netdev - Re: [PATCH RFC net-next 05/12] net: dsa: bcm_sf2: convert to phylink_generic

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]
Date:   Fri, 3 Dec 2021 20:18:22 -0800
From:   Florian Fainelli <f.fainelli@...il.com>
To:     "Russell King (Oracle)" <rmk+kernel@...linux.org.uk>,
        Andrew Lunn <andrew@...n.ch>,
        Vivien Didelot <vivien.didelot@...il.com>,
        Vladimir Oltean <olteanv@...il.com>,
        Alexandre Belloni <alexandre.belloni@...tlin.com>,
        Claudiu Manoil <claudiu.manoil@....com>,
        George McCollister <george.mccollister@...il.com>,
        Hauke Mehrtens <hauke@...ke-m.de>,
        Kurt Kanzenbach <kurt@...utronix.de>,
        Vladimir Oltean <vladimir.oltean@....com>,
        Woojung Huh <woojung.huh@...rochip.com>
Cc:     "David S. Miller" <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>, netdev@...r.kernel.org,
        UNGLinuxDriver@...rochip.com
Subject: Re: [PATCH RFC net-next 05/12] net: dsa: bcm_sf2: convert to
 phylink_generic_validate()

On 12/3/21 12:03 PM, Florian Fainelli wrote:
> On 11/24/21 9:52 AM, Russell King (Oracle) wrote:
>> Populate the supported interfaces and MAC capabilities for the bcm_sf2
>> DSA switch and remove the old validate implementation to allow DSA to
>> use phylink_generic_validate() for this switch driver.
>>
>> The exclusion of Gigabit linkmodes for MII and Reverse MII links is
>> handled within phylink_generic_validate() in phylink, so there is no
>> need to make them conditional on the interface mode in the driver.
>>
>> Signed-off-by: Russell King (Oracle) <rmk+kernel@...linux.org.uk>
> 
> Tested-by: Florian Fainelli <f.fainelli@...il.com>
> 
> but it looks like the fixed link ports are reporting some pretty strange
> advertisement values one of my two platforms running the same kernel image:

We would want to amend your patch with something that caters a bit more
towards how the ports have been configured:

diff --git a/drivers/net/dsa/bcm_sf2.c b/drivers/net/dsa/bcm_sf2.c
index d6ef0fb0d943..88933c3feddd 100644
--- a/drivers/net/dsa/bcm_sf2.c
+++ b/drivers/net/dsa/bcm_sf2.c
@@ -675,12 +675,18 @@ static u32 bcm_sf2_sw_get_phy_flags(struct
dsa_switch *ds, int port)
  static void bcm_sf2_sw_get_caps(struct dsa_switch *ds, int port,
                                 struct phylink_config *config)
  {
-       __set_bit(PHY_INTERFACE_MODE_MII, config->supported_interfaces);
-       __set_bit(PHY_INTERFACE_MODE_REVMII, config->supported_interfaces);
-       __set_bit(PHY_INTERFACE_MODE_GMII, config->supported_interfaces);
-       __set_bit(PHY_INTERFACE_MODE_INTERNAL,
config->supported_interfaces);
-       __set_bit(PHY_INTERFACE_MODE_MOCA, config->supported_interfaces);
-       phy_interface_set_rgmii(config->supported_interfaces);
+       struct bcm_sf2_priv *priv = bcm_sf2_to_priv(ds);
+
+       if (priv->int_phy_mask & BIT(port))
+               __set_bit(PHY_INTERFACE_MODE_INTERNAL,
config->supported_interfaces);
+       else if (priv->moca_port == port)
+               __set_bit(PHY_INTERFACE_MODE_MOCA,
config->supported_interfaces);
+       else {
+               __set_bit(PHY_INTERFACE_MODE_MII,
config->supported_interfaces);
+               __set_bit(PHY_INTERFACE_MODE_REVMII,
config->supported_interfaces);
+               __set_bit(PHY_INTERFACE_MODE_GMII,
config->supported_interfaces);
+               phy_interface_set_rgmii(config->supported_interfaces);
+       }

         config->mac_capabilities = MAC_ASYM_PAUSE | MAC_SYM_PAUSE |
                 MAC_10 | MAC_100 | MAC_1000;

Now, with respect to the fixed link ports reporting 1000baseKX/Full this 
is introduced by switching to your patch, it works before and it 
"breaks" after.

The first part that is a bit weird is that we seem to be calling 
phylink_generic_validate() twice in a row from the same call site.

For fixed link ports, instead of masking with what the fixed link 
actually supports, we seem to be using a supported mask which is all 1s 
which seems a bit excessive for a fixed link.

This is an excerpt with the internal PHY:

[    4.210890] brcm-sf2 f0b00000.ethernet_switch gphy (uninitialized): 
Calling phylink_generic_validate
[    4.220063] before phylink_get_linkmodes: 0000000,00000000,00010fc0
[    4.226357] phylink_get_linkmodes: caps: 0xffffffff mac_capabilities: 
0xff
[    4.233258] after phylink_get_linkmodes: c000018,00000200,00036fff
[    4.239463] before anding supported with mask: 0000000,00000000,000062ff
[    4.246189] after anding supported with mask: 0000000,00000000,000062ff
[    4.252829] before anding advertising with mask: 
c000018,00000200,00036fff
[    4.259729] after anding advertising with mask: c000018,00000200,00036fff
[    4.266546] brcm-sf2 f0b00000.ethernet_switch gphy (uninitialized): 
PHY [f0b403c0.mdio--1:05] driver [Broadcom BCM7445] (irq=POLL)

and this is what a fixed link port looks like:

[    4.430765] brcm-sf2 f0b00000.ethernet_switch rgmii_2 
(uninitialized): Calling phylink_generic_validate
[    4.440205] before phylink_get_linkmodes: 0000000,00000000,00010fc0
[    4.446500] phylink_get_linkmodes: caps: 0xff mac_capabilities: 0xff
[    4.452880] after phylink_get_linkmodes: c000018,00000200,00036fff
[    4.459085] before anding supported with mask: fffffff,ffffffff,ffffffff
[    4.465811] after anding supported with mask: c000018,00000200,00036fff
[    4.472450] before anding advertising with mask: 
c000018,00000200,00036fff
[    4.479349] after anding advertising with mask: c000018,00000200,00036fff

or maybe the problem is with phylink_get_ksettings... ran out of time 
tonight to look further into it.
-- 
Florian