linux-kernel - Re: [PATCH] phy: intel: fix enum type mismatch warning

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20200527164035.GA1073507@ubuntu-s3-xlarge-x86>
Date:   Wed, 27 May 2020 09:40:35 -0700
From:   Nathan Chancellor <natechancellor@...il.com>
To:     Arnd Bergmann <arnd@...db.de>
Cc:     Kishon Vijay Abraham I <kishon@...com>,
        Vinod Koul <vkoul@...nel.org>,
        Dilip Kota <eswara.kota@...ux.intel.com>,
        linux-kernel@...r.kernel.org, clang-built-linux@...glegroups.com
Subject: Re: [PATCH] phy: intel: fix enum type mismatch warning

On Wed, May 27, 2020 at 03:45:06PM +0200, Arnd Bergmann wrote:
> clang points out that a local variable is initialized with
> an enum value of the wrong type:
> 
> drivers/phy/intel/phy-intel-combo.c:202:34: error: implicit conversion from enumeration type 'enum intel_phy_mode' to different enumeration type 'enum intel_combo_mode' [-Werror,-Wenum-conversion]
>         enum intel_combo_mode cb_mode = PHY_PCIE_MODE;
>                               ~~~~~~~   ^~~~~~~~~~~~~
> 
> From reading the code, it seems that this was not only the
> wrong type, but not even supposed to be a code path that can
> happen in practice.
> 
> Change the code to have no default phy mode but instead return an
> error for invalid input.
> 
> Fixes: ac0a95a3ea78 ("phy: intel: Add driver support for ComboPhy")
> Signed-off-by: Arnd Bergmann <arnd@...db.de>
> ---
>  drivers/phy/intel/phy-intel-combo.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/phy/intel/phy-intel-combo.c b/drivers/phy/intel/phy-intel-combo.c
> index c2a35be4cdfb..04f7b0d08742 100644
> --- a/drivers/phy/intel/phy-intel-combo.c
> +++ b/drivers/phy/intel/phy-intel-combo.c
> @@ -199,7 +199,7 @@ static int intel_cbphy_pcie_dis_pad_refclk(struct intel_cbphy_iphy *iphy)
>  
>  static int intel_cbphy_set_mode(struct intel_combo_phy *cbphy)
>  {
> -	enum intel_combo_mode cb_mode = PHY_PCIE_MODE;
> +	enum intel_combo_mode cb_mode;
>  	enum aggregated_mode aggr = cbphy->aggr_mode;
>  	struct device *dev = cbphy->dev;
>  	enum intel_phy_mode mode;
> @@ -224,6 +224,8 @@ static int intel_cbphy_set_mode(struct intel_combo_phy *cbphy)
>  
>  		cb_mode = SATA0_SATA1_MODE;
>  		break;
> +	default:
> +		return -EINVAL;
>  	}
>  
>  	ret = regmap_write(cbphy->hsiocfg, REG_COMBO_MODE(cbphy->bid), cb_mode);
> -- 
> 2.26.2
> 

I sent an almost identical patch:

https://lore.kernel.org/lkml/20200523035043.3305846-1-natechancellor@gmail.com/

I left out the default case since clang warns when a switch on an enum
does not handle all the values (compile time scream) versus a run time
scream like yours.

I don't have a preference for either so:

Reviewed-by: Nathan Chancellor <natechancellor@...il.com>