linux-kernel - Re: [PATCH 1/1] USB: PHY: JZ4770: Fix uninitialized value written to HW register

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: <HG6QFQ.KLMIR92DB2D02@crapouillou.net>
Date:   Thu, 27 Aug 2020 15:50:41 +0200
From:   Paul Cercueil <paul@...pouillou.net>
To:     Felipe Balbi <balbi@...nel.org>
Cc:     周琰杰 <zhouyanjie@...yeetech.com>,
        周正 <sernia.zhou@...mail.com>,
        漆鹏振 <aric.pzqi@...enic.com>, od@...c.me,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        linux-usb@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/1] USB: PHY: JZ4770: Fix uninitialized value written to
 HW register



Le jeu. 27 août 2020 à 16:25, Felipe Balbi <balbi@...nel.org> a 
écrit :
> 
> Hi,
> 
> Paul Cercueil <paul@...pouillou.net> writes:
>>>>   @@ -172,7 +172,8 @@ static int ingenic_usb_phy_init(struct 
>>>> usb_phy
>>>>  *phy)
>>>>    		return err;
>>>>    	}
>>>> 
>>>>   -	priv->soc_info->usb_phy_init(phy);
>>>>   +	reg = priv->soc_info->usb_phy_init(phy);
>>>>   +	writel(reg, priv->base + REG_USBPCR_OFFSET);
>>> 
>>>  not fixing any bug.
>>> 
>>>  Looking at the code, the bug follows after this line. It would 
>>> suffice
>>>  to read REG_USBPCR_OFFSET in order to initialize reg. This bug fix
>>>  could
>>>  have been a one liner.
>> 
>>  There's no need to re-read a register when you have the value 
>> readily
>>  available. It just needs to be returned from the usb_phy_init
>>  callbacks. But yes, it's not a one-liner.
> 
> there's a difference between making a bug fix and reworking the 
> behavior
> of the driver ;-)

The one-liner is actually what changes the behavior of the driver, 
since previously the code did not read back the register.

In this case I guess it's fine, because the register does not have 
volatile bits.

>>>>   @@ -195,19 +196,15 @@ static void ingenic_usb_phy_remove(void 
>>>> *phy)
>>>>    	usb_remove_phy(phy);
>>>>    }
>>>> 
>>>>   -static void jz4770_usb_phy_init(struct usb_phy *phy)
>>>>   +static u32 jz4770_usb_phy_init(struct usb_phy *phy)
>>> 
>>>  not a bug fix
>>> 
>>>>    {
>>>>   -	struct jz4770_phy *priv = phy_to_jz4770_phy(phy);
>>>>   -	u32 reg;
>>>>   -
>>>>   -	reg = USBPCR_AVLD_REG | USBPCR_COMMONONN | 
>>>> USBPCR_IDPULLUP_ALWAYS
>>>>  |
>>>>   +	return USBPCR_AVLD_REG | USBPCR_COMMONONN |
>>>>  USBPCR_IDPULLUP_ALWAYS |
>>>>    		USBPCR_COMPDISTUNE_DFT | USBPCR_OTGTUNE_DFT |
>>>>  USBPCR_SQRXTUNE_DFT |
>>>>    		USBPCR_TXFSLSTUNE_DFT | USBPCR_TXRISETUNE_DFT |
>>>>  USBPCR_TXVREFTUNE_DFT |
>>>>    		USBPCR_POR;
>>>>   -	writel(reg, priv->base + REG_USBPCR_OFFSET);
>>> 
>>>  not a bug fix
>>> 
>>>>    }
>>>> 
>>>>   -static void jz4780_usb_phy_init(struct usb_phy *phy)
>>>>   +static u32 jz4780_usb_phy_init(struct usb_phy *phy)
>>> 
>>>  not a bug fix
>>> 
>>>>   @@ -216,11 +213,10 @@ static void jz4780_usb_phy_init(struct
>>>>  usb_phy *phy)
>>>>    		USBPCR1_WORD_IF_16BIT;
>>>>    	writel(reg, priv->base + REG_USBPCR1_OFFSET);
>>>> 
>>>>   -	reg = USBPCR_TXPREEMPHTUNE | USBPCR_COMMONONN | USBPCR_POR;
>>>>   -	writel(reg, priv->base + REG_USBPCR_OFFSET);
>>>>   +	return USBPCR_TXPREEMPHTUNE | USBPCR_COMMONONN | USBPCR_POR;
>>> 
>>>  not a bug fix
>>> 
>>>>    }
>>>> 
>>>>   -static void x1000_usb_phy_init(struct usb_phy *phy)
>>>>   +static u32 x1000_usb_phy_init(struct usb_phy *phy)
>>> 
>>>  not a bug fix
>>> 
>>>>    {
>>>>    	struct jz4770_phy *priv = phy_to_jz4770_phy(phy);
>>>>    	u32 reg;
>>>>   @@ -228,13 +224,12 @@ static void x1000_usb_phy_init(struct 
>>>> usb_phy
>>>>  *phy)
>>>>    	reg = readl(priv->base + REG_USBPCR1_OFFSET) |
>>>>  USBPCR1_WORD_IF_16BIT;
>>>>    	writel(reg, priv->base + REG_USBPCR1_OFFSET);
>>>> 
>>>>   -	reg = USBPCR_SQRXTUNE_DCR_20PCT | USBPCR_TXPREEMPHTUNE |
>>>>   +	return USBPCR_SQRXTUNE_DCR_20PCT | USBPCR_TXPREEMPHTUNE |
>>>>    		USBPCR_TXHSXVTUNE_DCR_15MV | USBPCR_TXVREFTUNE_INC_25PPT |
>>>>    		USBPCR_COMMONONN | USBPCR_POR;
>>>>   -	writel(reg, priv->base + REG_USBPCR_OFFSET);
>>> 
>>>  not a bug fix
>>> 
>>>>    }
>>>> 
>>>>   -static void x1830_usb_phy_init(struct usb_phy *phy)
>>>>   +static u32 x1830_usb_phy_init(struct usb_phy *phy)
>>> 
>>>  not a bug fix
>>> 
>>>>    {
>>>>    	struct jz4770_phy *priv = phy_to_jz4770_phy(phy);
>>>>    	u32 reg;
>>>>   @@ -246,9 +241,8 @@ static void x1830_usb_phy_init(struct usb_phy
>>>>  *phy)
>>>>    		USBPCR1_DMPD | USBPCR1_DPPD;
>>>>    	writel(reg, priv->base + REG_USBPCR1_OFFSET);
>>>> 
>>>>   -	reg = USBPCR_IDPULLUP_OTG | USBPCR_VBUSVLDEXT
>>>>  |	USBPCR_TXPREEMPHTUNE |
>>>>   +	return USBPCR_IDPULLUP_OTG | USBPCR_VBUSVLDEXT |
>>>>  USBPCR_TXPREEMPHTUNE |
>>>>    		USBPCR_COMMONONN | USBPCR_POR;
>>>>   -	writel(reg, priv->base + REG_USBPCR_OFFSET);
>>> 
>>>  not a bug fix
>> 
>>  Well, if you don't like my bug fix, next time wait for my 
>> Reviewed-by.
> 
> why so angry? Take a break every once in a while. Besides, someone 
> else
> already sent the oneliner before you ;-)

I'm just pissed that this patch has not been tested. I don't like 
sloppy work.

> In any case, why should I wait for your Reviewed-by? Get maintainer
> doesn't list you as the maintainer for it. Do you want to update
> MAINTAINERS by any chance?

Yes, I thought I was (I'm maintainer of all Ingenic drivers), that also 
explains why I wasn't Cc'd for the oneliner patch you mentioned...

IIRC Zhou has a patch to move the driver to drivers/phy/, I'll add 
myself as maintainer once it's moved there.

-Paul