_call_int_op(fwnode->secondary, get_reference_args,
> -					 prop, nargs_prop, nargs, index, args);
> -
> -	return ret;
> +	return fwnode_call_int_op(fwnode->secondary, get_reference_args, prop, nargs_prop,
> +				  nargs, index, args);
>  }
>  EXPORT_SYMBOL_GPL(fwnode_property_get_reference_args);

What happens before this patch is that sfp_bus_find_fwnode() will call
fwnode_property_get_reference_args() and the first calls return -ENOENT
which sfp_bus_find_fwnode() will handle in a special way. After your
patch, -EINVAL is returned, because fwnode_call_int_op() on
fwnode->secondary is always called regardless of the return value of
the original fwnode.

-michael