[PATCH 5.5 345/367] r8169: fix performance regression related to PCIe max read request size

From: Greg Kroah-Hartman
Date: Mon Feb 10 2020 - 07:47:19 EST


From: Heiner Kallweit <hkallweit1@xxxxxxxxx>

[ Upstream commit 21b5f672fb2eb1366dedc4ac9d32431146b378d3 ]

It turned out that on low performance systems the original change can
cause lower tx performance. On a N3450-based mini-PC tx performance
in iperf3 was reduced from 950Mbps to ~900Mbps. Therefore effectively
revert the original change, just use pcie_set_readrq() now instead of
changing the PCIe capability register directly.

Fixes: 2df49d365498 ("r8169: remove fiddling with the PCIe max read request size")
Signed-off-by: Heiner Kallweit <hkallweit1@xxxxxxxxx>
Signed-off-by: David S. Miller <davem@xxxxxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
---
drivers/net/ethernet/realtek/r8169_main.c | 6 ++++++
1 file changed, 6 insertions(+)

--- a/drivers/net/ethernet/realtek/r8169_main.c
+++ b/drivers/net/ethernet/realtek/r8169_main.c
@@ -3865,15 +3865,18 @@ static void rtl_hw_jumbo_enable(struct r
switch (tp->mac_version) {
case RTL_GIGA_MAC_VER_12:
case RTL_GIGA_MAC_VER_17:
+ pcie_set_readrq(tp->pci_dev, 512);
r8168b_1_hw_jumbo_enable(tp);
break;
case RTL_GIGA_MAC_VER_18 ... RTL_GIGA_MAC_VER_26:
+ pcie_set_readrq(tp->pci_dev, 512);
r8168c_hw_jumbo_enable(tp);
break;
case RTL_GIGA_MAC_VER_27 ... RTL_GIGA_MAC_VER_28:
r8168dp_hw_jumbo_enable(tp);
break;
case RTL_GIGA_MAC_VER_31 ... RTL_GIGA_MAC_VER_33:
+ pcie_set_readrq(tp->pci_dev, 512);
r8168e_hw_jumbo_enable(tp);
break;
default:
@@ -3903,6 +3906,9 @@ static void rtl_hw_jumbo_disable(struct
break;
}
rtl_lock_config_regs(tp);
+
+ if (pci_is_pcie(tp->pci_dev) && tp->supports_gmii)
+ pcie_set_readrq(tp->pci_dev, 4096);
}

static void rtl_jumbo_config(struct rtl8169_private *tp, int mtu)