* [PATCH v1 1/3] net: stmmac:sti: Add STi SOC glue driver.
From: srinivas.kandagatla @ 2014-02-03 12:01 UTC (permalink / raw)
To: netdev
Cc: Mark Rutland, devicetree, Russell King, kernel, Pawel Moll,
Ian Campbell, Srinivas Kandagatla, linux-doc, linux-kernel,
Stuart Menefy, Rob Herring, Rob Landley, Kumar Gala,
Giuseppe Cavallaro, davem, linux-arm-kernel
In-Reply-To: <1391428787-27143-1-git-send-email-srinivas.kandagatla@st.com>
From: Srinivas Kandagatla <srinivas.kandagatla@st.com>
STi series SOCs have a glue layer on top of the synopsis gmac IP, this
glue layer needs to be configured before the gmac driver starts using
the IP.
This patch adds a support to this glue layer which is configured via
stmmac setup, init, exit callbacks.
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@st.com>
---
.../devicetree/bindings/net/sti-dwmac.txt | 58 ++++
drivers/net/ethernet/stmicro/stmmac/Kconfig | 11 +
drivers/net/ethernet/stmicro/stmmac/Makefile | 1 +
drivers/net/ethernet/stmicro/stmmac/dwmac-sti.c | 331 ++++++++++++++++++++
drivers/net/ethernet/stmicro/stmmac/stmmac.h | 3 +
.../net/ethernet/stmicro/stmmac/stmmac_platform.c | 5 +
6 files changed, 409 insertions(+)
create mode 100644 Documentation/devicetree/bindings/net/sti-dwmac.txt
create mode 100644 drivers/net/ethernet/stmicro/stmmac/dwmac-sti.c
diff --git a/Documentation/devicetree/bindings/net/sti-dwmac.txt b/Documentation/devicetree/bindings/net/sti-dwmac.txt
new file mode 100644
index 0000000..3dd3d0b
--- /dev/null
+++ b/Documentation/devicetree/bindings/net/sti-dwmac.txt
@@ -0,0 +1,58 @@
+STMicroelectronics SoC DWMAC glue layer controller
+
+The device node has following properties.
+
+Required properties:
+ - compatible : Can be "st,stih415-dwmac", "st,stih416-dwmac" or
+ "st,stid127-dwmac".
+ - reg : Offset of the glue configuration register map in system
+ configuration regmap pointed by st,syscon property and size.
+
+ - reg-names : Should be "sti-ethconf".
+
+ - st,syscon : Should be phandle to system configuration node which
+ encompases this glue registers.
+
+ - st,tx-retime-src: On STi Parts for Giga bit speeds, 125Mhz clocks can be
+ wired up in from different sources. One via TXCLK pin and other via CLK_125
+ pin. This wiring is totally board dependent. However the retiming glue
+ logic should be configured accordingly. Possible values for this property
+
+ "txclk" - if 125Mhz clock is wired up via txclk line.
+ "clk_125" - if 125Mhz clock is wired up via clk_125 line.
+
+ This property is only valid for Giga bit setup( GMII, RGMII), and it is
+ un-used for non-giga bit (MII and RMII) setups. Also note that internal
+ clockgen can not generate stable 125Mhz clock.
+
+ - st,ext-phyclk: This boolean property indicates who is generating the clock
+ for tx and rx. This property is only valid for RMII case where the clock can
+ be generated from the MAC or PHY.
+
+ - clock-names: should be "sti-ethclk".
+ - clocks: Should point to ethernet clockgen which can generate phyclk.
+
+
+Example:
+
+ethernet0: dwmac@fe810000 {
+ device_type = "network";
+ compatible = "st,stih416-dwmac", "snps,dwmac", "snps,dwmac-3.710";
+ reg = <0xfe810000 0x8000>, <0x8bc 0x4>;
+ reg-names = "stmmaceth", "sti-ethconf";
+ interrupts = <0 133 0>, <0 134 0>, <0 135 0>;
+ interrupt-names = "macirq", "eth_wake_irq", "eth_lpi";
+ phy-mode = "mii";
+
+ st,syscon = <&syscfg_rear>;
+
+ snps,pbl = <32>;
+ snps,mixed-burst;
+
+ resets = <&softreset STIH416_ETH0_SOFTRESET>;
+ reset-names = "stmmaceth";
+ pinctrl-0 = <&pinctrl_mii0>;
+ pinctrl-names = "default";
+ clocks = <&CLK_S_GMAC0_PHY>;
+ clock-names = "stmmaceth";
+};
diff --git a/drivers/net/ethernet/stmicro/stmmac/Kconfig b/drivers/net/ethernet/stmicro/stmmac/Kconfig
index e2f202e..f2d7c70 100644
--- a/drivers/net/ethernet/stmicro/stmmac/Kconfig
+++ b/drivers/net/ethernet/stmicro/stmmac/Kconfig
@@ -37,6 +37,17 @@ config DWMAC_SUNXI
stmmac device driver. This driver is used for A20/A31
GMAC ethernet controller.
+config DWMAC_STI
+ bool "STi GMAC support"
+ depends on STMMAC_PLATFORM && ARCH_STI
+ default y
+ ---help---
+ Support for ethernet controller on STi SOCs.
+
+ This selects STi SoC glue layer support for the stmmac
+ device driver. This driver is used on for the STi series
+ SOCs GMAC ethernet controller.
+
config STMMAC_PCI
bool "STMMAC PCI bus support"
depends on STMMAC_ETH && PCI
diff --git a/drivers/net/ethernet/stmicro/stmmac/Makefile b/drivers/net/ethernet/stmicro/stmmac/Makefile
index ecadece..dcef287 100644
--- a/drivers/net/ethernet/stmicro/stmmac/Makefile
+++ b/drivers/net/ethernet/stmicro/stmmac/Makefile
@@ -2,6 +2,7 @@ obj-$(CONFIG_STMMAC_ETH) += stmmac.o
stmmac-$(CONFIG_STMMAC_PLATFORM) += stmmac_platform.o
stmmac-$(CONFIG_STMMAC_PCI) += stmmac_pci.o
stmmac-$(CONFIG_DWMAC_SUNXI) += dwmac-sunxi.o
+stmmac-$(CONFIG_DWMAC_STI) += dwmac-sti.o
stmmac-objs:= stmmac_main.o stmmac_ethtool.o stmmac_mdio.o ring_mode.o \
chain_mode.o dwmac_lib.o dwmac1000_core.o dwmac1000_dma.o \
dwmac100_core.o dwmac100_dma.o enh_desc.o norm_desc.o \
diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-sti.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-sti.c
new file mode 100644
index 0000000..d87584cb
--- /dev/null
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac-sti.c
@@ -0,0 +1,331 @@
+/**
+ * dwmac-sti.c - STMicroelectronics DWMAC Specific Glue layer
+ *
+ * Copyright (C) 2003-2014 STMicroelectronics (R&D) Limited
+ * Author: Srinivas Kandagatla <srinivas.kandagatla@st.com>
+ *
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/platform_device.h>
+#include <linux/stmmac.h>
+#include <linux/phy.h>
+#include <linux/mfd/syscon.h>
+#include <linux/regmap.h>
+#include <linux/clk.h>
+#include <linux/of.h>
+#include <linux/of_net.h>
+
+/**
+ * STi GMAC glue logic.
+ * --------------------
+ *
+ * _
+ * | \
+ * --------|0 \ ETH_SEL_INTERNAL_NOTEXT_PHYCLK
+ * phyclk | |___________________________________________
+ * | | | (phyclk-in)
+ * --------|1 / |
+ * int-clk |_ / |
+ * | _
+ * | | \
+ * |_______|0 \ ETH_SEL_TX_RETIME_CLK
+ * | |___________________________
+ * | | (tx-retime-clk)
+ * _______|1 /
+ * | |_ /
+ * _ |
+ * | \ |
+ * --------|0 \ |
+ * clk_125 | |__|
+ * | | ETH_SEL_TXCLK_NOT_CLK125
+ * --------|1 /
+ * txclk |_ /
+ *
+ *
+ * ETH_SEL_INTERNAL_NOTEXT_PHYCLK is valid only for RMII where PHY can
+ * generate 50MHz clock or MAC can generate it.
+ * This bit is configured by "st,ext-phyclk" property.
+ *
+ * ETH_SEL_TXCLK_NOT_CLK125 is only valid for gigabit modes, where the 125Mhz
+ * clock either comes from clk-125 pin or txclk pin. This configuration is
+ * totally driven by the board wiring. This bit is configured by
+ * "st,tx-retime-src" property.
+ *
+ * TXCLK configuration is different for different phy interface modes
+ * and changes according to link speed in modes like RGMII.
+ *
+ * Below table summarizes the clock requirement and clock sources for
+ * supported phy interface modes with link speeds.
+ * ________________________________________________
+ *| PHY_MODE | 1000 Mbit Link | 100 Mbit Link |
+ * ------------------------------------------------
+ *| MII | n/a | 25Mhz |
+ *| | | txclk |
+ * ------------------------------------------------
+ *| GMII | 125Mhz | 25Mhz |
+ *| | clk-125/txclk | txclk |
+ * ------------------------------------------------
+ *| RGMII | 125Mhz | 25Mhz |
+ *| | clk-125/txclk | clkgen |
+ * ------------------------------------------------
+ *| RMII | n/a | 25Mhz |
+ *| | |clkgen/phyclk-in |
+ * ------------------------------------------------
+ *
+ * TX lines are always retimed with a clk, which can vary depending
+ * on the board configuration. Below is the table of these bits
+ * in eth configuration register depending on source of retime clk.
+ *
+ *---------------------------------------------------------------
+ * src | tx_rt_clk | int_not_ext_phyclk | txclk_n_clk125|
+ *---------------------------------------------------------------
+ * txclk | 0 | n/a | 1 |
+ *---------------------------------------------------------------
+ * ck_125| 0 | n/a | 0 |
+ *---------------------------------------------------------------
+ * phyclk| 1 | 0 | n/a |
+ *---------------------------------------------------------------
+ * clkgen| 1 | 1 | n/a |
+ *---------------------------------------------------------------
+ */
+
+ /* Register definition */
+
+ /* 3 bits [8:6]
+ * [6:6] ETH_SEL_TXCLK_NOT_CLK125
+ * [7:7] ETH_SEL_INTERNAL_NOTEXT_PHYCLK
+ * [8:8] ETH_SEL_TX_RETIME_CLK
+ *
+ */
+
+#define TX_RETIME_SRC_MASK GENMASK(8, 6)
+#define ETH_SEL_TX_RETIME_CLK BIT(8)
+#define ETH_SEL_INTERNAL_NOTEXT_PHYCLK BIT(7)
+#define ETH_SEL_TXCLK_NOT_CLK125 BIT(6)
+
+#define ENMII_MASK GENMASK(5, 5)
+#define ENMII BIT(5)
+
+/**
+ * 3 bits [4:2]
+ * 000-GMII/MII
+ * 001-RGMII
+ * 010-SGMII
+ * 100-RMII
+*/
+#define MII_PHY_SEL_MASK GENMASK(4, 2)
+#define ETH_PHY_SEL_RMII BIT(4)
+#define ETH_PHY_SEL_SGMII BIT(3)
+#define ETH_PHY_SEL_RGMII BIT(2)
+#define ETH_PHY_SEL_GMII 0x0
+#define ETH_PHY_SEL_MII 0x0
+
+#define IS_PHY_IF_MODE_RGMII(iface) (iface == PHY_INTERFACE_MODE_RGMII || \
+ iface == PHY_INTERFACE_MODE_RGMII_ID || \
+ iface == PHY_INTERFACE_MODE_RGMII_RXID || \
+ iface == PHY_INTERFACE_MODE_RGMII_TXID)
+
+#define IS_PHY_IF_MODE_GBIT(iface) (IS_PHY_IF_MODE_RGMII(iface) || \
+ iface == PHY_INTERFACE_MODE_GMII)
+
+struct sti_dwmac {
+ int interface;
+ bool ext_phyclk;
+ bool is_tx_retime_src_clk_125;
+ struct clk *clk;
+ int reg;
+ struct device *dev;
+ struct regmap *regmap;
+};
+
+static u32 phy_intf_sels[] = {
+ [PHY_INTERFACE_MODE_MII] = ETH_PHY_SEL_MII,
+ [PHY_INTERFACE_MODE_GMII] = ETH_PHY_SEL_GMII,
+ [PHY_INTERFACE_MODE_RGMII] = ETH_PHY_SEL_RGMII,
+ [PHY_INTERFACE_MODE_RGMII_ID] = ETH_PHY_SEL_RGMII,
+ [PHY_INTERFACE_MODE_SGMII] = ETH_PHY_SEL_SGMII,
+ [PHY_INTERFACE_MODE_RMII] = ETH_PHY_SEL_RMII,
+};
+
+enum {
+ TX_RETIME_SRC_NA = 0,
+ TX_RETIME_SRC_TXCLK = 1,
+ TX_RETIME_SRC_CLK_125,
+ TX_RETIME_SRC_PHYCLK,
+ TX_RETIME_SRC_CLKGEN,
+};
+
+static const char * const tx_retime_srcs[] = {
+ [TX_RETIME_SRC_NA] = "",
+ [TX_RETIME_SRC_TXCLK] = "txclk",
+ [TX_RETIME_SRC_CLK_125] = "clk_125",
+ [TX_RETIME_SRC_PHYCLK] = "phyclk",
+ [TX_RETIME_SRC_CLKGEN] = "clkgen",
+};
+
+static u32 tx_retime_val[] = {
+ [TX_RETIME_SRC_TXCLK] = ETH_SEL_TXCLK_NOT_CLK125,
+ [TX_RETIME_SRC_CLK_125] = 0x0,
+ [TX_RETIME_SRC_PHYCLK] = ETH_SEL_TX_RETIME_CLK,
+ [TX_RETIME_SRC_CLKGEN] = ETH_SEL_TX_RETIME_CLK |
+ ETH_SEL_INTERNAL_NOTEXT_PHYCLK,
+};
+
+static void setup_retime_src(struct sti_dwmac *dwmac, u32 spd)
+{
+ u32 src = 0, freq = 0;
+
+ if (spd == SPEED_100) {
+ if (dwmac->interface == PHY_INTERFACE_MODE_MII ||
+ dwmac->interface == PHY_INTERFACE_MODE_GMII) {
+ src = TX_RETIME_SRC_TXCLK;
+ } else if (dwmac->interface == PHY_INTERFACE_MODE_RMII) {
+ if (dwmac->ext_phyclk) {
+ src = TX_RETIME_SRC_PHYCLK;
+ } else {
+ src = TX_RETIME_SRC_CLKGEN;
+ freq = 50000000;
+ }
+
+ } else if (IS_PHY_IF_MODE_RGMII(dwmac->interface)) {
+ src = TX_RETIME_SRC_CLKGEN;
+ freq = 25000000;
+ }
+
+ if (src == TX_RETIME_SRC_CLKGEN && dwmac->clk)
+ clk_set_rate(dwmac->clk, freq);
+
+ } else if (spd == SPEED_1000) {
+ if (dwmac->is_tx_retime_src_clk_125)
+ src = TX_RETIME_SRC_CLK_125;
+ else
+ src = TX_RETIME_SRC_TXCLK;
+ }
+
+ regmap_update_bits(dwmac->regmap, dwmac->reg,
+ TX_RETIME_SRC_MASK, tx_retime_val[src]);
+}
+
+static void sti_dwmac_exit(struct platform_device *pdev, void *priv)
+{
+ struct sti_dwmac *dwmac = priv;
+
+ if (dwmac->clk)
+ clk_disable_unprepare(dwmac->clk);
+}
+
+static void sti_fix_mac_speed(void *priv, unsigned int spd)
+{
+ struct sti_dwmac *dwmac = priv;
+ setup_retime_src(dwmac, spd);
+ return;
+}
+
+static int sti_dwmac_parse_data(struct sti_dwmac *dwmac,
+ struct platform_device *pdev)
+{
+ struct resource *res;
+ struct device *dev = &pdev->dev;
+ struct device_node *np = dev->of_node;
+ struct regmap *regmap;
+ int err;
+
+ if (!np)
+ return -EINVAL;
+
+ res = platform_get_resource_byname(pdev,
+ IORESOURCE_MEM, "sti-ethconf");
+ if (!res)
+ return -ENODATA;
+
+ regmap = syscon_regmap_lookup_by_phandle(np, "st,syscon");
+ if (IS_ERR(regmap))
+ return PTR_ERR(regmap);
+
+ dwmac->dev = dev;
+ dwmac->interface = of_get_phy_mode(np);
+ dwmac->regmap = regmap;
+ dwmac->reg = res->start;
+ dwmac->ext_phyclk = of_property_read_bool(np, "st,ext-phyclk");
+
+ dwmac->is_tx_retime_src_clk_125 = false;
+
+ if (IS_PHY_IF_MODE_GBIT(dwmac->interface)) {
+ const char *rs;
+ err = of_property_read_string(np, "st,tx-retime-src", &rs);
+ if (err < 0) {
+ dev_err(dev, "st,tx-retime-src not specified\n");
+ return err;
+ }
+
+ if (!strcasecmp(rs, "clk_125"))
+ dwmac->is_tx_retime_src_clk_125 = true;
+
+ }
+
+ dwmac->clk = devm_clk_get(dev, "sti-ethclk");
+
+ if (IS_ERR(dwmac->clk))
+ dwmac->clk = NULL;
+
+ return 0;
+}
+
+static int sti_dwmac_init(struct platform_device *pdev, void *priv)
+{
+ struct sti_dwmac *dwmac = priv;
+ struct regmap *regmap = dwmac->regmap;
+ int iface = dwmac->interface;
+ u32 reg = dwmac->reg;
+ u32 val, spd;
+
+ if (dwmac->clk)
+ clk_prepare_enable(dwmac->clk);
+
+ regmap_update_bits(regmap, reg, MII_PHY_SEL_MASK,
+ phy_intf_sels[iface]);
+
+ val = (iface == PHY_INTERFACE_MODE_REVMII) ? 0 : ENMII;
+ regmap_update_bits(regmap, reg, ENMII_MASK, val);
+
+ if (IS_PHY_IF_MODE_GBIT(iface))
+ spd = SPEED_1000;
+ else
+ spd = SPEED_100;
+
+ setup_retime_src(dwmac, spd);
+
+ return 0;
+}
+
+static void *sti_dwmac_setup(struct platform_device *pdev)
+{
+ struct sti_dwmac *dwmac;
+ int ret;
+
+ dwmac = devm_kzalloc(&pdev->dev, sizeof(*dwmac), GFP_KERNEL);
+ if (!dwmac)
+ return ERR_PTR(-ENOMEM);
+
+ ret = sti_dwmac_parse_data(dwmac, pdev);
+ if (ret) {
+ dev_err(&pdev->dev, "Unable to parse OF data\n");
+ return ERR_PTR(ret);
+ }
+
+ return dwmac;
+}
+
+const struct stmmac_of_data sti_gmac_data = {
+ .fix_mac_speed = sti_fix_mac_speed,
+ .setup = sti_dwmac_setup,
+ .init = sti_dwmac_init,
+ .exit = sti_dwmac_exit,
+};
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac.h b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
index d9af26e..f9e60d7 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
@@ -133,6 +133,9 @@ bool stmmac_eee_init(struct stmmac_priv *priv);
#ifdef CONFIG_DWMAC_SUNXI
extern const struct stmmac_of_data sun7i_gmac_data;
#endif
+#ifdef CONFIG_DWMAC_STI
+extern const struct stmmac_of_data sti_gmac_data;
+#endif
extern struct platform_driver stmmac_pltfr_driver;
static inline int stmmac_register_platform(void)
{
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
index 5884a7d..c61bc72b 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
@@ -33,6 +33,11 @@ static const struct of_device_id stmmac_dt_ids[] = {
#ifdef CONFIG_DWMAC_SUNXI
{ .compatible = "allwinner,sun7i-a20-gmac", .data = &sun7i_gmac_data},
#endif
+#ifdef CONFIG_DWMAC_STI
+ { .compatible = "st,stih415-dwmac", .data = &sti_gmac_data},
+ { .compatible = "st,stih416-dwmac", .data = &sti_gmac_data},
+ { .compatible = "st,stih127-dwmac", .data = &sti_gmac_data},
+#endif
/* SoC specific glue layers should come before generic bindings */
{ .compatible = "st,spear600-gmac"},
{ .compatible = "snps,dwmac-3.610"},
--
1.7.9.5
^ permalink raw reply related
* [PATCH v1 0/3] net: stmmac: Add STi GMAC ethernet
From: srinivas.kandagatla @ 2014-02-03 11:59 UTC (permalink / raw)
To: netdev
Cc: Mark Rutland, devicetree, Russell King, kernel, Pawel Moll,
Ian Campbell, Srinivas Kandagatla, linux-doc, linux-kernel,
Stuart Menefy, Rob Herring, Rob Landley, Kumar Gala,
Giuseppe Cavallaro, davem, linux-arm-kernel
From: Srinivas Kandagatla <srinivas.kandagatla@st.com>
Hi All,
This patch series adds Ethernet support to STi series SOCs STiH415 and STiH416.
STi SOC series integrates dwmac IP from synopsis, however there is a hardware
glue on top of this standard IP, this glue needs to configured before the
actual dwmac can be used. Also the glue logic needs re-configuring when the
link speed changes, This is because the clk source can change as the link
speed.
This patch just adds STi specific callbacks into of_data for configuring the
glue layer.
I have rebased my original patches (http://lkml.org/lkml/2013/11/12/243)
to latest stmmac which updates callbacks to suit glue drivers like this.
These patches are tested on b2000 and B2020 with STiH415 and STiH416.
Thanks,
srini
Srinivas Kandagatla (3):
net: stmmac:sti: Add STi SOC glue driver.
ARM: STi: Add STiH415 ethernet support.
ARM: STi: Add STiH416 ethernet support.
.../devicetree/bindings/net/sti-dwmac.txt | 58 ++++
arch/arm/boot/dts/stih415-clock.dtsi | 14 +
arch/arm/boot/dts/stih415-pinctrl.dtsi | 121 +++++++
arch/arm/boot/dts/stih415.dtsi | 48 +++
arch/arm/boot/dts/stih416-clock.dtsi | 14 +
arch/arm/boot/dts/stih416-pinctrl.dtsi | 109 +++++++
arch/arm/boot/dts/stih416.dtsi | 44 +++
arch/arm/boot/dts/stih41x-b2000.dtsi | 22 ++
arch/arm/boot/dts/stih41x-b2020.dtsi | 26 ++
drivers/net/ethernet/stmicro/stmmac/Kconfig | 11 +
drivers/net/ethernet/stmicro/stmmac/Makefile | 1 +
drivers/net/ethernet/stmicro/stmmac/dwmac-sti.c | 331 ++++++++++++++++++++
drivers/net/ethernet/stmicro/stmmac/stmmac.h | 3 +
.../net/ethernet/stmicro/stmmac/stmmac_platform.c | 5 +
14 files changed, 807 insertions(+)
create mode 100644 Documentation/devicetree/bindings/net/sti-dwmac.txt
create mode 100644 drivers/net/ethernet/stmicro/stmmac/dwmac-sti.c
--
1.7.9.5
^ permalink raw reply
* RE: [PATCH] drivers: staging: lustre: lustre: include: add "__attribute__((packed))" for the related union
From: David Laight @ 2014-02-03 11:54 UTC (permalink / raw)
To: 'James Hogan'
Cc: devel@driverdev.osuosl.org, andreas.dilger@intel.com, Chen Gang,
Greg KH, bergwolf@gmail.com, linux-kernel@vger.kernel.org,
linux-metag@vger.kernel.org, oleg.drokin@intel.com,
jacques-charles.lafoucriere@cea.fr, Antonio Quartulli, netdev,
jinshan.xiong@intel.com, David Miller, 'Dan Carpenter'
In-Reply-To: <52EF7762.2060809@imgtec.com>
From: James Hogan
> On 03/02/14 10:35, David Laight wrote:
> > From: James Hogan
> >> Combining __packed with __aligned(2) does the trick though (__packed
> >> alone sets __aligned(1) which is obviously going to be suboptimal).
...
>
> Meta is also one of those arches, and according to my quick tests,
> __packed alone does correctly make it fall back to byte loads/stores,
> but with __packed __aligned(2) it uses 16bit loads/stores. I've also
> confirmed that with an ARM toolchain (see below for example).
I would either:
1a) Add explicit padding to the relevant structures so that they are
multiple of 4 bytes.
or:
1b) #define some token to "__packed __aligned(2)" before all the structures
that require changing, and use that in there definitions.
This lets you comment on WHY you are doing it.
and:
2) Add a compile-time assert that the structures are the correct size.
Clearly you don't want to mark anything that contains a 32bit value
with __packed __aligned(2).
I'm not at all clear whether you are sometimes using a different compiler.
David
^ permalink raw reply
* Re: [Xen-devel] [PATCH v6] xen/grant-table: Avoid m2p_override during mapping
From: Konrad Rzeszutek Wilk @ 2014-02-03 11:50 UTC (permalink / raw)
To: David Vrabel
Cc: Zoltan Kiss, jonathan.davies, wei.liu2, ian.campbell,
Stefano Stabellini, netdev, Julien Grall, linux-kernel, xen-devel
In-Reply-To: <52EF7618.7030402@citrix.com>
On Mon, Feb 03, 2014 at 10:57:28AM +0000, David Vrabel wrote:
> On 02/02/14 18:52, Zoltan Kiss wrote:
> > On 02/02/14 11:29, Julien Grall wrote:
> >> Hello,
> >>
> >> This patch is breaking Linux compilation on ARM:
> >>
> >> drivers/xen/grant-table.c: In function ‘__gnttab_map_refs’:
> >> drivers/xen/grant-table.c:989:3: error: implicit declaration of
> >> function ‘FOREIGN_FRAME’ [-Werror=implicit-function-declaration]
> >> if (unlikely(!set_phys_to_machine(pfn, FOREIGN_FRAME(mfn)))) {
> >> ^
> >> drivers/xen/grant-table.c: In function ‘__gnttab_unmap_refs’:
> >> drivers/xen/grant-table.c:1054:3: error: implicit declaration of
> >> function ‘get_phys_to_machine’ [-Werror=implicit-function-declaration]
> >> mfn = get_phys_to_machine(pfn);
> >> ^
> >> drivers/xen/grant-table.c:1055:43: error: ‘FOREIGN_FRAME_BIT’
> >> undeclared (first use in this function)
> >> if (mfn == INVALID_P2M_ENTRY || !(mfn & FOREIGN_FRAME_BIT)) {
> >> ^
> >> drivers/xen/grant-table.c:1055:43: note: each undeclared identifier is
> >> reported only once for each function it appears in
> >> drivers/xen/grant-table.c:1068:9: error: too many arguments to
> >> function ‘m2p_remove_override’
> >> mfn);
> >> ^
> >> In file included from include/xen/page.h:4:0,
> >> from drivers/xen/grant-table.c:48:
> >> /local/home/julien/works/midway/linux/arch/arm/include/asm/xen/page.h:106:19:
> >> note: declared here
> >> static inline int m2p_remove_override(struct page *page, bool
> >> clear_pte)
> >> ^
> >> cc1: some warnings being treated as errors
> >
> > Hi,
> >
> > That's bad indeed. I think the best solution is to put those parts
> > behind an #ifdef x86. The ones moved from x86/p2m.c to grant-table.c.
> > David, Stefano, what do you think?
>
> I don't think we want (more) #ifdef CONFIG_X86 in grant-table.c and the
> arch-specific bits will have to factored out into their own functions
> with suitable stubs provided for ARM.
>
> But, this patch went in late and it's clearly not ready. So I think it
> should be reverted and we should aim to get it sorted out for 3.15.
>
> Konrad/Stefano (if you agree) please revert
> 08ece5bb2312b4510b161a6ef6682f37f4eac8a1 and send a pull request.
OK, queued up. I also put on the xen/cr4 patch on the queue - just sent
an email with it.
>
> Konrad, I also think you should look at adding an ARM build to your test
> system (I thought you had this already).
>
> David
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel
^ permalink raw reply
* Re: [PATCH] drivers: staging: lustre: lustre: include: add "__attribute__((packed))" for the related union
From: Chen Gang @ 2014-02-03 11:35 UTC (permalink / raw)
To: Dan Carpenter
Cc: James Hogan, devel-gWbeCf7V1WCQmaza687I9mD2FQJk+8+b,
andreas.dilger-ral2JQCrhuEAvxtiuMwx3w, Antonio Quartulli, Greg KH,
bergwolf-Re5JQEeQqe8AvxtiuMwx3w,
linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
David Miller, oleg.drokin-ral2JQCrhuEAvxtiuMwx3w,
jacques-charles.lafoucriere-KCE40YydGKI,
jinshan.xiong-ral2JQCrhuEAvxtiuMwx3w, netdev,
linux-metag-u79uwXL29TY76Z2rM5mHXA, David Laight
In-Reply-To: <52EF6965.6040406-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
On 02/03/2014 06:03 PM, Chen Gang wrote:
> On 02/03/2014 04:58 PM, Dan Carpenter wrote:
>> On Sat, Feb 01, 2014 at 09:57:39PM +0800, Chen Gang wrote:
>>> It seems, our kernel still stick to treate 'pack' region have effect
>>> with both 'align' and 'sizeof'.
>>>
>>
>> It's not about packed regions. It's about unions. It's saying the
>> sizeof() a union is a multiple of 4 unless it's packed.
>>
>> union foo {
>> short x;
>> short y;
>> };
>>
>> The author intended the sizeof(union foo) to be 2 but on metag arch then
>> it is 4.
>>
>
> Yeah, just like your original discussion. :-)
>
>
> Hmm... can we say: "for metag compiler, in a pack region, it considers
> variables alignment, but does not consider about struct/union alignment
> (except append packed to each related struct/union)".
>
> For compatible (consider about its ABI), it has to keep this features,
> but for kernel, it needs be changed.
>
> So, I suggest to add one parameter to compiler to switch this feature,
> and append this parameter to KBUILD_CFLAGS in "arch/metag/Makefile"
> which can satisfy both ABI and kernel.
>
After append the parameter to KBUILD_CFLAGS in "arch/metag/Makefile",
- I guess/assume "include/uapi/*" should/will not need be modified.
- but need check all files in "arch/metag/include/uapi/*".
(add padding data for packed struct/union when __KERNEL__ defined)
- maybe we have to process metag related ABI which not in "*/uapi/*"
(add padding data for packed struct/union when __KERNEL__ defined)
and when we find them, recommend to move all of them to "*/uapi/*".
Sorry, I don't know whether this way is the best way or not, but for me
it is an executable way to solve this feature issue and satisfy both
kernel and ABI.
Thanks.
--
Chen Gang
Open, share and attitude like air, water and life which God blessed
--
To unsubscribe from this list: send the line "unsubscribe linux-metag" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* Bug in IPv4 version of ping...
From: Vic @ 2014-02-03 10:31 UTC (permalink / raw)
To: netdev
Hi All.
I hope this is the right place to post this - this list is mentioned as
the mailing list for iputils.
I believe I've found a bug in ping.c - there is an ancient work-around for
an IP_RECVERR bug in raw sockets. The problem is as follows :-
ping_common.c defines a variable working_recverr
ping.c uses this variable to work out what it should do under certain
failure circumstances.
As far as I can see, working_recverr is undefined when first used; the
initial response to it could be either way. This leads to an erroneous
report that the kernel is "not very fresh", and needs to be upgraded.
I have seen this in the current builds for RHEL5 and 6 (s20071127) and in
the current release (s20121221).
I suspect this variable should be set somewhere after option parsing...
HTH
Vic.
^ permalink raw reply
* Re: [Xen-devel] [PATCH v6] xen/grant-table: Avoid m2p_override during mapping
From: Stefano Stabellini @ 2014-02-03 11:13 UTC (permalink / raw)
To: David Vrabel
Cc: Zoltan Kiss, Julien Grall, Stefano Stabellini, jonathan.davies,
wei.liu2, ian.campbell, netdev, linux-kernel, xen-devel
In-Reply-To: <52EF7618.7030402@citrix.com>
[-- Attachment #1: Type: text/plain, Size: 2936 bytes --]
On Mon, 3 Feb 2014, David Vrabel wrote:
> On 02/02/14 18:52, Zoltan Kiss wrote:
> > On 02/02/14 11:29, Julien Grall wrote:
> >> Hello,
> >>
> >> This patch is breaking Linux compilation on ARM:
> >>
> >> drivers/xen/grant-table.c: In function ‘__gnttab_map_refs’:
> >> drivers/xen/grant-table.c:989:3: error: implicit declaration of
> >> function ‘FOREIGN_FRAME’ [-Werror=implicit-function-declaration]
> >> if (unlikely(!set_phys_to_machine(pfn, FOREIGN_FRAME(mfn)))) {
> >> ^
> >> drivers/xen/grant-table.c: In function ‘__gnttab_unmap_refs’:
> >> drivers/xen/grant-table.c:1054:3: error: implicit declaration of
> >> function ‘get_phys_to_machine’ [-Werror=implicit-function-declaration]
> >> mfn = get_phys_to_machine(pfn);
> >> ^
> >> drivers/xen/grant-table.c:1055:43: error: ‘FOREIGN_FRAME_BIT’
> >> undeclared (first use in this function)
> >> if (mfn == INVALID_P2M_ENTRY || !(mfn & FOREIGN_FRAME_BIT)) {
> >> ^
> >> drivers/xen/grant-table.c:1055:43: note: each undeclared identifier is
> >> reported only once for each function it appears in
> >> drivers/xen/grant-table.c:1068:9: error: too many arguments to
> >> function ‘m2p_remove_override’
> >> mfn);
> >> ^
> >> In file included from include/xen/page.h:4:0,
> >> from drivers/xen/grant-table.c:48:
> >> /local/home/julien/works/midway/linux/arch/arm/include/asm/xen/page.h:106:19:
> >> note: declared here
> >> static inline int m2p_remove_override(struct page *page, bool
> >> clear_pte)
> >> ^
> >> cc1: some warnings being treated as errors
> >
> > Hi,
> >
> > That's bad indeed. I think the best solution is to put those parts
> > behind an #ifdef x86. The ones moved from x86/p2m.c to grant-table.c.
> > David, Stefano, what do you think?
>
> I don't think we want (more) #ifdef CONFIG_X86 in grant-table.c and the
> arch-specific bits will have to factored out into their own functions
> with suitable stubs provided for ARM.
We certainly don't want more ifdefs like that.
> But, this patch went in late and it's clearly not ready. So I think it
> should be reverted and we should aim to get it sorted out for 3.15.
>
> Konrad/Stefano (if you agree) please revert
> 08ece5bb2312b4510b161a6ef6682f37f4eac8a1 and send a pull request.
Unfortunately I have to agree: fixing the prototype of
m2p_remove_override and replacing get_phys_to_machine with pfn_to_mfn is
easy.
However FOREIGN_FRAME is an x86-ism and I don't feel confortable with
adding yet another #define under arch/arm/xen just to deal with x86
stuff that spill on common code.
Sorry for not spotting this earlier.
> Konrad, I also think you should look at adding an ARM build to your test
> system (I thought you had this already).
Let's talk about how to set it up offline.
^ permalink raw reply
* Re: [PATCH] drivers: staging: lustre: lustre: include: add "__attribute__((packed))" for the related union
From: James Hogan @ 2014-02-03 11:02 UTC (permalink / raw)
To: David Laight
Cc: devel@driverdev.osuosl.org, andreas.dilger@intel.com, Chen Gang,
Greg KH, bergwolf@gmail.com, linux-kernel@vger.kernel.org,
linux-metag@vger.kernel.org, oleg.drokin@intel.com,
jacques-charles.lafoucriere@cea.fr, Antonio Quartulli, netdev,
jinshan.xiong@intel.com, David Miller, 'Dan Carpenter'
In-Reply-To: <063D6719AE5E284EB5DD2968C1650D6D0F6B777A@AcuExch.aculab.com>
On 03/02/14 10:35, David Laight wrote:
> From: James Hogan
>> On 03/02/14 10:05, David Laight wrote:
>>> Architectures that define such alignment rules are a right PITA.
>>> You either need to get the size to 2 without using 'packed', or
>>> just not define such structures.
>>> It is worth seeing if adding aligned(2) will change the size - I'm
>>> not sure.
>>
>> __aligned(2) alone doesn't seem to have any effect on sizeof() or
>> __alignof__() unless it is accompanied by __packed. x86_64 is similar in
>> that respect (it just packs sanely in the first place).
>>
>> Combining __packed with __aligned(2) does the trick though (__packed
>> alone sets __aligned(1) which is obviously going to be suboptimal).
>
> Compile some code for a cpu that doesn't support misaligned transfers
> (probably one of sparc, arm, ppc) and see if the compiler generates a
> single 16bit request or two 8 bits ones.
> You don't want the compiler generating multiple byte-sized memory transfers.
Meta is also one of those arches, and according to my quick tests,
__packed alone does correctly make it fall back to byte loads/stores,
but with __packed __aligned(2) it uses 16bit loads/stores. I've also
confirmed that with an ARM toolchain (see below for example).
Cheers
James
input:
#define __aligned(x) __attribute__((aligned(x)))
#define __packed __attribute__((packed))
union a {
short x, y;
} __aligned(2) __packed;
struct b {
short x;
} __aligned(2) __packed;
unsigned int soa = sizeof(union a);
unsigned int aoa = __alignof__(union a);
unsigned int sob = sizeof(struct b);
unsigned int aob = __alignof__(struct b);
void t(struct b *x, union a *y)
{
++x->x;
++y->x;
}
ARM output (-O2):
.cpu arm10tdmi
.fpu softvfp
.eabi_attribute 20, 1
.eabi_attribute 21, 1
.eabi_attribute 23, 3
.eabi_attribute 24, 1
.eabi_attribute 25, 1
.eabi_attribute 26, 2
.eabi_attribute 30, 2
.eabi_attribute 34, 0
.eabi_attribute 18, 4
.file "alignment4.c"
.text
.align 2
.global t
.type t, %function
t:
@ args = 0, pretend = 0, frame = 0
@ frame_needed = 0, uses_anonymous_args = 0
@ link register save eliminated.
ldrh r3, [r0, #0]
add r3, r3, #1
strh r3, [r0, #0] @ movhi
ldrh r3, [r1, #0]
add r3, r3, #1
strh r3, [r1, #0] @ movhi
bx lr
.size t, .-t
.global aob
.global sob
.global aoa
.global soa
.data
.align 2
.type aob, %object
.size aob, 4
aob:
.word 2
.type sob, %object
.size sob, 4
sob:
.word 2
.type aoa, %object
.size aoa, 4
aoa:
.word 2
.type soa, %object
.size soa, 4
soa:
.word 2
.ident "GCC: (GNU) 4.7.1 20120606 (Red Hat 4.7.1-0.1.20120606)"
.section .note.GNU-stack,"",%progbits
^ permalink raw reply
* Re: [Xen-devel] [PATCH v6] xen/grant-table: Avoid m2p_override during mapping
From: David Vrabel @ 2014-02-03 10:57 UTC (permalink / raw)
To: Zoltan Kiss
Cc: Julien Grall, Stefano Stabellini, jonathan.davies, wei.liu2,
ian.campbell, netdev, linux-kernel, David Vrabel, xen-devel
In-Reply-To: <52EE93F0.1020508@citrix.com>
On 02/02/14 18:52, Zoltan Kiss wrote:
> On 02/02/14 11:29, Julien Grall wrote:
>> Hello,
>>
>> This patch is breaking Linux compilation on ARM:
>>
>> drivers/xen/grant-table.c: In function ‘__gnttab_map_refs’:
>> drivers/xen/grant-table.c:989:3: error: implicit declaration of
>> function ‘FOREIGN_FRAME’ [-Werror=implicit-function-declaration]
>> if (unlikely(!set_phys_to_machine(pfn, FOREIGN_FRAME(mfn)))) {
>> ^
>> drivers/xen/grant-table.c: In function ‘__gnttab_unmap_refs’:
>> drivers/xen/grant-table.c:1054:3: error: implicit declaration of
>> function ‘get_phys_to_machine’ [-Werror=implicit-function-declaration]
>> mfn = get_phys_to_machine(pfn);
>> ^
>> drivers/xen/grant-table.c:1055:43: error: ‘FOREIGN_FRAME_BIT’
>> undeclared (first use in this function)
>> if (mfn == INVALID_P2M_ENTRY || !(mfn & FOREIGN_FRAME_BIT)) {
>> ^
>> drivers/xen/grant-table.c:1055:43: note: each undeclared identifier is
>> reported only once for each function it appears in
>> drivers/xen/grant-table.c:1068:9: error: too many arguments to
>> function ‘m2p_remove_override’
>> mfn);
>> ^
>> In file included from include/xen/page.h:4:0,
>> from drivers/xen/grant-table.c:48:
>> /local/home/julien/works/midway/linux/arch/arm/include/asm/xen/page.h:106:19:
>> note: declared here
>> static inline int m2p_remove_override(struct page *page, bool
>> clear_pte)
>> ^
>> cc1: some warnings being treated as errors
>
> Hi,
>
> That's bad indeed. I think the best solution is to put those parts
> behind an #ifdef x86. The ones moved from x86/p2m.c to grant-table.c.
> David, Stefano, what do you think?
I don't think we want (more) #ifdef CONFIG_X86 in grant-table.c and the
arch-specific bits will have to factored out into their own functions
with suitable stubs provided for ARM.
But, this patch went in late and it's clearly not ready. So I think it
should be reverted and we should aim to get it sorted out for 3.15.
Konrad/Stefano (if you agree) please revert
08ece5bb2312b4510b161a6ef6682f37f4eac8a1 and send a pull request.
Konrad, I also think you should look at adding an ARM build to your test
system (I thought you had this already).
David
^ permalink raw reply
* RE: [PATCH] drivers: staging: lustre: lustre: include: add "__attribute__((packed))" for the related union
From: David Laight @ 2014-02-03 10:35 UTC (permalink / raw)
To: 'James Hogan'
Cc: 'Dan Carpenter', Chen Gang,
devel-gWbeCf7V1WCQmaza687I9mD2FQJk+8+b@public.gmane.org,
andreas.dilger-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org,
Antonio Quartulli, Greg KH,
bergwolf-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org,
linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
David Miller, oleg.drokin-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org,
jacques-charles.lafoucriere-KCE40YydGKI@public.gmane.org,
jinshan.xiong-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org, netdev,
linux-metag-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
In-Reply-To: <52EF6DCC.6040807-1AXoQHu6uovQT0dZR+AlfA@public.gmane.org>
From: James Hogan
> On 03/02/14 10:05, David Laight wrote:
> > From: Dan Carpenter
> >> On Sat, Feb 01, 2014 at 09:57:39PM +0800, Chen Gang wrote:
> >>> It seems, our kernel still stick to treate 'pack' region have effect
> >>> with both 'align' and 'sizeof'.
> >>
> >> It's not about packed regions. It's about unions. It's saying the
> >> sizeof() a union is a multiple of 4 unless it's packed.
> >>
> >> union foo {
> >> short x;
> >> short y;
> >> };
> >>
> >> The author intended the sizeof(union foo) to be 2 but on metag arch then
> >> it is 4.
> >
> > The same is probably be true of: struct foo { _u16 bar; };
>
> Yes indeed.
>
> > Architectures that define such alignment rules are a right PITA.
> > You either need to get the size to 2 without using 'packed', or
> > just not define such structures.
> > It is worth seeing if adding aligned(2) will change the size - I'm
> > not sure.
>
> __aligned(2) alone doesn't seem to have any effect on sizeof() or
> __alignof__() unless it is accompanied by __packed. x86_64 is similar in
> that respect (it just packs sanely in the first place).
>
> Combining __packed with __aligned(2) does the trick though (__packed
> alone sets __aligned(1) which is obviously going to be suboptimal).
Compile some code for a cpu that doesn't support misaligned transfers
(probably one of sparc, arm, ppc) and see if the compiler generates a
single 16bit request or two 8 bits ones.
You don't want the compiler generating multiple byte-sized memory transfers.
David
--
To unsubscribe from this list: send the line "unsubscribe linux-metag" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* Re: [PATCH] drivers: staging: lustre: lustre: include: add "__attribute__((packed))" for the related union
From: Chen Gang @ 2014-02-03 10:30 UTC (permalink / raw)
To: James Hogan
Cc: devel@driverdev.osuosl.org, andreas.dilger@intel.com,
jinshan.xiong@intel.com, Greg KH, bergwolf@gmail.com,
linux-kernel@vger.kernel.org, linux-metag@vger.kernel.org,
oleg.drokin@intel.com, David Laight,
jacques-charles.lafoucriere@cea.fr, Antonio Quartulli, netdev,
David Miller, 'Dan Carpenter'
In-Reply-To: <52EF6DCC.6040807@imgtec.com>
On 02/03/2014 06:22 PM, James Hogan wrote:
> On 03/02/14 10:05, David Laight wrote:
>> From: Dan Carpenter
>>> On Sat, Feb 01, 2014 at 09:57:39PM +0800, Chen Gang wrote:
>>>> It seems, our kernel still stick to treate 'pack' region have effect
>>>> with both 'align' and 'sizeof'.
>>>
>>> It's not about packed regions. It's about unions. It's saying the
>>> sizeof() a union is a multiple of 4 unless it's packed.
>>>
>>> union foo {
>>> short x;
>>> short y;
>>> };
>>>
>>> The author intended the sizeof(union foo) to be 2 but on metag arch then
>>> it is 4.
>>
>> The same is probably be true of: struct foo { _u16 bar; };
>
> Yes indeed.
>
>> Architectures that define such alignment rules are a right PITA.
>> You either need to get the size to 2 without using 'packed', or
>> just not define such structures.
>> It is worth seeing if adding aligned(2) will change the size - I'm
>> not sure.
>
> __aligned(2) alone doesn't seem to have any effect on sizeof() or
> __alignof__() unless it is accompanied by __packed. x86_64 is similar in
> that respect (it just packs sanely in the first place).
>
> Combining __packed with __aligned(2) does the trick though (__packed
> alone sets __aligned(1) which is obviously going to be suboptimal).
>
Oh, thank you for your explanation.
And hope this feature issue can be fixed, and satisfy both kernel and
ABI. :-)
Thanks.
--
Chen Gang
Open, share and attitude like air, water and life which God blessed
^ permalink raw reply
* Re: [PATCH] drivers: staging: lustre: lustre: include: add "__attribute__((packed))" for the related union
From: Chen Gang @ 2014-02-03 10:25 UTC (permalink / raw)
To: David Laight
Cc: 'Dan Carpenter', James Hogan,
devel-gWbeCf7V1WCQmaza687I9mD2FQJk+8+b@public.gmane.org,
andreas.dilger-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org,
Antonio Quartulli, Greg KH,
bergwolf-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org,
linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
David Miller, oleg.drokin-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org,
jacques-charles.lafoucriere-KCE40YydGKI@public.gmane.org,
jinshan.xiong-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org, netdev,
linux-metag-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
In-Reply-To: <063D6719AE5E284EB5DD2968C1650D6D0F6B772B-VkEWCZq2GCInGFn1LkZF6NBPR1lH4CV8@public.gmane.org>
On 02/03/2014 06:05 PM, David Laight wrote:
> From: Dan Carpenter
>> On Sat, Feb 01, 2014 at 09:57:39PM +0800, Chen Gang wrote:
>>> It seems, our kernel still stick to treate 'pack' region have effect
>>> with both 'align' and 'sizeof'.
>>
>> It's not about packed regions. It's about unions. It's saying the
>> sizeof() a union is a multiple of 4 unless it's packed.
>>
>> union foo {
>> short x;
>> short y;
>> };
>>
>> The author intended the sizeof(union foo) to be 2 but on metag arch then
>> it is 4.
>
> The same is probably be true of: struct foo { _u16 bar; };
>
I guess so.
> Architectures that define such alignment rules are a right PITA.
Sorry, I do not know about PITA (after google or wiki, I can not get
more related information).
Could you provide more information about PITA, thanks?
> You either need to get the size to 2 without using 'packed', or
> just not define such structures.
Excuse me, I don't quite understand your meaning. I guess your meaning
is:
"normally, we should not use a struct/union like that, no matter what it is (2 or 4)".
Is it correct.
> It is worth seeing if adding aligned(2) will change the size - I'm
> not sure.
>
Yes, it will/should make sure that it must be 2.
Thanks.
--
Chen Gang
Open, share and attitude like air, water and life which God blessed
--
To unsubscribe from this list: send the line "unsubscribe linux-metag" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* Re: [PATCH] drivers: staging: lustre: lustre: include: add "__attribute__((packed))" for the related union
From: James Hogan @ 2014-02-03 10:22 UTC (permalink / raw)
To: David Laight
Cc: devel@driverdev.osuosl.org, andreas.dilger@intel.com, Chen Gang,
Greg KH, bergwolf@gmail.com, linux-kernel@vger.kernel.org,
linux-metag@vger.kernel.org, oleg.drokin@intel.com,
jacques-charles.lafoucriere@cea.fr, Antonio Quartulli, netdev,
jinshan.xiong@intel.com, David Miller, 'Dan Carpenter'
In-Reply-To: <063D6719AE5E284EB5DD2968C1650D6D0F6B772B@AcuExch.aculab.com>
On 03/02/14 10:05, David Laight wrote:
> From: Dan Carpenter
>> On Sat, Feb 01, 2014 at 09:57:39PM +0800, Chen Gang wrote:
>>> It seems, our kernel still stick to treate 'pack' region have effect
>>> with both 'align' and 'sizeof'.
>>
>> It's not about packed regions. It's about unions. It's saying the
>> sizeof() a union is a multiple of 4 unless it's packed.
>>
>> union foo {
>> short x;
>> short y;
>> };
>>
>> The author intended the sizeof(union foo) to be 2 but on metag arch then
>> it is 4.
>
> The same is probably be true of: struct foo { _u16 bar; };
Yes indeed.
> Architectures that define such alignment rules are a right PITA.
> You either need to get the size to 2 without using 'packed', or
> just not define such structures.
> It is worth seeing if adding aligned(2) will change the size - I'm
> not sure.
__aligned(2) alone doesn't seem to have any effect on sizeof() or
__alignof__() unless it is accompanied by __packed. x86_64 is similar in
that respect (it just packs sanely in the first place).
Combining __packed with __aligned(2) does the trick though (__packed
alone sets __aligned(1) which is obviously going to be suboptimal).
Cheers
James
^ permalink raw reply
* RE: [PATCH] drivers: staging: lustre: lustre: include: add "__attribute__((packed))" for the related union
From: David Laight @ 2014-02-03 10:05 UTC (permalink / raw)
To: 'Dan Carpenter', Chen Gang
Cc: James Hogan, devel@driverdev.osuosl.org, andreas.dilger@intel.com,
Antonio Quartulli, Greg KH, bergwolf@gmail.com,
linux-kernel@vger.kernel.org, David Miller, oleg.drokin@intel.com,
jacques-charles.lafoucriere@cea.fr, jinshan.xiong@intel.com,
netdev, linux-metag@vger.kernel.org
In-Reply-To: <20140203085855.GA26722@mwanda>
From: Dan Carpenter
> On Sat, Feb 01, 2014 at 09:57:39PM +0800, Chen Gang wrote:
> > It seems, our kernel still stick to treate 'pack' region have effect
> > with both 'align' and 'sizeof'.
>
> It's not about packed regions. It's about unions. It's saying the
> sizeof() a union is a multiple of 4 unless it's packed.
>
> union foo {
> short x;
> short y;
> };
>
> The author intended the sizeof(union foo) to be 2 but on metag arch then
> it is 4.
The same is probably be true of: struct foo { _u16 bar; };
Architectures that define such alignment rules are a right PITA.
You either need to get the size to 2 without using 'packed', or
just not define such structures.
It is worth seeing if adding aligned(2) will change the size - I'm
not sure.
David
^ permalink raw reply
* Re: [PATCH] drivers: staging: lustre: lustre: include: add "__attribute__((packed))" for the related union
From: Chen Gang @ 2014-02-03 10:03 UTC (permalink / raw)
To: Dan Carpenter
Cc: devel, James Hogan, andreas.dilger, jinshan.xiong, Greg KH,
bergwolf, linux-kernel@vger.kernel.org, linux-metag, oleg.drokin,
jacques-charles.lafoucriere, Antonio Quartulli, netdev,
David Miller
In-Reply-To: <20140203085855.GA26722@mwanda>
On 02/03/2014 04:58 PM, Dan Carpenter wrote:
> On Sat, Feb 01, 2014 at 09:57:39PM +0800, Chen Gang wrote:
>> It seems, our kernel still stick to treate 'pack' region have effect
>> with both 'align' and 'sizeof'.
>>
>
> It's not about packed regions. It's about unions. It's saying the
> sizeof() a union is a multiple of 4 unless it's packed.
>
> union foo {
> short x;
> short y;
> };
>
> The author intended the sizeof(union foo) to be 2 but on metag arch then
> it is 4.
>
Yeah, just like your original discussion. :-)
Hmm... can we say: "for metag compiler, in a pack region, it considers
variables alignment, but does not consider about struct/union alignment
(except append packed to each related struct/union)".
For compatible (consider about its ABI), it has to keep this features,
but for kernel, it needs be changed.
So, I suggest to add one parameter to compiler to switch this feature,
and append this parameter to KBUILD_CFLAGS in "arch/metag/Makefile"
which can satisfy both ABI and kernel.
Thanks.
--
Chen Gang
Open, share and attitude like air, water and life which God blessed
^ permalink raw reply
* RE: [PATCH RFC 1/1] usb: Tell xhci when usb data might be misaligned
From: David Laight @ 2014-02-03 9:54 UTC (permalink / raw)
To: 'Mark Lord', Ming Lei
Cc: Sarah Sharp, Bjørn Mork,
linux-usb-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
netdev-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
Greg Kroah-Hartman, David Miller, Dan Williams, Nyman, Mathias,
Alan Stern, Freddy Xin
In-Reply-To: <52ED5381.2010106-e+AXbWqSrlAAvxtiuMwx3w@public.gmane.org>
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="utf-8", Size: 1359 bytes --]
From: Mark Lord
> On 14-02-01 09:18 AM, Ming Lei wrote:
> >
> > Even real regressions are easily/often introduced, and we are discussing
> > how to fix that. I suggest to unset the flag only for the known buggy
> > controllers.
>
> It is not the controllers that are particularly "buggy" here.
> But rather the drivers and design of parts of the kernel.
I suspect that the documentation is describing the actual implementation
of a specific hardware implementation, not necessarily how the hardware was
intended to behave.
The requirement for two 32bit accesses to a 64bit register is very similar.
This also means that implementations of the hardware that claim conformance
to the 0.96 specification might have similar issues.
Given the small number of xhci controllers and the even smaller number of
VHDL (or similar) sources they will be based on, it really ought to be
possible to tabulate the controller versions and families to get a much
better idea of their behaviour.
I've got two systems with Intel USB3 controllers, linux reports one as
'panther point', the other as '7 Series/C210 Series' (seems to be a Xeon
chipset). I've no idea how the latter relates to the former.
David
N§²æìr¸yúèØb²X¬¶Ç§vØ^)Þº{.nÇ+·¥{±ºÆâØ^nr¡ö¦zË\x1aëh¨èÚ&¢îý»\x05ËÛÔØï¦v¬Îf\x1dp)¹¹br ê+Ê+zf£¢·h§~Ûiÿûàz¹\x1e®w¥¢¸?¨èÚ&¢)ߢ^[f
^ permalink raw reply
* Re: [PATCH] staging: r8188eu: Fix typo in USB_DEVICE list
From: Dan Carpenter @ 2014-02-03 9:17 UTC (permalink / raw)
To: Larry Finger; +Cc: devel, gregkh, netdev
In-Reply-To: <1391371626-13551-1-git-send-email-Larry.Finger@lwfinger.net>
On Sun, Feb 02, 2014 at 02:07:06PM -0600, Larry Finger wrote:
> There is a typo in the device list that interchanges the vendor and
> product codes for one of the entries.
>
> Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
> ---
> drivers/staging/rtl8188eu/os_dep/usb_intf.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/staging/rtl8188eu/os_dep/usb_intf.c b/drivers/staging/rtl8188eu/os_dep/usb_intf.c
> index 0a341d6..e9e3c76 100644
> --- a/drivers/staging/rtl8188eu/os_dep/usb_intf.c
> +++ b/drivers/staging/rtl8188eu/os_dep/usb_intf.c
> @@ -53,7 +53,7 @@ static struct usb_device_id rtw_usb_id_tbl[] = {
> {USB_DEVICE(USB_VENDER_ID_REALTEK, 0x0179)}, /* 8188ETV */
> /*=== Customer ID ===*/
> /****** 8188EUS ********/
> - {USB_DEVICE(0x8179, 0x07B8)}, /* Abocom - Abocom */
> + {USB_DEVICE(0x07bb, 0x8179)}, /* Abocom - Abocom */
^^^^^^
Should this be 0x07b8?
regards,
dan carpenter
> {USB_DEVICE(0x2001, 0x330F)}, /* DLink DWA-125 REV D1 */
> {} /* Terminating entry */
> };
> --
> 1.8.4
>
> _______________________________________________
> devel mailing list
> devel@linuxdriverproject.org
> http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
^ permalink raw reply
* Re: [PATCH] drivers: staging: lustre: lustre: include: add "__attribute__((packed))" for the related union
From: Dan Carpenter @ 2014-02-03 8:58 UTC (permalink / raw)
To: Chen Gang
Cc: devel, James Hogan, andreas.dilger, jinshan.xiong, Greg KH,
bergwolf, linux-kernel@vger.kernel.org, linux-metag, oleg.drokin,
jacques-charles.lafoucriere, Antonio Quartulli, netdev,
David Miller
In-Reply-To: <52ECFD53.7010401@gmail.com>
On Sat, Feb 01, 2014 at 09:57:39PM +0800, Chen Gang wrote:
> It seems, our kernel still stick to treate 'pack' region have effect
> with both 'align' and 'sizeof'.
>
It's not about packed regions. It's about unions. It's saying the
sizeof() a union is a multiple of 4 unless it's packed.
union foo {
short x;
short y;
};
The author intended the sizeof(union foo) to be 2 but on metag arch then
it is 4.
regards,
dan carpenter
^ permalink raw reply
* [PATCH 1/3] net: ethernet: sunxi: Add new compatibles
From: Maxime Ripard @ 2014-02-02 13:49 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-arm-kernel, Maxime Ripard
The Allwinner A10 compatibles were following a slightly different compatible
patterns than the rest of the SoCs for historical reasons. Add compatibles
matching the other pattern to the ethernet driver for consistency, and keep the
older one for backward compatibility.
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
---
Documentation/devicetree/bindings/net/allwinner,sun4i-emac.txt | 5 +++--
drivers/net/ethernet/allwinner/sun4i-emac.c | 3 +++
2 files changed, 6 insertions(+), 2 deletions(-)
diff --git a/Documentation/devicetree/bindings/net/allwinner,sun4i-emac.txt b/Documentation/devicetree/bindings/net/allwinner,sun4i-emac.txt
index b90bfcd..863d5b81 100644
--- a/Documentation/devicetree/bindings/net/allwinner,sun4i-emac.txt
+++ b/Documentation/devicetree/bindings/net/allwinner,sun4i-emac.txt
@@ -1,7 +1,8 @@
* Allwinner EMAC ethernet controller
Required properties:
-- compatible: should be "allwinner,sun4i-emac".
+- compatible: should be "allwinner,sun4i-a10-emac" (Deprecated:
+ "allwinner,sun4i-emac")
- reg: address and length of the register set for the device.
- interrupts: interrupt for the device
- phy: A phandle to a phy node defining the PHY address (as the reg
@@ -14,7 +15,7 @@ Optional properties:
Example:
emac: ethernet@01c0b000 {
- compatible = "allwinner,sun4i-emac";
+ compatible = "allwinner,sun4i-a10-emac";
reg = <0x01c0b000 0x1000>;
interrupts = <55>;
clocks = <&ahb_gates 17>;
diff --git a/drivers/net/ethernet/allwinner/sun4i-emac.c b/drivers/net/ethernet/allwinner/sun4i-emac.c
index 46dfb13..6673106 100644
--- a/drivers/net/ethernet/allwinner/sun4i-emac.c
+++ b/drivers/net/ethernet/allwinner/sun4i-emac.c
@@ -930,6 +930,9 @@ static int emac_resume(struct platform_device *dev)
}
static const struct of_device_id emac_of_match[] = {
+ {.compatible = "allwinner,sun4i-a10-emac",},
+
+ /* Deprecated */
{.compatible = "allwinner,sun4i-emac",},
{},
};
--
1.8.4.2
^ permalink raw reply related
* [PATCH 2/3] net: phy: sunxi: Add new compatibles
From: Maxime Ripard @ 2014-02-02 13:49 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-arm-kernel, Maxime Ripard
In-Reply-To: <1391348953-10662-1-git-send-email-maxime.ripard@free-electrons.com>
The Allwinner A10 compatibles were following a slightly different compatible
patterns than the rest of the SoCs for historical reasons. Add compatibles
matching the other pattern to the mdio driver for consistency, and keep the
older one for backward compatibility.
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
---
Documentation/devicetree/bindings/net/allwinner,sun4i-mdio.txt | 5 +++--
drivers/net/phy/mdio-sun4i.c | 3 +++
2 files changed, 6 insertions(+), 2 deletions(-)
diff --git a/Documentation/devicetree/bindings/net/allwinner,sun4i-mdio.txt b/Documentation/devicetree/bindings/net/allwinner,sun4i-mdio.txt
index 00b9f9a..4ec5641 100644
--- a/Documentation/devicetree/bindings/net/allwinner,sun4i-mdio.txt
+++ b/Documentation/devicetree/bindings/net/allwinner,sun4i-mdio.txt
@@ -1,7 +1,8 @@
* Allwinner A10 MDIO Ethernet Controller interface
Required properties:
-- compatible: should be "allwinner,sun4i-mdio".
+- compatible: should be "allwinner,sun4i-a10-mdio"
+ (Deprecated: "allwinner,sun4i-mdio").
- reg: address and length of the register set for the device.
Optional properties:
@@ -9,7 +10,7 @@ Optional properties:
Example at the SoC level:
mdio@01c0b080 {
- compatible = "allwinner,sun4i-mdio";
+ compatible = "allwinner,sun4i-a10-mdio";
reg = <0x01c0b080 0x14>;
#address-cells = <1>;
#size-cells = <0>;
diff --git a/drivers/net/phy/mdio-sun4i.c b/drivers/net/phy/mdio-sun4i.c
index 18969b3..7f4a66d 100644
--- a/drivers/net/phy/mdio-sun4i.c
+++ b/drivers/net/phy/mdio-sun4i.c
@@ -171,6 +171,9 @@ static int sun4i_mdio_remove(struct platform_device *pdev)
}
static const struct of_device_id sun4i_mdio_dt_ids[] = {
+ { .compatible = "allwinner,sun4i-a10-mdio" },
+
+ /* Deprecated */
{ .compatible = "allwinner,sun4i-mdio" },
{ }
};
--
1.8.4.2
^ permalink raw reply related
* [PATCH 3/3] ARM: sunxi: dt: Convert to the new net compatibles
From: Maxime Ripard @ 2014-02-02 13:49 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-arm-kernel, Maxime Ripard
In-Reply-To: <1391348953-10662-1-git-send-email-maxime.ripard@free-electrons.com>
Switch the device tree to the new compatibles introduced in the ethernet and
mdio drivers to have a common pattern accross all Allwinner SoCs.
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
---
arch/arm/boot/dts/sun4i-a10.dtsi | 4 ++--
arch/arm/boot/dts/sun5i-a10s.dtsi | 4 ++--
arch/arm/boot/dts/sun7i-a20.dtsi | 4 ++--
3 files changed, 6 insertions(+), 6 deletions(-)
diff --git a/arch/arm/boot/dts/sun4i-a10.dtsi b/arch/arm/boot/dts/sun4i-a10.dtsi
index 040bb0e..10666ca 100644
--- a/arch/arm/boot/dts/sun4i-a10.dtsi
+++ b/arch/arm/boot/dts/sun4i-a10.dtsi
@@ -315,7 +315,7 @@
ranges;
emac: ethernet@01c0b000 {
- compatible = "allwinner,sun4i-emac";
+ compatible = "allwinner,sun4i-a10-emac";
reg = <0x01c0b000 0x1000>;
interrupts = <55>;
clocks = <&ahb_gates 17>;
@@ -323,7 +323,7 @@
};
mdio@01c0b080 {
- compatible = "allwinner,sun4i-mdio";
+ compatible = "allwinner,sun4i-a10-mdio";
reg = <0x01c0b080 0x14>;
status = "disabled";
#address-cells = <1>;
diff --git a/arch/arm/boot/dts/sun5i-a10s.dtsi b/arch/arm/boot/dts/sun5i-a10s.dtsi
index ea16054..6496159 100644
--- a/arch/arm/boot/dts/sun5i-a10s.dtsi
+++ b/arch/arm/boot/dts/sun5i-a10s.dtsi
@@ -278,7 +278,7 @@
ranges;
emac: ethernet@01c0b000 {
- compatible = "allwinner,sun4i-emac";
+ compatible = "allwinner,sun4i-a10-emac";
reg = <0x01c0b000 0x1000>;
interrupts = <55>;
clocks = <&ahb_gates 17>;
@@ -286,7 +286,7 @@
};
mdio@01c0b080 {
- compatible = "allwinner,sun4i-mdio";
+ compatible = "allwinner,sun4i-a10-mdio";
reg = <0x01c0b080 0x14>;
status = "disabled";
#address-cells = <1>;
diff --git a/arch/arm/boot/dts/sun7i-a20.dtsi b/arch/arm/boot/dts/sun7i-a20.dtsi
index 119f066..9ff0948 100644
--- a/arch/arm/boot/dts/sun7i-a20.dtsi
+++ b/arch/arm/boot/dts/sun7i-a20.dtsi
@@ -340,7 +340,7 @@
ranges;
emac: ethernet@01c0b000 {
- compatible = "allwinner,sun4i-emac";
+ compatible = "allwinner,sun4i-a10-emac";
reg = <0x01c0b000 0x1000>;
interrupts = <0 55 4>;
clocks = <&ahb_gates 17>;
@@ -348,7 +348,7 @@
};
mdio@01c0b080 {
- compatible = "allwinner,sun4i-mdio";
+ compatible = "allwinner,sun4i-a10-mdio";
reg = <0x01c0b080 0x14>;
status = "disabled";
#address-cells = <1>;
--
1.8.4.2
^ permalink raw reply related
* Re: [PATCH v2 5/6] Documentation: devicetree: sja1000: add reg-io-width binding
From: Florian Vaussard @ 2014-02-03 8:11 UTC (permalink / raw)
To: Marc Kleine-Budde, Wolfgang Grandegger
Cc: linux-can, netdev, linux-kernel, Grant Likely, Rob Herring,
Pawel Moll, Mark Rutland, Ian Campbell, Kumar Gala, devicetree
In-Reply-To: <52EB9723.3080008@pengutronix.de>
On 01/31/2014 01:29 PM, Marc Kleine-Budde wrote:
> On 01/31/2014 11:35 AM, Florian Vaussard wrote:
>> Add the reg-io-width property to describe the width of the memory
>> accesses.
>>
>> Cc: Grant Likely <grant.likely@linaro.org>
>> Cc: Rob Herring <robh+dt@kernel.org>
>> Cc: Pawel Moll <pawel.moll@arm.com>
>> Cc: Mark Rutland <mark.rutland@arm.com>
>> Cc: Ian Campbell <ijc+devicetree@hellion.org.uk>
>> Cc: Kumar Gala <galak@codeaurora.org>
>> Cc: devicetree@vger.kernel.org
>> Acked-by: Rob Herring <robh@kernel.org>
>> Signed-off-by: Florian Vaussard <florian.vaussard@epfl.ch>
>
> I think it makes sense to squash into patch 6.
>
According to recent guidelines from the DT maintainers [1], the
documentation part should be a separate patch. I guess it makes sense to
ease the review process for new bindings.
Regards
Florian
[1]
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/Documentation/devicetree/bindings/submitting-patches.txt
^ permalink raw reply
* Re: [PATCH net-stable 3.3..3.12] inet_diag: fix inet_diag_dump_icsk() timewait socket state logic
From: David Miller @ 2014-02-03 7:30 UTC (permalink / raw)
To: eric.dumazet; +Cc: ncardwell, netdev, edumazet
In-Reply-To: <1391399657.28432.93.camel@edumazet-glaptop2.roam.corp.google.com>
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Sun, 02 Feb 2014 19:54:17 -0800
> On Sun, 2014-02-02 at 20:40 -0500, Neal Cardwell wrote:
> ...
>> An analogous fix is in v3.13: 70315d22d3c7383f9a508d0aab21e2eb35b2303a
>> ("inet_diag: fix inet_diag_dump_icsk() to use correct state for
>> timewait sockets") but that patch is quite different because 3.13 code
>> is very different in this area due to the unification of TCP hash
>> tables in 05dbc7b ("tcp/dccp: remove twchain") in v3.13-rc1.
> ...
>> Signed-off-by: Neal Cardwell <ncardwell@google.com>
>> Cc: Eric Dumazet <edumazet@google.com>
>> ---
>> net/ipv4/inet_diag.c | 4 +++-
>> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> Acked-by: Eric Dumazet <edumazet@google.com>
Queued up, thanks Neal.
^ permalink raw reply
* Re: [PATCH] ipv6: default route for link local address is not added while assigning a address
From: Sohny Thomas @ 2014-02-03 7:19 UTC (permalink / raw)
To: Nicolas Dichtel, netdev, linux-kernel, yoshfuji, davem, kumuda
In-Reply-To: <20140130232909.GH25336@order.stressinduktion.org>
> Actually I am not so sure, there is no defined semantic of flush. I would
> be ok with all three solutions: leave it as is, always add link-local
> address (it does not matter if we don't have a link-local address on
> that interface, as a global scoped one is just fine enough) or make flush not
> remove the link-local address (but this seems a bit too special cased for me).
1) In case if we leave it as it is, there is rfc 6724 rule 2 to be
considered ( previously rfc 3484)
Rule 2: Prefer appropriate scope.
If Scope(SA) < Scope(SB): If Scope(SA) < Scope(D), then prefer SB and
otherwise prefer SA. Similarly, if Scope(SB) < Scope(SA): If
Scope(SB) < Scope(D), then prefer SA and otherwise prefer SB.
Test:
Destination: fe80::2(LS)
Candidate Source Addresses: 3ffe::1(GS) or fec0::1(SS) or LLA(LS)
Result: LLA(LS)
Scope(LLA) < Scope(fec0::1): If Scope(LLA) < Scope(fe80::2), no,
prefer LLA
Scope(LLA) < Scope(3ffe::1): If Scope(LLA) < Scope(fe80::2), no,
prefer LLA
Now the above test fails since the route itself is not present, and the
test assumes that the route gets added since the LLA is not removed
during the test
2) having a LLA always helps in NDP i think
3) making flush not remove link-local address will be chnaging
functionality of ip flush command
Regards,
Sohny
>
> Greetings,
>
> Hannes
>
>
>
^ permalink raw reply
* Re: [PATCH] ipv6: default route for link local address is not added while assigning a address
From: Sohny Thomas @ 2014-02-03 7:02 UTC (permalink / raw)
To: nicolas.dichtel, netdev, linux-kernel, yoshfuji, davem, kumuda
In-Reply-To: <52E8DA37.7010208@6wind.com>
On Wednesday 29 January 2014 04:08 PM, Nicolas Dichtel wrote:
> Le 29/01/2014 07:41, Sohny Thomas a écrit :
>> Resending this on netdev mailing list:
>> Default route for link local address is configured automatically if
>> NETWORKING_IPV6=yes is in ifcfg-eth*.
>> When the route table for the interface is flushed and a new address is
>> added to
>> the same device with out removing linklocal addr, default route for
>> link local
>> address has to added by default.
>>
>> I have found the issue to be caused by this checkin
>>
>> http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/net/ipv6?id=62b54dd91567686a1cb118f76a72d5f4764a86dd
>>
>>
>>
>> According to this change :
>> He removes adding a link local route if any other address is added ,
>> applicable
>> across all interfaces though there's mentioned only lo interface
>> So below patch fixes for other devices
>>
>> Signed-off-by: Sohny THomas <sohthoma@linux.vnet.ibm.com>
> Your email client has corrupted the patch, it cannot be applied.
> Please read Documentation/email-clients.txt
Sorry about that. Will resend again
>
> About the patch, I still think that the flush is too agressive. Link local
> routes are marked as 'proto kernel', removing them without the link local
> address is wrong.
>
> With this patch, you will add a link local route even if you don't have
> a link local address.
I think it wouldn't hurt to have a Link local route for NDP in case a
the routes become unreachable
-Regards,
Sohny
>
>
>
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox