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  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date:   Tue, 19 May 2020 11:00:13 -0500
From:   Jeremy Linton <jeremy.linton@....com>
To:     netdev@...r.kernel.org
Cc:     andrew@...n.ch, f.fainelli@...il.com, hkallweit1@...il.com,
        linux@...linux.org.uk, davem@...emloft.net,
        linux-kernel@...r.kernel.org,
        Calvin Johnson <calvin.johnson@....nxp.com>
Subject: Re: [PATCH] net: phy: Fix c45 no phy detected logic

Hi,

On 5/14/20 12:00 PM, Jeremy Linton wrote:
> The commit "disregard Clause 22 registers present bit..." clears
> the low bit of the devices_in_package data which is being used
> in get_phy_c45_ids() to determine if a phy/register is responding
> correctly. That check is against 0x1FFFFFFF, but since the low
> bit is always cleared, the check can never be true. This leads to
> detecting c45 phy devices where none exist.
> 
> Lets fix this by also clearing the low bit in the mask to 0x1FFFFFFE.
> This allows us to continue to autoprobe standards compliant devices
> without also gaining a large number of bogus ones.

So, I've been reworking the c45 ID detection logic, with an aim to 
hinting to the scanner that it should fallback to c22 for a given phy 
address (as well as giving it some additional standardized areas to 
probe for phy ids). It turns out that the c22 registers present bit is a 
pretty useful signal that this needs to happen. So, I think this patch 
really should move the BIT(0) sanitation after the MMD detection loop in 
get_phy_c45_ids().

But having dug into this code for a while now, I'm hard pressed to 
understand the case that the original 3b5e74e0afe3 commit fixed. The 
only thing I can see is that the "bug" i'm fixing here was intentionally 
creating bogus phy nodes when the MMDs weren't responding.


Thanks,

> 
> Fixes: 3b5e74e0afe3 ("net: phy: disregard "Clause 22 registers present" bit in get_phy_c45_devs_in_pkg")
> Cc: Heiner Kallweit <hkallweit1@...il.com>
> Signed-off-by: Jeremy Linton <jeremy.linton@....com>
> ---
>   drivers/net/phy/phy_device.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/phy/phy_device.c b/drivers/net/phy/phy_device.c
> index ac2784192472..b93d984d35cc 100644
> --- a/drivers/net/phy/phy_device.c
> +++ b/drivers/net/phy/phy_device.c
> @@ -723,7 +723,7 @@ static int get_phy_c45_ids(struct mii_bus *bus, int addr, u32 *phy_id,
>   		if (phy_reg < 0)
>   			return -EIO;
>   
> -		if ((*devs & 0x1fffffff) == 0x1fffffff) {
> +		if ((*devs & 0x1ffffffe) == 0x1ffffffe) {
>   			/*  If mostly Fs, there is no device there,
>   			 *  then let's continue to probe more, as some
>   			 *  10G PHYs have zero Devices In package,
> @@ -733,7 +733,7 @@ static int get_phy_c45_ids(struct mii_bus *bus, int addr, u32 *phy_id,
>   			if (phy_reg < 0)
>   				return -EIO;
>   			/* no device there, let's get out of here */
> -			if ((*devs & 0x1fffffff) == 0x1fffffff) {
> +			if ((*devs & 0x1ffffffe) == 0x1ffffffe) {
>   				*phy_id = 0xffffffff;
>   				return 0;
>   			} else {
> 

Powered by blists - more mailing lists