Keyboard Shortcuts
ctrl + shift + ? :
Show all keyboard shortcuts
ctrl + g :
Navigate to a group
ctrl + shift + f :
Find
ctrl + / :
Quick actions
esc to dismiss
Likes
- U-Boot-Amlogic
- Messages
Search
|
[RFC v1 0/7] common: usb_onboard_hub: Add support for Odroid onboard USB hub reset
Odroid C1+/C2/C4/N2/N2-plus devices have an onboard USB hub
controlled by a reset GPIO pin. Adds changes to the onboard USB hub driver to perform a proper reset of the USB hub. Added conditional compilation for I2C initialization based on CONFIG_DM_I2C for failed to compile. Thanks -Anand Anand Moon (7): usb: onboard-hub: Add support for Genesys GL852G hub configs: odorid-c2: Enable Onboard HUB driver usb: onboard-hub: Add support for Genesys GL853G configs: odorid-n2: Enable Oboard HUB driver usb: onboard-hub: Add support for VL817 USB hub configs: odorid-c4: Enable Oboard HUB driver usb: onboard-hub: Add conditional compilation for I2C initialization common/usb_onboard_hub.c | 28 ++++++++++++++++++++++++++++ configs/odroid-c2_defconfig | 1 + configs/odroid-c4_defconfig | 1 + configs/odroid-n2_defconfig | 1 + 4 files changed, 31 insertions(+) base-commit: 48db49b0977cc1c9c9abf82c0fb704238fcef4fd -- 2.49.0 |
|
Re: Odroid n2+ usb broken on reboot
On Wed May 7, 2025 at 1:03 PM UTC, Wayne Schroeder wrote:
It's hard for me to quantify if usb works in u-boot as I'm not loadingYou can send bytes via UART to abort boot, if you happen to have a UART adapter around. |
|
Re: [PATCH v2 12/22] amlogic: Include env.h to permit reading the environment
01/05/2025 04.04, Simon Glass via groups.io wrote:
This file reads from the environment but does not include the correctReviewed-by: Viacheslav Bocharov <adeep@...> |
|
Re: Odroid n2+ usb broken on reboot
Hi Wayne,
On Wed, 7 May 2025 at 17:31, Neil Armstrong via groups.io <neil.armstrong@...> wrote: I wanted to enable CONFIG_USB_ONBOARD_HUB on Odroid platform on u-boot I think this should help USB hub reset, I'll check this and let you know. Thanks -Anand |
|
Re: Odroid n2+ usb broken on reboot
Wayne Schroeder
It's hard for me to quantify if usb works in u-boot as I'm not loading
toggle quoted message
Show quoted text
off of any usb storage devices. It certainly never lets me use the keyboard to abort auto boot, and never has. On reboot, to reliably get it to work, I have to do a usb reset and sleep some amount of time before doing distro boot. I picked 10 seconds. If you would like for me to do some scripted boot command additions before my extlinux to gather data, I can. Right now, what I have done is make a reboot hook in linux to touch a file on the boot partition that my bootcommand fatrms and if it was present, it resets usb and sleeps 10 seconds, so it's reliable to reboot remotely like that. Let me know how else I can gather any data. On Wed, May 7, 2025 at 7:01?AM <neil.armstrong@...> wrote:
|
|
Re: Odroid n2+ usb broken on reboot
On 05/05/2025 15:12, Wayne Schroeder wrote:
If I add the usb reset command, it works reliably at reboot, but itOK this means there's a dangling resource, Does USB always work within U-Boot or it breaks U-Boot and Linux in those cases ? On Mon, May 5, 2025 at 2:07?AM <neil.armstrong@...> wrote:I think it's a very long term issue, USB lacks a proper reset at probe on |
|
Re: Odroid n2+ usb broken on reboot
Wayne Schroeder
If I add the usb reset command, it works reliably at reboot, but it
toggle quoted message
Show quoted text
does not work on a fresh power-up with that change. On Mon, May 5, 2025 at 2:07?AM <neil.armstrong@...> wrote:
I think it's a very long term issue, USB lacks a proper reset at probe on |
|
Re: Odroid n2+ usb broken on reboot
Wayne Schroeder
It works at reboot with USB reset, but fails at a fresh power on now. I
toggle quoted message
Show quoted text
had to remove it unfortunately. On Mon, May 5, 2025, 2:07?AM <neil.armstrong@...> wrote:
Hi, |
|
Re: Odroid n2+ usb broken on reboot
Hi,
+CC [email protected] +CC linux-amlogic@... On 05/05/2025 00:34, Wayne Schroeder wrote: Hello. I am not sure if this is a bug, or even related to u-boot, but I amI think it's a very long term issue, USB lacks a proper reset at probe on warm reset, does USB still work from U-Boot after the reboot ? My apologies if this is not applicable to u-boot because it is a linuxI'm pretty sure there's driver callbacks we should implement to do the same here, I need to check. This has allowed me to reboot the system and successfully have access to |
|
[PATCH] net: mdio: mux-meson-gxl: set reversed bit when using internal phy
This bit is necessary to receive packets from the internal PHY.
Without this bit set, no activity occurs on the interface. Normally u-boot sets this bit, but if u-boot is compiled without net support, the interface will be up but without any activity. The vendor SDK sets this bit along with the PHY_ID bits. Ported from the Linux change at [1] from Da Xu merged in commit [2]. [1] [2] b23285e93bef ("net: mdio: mux-meson-gxl: set reversed bit when using internal phy") Suggested-by: Da Xue <da@...> Signed-off-by: Neil Armstrong <neil.armstrong@...> --- drivers/net/mdio_mux_meson_gxl.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/net/mdio_mux_meson_gxl.c b/drivers/net/mdio_mux_meson_gxl.c index 8ef3ae598b77fabf7b992900141d309708428b13..31898ed437e4857dd7027d311ab496be3e448d16 100644 --- a/drivers/net/mdio_mux_meson_gxl.c +++ b/drivers/net/mdio_mux_meson_gxl.c @@ -19,6 +19,7 @@ #define REG2_LEDACT GENMASK(23, 22) #define REG2_LEDLINK GENMASK(25, 24) #define REG2_DIV4SEL BIT(27) +#define REG2_REVERSED BIT(28) #define REG2_ADCBYPASS BIT(30) #define REG2_CLKINSEL BIT(31) #define ETH_REG3 0x4 @@ -66,7 +67,7 @@ static int meson_gxl_enable_internal_mdio(struct mdio_mux_meson_gxl_priv *priv) * The only constraint is that it must match the one in * drivers/net/phy/meson-gxl.c to properly match the PHY. */ - writel(FIELD_PREP(REG2_PHYID, EPHY_GXL_ID), + writel(REG2_REVERSED | FIELD_PREP(REG2_PHYID, EPHY_GXL_ID), priv->regs + ETH_REG2); /* Enable the internal phy */ --- base-commit: 98a898e58499c78245e4fe36ee404b1b1208e32b change-id: 20250502-u-boot-topic-mdio-mux-gxl-bit28-a6d03bf8e383 Best regards, -- Neil Armstrong <neil.armstrong@...> |
|
[PATCH v2 00/22] Deal with exfat versus byteorder
Simon Glass
Since exfat has its own byteorder functions we need to be careful about
including the normal U-Boot byteorder headers in code included by exfat. One chain for include/linux/byteorder/little_endian.h is like this: In file included from arch/sandbox/include/asm/byteorder.h:19, from arch/sandbox/include/asm/byteorder.h:19, from include/compiler.h:132, from include/env.h:12, from include/command.h:13, from include/net-common.h:7, from include/net.h:6, from include/efi.h:23, from include/blk.h:12, from include/part.h:9, from include/fs_internal.h:11, from fs/exfat/io.c:52: This series tidies things up so that efi.h can include net.h as required. Changes in v2: - Add patches for other files exposed by this change Simon Glass (22): net: Use a forward declaration for cmd_tbl in net-common.h congatec: Include env.h to permit reading the environment dhelectronics: Include env.h to permit reading the environment imx8ulp_evk: Include env.h to permit reading the environment venice: Include env.h to permit reading the environment phytec: Include env.h to permit reading the environment ronetix: Include env.h to permit reading the environment toradex: Include env.h to permit reading the environment advantech: Include env.h in imx8qm_dmsse20_a1 tegra: Include env.h to permit reading the environment synology: Include env.h to permit reading the environment amlogic: Include env.h to permit reading the environment freescale: Include env.h to permit reading the environment google: Include env.h to permit reading the environment liebherr: Include env.h to permit reading the environment technexion: Include env.h to permit reading the environment elf: Only use network environment-variables if net enabled net: Include byteorder in net6.h net: Include string.h in net-legacy.h net: Include env.h in pcap.c net: dc2114x: Include env.h to permit reading the environment net: Move env_get_ip() out of the header file arch/arm/mach-tegra/dt-setup.c | 1 + board/Synology/ds414/ds414.c | 1 + .../imx8qm_dmsse20_a1/imx8qm_dmsse20_a1.c | 1 + board/amlogic/jethub-j100/jethub-j100.c | 1 + board/congatec/cgtqmx8/cgtqmx8.c | 3 +- board/dhelectronics/common/dh_common.c | 1 + board/freescale/imx8ulp_evk/imx8ulp_evk.c | 1 + board/freescale/mx7dsabresd/mx7dsabresd.c | 1 + board/gateworks/venice/venice.c | 1 + board/google/chameleonv3/board.c | 2 + board/liebherr/xea/xea.c | 1 + .../common/phytec_som_detection_blocks.c | 1 + board/ronetix/imx8mq-cm/imx8mq_cm.c | 1 + board/technexion/pico-imx7d/pico-imx7d.c | 1 + board/toradex/colibri_imx7/colibri_imx7.c | 1 + board/toradex/verdin-imx8mm/verdin-imx8mm.c | 2 +- cmd/elf.c | 41 ++++++++++--------- drivers/net/dc2114x.c | 1 + include/net-common.h | 9 ++-- include/net-legacy.h | 1 + include/net6.h | 1 + net/net-common.c | 6 +++ net/pcap.c | 1 + 23 files changed, 53 insertions(+), 27 deletions(-) -- 2.43.0 base-commit: b5d6220dd2f4612912989f3c2b5a710f2248cb36 branch: fix-net2 |
|
[PATCH v2 12/22] amlogic: Include env.h to permit reading the environment
Simon Glass
This file reads from the environment but does not include the correct
header. Update it. Signed-off-by: Simon Glass <sjg@...> --- (no changes since v1) board/amlogic/jethub-j100/jethub-j100.c | 1 + 1 file changed, 1 insertion(+) diff --git a/board/amlogic/jethub-j100/jethub-j100.c b/board/amlogic/jethub-j100/jethub-j100.c index b770a1f8c53..9e87fb9f9d7 100644 --- a/board/amlogic/jethub-j100/jethub-j100.c +++ b/board/amlogic/jethub-j100/jethub-j100.c @@ -5,6 +5,7 @@ */ #include <dm.h> +#include <env.h> #include <init.h> #include <net.h> #include <asm/io.h> -- 2.43.0 |
|
[PATCH 2/2] ARM: mach-meson: use livetree by default
Finally enable OF_LIVE by default on the Amlogic support,
this add some cost to build the live tree, but makes DT parsing from driver much faster. Signed-off-by: Neil Armstrong <neil.armstrong@...> --- arch/arm/mach-meson/Kconfig | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/arm/mach-meson/Kconfig b/arch/arm/mach-meson/Kconfig index 7570f48e25fe477ce4dfb1eb608ed07b3a83d27f..10cef2ad6623180e2ff1ed2802543f3ba0714bd3 100644 --- a/arch/arm/mach-meson/Kconfig +++ b/arch/arm/mach-meson/Kconfig @@ -12,6 +12,7 @@ config MESON64_COMMON select MMC_PWRSEQ select BOARD_LATE_INIT select MESON_SM + imply OF_LIVE imply CMD_DM config MESON_GX -- 2.34.1 |
|
[PATCH 1/2] pinctrl: meson: convert to livetree
From: Beniamino Galvani <b.galvani@...>
Update the Meson pinctrl/gpio driver to support a live device tree. Signed-off-by: Beniamino Galvani <b.galvani@...> Link: Signed-off-by: Neil Armstrong <neil.armstrong@...> --- drivers/pinctrl/meson/pinctrl-meson.c | 73 +++++++++++++++++++---------------- 1 file changed, 39 insertions(+), 34 deletions(-) diff --git a/drivers/pinctrl/meson/pinctrl-meson.c b/drivers/pinctrl/meson/pinctrl-meson.c index babf1bccc9697596522dc93d8906ff0454cfebf5..21409063ee89cd2f8aa3d511f3fed6b64f800e51 100644 --- a/drivers/pinctrl/meson/pinctrl-meson.c +++ b/drivers/pinctrl/meson/pinctrl-meson.c @@ -10,6 +10,8 @@ #include <dm/device-internal.h> #include <dm/device_compat.h> #include <dm/lists.h> +#include <dm/of_addr.h> +#include <linux/ioport.h> #include <dm/pinctrl.h> #include <fdt_support.h> #include <linux/bitops.h> @@ -319,88 +321,91 @@ int meson_gpio_probe(struct udevice *dev) return 0; } -static fdt_addr_t parse_address(int offset, const char *name, int na, int ns) +static phys_addr_t parse_address(struct udevice *dev, ofnode node, + const char *name) { - int index, len = 0; - const fdt32_t *reg; + struct resource r; + fdt_size_t sz; + int na, ns, index; - index = fdt_stringlist_search(gd->fdt_blob, offset, "reg-names", name); + index = ofnode_stringlist_search(node, "reg-names", name); if (index < 0) return FDT_ADDR_T_NONE; - reg = fdt_getprop(gd->fdt_blob, offset, "reg", &len); - if (!reg || (len <= (index * sizeof(fdt32_t) * (na + ns)))) + if (of_live_active()) { + if (of_address_to_resource(ofnode_to_np(node), index, &r)) + return FDT_ADDR_T_NONE; + else + return r.start; + } + + na = dev_read_addr_cells(dev->parent); + if (na < 1) { + debug("bad #address-cells\n"); return FDT_ADDR_T_NONE; + } - reg += index * (na + ns); + ns = dev_read_size_cells(dev->parent); + if (ns < 1) { + debug("bad #size-cells\n"); + return FDT_ADDR_T_NONE; + } - return fdt_translate_address((void *)gd->fdt_blob, offset, reg); + return fdtdec_get_addr_size_fixed(gd->fdt_blob, ofnode_to_offset(node), + "reg", index, na, ns, &sz, true); } int meson_pinctrl_probe(struct udevice *dev) { struct meson_pinctrl *priv = dev_get_priv(dev); + ofnode node, gpio = ofnode_null(); struct uclass_driver *drv; struct udevice *gpio_dev; - fdt_addr_t addr; - int node, gpio = -1, len; - int na, ns; + phys_addr_t addr; char *name; + int len; - /* FIXME: Should use livetree */ - na = fdt_address_cells(gd->fdt_blob, dev_of_offset(dev->parent)); - if (na < 1) { - debug("bad #address-cells\n"); - return -EINVAL; - } - - ns = fdt_size_cells(gd->fdt_blob, dev_of_offset(dev->parent)); - if (ns < 1) { - debug("bad #size-cells\n"); - return -EINVAL; - } - - fdt_for_each_subnode(node, gd->fdt_blob, dev_of_offset(dev)) { - if (fdt_getprop(gd->fdt_blob, node, "gpio-controller", &len)) { + dev_for_each_subnode(node, dev) { + if (ofnode_read_prop(node, "gpio-controller", &len)) { gpio = node; break; } } - if (!gpio) { + if (!ofnode_valid(gpio)) { debug("gpio node not found\n"); return -EINVAL; } - addr = parse_address(gpio, "mux", na, ns); + addr = parse_address(dev, gpio, "mux"); if (addr == FDT_ADDR_T_NONE) { debug("mux address not found\n"); return -EINVAL; } priv->reg_mux = (void __iomem *)addr; - addr = parse_address(gpio, "gpio", na, ns); + addr = parse_address(dev, gpio, "gpio"); if (addr == FDT_ADDR_T_NONE) { debug("gpio address not found\n"); return -EINVAL; } priv->reg_gpio = (void __iomem *)addr; - addr = parse_address(gpio, "pull", na, ns); + addr = parse_address(dev, gpio, "pull"); /* Use gpio region if pull one is not present */ if (addr == FDT_ADDR_T_NONE) priv->reg_pull = priv->reg_gpio; else priv->reg_pull = (void __iomem *)addr; - addr = parse_address(gpio, "pull-enable", na, ns); + addr = parse_address(dev, gpio, "pull-enable"); /* Use pull region if pull-enable one is not present */ if (addr == FDT_ADDR_T_NONE) priv->reg_pullen = priv->reg_pull; else priv->reg_pullen = (void __iomem *)addr; - addr = parse_address(gpio, "ds", na, ns); + addr = parse_address(dev, gpio, "ds"); /* Drive strength region is optional */ if (addr == FDT_ADDR_T_NONE) priv->reg_ds = NULL; @@ -420,8 +425,8 @@ int meson_pinctrl_probe(struct udevice *dev) sprintf(name, "meson-gpio"); /* Create child device UCLASS_GPIO and bind it */ - device_bind(dev, priv->data->gpio_driver, name, NULL, - offset_to_ofnode(gpio), &gpio_dev); + device_bind_with_driver_data(dev, priv->data->gpio_driver, name, 0, + gpio, &gpio_dev); return 0; } -- 2.34.1 |
|
[PATCH 0/2] ARM: meson: finally switch to livetree
Never late than never, finally try to land Beniamino's
"pinctrl: meson: convert to livetree" from 2017 to finally be a modern platform with OF_LIVE ! Signed-off-by: Neil Armstrong <neil.armstrong@...> --- Beniamino Galvani (1): pinctrl: meson: convert to livetree Neil Armstrong (1): ARM: mach-meson: use livetree by default arch/arm/mach-meson/Kconfig | 1 + drivers/pinctrl/meson/pinctrl-meson.c | 73 +++++++++++++++++++---------------- 2 files changed, 40 insertions(+), 34 deletions(-) --- base-commit: f892a7f397a66d8d09f418d1e0e06dfb48bac27d change-id: 20250411-u-boot-meson-livetree-8fd1b0a18066 Best regards, -- Neil Armstrong <neil.armstrong@...> |
|
[PATCH 2/3] designware: Use the remove() method with related drivers
Simon Glass
Several drivers make use of the designware Ethernet driver but do not
implement the remove() method. Add this so that DMA is stopped when the OS is booted to avoid memory corruption, etc. Signed-off-by: Simon Glass <sjg@...> Reported-by: Christian Kohlschütter <christian@...> --- drivers/net/designware.c | 2 +- drivers/net/designware.h | 12 ++++++++++++ drivers/net/dwmac_meson8b.c | 6 ++++++ drivers/net/dwmac_s700.c | 6 ++++++ drivers/net/dwmac_socfpga.c | 6 ++++++ 5 files changed, 31 insertions(+), 1 deletion(-) diff --git a/drivers/net/designware.c b/drivers/net/designware.c index eebf14bd51a..d4d31925fca 100644 --- a/drivers/net/designware.c +++ b/drivers/net/designware.c @@ -805,7 +805,7 @@ clk_err: return err; } -static int designware_eth_remove(struct udevice *dev) +int designware_eth_remove(struct udevice *dev) { struct dw_eth_dev *priv = dev_get_priv(dev); diff --git a/drivers/net/designware.h b/drivers/net/designware.h index e47101ccaf6..8c9d0190e03 100644 --- a/drivers/net/designware.h +++ b/drivers/net/designware.h @@ -247,6 +247,18 @@ struct dw_eth_dev { int designware_eth_of_to_plat(struct udevice *dev); int designware_eth_probe(struct udevice *dev); + +/** + * designware_eth_remove() - Remove the device + * + * Disables DMA and marks the device as remove. This must be called before + * booting an OS, to ensure that DMA is inactive. + * + * @dev: Device to remove + * Return 0 if OK, -ve on error + */ +int designware_eth_remove(struct udevice *dev); + extern const struct eth_ops designware_eth_ops; struct dw_eth_pdata { diff --git a/drivers/net/dwmac_meson8b.c b/drivers/net/dwmac_meson8b.c index fde4aabbace..b2152f8da9b 100644 --- a/drivers/net/dwmac_meson8b.c +++ b/drivers/net/dwmac_meson8b.c @@ -145,6 +145,11 @@ static int dwmac_meson8b_probe(struct udevice *dev) return designware_eth_probe(dev); } +static int dwmac_meson8b_remove(struct udevice *dev) +{ + return designware_eth_remove(dev); +} + static const struct udevice_id dwmac_meson8b_ids[] = { { .compatible = "amlogic,meson-gxbb-dwmac", .data = (ulong)dwmac_setup_gx }, { .compatible = "amlogic,meson-g12a-dwmac", .data = (ulong)dwmac_setup_axg }, @@ -158,6 +163,7 @@ U_BOOT_DRIVER(dwmac_meson8b) = { .of_match = dwmac_meson8b_ids, .of_to_plat = dwmac_meson8b_of_to_plat, .probe = dwmac_meson8b_probe, + .remove = dwmac_meson8b_remove, .ops = &designware_eth_ops, .priv_auto = sizeof(struct dw_eth_dev), .plat_auto = sizeof(struct dwmac_meson8b_plat), diff --git a/drivers/net/dwmac_s700.c b/drivers/net/dwmac_s700.c index 969d247b4f3..cfb37c3aa71 100644 --- a/drivers/net/dwmac_s700.c +++ b/drivers/net/dwmac_s700.c @@ -44,6 +44,11 @@ static int dwmac_s700_probe(struct udevice *dev) return designware_eth_probe(dev); } +static int dwmac_s700_remove(struct udevice *dev) +{ + return designware_eth_remove(dev); +} + static int dwmac_s700_of_to_plat(struct udevice *dev) { return designware_eth_of_to_plat(dev); @@ -60,6 +65,7 @@ U_BOOT_DRIVER(dwmac_s700) = { .of_match = dwmac_s700_ids, .of_to_plat = dwmac_s700_of_to_plat, .probe = dwmac_s700_probe, + .remove = dwmac_s700_remove, .ops = &designware_eth_ops, .priv_auto = sizeof(struct dw_eth_dev), .plat_auto = sizeof(struct eth_pdata), diff --git a/drivers/net/dwmac_socfpga.c b/drivers/net/dwmac_socfpga.c index a9e2d8c0972..8e7a9975d28 100644 --- a/drivers/net/dwmac_socfpga.c +++ b/drivers/net/dwmac_socfpga.c @@ -130,6 +130,11 @@ static int dwmac_socfpga_probe(struct udevice *dev) return designware_eth_probe(dev); } +static int dwmac_socfpga_remove(struct udevice *dev) +{ + return designware_eth_remove(dev); +} + static const struct udevice_id dwmac_socfpga_ids[] = { { .compatible = "altr,socfpga-stmmac" }, { } @@ -141,6 +146,7 @@ U_BOOT_DRIVER(dwmac_socfpga) = { .of_match = dwmac_socfpga_ids, .of_to_plat = dwmac_socfpga_of_to_plat, .probe = dwmac_socfpga_probe, + .remove = dwmac_socfpga_remove, .ops = &designware_eth_ops, .priv_auto = sizeof(struct dw_eth_dev), .plat_auto = sizeof(struct dwmac_socfpga_plat), -- 2.43.0 base-commit: bcfa94be955ef263d37918ef4b8f898ee8964c61 branch: eth |
|
[PATCH v2 2/4] net: designware: Mark drivers as having active DMA
Simon Glass
Network devices which use this driver have DMA running while receiving
packages. Add the required flag to driver model so that they will be removed before starting the OS. Signed-off-by: Simon Glass <sjg@...> --- Changes in v2: - Add a new patch to mark drivers as having active DMA drivers/net/designware.c | 2 +- drivers/net/dwmac_meson8b.c | 2 +- drivers/net/dwmac_s700.c | 2 +- drivers/net/dwmac_socfpga.c | 2 +- drivers/net/gmac_rockchip.c | 2 +- 5 files changed, 5 insertions(+), 5 deletions(-) diff --git a/drivers/net/designware.c b/drivers/net/designware.c index 5385849ea6b..9b2b0a5d602 100644 --- a/drivers/net/designware.c +++ b/drivers/net/designware.c @@ -885,7 +885,7 @@ U_BOOT_DRIVER(eth_designware) = { .ops = &designware_eth_ops, .priv_auto = sizeof(struct dw_eth_dev), .plat_auto = sizeof(struct dw_eth_pdata), - .flags = DM_FLAG_ALLOC_PRIV_DMA, + .flags = DM_FLAG_ALLOC_PRIV_DMA | DM_FLAG_ACTIVE_DMA, }; static struct pci_device_id supported[] = { diff --git a/drivers/net/dwmac_meson8b.c b/drivers/net/dwmac_meson8b.c index bf94e50842d..95a99a98e99 100644 --- a/drivers/net/dwmac_meson8b.c +++ b/drivers/net/dwmac_meson8b.c @@ -162,5 +162,5 @@ U_BOOT_DRIVER(dwmac_meson8b) = { .ops = &designware_eth_ops, .priv_auto = sizeof(struct dw_eth_dev), .plat_auto = sizeof(struct dwmac_meson8b_plat), - .flags = DM_FLAG_ALLOC_PRIV_DMA, + .flags = DM_FLAG_ALLOC_PRIV_DMA | DM_FLAG_ACTIVE_DMA, }; diff --git a/drivers/net/dwmac_s700.c b/drivers/net/dwmac_s700.c index e78193f3376..21df8e3ce81 100644 --- a/drivers/net/dwmac_s700.c +++ b/drivers/net/dwmac_s700.c @@ -64,5 +64,5 @@ U_BOOT_DRIVER(dwmac_s700) = { .ops = &designware_eth_ops, .priv_auto = sizeof(struct dw_eth_dev), .plat_auto = sizeof(struct eth_pdata), - .flags = DM_FLAG_ALLOC_PRIV_DMA, + .flags = DM_FLAG_ALLOC_PRIV_DMA | DM_FLAG_ACTIVE_DMA, }; diff --git a/drivers/net/dwmac_socfpga.c b/drivers/net/dwmac_socfpga.c index dda33f90d65..9f262aaee60 100644 --- a/drivers/net/dwmac_socfpga.c +++ b/drivers/net/dwmac_socfpga.c @@ -145,5 +145,5 @@ U_BOOT_DRIVER(dwmac_socfpga) = { .ops = &designware_eth_ops, .priv_auto = sizeof(struct dw_eth_dev), .plat_auto = sizeof(struct dwmac_socfpga_plat), - .flags = DM_FLAG_ALLOC_PRIV_DMA, + .flags = DM_FLAG_ALLOC_PRIV_DMA | DM_FLAG_ACTIVE_DMA, }; diff --git a/drivers/net/gmac_rockchip.c b/drivers/net/gmac_rockchip.c index 5c542d62bd8..52be3a5687f 100644 --- a/drivers/net/gmac_rockchip.c +++ b/drivers/net/gmac_rockchip.c @@ -760,5 +760,5 @@ U_BOOT_DRIVER(eth_gmac_rockchip) = { .ops = &gmac_rockchip_eth_ops, .priv_auto = sizeof(struct dw_eth_dev), .plat_auto = sizeof(struct gmac_rockchip_plat), - .flags = DM_FLAG_ALLOC_PRIV_DMA, + .flags = DM_FLAG_ALLOC_PRIV_DMA | DM_FLAG_ACTIVE_DMA, }; -- 2.43.0 |
|
Re: [PATCH 2/3] designware: Use the remove() method with related drivers
Jonas Karlman
Hi Simon,
On 2025-04-06 23:10, Simon Glass wrote: Hi Jonas,Ahh, I see, remove() is only called for devices with ACTIVE_DMA or OS_PREPARE. So I guess that was probably the issue all along? There is some confusion that dm_remove_devices_active() does not remove device_active()=true devices and instead only remove ACTIVE_DMA and/or OS_PREPARE flagged devices. Maybe eth_post_probe() or similar should add the OS_PREPARE orSide note, this also misses the gmac rockchip driver that uses its ownHmm yes I saw that but then apparently forgot to include it. ACTIVE_DMA flag for all eth-uclass, to ensure remove() and thus stop() gets called when dm_remove_devices_active() is called. There is probably no need to add all these remove() ops, as they would still not have been called by dm_remove_devices_active(). Regards, Jonas Yes, I wasn't sure which way to go on that. I'll remove them.There is no reason to add the dwmac_meson8b_remove function above, you |
|
Re: [PATCH 2/3] designware: Use the remove() method with related drivers
Jonas Karlman
Hi Simon,
On 2025-04-06 00:12, Simon Glass wrote: Several drivers make use of the designware Ethernet driver but do notThe designware Ethernet driver core should not need the remove() ops as the eth_ops.stop() ops already should stop DMA. And the eth-uclass pre_remove() ops already call the eth_ops.stop() ops. All these drivers seem to use designware_eth_ops and thus have correct stop() ops configured and should really not need the remove() ops. Side note, this also misses the gmac rockchip driver that uses its own eth_ops instead of designware_eth_ops, however it also use same stop() ops as all these drivers. There is no reason to add the dwmac_meson8b_remove function above, you could just use designware_eth_remove directly here. Same for remaining drivers below. Regards, Jonas .ops = &designware_eth_ops, |
|
Re: [PATCH 2/3] designware: Use the remove() method with related drivers
Simon Glass
Hi Jonas,
On Mon, 7 Apr 2025 at 09:36, Jonas Karlman <jonas@...> wrote: I suppose so, yes. But not all network drivers use DMA. It should really be set by the driver. Well, since we can rely on the eth uclass, that is correct, but I still think it is better to add it, as an example to other drivers. Regards, Simon |
to navigate to use esc to dismiss