| Message ID | 20200717132135.361267-1-mark.einon@gmail.com |
|---|---|
| State | Accepted |
| Delegated to: | David Miller |
| Headers | show |
| Series | net: ethernet: et131x: Remove redundant register read | expand |
On Fri, Jul 17, 2020 at 02:21:35PM +0100, Mark Einon wrote: > Following the removal of an unused variable assignment (remove > unused variable 'pm_csr') the associated register read can also go, > as the read also occurs in the subsequent et1310_in_phy_coma() > call. Hi Mark Do you have any hardware documentation which indicates these read are not required? Have you looked back through the git history to see if there are any comments about these read? Hardware reads which appear pointless are sometimes very important to actually make the hardware work. Andrew
On Fri, 2020-07-17 at 15:40 +0200, Andrew Lunn wrote: > On Fri, Jul 17, 2020 at 02:21:35PM +0100, Mark Einon wrote: > > Following the removal of an unused variable assignment (remove > > unused variable 'pm_csr') the associated register read can also go, > > as the read also occurs in the subsequent > > call. > > Hi Mark > > Do you have any hardware documentation which indicates these read are > not required? Have you looked back through the git history to see if > there are any comments about these read? > > Hardware reads which appear pointless are sometimes very important to > actually make the hardware work. > > Andrew Hi Andrew, Yes - I'm aware of such effects. In the original vendor driver ( https://gitlab.com/einonm/Legacy-et131x) the read of this register ( pm_phy_sw_coma) is not wrapped in a function call and is always called once when needed. Also in the current kernel driver et1310_in_phy_coma() is called a few other times without the removed read being made. The datasheet I have for a similar device (et1011) doesn't say anything other than the register should be read/write. So I think this is a safe thing to do. Best regards, Mark
From: Mark Einon <mark.einon@gmail.com> Date: Fri, 17 Jul 2020 14:21:35 +0100 > Following the removal of an unused variable assignment (remove > unused variable 'pm_csr') the associated register read can also go, > as the read also occurs in the subsequent et1310_in_phy_coma() > call. > > Signed-off-by: Mark Einon <mark.einon@gmail.com> Applied to net-next.
On Fri, Jul 17, 2020 at 04:21:51PM +0100, Mark Einon wrote: > On Fri, 2020-07-17 at 15:40 +0200, Andrew Lunn wrote: > > On Fri, Jul 17, 2020 at 02:21:35PM +0100, Mark Einon wrote: > > > Following the removal of an unused variable assignment (remove > > > unused variable 'pm_csr') the associated register read can also go, > > > as the read also occurs in the subsequent > > > call. > > > > Hi Mark > > > > Do you have any hardware documentation which indicates these read are > > not required? Have you looked back through the git history to see if > > there are any comments about these read? > > > > Hardware reads which appear pointless are sometimes very important to > > actually make the hardware work. > > > > Andrew > > Hi Andrew, > > Yes - I'm aware of such effects. In the original vendor driver ( > https://gitlab.com/einonm/Legacy-et131x) the read of this register ( > pm_phy_sw_coma) is not wrapped in a function call and is always called > once when needed. > > Also in the current kernel driver et1310_in_phy_coma() is called a few > other times without the removed read being made. > > The datasheet I have for a similar device (et1011) doesn't say anything > other than the register should be read/write. > > So I think this is a safe thing to do. Hi Mark It is good to include this sort of information in the commit message. It makes it clear you have considered this, and it stops people like me asking... Andrew
diff --git a/drivers/net/ethernet/agere/et131x.c b/drivers/net/ethernet/agere/et131x.c index 8806e1e4c20f..41f8821f792d 100644 --- a/drivers/net/ethernet/agere/et131x.c +++ b/drivers/net/ethernet/agere/et131x.c @@ -983,7 +983,6 @@ static void et1310_setup_device_for_multicast(struct et131x_adapter *adapter) } /* Write out the new hash to the device */ - readl(&adapter->regs->global.pm_csr); if (!et1310_in_phy_coma(adapter)) { writel(hash1, &rxmac->multi_hash1); writel(hash2, &rxmac->multi_hash2); @@ -1023,7 +1022,6 @@ static void et1310_setup_device_for_unicast(struct et131x_adapter *adapter) (adapter->addr[4] << ET_RX_UNI_PF_ADDR1_5_SHIFT) | adapter->addr[5]; - readl(&adapter->regs->global.pm_csr); if (!et1310_in_phy_coma(adapter)) { writel(uni_pf1, &rxmac->uni_pf_addr1); writel(uni_pf2, &rxmac->uni_pf_addr2); @@ -3444,7 +3442,6 @@ static irqreturn_t et131x_isr(int irq, void *dev_id) /* Tell the device to send a pause packet via the back * pressure register (bp req and bp xon/xoff) */ - readl(&iomem->global.pm_csr); if (!et1310_in_phy_coma(adapter)) writel(3, &iomem->txmac.bp_ctrl); }
Following the removal of an unused variable assignment (remove unused variable 'pm_csr') the associated register read can also go, as the read also occurs in the subsequent et1310_in_phy_coma() call. Signed-off-by: Mark Einon <mark.einon@gmail.com> --- drivers/net/ethernet/agere/et131x.c | 3 --- 1 file changed, 3 deletions(-)