e1000: Initialize phy_data to avoid unexpected values
Subsystems (3)
PATCH 1/1 e1000: Initialize phy_data to avoid unexpected values
From: Rongguang Wei <[email protected]>
The phy_data variable is not initialized. If e1000_read_phy_reg returns an error, phy_data will not point to a valid value from the PHY register, which may cause the regs_buff array to be populated with unexpected values.
--- drivers/net/ethernet/intel/e1000/e1000_ethtool.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/intel/e1000/e1000_ethtool.c +1 -1
@@ -327,7 +327,7 @@ static void e1000_get_regs(struct net_device *netdev, struct ethtool_regs *regs,struct e1000_adapter *adapter =netdev_priv (netdev);struct e1000_hw *hw = &adapter->hw;u32 *regs_buff = p; -u16 phy_data; +u16 phy_data =0 ;memset (p,0 ,E1000_REGS_LEN *sizeof (u32));
Signature · 1 line
2.25.1
From: Rongguang Wei <[email protected]>
Sent: Friday, June 12, 2026 10:04 AM
Quoted 6 lines
From: Rongguang Wei <[email protected]>
The
phy_datavariable is not initialized. Ife1000_read_phy_regreturns an error,phy_datawill not point to a valid value from the PHY register, which may cause theregs_buffarray to be populated with unexpected values.
Hi,
Sounds like a fix, but i believe we would like to have any real scenario when the issue occurs and how it can be reproduced. If such is provided please target the patch against net tree and add fixes tag.
Quoted 3 lines
Signed-off-by Rongguang Wei <[email protected]>Change-Id I46071b3b21a566f8da650168d38d6968251b077d
i doubt this is a correct kernel commit tag
Quoted 19 lines
---
drivers/net/ethernet/intel/e1000/e1000_ethtool.c| 2 +-1 file changed, 1 insertion(+), 1 deletion(-)drivers/net/ethernet/intel/e1000/e1000_ethtool.c +1
-1@@ -327,7 +327,7 @@ static void e1000_get_regs(struct net_device *netdev, struct ethtool_regs *regs,struct e1000_adapter *adapter =netdev_priv (netdev);struct e1000_hw *hw = &adapter->hw;u32 *regs_buff = p; -u16 phy_data; +u16 phy_data =0 ;memset (p,0 ,E1000_REGS_LEN *sizeof (u32));Signature · 1 line
2.25.1
Quoted 10 lines · Jedrzej
From: Rongguang Wei <[email protected]>
Sent: Friday, June 12, 2026 10:04 AMQuoted 6 lines
From: Rongguang Wei <[email protected]>
The
phy_datavariable is not initialized. Ife1000_read_phy_regreturns an error,phy_datawill not point to a valid value from the PHY register, which may cause theregs_buffarray to be populated with unexpected values.
Quoted 7 lines
Hi,
Sounds like a fix, but i believe we would like to have any real scenario when the issue occurs and how it can be reproduced. If such is provided please target the patch against net tree and add fixes tag.
Hi,
I was not face a real scenario. I just found out there is no check for e1000_read_phy_reg return value when I reading the driver code.
Maybe is better to initialized the value or check the return value of e1000_read_phy_reg.
Quoted 27 lines
Quoted 3 lines
Signed-off-by Rongguang Wei <[email protected]>Change-Id I46071b3b21a566f8da650168d38d6968251b077di doubt this is a correct kernel commit tag
Quoted 19 lines
---
drivers/net/ethernet/intel/e1000/e1000_ethtool.c| 2 +-1 file changed, 1 insertion(+), 1 deletion(-)drivers/net/ethernet/intel/e1000/e1000_ethtool.c +1
-1@@ -327,7 +327,7 @@ static void e1000_get_regs(struct net_device *netdev, struct ethtool_regs *regs,struct e1000_adapter *adapter =netdev_priv (netdev);struct e1000_hw *hw = &adapter->hw;u32 *regs_buff = p; -u16 phy_data; +u16 phy_data =0 ;memset (p,0 ,E1000_REGS_LEN *sizeof (u32));Signature · 1 line
2.25.1
Quoted 9 lines · Rongguang Wei
From: Rongguang Wei <[email protected]>
The
phy_datavariable is not initialized. Ife1000_read_phy_regreturns an error,phy_datawill not point to a valid value from the PHY register, which may cause theregs_buffarray to be populated with unexpected values.Signed-off-by Rongguang Wei <[email protected]>Change-Id I46071b3b21a566f8da650168d38d6968251b077d
What does this Change-Id mean?
Quoted 9 lines
--- drivers/net/ethernet/intel/e1000/e1000_ethtool.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/net/ethernet/intel/e1000/e1000_ethtool.c +0
-0
drivers/net/ethernet/intel/e1000/e1000_ethtool.c static void e1000_get_regs(struct net_device *netdev, struct ethtool_regs *regs, · line 330
Quoted 5 lines
drivers/net/ethernet/intel/e1000/e1000_ethtool.c +1
-1@@ -327,7 +327,7 @@ static void e1000_get_regs(struct net_device *netdev, struct ethtool_regs *regs,struct e1000_adapter *adapter =netdev_priv (netdev);struct e1000_hw *hw = &adapter->hw;u32 *regs_buff = p; -u16 phy_data; +u16 phy_data =0 ;
if (hw->phy_type == e1000_phy_igp) {
e1000_write_phy_reg(hw, IGP01E1000_PHY_PAGE_SELECT,
IGP01E1000_PHY_AGC_A);
e1000_read_phy_reg(hw, IGP01E1000_PHY_AGC_A &
IGP01E1000_PHY_PAGE_SELECT, &phy_data);
regs_buff[13] = (u32)phy_data; /* cable length */Isn't a cable length of 0 also unexpected?
How does this patch actually make the situation better?
Andrew
---
pw-bot: cr
Quoted 8 lines · Andrew Lunn
Quoted 7 lines · Rongguang Wei
From: Rongguang Wei <[email protected]>
The
phy_datavariable is not initialized. Ife1000_read_phy_regreturns an error,phy_datawill not point to a valid value from the PHY register, which may cause theregs_buffarray to be populated with unexpected values.
Quoted 5 lines
Quoted 2 lines
Signed-off-by Rongguang Wei <[email protected]>Change-Id I46071b3b21a566f8da650168d38d6968251b077dWhat does this Change-Id mean?
Sorry, it just a auto generate id when I push this patch on my own repos. I forget to delete.
Quoted 22 lines
Quoted 9 lines
--- drivers/net/ethernet/intel/e1000/e1000_ethtool.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/net/ethernet/intel/e1000/e1000_ethtool.c +0
-0drivers/net/ethernet/intel/e1000/e1000_ethtool.c static void e1000_get_regs(struct net_device *netdev, struct ethtool_regs *regs, · line 330
Quoted 5 lines
drivers/net/ethernet/intel/e1000/e1000_ethtool.c +1
-1@@ -327,7 +327,7 @@ static void e1000_get_regs(struct net_device *netdev, struct ethtool_regs *regs,struct e1000_adapter *adapter =netdev_priv (netdev);struct e1000_hw *hw = &adapter->hw;u32 *regs_buff = p; -u16 phy_data; +u16 phy_data =0 ;if (hw->phy_type == e1000_phy_igp) { e1000_write_phy_reg(hw, IGP01E1000_PHY_PAGE_SELECT, IGP01E1000_PHY_AGC_A); e1000_read_phy_reg(hw, IGP01E1000_PHY_AGC_A & IGP01E1000_PHY_PAGE_SELECT, &phy_data); regs_buff[13] = (u32)phy_data; /* cable length */
Quoted 4 lines
Isn't a cable length of 0 also unexpected?
How does this patch actually make the situation better?
Uninitialized variables may be initialized to 0 by the system, explicit initialization is performed to avoid accidents.
The 0 is from e1000_read_phy_reg_ex in e1000_main.c and e1000_power_down_phy when use e1000_read_phy_reg function the last paramenters is initialized 0. So I used this value.
There are many other function which use e1000_read_phy_reg also not initialize the last paramenters eg. e1000_phy_reset_clk_and_crs. I can do it in V2.
Quoted 5 lines
Andrew
---
pw-bot: cr
Quoted 6 lines
Quoted 4 lines
Isn't a cable length of 0 also unexpected?
How does this patch actually make the situation better?
Uninitialized variables may be initialized to 0 by the system, explicit initialization is performed to avoid accidents.
That is still not an explanation why 0 is better.
Maybe you want to change this from a void function to something which can return an error code?
But also consider the history of this driver. When did anybody do any serious development work on it? Does it makes sense to do work on it? I assume you have the hardware and can test the changes you are making.
Andrew