| 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
| ||
|
Message-ID: <f59f9386-8c65-4e56-b77a-82366e2d2821@intel.com>
Date: Tue, 26 Sep 2023 08:23:06 +0300
From: "Neftin, Sasha" <sasha.neftin@...el.com>
To: Prasad Koya <prasad@...sta.com>, Andrew Lunn <andrew@...n.ch>
CC: <intel-wired-lan@...ts.osuosl.org>, <kuba@...nel.org>,
<davem@...emloft.net>, <pabeni@...hat.com>, <jesse.brandeburg@...el.com>,
<anthony.l.nguyen@...el.com>, <netdev@...r.kernel.org>, "lifshits, Vitaly"
<vitaly.lifshits@...el.com>, <edumazet@...gle.com>, "Ruinskiy, Dima"
<dima.ruinskiy@...el.com>
Subject: Re: [PATCH] [iwl-net] Revert "igc: set TP bit in 'supported' and
'advertising' fields of ethtool_link_ksettings"
On 25/09/2023 22:40, Prasad Koya wrote:
> Hi,
>
> Here is the ethtool output before and after changing the speed with the
> commit 9ac3fc2f42e5ffa1e927dcbffb71b15fa81459e2:
>
> -bash-4.2# ethtool ma1
> Settings for ma1:
> Supported ports: [ TP ]
> Supported link modes: 10baseT/Half 10baseT/Full
> 100baseT/Half 100baseT/Full
> 1000baseT/Full
> 2500baseT/Full
> Supported pause frame use: Symmetric
> Supports auto-negotiation: Yes
> Supported FEC modes: Not reported
> Advertised link modes: 10baseT/Half 10baseT/Full
> 100baseT/Half 100baseT/Full
> 1000baseT/Full
> 2500baseT/Full
> Advertised pause frame use: Symmetric
> Advertised auto-negotiation: Yes
> Advertised FEC modes: Not reported
> Speed: 1000Mb/s
> Duplex: Full
> Auto-negotiation: on
> Port: Twisted Pair
> PHYAD: 0
> Transceiver: internal
> MDI-X: off (auto)
> Supports Wake-on: pumbg
> Wake-on: d
> Current message level: 0x00000007 (7)
> drv probe link
> Link detected: yes
> -bash-4.2#
> -bash-4.2# ethtool -s ma1 speed 100 duplex full autoneg on
> -bash-4.2#
> -bash-4.2# ethtool ma1
> Settings for ma1:
> Supported ports: [ TP ]
> Supported link modes: 10baseT/Half 10baseT/Full
> 100baseT/Half 100baseT/Full
> 1000baseT/Full
> 2500baseT/Full
> Supported pause frame use: Symmetric
> Supports auto-negotiation: Yes
> Supported FEC modes: Not reported
> Advertised link modes: 100baseT/Full
> 2500baseT/Full
> Advertised pause frame use: Symmetric
> Advertised auto-negotiation: Yes
> Advertised FEC modes: Not reported
> Speed: 100Mb/s
> Duplex: Full
> Auto-negotiation: on
> Port: Twisted Pair
> PHYAD: 0
> Transceiver: internal
> MDI-X: off (auto)
> Supports Wake-on: pumbg
> Wake-on: d
> Current message level: 0x00000007 (7)
> drv probe link
> Link detected: yes
> -bash-4.2#
>
> With the patch reverted:
>
> -bash-4.2# ethtool -s ma1 speed 100 duplex full autoneg on
> -bash-4.2#
> -bash-4.2# ethtool ma1
> Settings for ma1:
> Supported ports: [ TP ]
> Supported link modes: 10baseT/Half 10baseT/Full
> 100baseT/Half 100baseT/Full
> 1000baseT/Full
> 2500baseT/Full
> Supported pause frame use: Symmetric
> Supports auto-negotiation: Yes
> Supported FEC modes: Not reported
> Advertised link modes: 100baseT/Full
> Advertised pause frame use: Symmetric
> Advertised auto-negotiation: Yes
> Advertised FEC modes: Not reported
> Speed: 100Mb/s
> Duplex: Full
> Port: Twisted Pair
> PHYAD: 0
> Transceiver: internal
> Auto-negotiation: on
> MDI-X: off (auto)
> Supports Wake-on: pumbg
> Wake-on: d
> Current message level: 0x00000007 (7)
> drv probe link
> Link detected: yes
> -bash-4.2#
>
> with the patch enabled:
> ==================
>
> Default 'advertising' field is: 0x8000000020ef
> ie., 10Mbps_half, 10Mbps_full, 100Mbps_half, 100Mbps_full,
> 1000Mbps_full, Autoneg, TP, Pause and 2500Mbps_full bits set.
>
> and 'hw->phy.autoneg_advertised' is 0xaf
>
> During "ethtool -s ma1 speed 100 duplex full autoneg on"
>
> ethtool sends 'advertising' as 0x20c8 ie., 100Mbps_full, Autoneg, TP,
> Pause bits set which are correct.
>
> However, to reset the link with new 'advertising' bits, code takes this
> path:
>
> [ 255.073847] igc_setup_copper_link+0x73c/0x750
> [ 255.073851] igc_setup_link+0x4a/0x170
> [ 255.073852] igc_init_hw_base+0x98/0x100
> [ 255.073855] igc_reset+0x69/0xe0
> [ 255.073857] igc_down+0x22b/0x230
> [ 255.073859] igc_ethtool_set_link_ksettings+0x25f/0x270
> [ 255.073863] ethtool_set_link_ksettings+0xa9/0x140
> [ 255.073866] dev_ethtool+0x1236/0x2570
>
> igc_setup_copper_link() calls igc_copper_link_autoneg().
> igc_copper_link_autoneg() changes phy->autoneg_advertised
>
> phy->autoneg_advertised &= phy->autoneg_mask;
>
> and autoneg_mask is IGC_ALL_SPEED_DUPLEX_2500 which is 0xaf:
>
> /* 1Gbps and 2.5Gbps half duplex is not supported, nor spec-compliant. */
> #define ADVERTISE_10_HALF 0x0001
> #define ADVERTISE_10_FULL 0x0002
> #define ADVERTISE_100_HALF 0x0004
> #define ADVERTISE_100_FULL 0x0008
> #define ADVERTISE_1000_HALF 0x0010 /* Not used, just FYI */
> #define ADVERTISE_1000_FULL 0x0020
> #define ADVERTISE_2500_HALF 0x0040 /* Not used, just FYI */
> #define ADVERTISE_2500_FULL 0x0080
>
> #define IGC_ALL_SPEED_DUPLEX_2500 ( \
> ADVERTISE_10_HALF | ADVERTISE_10_FULL | ADVERTISE_100_HALF | \
> ADVERTISE_100_FULL | ADVERTISE_1000_FULL | ADVERTISE_2500_FULL)
>
> so 0x20c8 & 0xaf becomes 0x88 ie., the TP bit (bit 7
> of ethtool_link_mode_bit_indices) in 0x20c8 got interpreted as
> ADVERTISE_2500_FULL. so after igc_reset(), hw->phy.autoneg_advertised is
> 0x88. Post that, 'ethtool <interface>' reports 2500Mbps can also be
> advertised.
>
> @@ -445,9 +451,19 @@ static s32 igc_copper_link_autoneg(struct igc_hw *hw)
> u16 phy_ctrl;
> s32 ret_val;
>
> /* Perform some bounds checking on the autoneg advertisement
> * parameter.
> */
> + if (!(phy->autoneg_advertised & ADVERTISED_2500baseX_Full))
> + phy->autoneg_advertised &= ~ADVERTISE_2500_FULL;
> + if ((phy->autoneg_advertised & ADVERTISED_2500baseX_Full))
> + phy->autoneg_advertised |= ADVERTISE_2500_FULL;
> +
It will introduce more ambiguity. ADVERTISED_2500baseX_Full (is bit 15),
2500 Base-X is a different type not supported by i225/6 parts. i225/6
parts support 2500baseT_Full_BIT (bit 47 in new structure).
Look, ethtool used (same as igb) ethtool_convert_link_mode_to_legacy_u32
method, but there is no option for 2500baseT_Full_BIT. (since i225 only
copper mode, the TP advertisement was omitted intentionally in an
original code, I thought).
> phy->autoneg_advertised &= phy->autoneg_mask;
>
> I see phy->autoneg_advertised modified similarly
> in igc_phy_setup_autoneg() as well.
>
> Above diff works for:
>
> ethtool -s <intf> speed 10/100/1000 duplex full autoneg on
> or
> ethtool -s <intf> advertise 0x3f (0x03 or 0x0f etc)
>
> but I haven't tested on a 2500 Mbps link. ADVERTISE_2500_FULL is there
> only for igc.
>
> Thanks
> Prasad
>
>
> On Sun, Sep 24, 2023 at 7:51 AM Andrew Lunn <andrew@...n.ch
> <mailto:andrew@...n.ch>> wrote:
>
> On Sun, Sep 24, 2023 at 10:09:17AM +0300, Neftin, Sasha wrote:
> > On 22/09/2023 19:38, Prasad Koya wrote:
> > > This reverts commit 9ac3fc2f42e5ffa1e927dcbffb71b15fa81459e2.
> > >
> > > After the command "ethtool -s enps0 speed 100 duplex full
> autoneg on",
> > > i.e., advertise only 100Mbps speed to the peer, "ethtool enps0"
> shows
> > > advertised speeds as 100Mbps and 2500Mbps. Same behavior is seen
> > > when changing the speed to 10Mbps or 1000Mbps.
> > >
> > > This applies to I225/226 parts, which only supports copper mode.
> > > Reverting to original till the ambiguity is resolved.
> > >
> > > Fixes: 9ac3fc2f42e5 ("igc: set TP bit in 'supported' and
> > > 'advertising' fields of ethtool_link_ksettings")
> > > Signed-off-by: Prasad Koya <prasad@...sta.com
> <mailto:prasad@...sta.com>>
> >
> > Acked-by: Sasha Neftin <sasha.neftin@...el.com
> <mailto:sasha.neftin@...el.com>>
> >
> > > ---
> > > drivers/net/ethernet/intel/igc/igc_ethtool.c | 2 --
> > > 1 file changed, 2 deletions(-)
> > >
> > > diff --git a/drivers/net/ethernet/intel/igc/igc_ethtool.c
> b/drivers/net/ethernet/intel/igc/igc_ethtool.c
> > > index 93bce729be76..0e2cb00622d1 100644
> > > --- a/drivers/net/ethernet/intel/igc/igc_ethtool.c
> > > +++ b/drivers/net/ethernet/intel/igc/igc_ethtool.c
> > > @@ -1708,8 +1708,6 @@ static int
> igc_ethtool_get_link_ksettings(struct net_device *netdev,
> > > /* twisted pair */
> > > cmd->base.port = PORT_TP;
> > > cmd->base.phy_address = hw->phy.addr;
> > > - ethtool_link_ksettings_add_link_mode(cmd, supported, TP);
> > > - ethtool_link_ksettings_add_link_mode(cmd, advertising, TP);
>
> This looks very odd. Please can you confirm this revert really does
> make ethtool report the correct advertisement when it has been limited
> to 100Mbps. Because looking at this patch, i have no idea how this is
> going wrong.
Andrew, yes, I can confirm. (revert works).
We need a process fix, but it will be a different patch. Also, I prefer
not to leave upstream code with a regression.
ethtool_convert_link_mode_to_legacy_u32 have no option to work with
2500baseT_Full_BIT (47 > 32). Need to use a new structure for
auto-negotiation advertisement, mask... (not u16). We need to think how
to fix it right.
>
> Andrew
>
Powered by blists - more mailing lists