bootstd: Update ARM QEMU for standard boot and environment
Subsystems (3)
cover bootstd: Update ARM QEMU for standard boot and environment
This series fixes a few reported problems with virtio block-device handling.
It also moves QEMU over to use standard boot on ARM, by adding a suitable bootdev and bootmeth for qfw.
Finally, it moves the boards to use a text-based environment. The only remaining item in the config.h header file is CFG_SYS_SDRAM_BASE so it would be nice to move that to Kconfig, if possible.
Simon Glass (13):
log: Add a category for filesystems
virtio: Add some debugging
bootstd: Allow enabling BOOTSTD_FULL without needing EXPO
bootstd: Probe the block device before use
bootstd: Correct virtio block-device handling
bootstd: Add some default filesystems and commands
qemu: Update qfw command to use addresses
qemu: Move qfw kernel setup into a common file
qemu: Add a bootdev for qfw
qemu: Add a bootmeth for qfw
arm: qemu: Switch to standard boot
arm: qemu: Switch to a text environment
arm: qemu: Move GUIDs to the C file
board/emulation/qemu-arm/qemu-arm.c | 10 +++ board/emulation/qemu-arm/qemu-arm.env | 12 +++ boot/Kconfig | 28 +++++++ boot/Makefile | 5 +- boot/bootdev-uclass.c | 3 + boot/bootflow.c | 6 +- boot/bootmeth_qfw.c | 102 ++++++++++++++++++++++++++ cmd/bootflow.c | 21 ++++-- cmd/qfw.c | 89 +++------------------- common/log.c | 1 + common/qfw.c | 66 ++++++++++++++++- configs/qemu_arm64_defconfig | 5 +- configs/qemu_arm_defconfig | 5 +- configs/tools-only_defconfig | 1 + doc/develop/bootstd.rst | 4 + drivers/misc/qfw.c | 87 ++++++++++++++++++++++ drivers/virtio/virtio-uclass.c | 8 +- drivers/virtio/virtio_blk.c | 7 ++ fs/fat/fat_write.c | 2 + include/configs/qemu-arm.h | 75 ------------------- include/log.h | 2 + include/qfw.h | 13 ++++ test/boot/bootdev.c | 25 ++++--- test/boot/bootstd_common.h | 2 +- 24 files changed, 398 insertions(+), 181 deletions(-) create mode 100644 board/emulation/qemu-arm/qemu-arm.env create mode 100644 boot/bootmeth_qfw.c
Signature · 1 line
2.39.1.456.gfc5497dd1b-goog
PATCH 1/13 log: Add a category for filesystems
Sometimes it is useful to log things related to filesystems. Add a new category and place it at the top of one of the FAT files.
---
common/log.c | 1 + fs/fat/fat_write.c | 2 ++ include/log.h | 2 ++ 3 files changed, 5 insertions(+)
common/log.c +1 -0
@@ -30,6 +30,7 @@ static const char *const log_cat_name[] = {"acpi" ,"boot" ,"event" , +"fs" , }; _Static_assert(ARRAY_SIZE(log_cat_name) == LOGC_COUNT - LOGC_NONE,
fs/fat/fat_write.c +2 -0
@@ -5,6 +5,8 @@* R/W (V)FAT 12/16/32 filesystem implementation by Donggeun Kim */ +#define LOG_CATEGORY LOGC_FS +#include <common.h> #include <command.h> #include <config.h>
include/log.h +2 -0
@@ -100,6 +100,8 @@ enum log_category_t {LOGC_BOOT ,/** @LOGC_EVENT: Related to event and event handling */ LOGC_EVENT , +/** @LOGC_FS: Related to filesystems */ +LOGC_FS ,/** @LOGC_COUNT: Number of log categories */ LOGC_COUNT,/** @LOGC_END: Sentinel value for lists of log categories */
Signature · 1 line
2.39.1.456.gfc5497dd1b-goog
Re: [PATCH 01/13] log: Add a category for filesystems
Quoted 4 lines · Simon Glass
Sometimes it is useful to log things related to filesystems. Add a new category and place it at the top of one of the FAT files.
Signed-off-by Simon Glass <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom
PATCH 2/13 virtio: Add some debugging
In reply to Simon GlassWhen QEMU does not respond for some reason, it is helpful to have debugging info to show. Add some.
---
drivers/virtio/virtio_blk.c | 7 +++++++ 1 file changed, 7 insertions(+)
drivers/virtio/virtio_blk.c +7 -0
@@ -4,6 +4,8 @@* Copyright (C) 2018, Bin Meng <[email protected]> */ +#define LOG_CATEGORY UCLASS_VIRTIO +#include <common.h> #include <blk.h> #include <dm.h>
@@ -42,6 +44,8 @@ static ulong virtio_blk_do_req(struct udevice *dev, u64 sector, sgs[num_out + num_in++] = &data_sg; sgs[num_out + num_in++] = &status_sg; +log_debug ("dev=%s, active=%d, priv=%p, priv->vq=%p\n" , dev->name, +device_active (dev), priv, priv->vq); ret =virtqueue_add (priv->vq, sgs, num_out, num_in);if (ret)
@@ -49,8 +53,10 @@ static ulong virtio_blk_do_req(struct udevice *dev, u64 sector,virtqueue_kick (priv->vq); +log_debug ("wait..." );while (!virtqueue_get_buf (priv->vq,NULL )) ; +log_debug ("done\n" );return status ==VIRTIO_BLK_S_OK ? blkcnt : -EIO ; }
@@ -58,6 +64,7 @@ static ulong virtio_blk_do_req(struct udevice *dev, u64 sector,static ulong virtio_blk_read (struct udevice *dev,lbaint_t start,lbaint_t blkcnt,void *buffer) { +log_debug ("read %s\n" , dev->name);return virtio_blk_do_req (dev, start, blkcnt, buffer,VIRTIO_BLK_T_IN ); }
Signature · 1 line
2.39.1.456.gfc5497dd1b-goog
Re: [PATCH 02/13] virtio: Add some debugging
Quoted 4 lines · Simon Glass
When QEMU does not respond for some reason, it is helpful to have debugging info to show. Add some.
Signed-off-by Simon Glass <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom
PATCH 3/13 bootstd: Allow enabling BOOTSTD_FULL without needing EXPO
In reply to Simon GlassIt is sometimes useful to have one without the other, e.g. on a device without a display, since at present the expo feature requires CONFIG_VIDEO to be enabled.
Update the Makefile and bootflow command to support this, as well as the EXPO dependency.
---
boot/Kconfig | 1 + boot/Makefile | 4 ++-- cmd/bootflow.c | 21 +++++++++++++-------- 3 files changed, 16 insertions(+), 10 deletions(-)
boot/Kconfig +1 -0
@@ -561,6 +561,7 @@ endif # BOOTMETH_VBE config EXPO bool "Support for expos - groups of scenes displaying a UI" + depends on VIDEO default y if BOOTMETH_VBE help An expo is a way of presenting and collecting information from the
boot/Makefile +2 -2
@@ -30,7 +30,7 @@ obj-$(CONFIG_$(SPL_TPL_)BOOTMETH_SANDBOX) += bootmeth_sandbox.o obj-$(CONFIG_$(SPL_TPL_)BOOTMETH_SCRIPT) += bootmeth_script.o ifdef CONFIG_$(SPL_TPL_)BOOTSTD_FULL obj-$(CONFIG_$(SPL_TPL_)CMD_BOOTEFI_BOOTMGR) += bootmeth_efi_mgr.o -obj-$(CONFIG_$(SPL_TPL_)BOOTSTD) += bootflow_menu.o +obj-$(CONFIG_$(SPL_TPL_)EXPO) += bootflow_menu.o endif obj-$(CONFIG_$(SPL_TPL_)OF_LIBFDT) += image-fdt.o
@@ -48,7 +48,7 @@ ifdef CONFIG_SPL_BUILD obj-$(CONFIG_SPL_LOAD_FIT) += common_fit.o endif -obj-$(CONFIG_$(SPL_TPL_)BOOTMETH_VBE) += expo.o scene.o scene_menu.o +obj-$(CONFIG_$(SPL_TPL_)EXPO) += expo.o scene.o scene_menu.o obj-$(CONFIG_$(SPL_TPL_)BOOTMETH_VBE) += vbe.o vbe_request.o obj-$(CONFIG_$(SPL_TPL_)BOOTMETH_VBE_SIMPLE) += vbe_simple.o
cmd/bootflow.c +13 -8
@@ -394,15 +394,20 @@ static int do_bootflow_menu(struct cmd_tbl *cmdtp, int flag, int argc,if (ret)return CMD_RET_FAILURE ; - ret =bootflow_menu_run (std, text_mode, &bflow); -if (ret) { -if (ret == -EAGAIN ) -printf ("Nothing chosen\n" ); -else -printf ("Menu failed (err=%d)\n" , ret); - -return CMD_RET_FAILURE ; +if (IS_ENABLED (CONFIG_EXPO )) { + ret =bootflow_menu_run (std, text_mode, &bflow); +if (ret) { +if (ret == -EAGAIN ) +printf ("Nothing chosen\n" ); +else +printf ("Menu failed (err=%d)\n" , ret); + } + }else { +printf ("Menu not supported\n" ); + ret = -ENOSYS ; } +if (ret) +return CMD_RET_FAILURE ;printf ("Selected: %s\n" , bflow->os_name ? bflow->os_name : bflow->name); std->cur_bootflow = bflow;
Signature · 1 line
2.39.1.456.gfc5497dd1b-goog
Re: [PATCH 03/13] bootstd: Allow enabling BOOTSTD_FULL without needing EXPO
Quoted 8 lines · Simon Glass
It is sometimes useful to have one without the other, e.g. on a device without a display, since at present the expo feature requires
CONFIG_VIDEOto be enabled.Update the
Makefileand bootflow command to support this, as well as the EXPO dependency.Signed-off-by Simon Glass <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom
PATCH 4/13 bootstd: Probe the block device before use
In reply to Simon GlassIn some cases the block device is obtained but is not probed, since it is a sibling of the bootdev. Make sure it is probed, so it can be used without any trouble.
This fixes a bug with virtio, where the device is accessed before it has been set up by the virtio uclass.
---
boot/bootdev-uclass.c | 3 +++ boot/bootflow.c | 4 ++++ 2 files changed, 7 insertions(+)
boot/bootdev-uclass.c +3 -0
@@ -309,6 +309,9 @@ int bootdev_get_sibling_blk(struct udevice *dev, struct udevice **blkp)if (ret)return log_msg_ret ("find" , ret); } + ret =device_probe (blk); +if (ret) +return log_msg_ret ("act" , ret); *blkp = blk;return 0 ;
boot/bootflow.c +4 -0
@@ -270,6 +270,10 @@ static int iter_incr(struct bootflow_iter *iter)if (ret) {bootflow_iter_set_dev (iter,NULL ,0 ); }else { +/* +* Probe the bootdev. This does not probe any attached +* block device, since they are siblings +*/ ret =device_probe (dev);log_debug ("probe %s %d\n" , dev->name, ret);if (!log_msg_ret ("probe" , ret))
Signature · 1 line
2.39.1.456.gfc5497dd1b-goog
Re: [PATCH 04/13] bootstd: Probe the block device before use
Quoted 10 lines · Simon Glass
In some cases the block device is obtained but is not probed, since it is a sibling of the bootdev. Make sure it is probed, so it can be used without any trouble.
This fixes a bug with virtio, where the device is accessed before it has been set up by the virtio uclass.
Signed-off-by Simon Glass <[email protected]>Fixes 201417d700a ("bootstd: Add the bootdev uclass")Reported-by Ilias Apalodimas <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom
PATCH 5/13 bootstd: Correct virtio block-device handling
In reply to Simon GlassAt present virtio tries to attach QEMU services to a bootdev device, which cannot work. Add a check for this.
Also use bootdev_setup_sibling_blk() to create the bootdev device, since it allows the correct name to be used and bootdev_get_sibling_blk() to work as expected.
The bootdev is not created on sandbox since it does have a real virtio device and it is not possible to read blocks.
---
drivers/virtio/virtio-uclass.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/virtio/virtio-uclass.c +6 -2
@@ -247,8 +247,8 @@ static int virtio_uclass_post_probe(struct udevice *udev) }device_set_name_alloced (vdev); -if (uc_priv->device ==VIRTIO_ID_BLOCK ) { - ret =bootdev_setup_for_dev (udev, name); +if (uc_priv->device ==VIRTIO_ID_BLOCK && !IS_ENABLED (CONFIG_SANDBOX )) { + ret =bootdev_setup_sibling_blk (vdev,"virtio_bootdev" );if (ret)return log_msg_ret ("bootdev" , ret); }
@@ -275,6 +275,10 @@ static int virtio_uclass_child_pre_probe(struct udevice *vdev)int i;int ret; +/* bootdevs are not virtio devices */ +if (device_get_uclass_id (vdev) ==UCLASS_BOOTDEV ) +return 0 ; +/* * Save the real virtio device (eg: virtio-net, virtio-blk) to * the transport (parent) device's uclass priv for future use.
Signature · 1 line
2.39.1.456.gfc5497dd1b-goog
Re: [PATCH 05/13] bootstd: Correct virtio block-device handling
Quoted 7 lines · Simon Glass
At present virtio tries to attach QEMU services to a bootdev device, which cannot work. Add a check for this.
Also use
bootdev_setup_sibling_blk()to create the bootdev device, since it allows the correct name to be used andbootdev_get_sibling_blk()to work as expected.
Quoted 6 lines
The bootdev is not created on sandbox since it does have a real virtio device and it is not possible to read blocks.
Signed-off-by Simon Glass <[email protected]>Fixes a60f7a3e35b ("bootstd: Add a virtio bootdev")Reported-by Ilias Apalodimas <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom
PATCH 6/13 bootstd: Add some default filesystems and commands
In reply to Simon GlassWe need to support a basic set of filesystems for booting to work in most cases. Add these in via a new option, letting the board disable them individually (for space reasons) if desired.
This enables the filesystem commands as well as the actual functionality, even though bootstd is quite happy to use ext4 without the ext4 command. Further work would be needed to disintangle this and reduce code size.
Add several other options as well, providing sensible defaults.
We cannot enable this by default, since it expands the size of many boards quite a lot.
---
boot/Kconfig | 27 +++++++++++++++++++++++++++ configs/tools-only_defconfig | 1 + doc/develop/bootstd.rst | 4 ++++ 3 files changed, 32 insertions(+)
boot/Kconfig +27 -0
@@ -406,6 +406,33 @@ config VPL_BOOTSTD if BOOTSTD +config BOOTSTD_DEFAULTS + bool "Select some common defaults for standard boot" + depends on BOOTSTD + imply USE_BOOTCOMMAND + # Bring in some defaults which are generally needed. Boards can drop + # these as needed to save code space. Bootstd does not generally require + # the commands themselves to be enabled, but this is how some of the + # functionality is controlled at present + imply CMD_EXT2 + imply CMD_EXT4 + imply CMD_FAT + imply CMD_FS_GENERIC + imply CMD_PART + imply CMD_DHCP if NET + imply CMD_MII if NET + imply CMD_PING if NET + imply CMD_PXE if NET + imply USB_STORAGE + imply SUPPORT_RAW_INITRD + imply ENV_VARS_UBOOT_CONFIG + imply EFI_PARTITION + imply ISO_PARTITION + help + These are not required but are commonly needed to support a good + selection of booting methods. Enable this to improve the capability + of U-Boot to boot various images. + config BOOTSTD_BOOTCOMMAND bool "Use bootstd to boot" default y if !DISTRO_DEFAULTS
configs/tools-only_defconfig +1 -0
@@ -12,6 +12,7 @@ CONFIG_FIT_SIGNATURE=y # CONFIG_BOOTMETH_VBE is not set CONFIG_USE_BOOTCOMMAND=y CONFIG_BOOTCOMMAND="run distro_bootcmd" +# CONFIG_AVB_VERIFY is not set # CONFIG_CMD_BOOTD is not set # CONFIG_CMD_BOOTM is not set # CONFIG_CMD_ELF is not set
doc/develop/bootstd.rst +4 -0
@@ -396,6 +396,10 @@ To enable all feature sof standard boot, use `CONFIG_BOOTSTD_FULL`. This includes the full set of commands, more error messages when things go wrong and bootmeth ordering with the bootmeths environment variable. +You should probably also enable `CONFIG_BOOTSTD_DEFAULTS`, which provides +several filesystem and network features (if `CONFIG_NET` is enabled) so that +a good selection of boot options is available. + Available bootmeth drivers --------------------------
Signature · 1 line
2.39.1.456.gfc5497dd1b-goog
Re: [PATCH 06/13] bootstd: Add some default filesystems and commands
Quoted 8 lines · Simon Glass
We need to support a basic set of filesystems for booting to work in most cases. Add these in via a new option, letting the board disable them individually (for space reasons) if desired.
This enables the filesystem commands as well as the actual functionality, even though bootstd is quite happy to use ext4 without the ext4 command. Further work would be needed to disintangle this and reduce code size.
Quoted 6 lines
Add several other options as well, providing sensible defaults.
We cannot enable this by default, since it expands the size of many boards quite a lot.
Signed-off-by Simon Glass <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom
PATCH 7/13 qemu: Update qfw command to use addresses
In reply to Simon GlassThis uses casts all over the place. Use the correct type so that these can be avoided, as is done with other commands. Also simplify a few conditionals.
---
cmd/qfw.c | 42 +++++++++++++++++++++--------------------- 1 file changed, 21 insertions(+), 21 deletions(-)
cmd/qfw.c +21 -21
@@ -7,6 +7,7 @@#include <command.h> #include <env.h> #include <errno.h> +#include <mapmem.h> #include <qfw.h> #include <dm.h>
@@ -17,7 +18,7 @@ static struct udevice *qfw_dev;* to 'load_addr', initrd to 'initrd_addr' and kernel command * line using qemu fw_cfg interface. */ -static int qemu_fwcfg_cmd_setup_kernel (void *load_addr,void *initrd_addr) +int qemu_fwcfg_cmd_setup_kernel (ulong load_addr,ulong initrd_addr) {char *data_addr;uint32_t setup_size, kernel_size, cmdline_size, initrd_size;
@@ -25,13 +26,13 @@ static int qemu_fwcfg_cmd_setup_kernel(void *load_addr, void *initrd_addr)qfw_read_entry (qfw_dev,FW_CFG_SETUP_SIZE ,4 , &setup_size);qfw_read_entry (qfw_dev,FW_CFG_KERNEL_SIZE ,4 , &kernel_size); -if (kernel_size ==0 ) { +if (!kernel_size) {printf ("fatal: no kernel available\n" ); -return CMD_RET_FAILURE ; +return -ENOENT ; } - data_addr = load_addr; -if (setup_size !=0 ) { + data_addr =map_sysmem (load_addr,0 ); +if (setup_size) {qfw_read_entry (qfw_dev,FW_CFG_SETUP_DATA ,le32_to_cpu (setup_size), data_addr); data_addr +=le32_to_cpu (setup_size);
@@ -42,9 +43,9 @@ static int qemu_fwcfg_cmd_setup_kernel(void *load_addr, void *initrd_addr) data_addr +=le32_to_cpu (kernel_size);env_set_hex ("filesize" ,le32_to_cpu (kernel_size)); - data_addr = initrd_addr; + data_addr =map_sysmem (initrd_addr,0 );qfw_read_entry (qfw_dev,FW_CFG_INITRD_SIZE ,4 , &initrd_size); -if (initrd_size ==0 ) { +if (!initrd_size) {printf ("warning: no initrd available\n" ); }else { qfw_read_entry(qfw_dev, FW_CFG_INITRD_DATA,
@@ -61,17 +62,16 @@ static int qemu_fwcfg_cmd_setup_kernel(void *load_addr, void *initrd_addr)* if kernel cmdline only contains '\0', (e.g. no -append * when invoking qemu), do not update bootargs */ -if (*data_addr !='\0' ) { +if (*data_addr) {if (env_set ("bootargs" , data_addr) <0 )printf ("warning: unable to change bootargs\n" ); } } -printf ("loading kernel to address %p size %x" , load_addr, +printf ("loading kernel to address %lx size %x" , load_addr,le32_to_cpu (kernel_size));if (initrd_size) -printf (" initrd %p size %x\n" , - initrd_addr, +printf (" initrd %lx size %x\n" , initrd_addr,le32_to_cpu (initrd_size));else printf ("\n" );
@@ -119,28 +119,28 @@ static int qemu_fwcfg_do_load(struct cmd_tbl *cmdtp, int flag, int argc, char *const argv[]) { char *env; -void *load_addr; -void *initrd_addr; +ulong load_addr; +ulong initrd_addr; env =env_get ("loadaddr" ); load_addr = env ? - (void *)hextoul (env,NULL ) : - (void *)CONFIG_SYS_LOAD_ADDR ; +hextoul (env,NULL ) : +CONFIG_SYS_LOAD_ADDR ; env =env_get ("ramdiskaddr" ); initrd_addr = env ? - (void *)hextoul (env,NULL ) : +hextoul (env,NULL ) : #ifdef CFG_RAMDISK_ADDR - (void *)CFG_RAMDISK_ADDR; +CFG_RAMDISK_ADDR ;#else -NULL ; +0 ;#endif if (argc ==2 ) { - load_addr = (void *)hextoul (argv[0 ],NULL ); - initrd_addr = (void *)hextoul (argv[1 ],NULL ); + load_addr =hextoul (argv[0 ],NULL ); + initrd_addr =hextoul (argv[1 ],NULL ); }else if (argc ==1 ) { - load_addr = (void *)hextoul (argv[0 ],NULL ); + load_addr =hextoul (argv[0 ],NULL ); }if (!load_addr || !initrd_addr) {
Signature · 1 line
2.39.1.456.gfc5497dd1b-goog
Re: [PATCH 07/13] qemu: Update qfw command to use addresses
Quoted 5 lines · Simon Glass
This uses casts all over the place. Use the correct type so that these can be avoided, as is done with other commands. Also simplify a few conditionals.
Signed-off-by Simon Glass <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom
PATCH 8/13 qemu: Move qfw kernel setup into a common file
In reply to Simon GlassThis is currently in the cmd/ file but we want to call it from a driver. Move it into a common place. Tidy up the header-file order while we are here.
---
cmd/qfw.c | 69 +-------------------------------------------------- common/qfw.c | 66 +++++++++++++++++++++++++++++++++++++++++++++++- include/qfw.h | 13 ++++++++++ 3 files changed, 79 insertions(+), 69 deletions(-)
cmd/qfw.c +1 -68
@@ -7,78 +7,11 @@#include <command.h> #include <env.h> #include <errno.h> -#include <mapmem.h> #include <qfw.h> #include <dm.h> static struct udevice *qfw_dev; -/* -* This function prepares kernel for zboot. It loads kernel data -* to 'load_addr', initrd to 'initrd_addr' and kernel command -* line using qemu fw_cfg interface. -*/ -int qemu_fwcfg_cmd_setup_kernel (ulong load_addr,ulong initrd_addr) -{ -char *data_addr; -uint32_t setup_size, kernel_size, cmdline_size, initrd_size; - -qfw_read_entry (qfw_dev,FW_CFG_SETUP_SIZE ,4 , &setup_size); -qfw_read_entry (qfw_dev,FW_CFG_KERNEL_SIZE ,4 , &kernel_size); - -if (!kernel_size) { -printf ("fatal: no kernel available\n" ); -return -ENOENT ; - } - - data_addr =map_sysmem (load_addr,0 ); -if (setup_size) { -qfw_read_entry (qfw_dev,FW_CFG_SETUP_DATA , -le32_to_cpu (setup_size), data_addr); - data_addr +=le32_to_cpu (setup_size); - } - -qfw_read_entry (qfw_dev,FW_CFG_KERNEL_DATA , -le32_to_cpu (kernel_size), data_addr); - data_addr +=le32_to_cpu (kernel_size); -env_set_hex ("filesize" ,le32_to_cpu (kernel_size)); - - data_addr =map_sysmem (initrd_addr,0 ); -qfw_read_entry (qfw_dev,FW_CFG_INITRD_SIZE ,4 , &initrd_size); -if (!initrd_size) { -printf ("warning: no initrd available\n" ); - }else { -qfw_read_entry (qfw_dev,FW_CFG_INITRD_DATA , -le32_to_cpu (initrd_size), data_addr); - data_addr +=le32_to_cpu (initrd_size); -env_set_hex ("filesize" ,le32_to_cpu (initrd_size)); - } - -qfw_read_entry (qfw_dev,FW_CFG_CMDLINE_SIZE ,4 , &cmdline_size); -if (cmdline_size) { -qfw_read_entry (qfw_dev,FW_CFG_CMDLINE_DATA , -le32_to_cpu (cmdline_size), data_addr); -/* -* if kernel cmdline only contains '\0', (e.g. no -append -* when invoking qemu), do not update bootargs -*/ -if (*data_addr) { -if (env_set ("bootargs" , data_addr) <0 ) -printf ("warning: unable to change bootargs\n" ); - } - } - -printf ("loading kernel to address %lx size %x" , load_addr, -le32_to_cpu (kernel_size)); -if (initrd_size) -printf (" initrd %lx size %x\n" , initrd_addr, -le32_to_cpu (initrd_size)); -else -printf ("\n" ); - -return 0 ; -} -static int qemu_fwcfg_cmd_list_firmware (void ) {int ret;
@@ -148,7 +81,7 @@ static int qemu_fwcfg_do_load(struct cmd_tbl *cmdtp, int flag,return CMD_RET_FAILURE ; } -return qemu_fwcfg_cmd_setup_kernel (load_addr, initrd_addr); +return qemu_fwcfg_setup_kernel (qfw_dev, load_addr, initrd_addr); }static struct cmd_tbl fwcfg_commands[] = {
common/qfw.c +65 -1
77 lines not on this page · raw message
include/qfw.h +13 -0
@@ -316,4 +316,17 @@ bool qfw_file_iter_end(struct fw_cfg_file_iter *iter);*/ int qemu_cpu_fixup (void ); +/* +* qemu_fwcfg_setup_kernel() - Prepare the kernel for zboot +* +* Loads kernel data to 'load_addr', initrd to 'initrd_addr' and kernel command +* line using qemu fw_cfg interface +* +* @load_addr: Load address for kernel +* @initrd_addr: Load address for ramdisk +* @return 0 if OK, -ENOENT if no kernel +*/ +int qemu_fwcfg_setup_kernel (struct udevice *qfw_dev,ulong load_addr, +ulong initrd_addr); +#endif
Signature · 1 line
2.39.1.456.gfc5497dd1b-goog
Re: [PATCH 08/13] qemu: Move qfw kernel setup into a common file
Quoted 5 lines · Simon Glass
This is currently in the cmd/ file but we want to call it from a driver. Move it into a common place. Tidy up the header-file order while we are here.
Signed-off-by Simon Glass <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom
PATCH 9/13 qemu: Add a bootdev for qfw
In reply to Simon GlassAdd a bootdev device for qfw so that it can be used with standard boot. This simply checks for the correct method and then does the read. Most of the other logic is handed in a new bootmeth driver.
---
drivers/misc/qfw.c | 87 ++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 87 insertions(+)
drivers/misc/qfw.c +87 -0
@@ -7,6 +7,9 @@#define LOG_CATEGORY UCLASS_QFW#include <common.h> +#include <bootdev.h> +#include <bootflow.h> +#include <bootmeth.h> #include <command.h> #include <errno.h> #include <log.h>
@@ -310,8 +313,92 @@ int qfw_register(struct udevice *dev)return 0 ; } +static int qfw_post_bind (struct udevice *dev) +{ +int ret; + + ret =bootdev_setup_for_dev (dev,"qfw_bootdev" ); +if (ret) +return log_msg_ret ("dev" , ret); + +return 0 ; +} + +static int qfw_get_bootflow (struct udevice *dev,struct bootflow_iter *iter, +struct bootflow *bflow) +{ +const struct udevice *media =dev_get_parent (dev); +int ret; + +if (!CONFIG_IS_ENABLED (BOOTSTD )) +return -ENOSYS ; + +log_debug ("media=%s\n" , media->name); + ret =bootmeth_check (bflow->method, iter); +if (ret) +return log_msg_ret ("check" , ret); + +log_debug ("iter->part=%d\n" , iter->part); + +/* We only support the whole device, not partitions */ +if (iter->part) +return log_msg_ret ("max" , -ESHUTDOWN ); + +log_debug ("reading bootflow with method: %s\n" , bflow->method->name); + ret =bootmeth_read_bootflow (bflow->method, bflow); +if (ret) +return log_msg_ret ("method" , ret); + +return 0 ; +} + +static int qfw_bootdev_bind (struct udevice *dev) +{ +struct bootdev_uc_plat *ucp =dev_get_uclass_plat (dev); + + ucp->prio =BOOTDEVP_4_SCAN_FAST ; + +return 0 ; +} + +static int qfw_bootdev_hunt (struct bootdev_hunter *info,bool show) +{ +int ret; + + ret =uclass_probe_all (UCLASS_QFW ); +if (ret && ret != -ENOENT ) +return log_msg_ret ("vir" , ret); + +return 0 ; +} +UCLASS_DRIVER (qfw) = { .id =UCLASS_QFW , .name ="qfw" , + .post_bind = qfw_post_bind, .per_device_auto =sizeof (struct qfw_dev ), }; + +struct bootdev_ops qfw_bootdev_ops = { + .get_bootflow = qfw_get_bootflow, +}; + +static const struct udevice_id qfw_bootdev_ids[] = { + { .compatible ="u-boot,bootdev-qfw" }, + { } +}; + +U_BOOT_DRIVER (qfw_bootdev) = { + .name ="qfw_bootdev" , + .id =UCLASS_BOOTDEV , + .ops = &qfw_bootdev_ops, + .bind = qfw_bootdev_bind, + .of_match = qfw_bootdev_ids, +}; + +BOOTDEV_HUNTER (qfw_bootdev_hunter) = { + .prio =BOOTDEVP_4_SCAN_FAST , + .uclass =UCLASS_QFW , + .hunt = qfw_bootdev_hunt, + .drv =DM_DRIVER_REF (qfw_bootdev), +};
Signature · 1 line
2.39.1.456.gfc5497dd1b-goog
Re: [PATCH 09/13] qemu: Add a bootdev for qfw
Quoted 5 lines · Simon Glass
Add a bootdev device for qfw so that it can be used with standard boot. This simply checks for the correct method and then does the read. Most of the other logic is handed in a new bootmeth driver.
Signed-off-by Simon Glass <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom
PATCH 10/13 qemu: Add a bootmeth for qfw
In reply to Simon GlassThis supports reading a kernel and ramdisk from qfw, then loading it with either the booti or bootz commands.
For now this uses the existing booti and bootz commands, rather than trying to call that functionality directly (e.g. do_bootm_states()). It does not require the HUSH parser though, which helps a little with size.
---
boot/Makefile | 1 + boot/bootflow.c | 2 +- boot/bootmeth_qfw.c | 102 ++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 104 insertions(+), 1 deletion(-) create mode 100644 boot/bootmeth_qfw.c
boot/Makefile +1 -0
@@ -26,6 +26,7 @@ obj-$(CONFIG_$(SPL_TPL_)BOOTSTD) += bootstd-uclass.o obj-$(CONFIG_$(SPL_TPL_)BOOTMETH_DISTRO) += bootmeth_distro.o obj-$(CONFIG_$(SPL_TPL_)BOOTMETH_DISTRO_PXE) += bootmeth_pxe.o obj-$(CONFIG_$(SPL_TPL_)BOOTMETH_EFILOADER) += bootmeth_efi.o +obj-$(CONFIG_$(SPL_TPL_)QFW) += bootmeth_qfw.o obj-$(CONFIG_$(SPL_TPL_)BOOTMETH_SANDBOX) += bootmeth_sandbox.o obj-$(CONFIG_$(SPL_TPL_)BOOTMETH_SCRIPT) += bootmeth_script.o ifdef CONFIG_$(SPL_TPL_)BOOTSTD_FULL
boot/bootflow.c +1 -1
@@ -508,7 +508,7 @@ int bootflow_iter_check_blk(const struct bootflow_iter *iter)enum uclass_id id =device_get_uclass_id (media);log_debug ("uclass %d: %s\n" , id,uclass_get_name (id)); -if (id !=UCLASS_ETH && id !=UCLASS_BOOTSTD ) +if (id !=UCLASS_ETH && id !=UCLASS_BOOTSTD && id !=UCLASS_QFW )return 0 ;return -ENOTSUPP ;
boot/bootmeth_qfw.c new file +102 -0
102 lines not on this page · raw message
Signature · 1 line
2.39.1.456.gfc5497dd1b-goog
Re: [PATCH 10/13] qemu: Add a bootmeth for qfw
Quoted 8 lines · Simon Glass
This supports reading a kernel and ramdisk from qfw, then loading it with either the booti or bootz commands.
For now this uses the existing booti and bootz commands, rather than trying to call that functionality directly (e.g.
do_bootm_states()). It does not require the HUSH parser though, which helps a little with size.Signed-off-by Simon Glass <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom
PATCH 11/13 arm: qemu: Switch to standard boot
In reply to Simon GlassDrop use of the distro scripts and use standard boot instead.
Enable BOOTDEV_FULL just for convenience, although this does add quite a bit to the size.
---
configs/qemu_arm64_defconfig | 4 ++- configs/qemu_arm_defconfig | 4 ++- include/configs/qemu-arm.h | 54 ++---------------------------------- 3 files changed, 8 insertions(+), 54 deletions(-)
configs/qemu_arm64_defconfig +3 -1
@@ -13,13 +13,14 @@ CONFIG_SYS_LOAD_ADDR=0x40200000 CONFIG_ENV_ADDR=0x4000000 CONFIG_DEBUG_UART=y CONFIG_AHCI=y -CONFIG_DISTRO_DEFAULTS=y CONFIG_HAS_CUSTOM_SYS_INIT_SP_ADDR=y CONFIG_CUSTOM_SYS_INIT_SP_ADDR=0x40200000 CONFIG_FIT=y CONFIG_FIT_SIGNATURE=y CONFIG_FIT_VERBOSE=y CONFIG_FIT_BEST_MATCH=y +CONFIG_BOOTSTD_FULL=y +CONFIG_BOOTSTD_DEFAULTS=y CONFIG_LEGACY_IMAGE_FORMAT=y CONFIG_USE_PREBOOT=y # CONFIG_DISPLAY_CPUINFO is not set
@@ -27,6 +28,7 @@ CONFIG_USE_PREBOOT=y CONFIG_PCI_INIT_R=y CONFIG_SYS_CBSIZE=512 CONFIG_SYS_PBSIZE=532 +CONFIG_CMD_BOOTZ=y CONFIG_CMD_BOOTEFI_SELFTEST=y CONFIG_CMD_NVEDIT_EFI=y CONFIG_CMD_DFU=y
configs/qemu_arm_defconfig +3 -1
@@ -14,13 +14,14 @@ CONFIG_SYS_LOAD_ADDR=0x40200000 CONFIG_ENV_ADDR=0x4000000 CONFIG_DEBUG_UART=y CONFIG_AHCI=y -CONFIG_DISTRO_DEFAULTS=y CONFIG_HAS_CUSTOM_SYS_INIT_SP_ADDR=y CONFIG_CUSTOM_SYS_INIT_SP_ADDR=0x40200000 CONFIG_FIT=y CONFIG_FIT_SIGNATURE=y CONFIG_FIT_VERBOSE=y CONFIG_FIT_BEST_MATCH=y +CONFIG_BOOTSTD_FULL=y +CONFIG_BOOTSTD_DEFAULTS=y CONFIG_LEGACY_IMAGE_FORMAT=y CONFIG_USE_PREBOOT=y # CONFIG_DISPLAY_CPUINFO is not set
@@ -28,6 +29,7 @@ CONFIG_USE_PREBOOT=y CONFIG_PCI_INIT_R=y CONFIG_SYS_CBSIZE=512 CONFIG_SYS_PBSIZE=532 +CONFIG_CMD_BOOTZ=y CONFIG_SYS_BOOTM_LEN=0x4000000 CONFIG_CMD_BOOTEFI_SELFTEST=y CONFIG_CMD_NVEDIT_EFI=y
include/configs/qemu-arm.h +2 -52
68 lines not on this page · raw message
Signature · 1 line
2.39.1.456.gfc5497dd1b-goog
Re: [PATCH 11/13] arm: qemu: Switch to standard boot
Quoted 6 lines · Simon Glass
Drop use of the distro scripts and use standard boot instead.
Enable
BOOTDEV_FULLjust for convenience, although this does add quite a bit to the size.Signed-off-by Simon Glass <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom
PATCH 12/13 arm: qemu: Switch to a text environment
In reply to Simon GlassUse the new environment format so we can drop most of the config.h file.
---
board/emulation/qemu-arm/qemu-arm.env | 12 ++++++++++++ include/configs/qemu-arm.h | 13 ------------- 2 files changed, 12 insertions(+), 13 deletions(-) create mode 100644 board/emulation/qemu-arm/qemu-arm.env
board/emulation/qemu-arm/qemu-arm.env new file +12 -0
@@ -0,0 +1,12 @@ +/* SPDX-License-Identifier: GPL-2.0+ */ + +/* environment for qemu-arm and qemu-arm64 */ + +fdt_high=0xffffffff +initrd_high=0xffffffff +fdt_addr=0x40000000 +scriptaddr=0x40200000 +pxefile_addr_r=0x40300000 +kernel_addr_r=0x40400000 +ramdisk_addr_r=0x44000000 +boot_targets=qfw usb scsi virtio nvme dhcp
include/configs/qemu-arm.h +0 -13
@@ -23,17 +23,4 @@/* For timer, QEMU emulates an ARMv7/ARMv8 architected timer */ -/* Environment options */ -#define BOOT_TARGETS "qfw usb scsi virtio nvme dhcp" - -#define CFG_EXTRA_ENV_SETTINGS \ - "fdt_high=0xffffffff\0" \ - "initrd_high=0xffffffff\0" \ - "fdt_addr=0x40000000\0" \ - "scriptaddr=0x40200000\0" \ - "pxefile_addr_r=0x40300000\0" \ - "kernel_addr_r=0x40400000\0" \ - "ramdisk_addr_r=0x44000000\0" \ - "boot_targets=" BOOT_TARGETS "\0" -#endif /* __CONFIG_H */
Signature · 1 line
2.39.1.456.gfc5497dd1b-goog
Re: [PATCH 12/13] arm: qemu: Switch to a text environment
Quoted 3 lines · Simon Glass
Use the new environment format so we can drop most of the
config.hfile.Signed-off-by Simon Glass <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom
PATCH 13/13 arm: qemu: Move GUIDs to the C file
In reply to Simon GlassThese are only used in one place, so move them there.
---
board/emulation/qemu-arm/qemu-arm.c | 10 ++++++++++ configs/qemu_arm64_defconfig | 1 + configs/qemu_arm_defconfig | 1 + include/configs/qemu-arm.h | 12 ------------ test/boot/bootdev.c | 25 ++++++++++++++----------- test/boot/bootstd_common.h | 2 +- 6 files changed, 27 insertions(+), 24 deletions(-)
board/emulation/qemu-arm/qemu-arm.c +10 -0
@@ -15,6 +15,16 @@#include <virtio.h> #include <linux/kernel.h> +#include <linux/sizes.h> + +/* GUIDs for capsule updatable firmware images */ +#define QEMU_ARM_UBOOT_IMAGE_GUID \ + EFI_GUID(0xf885b085, 0x99f8, 0x45af, 0x84, 0x7d, \ + 0xd5, 0x14, 0x10, 0x7a, 0x4a, 0x2c) + +#define QEMU_ARM64_UBOOT_IMAGE_GUID \ + EFI_GUID(0x058b7d83, 0x50d5, 0x4c47, 0xa1, 0x95, \ + 0x60, 0xd8, 0x6a, 0xd3, 0x41, 0xc4)#ifdef CONFIG_ARM64 #include <asm/armv8/mmu.h>
configs/qemu_arm64_defconfig +1 -0
@@ -72,3 +72,4 @@ CONFIG_USB=y CONFIG_USB_EHCI_HCD=y CONFIG_USB_EHCI_PCI=y CONFIG_TPM=y +CONFIG_HUSH_PARSER=y
configs/qemu_arm_defconfig +1 -0
@@ -74,3 +74,4 @@ CONFIG_USB=y CONFIG_USB_EHCI_HCD=y CONFIG_USB_EHCI_PCI=y CONFIG_TPM=y +CONFIG_HUSH_PARSER=y
include/configs/qemu-arm.h +0 -12
@@ -6,21 +6,9 @@#ifndef __CONFIG_H#define __CONFIG_H -#include <linux/sizes.h> -/* Physical memory map */ -#define CFG_SYS_SDRAM_BASE 0x40000000 -/* GUIDs for capsule updatable firmware images */ -#define QEMU_ARM_UBOOT_IMAGE_GUID \ - EFI_GUID(0xf885b085, 0x99f8, 0x45af, 0x84, 0x7d, \ - 0xd5, 0x14, 0x10, 0x7a, 0x4a, 0x2c) - -#define QEMU_ARM64_UBOOT_IMAGE_GUID \ - EFI_GUID(0x058b7d83, 0x50d5, 0x4c47, 0xa1, 0x95, \ - 0x60, 0xd8, 0x6a, 0xd3, 0x41, 0xc4) -/* For timer, QEMU emulates an ARMv7/ARMv8 architected timer */ #endif /* __CONFIG_H */
test/boot/bootdev.c +14 -11
@@ -316,11 +316,12 @@ static int bootdev_test_hunter(struct unit_test_state *uts)ut_assert_nextline (" 5 ide ide_bootdev" );ut_assert_nextline (" 2 mmc mmc_bootdev" );ut_assert_nextline (" 4 nvme nvme_bootdev" ); +ut_assert_nextline (" 4 qfw qfw_bootdev" );ut_assert_nextline (" 4 scsi scsi_bootdev" );ut_assert_nextline (" 4 spi_flash sf_bootdev" );ut_assert_nextline (" 5 usb usb_bootdev" );ut_assert_nextline (" 4 virtio virtio_bootdev" ); -ut_assert_nextline ("(total hunters: 9)" ); +ut_assert_nextline ("(total hunters: 10)" );ut_assert_console_end ();ut_assertok (bootdev_hunt ("usb1" , false));
@@ -328,8 +329,8 @@ static int bootdev_test_hunter(struct unit_test_state *uts) "Bus usb@1: scanning bus usb@1 for devices... 5 USB Device(s) found");ut_assert_console_end (); -/* USB is sixth in the list, so bit 7 */ -ut_asserteq (BIT (7 ), std->hunters_used); +/* USB is 7th in the list, so bit 8 */ +ut_asserteq (BIT (8 ), std->hunters_used);return 0 ; }
@@ -350,7 +351,7 @@ static int bootdev_test_cmd_hunt(struct unit_test_state *uts)ut_assert_nextline ("Prio Used Uclass Hunter" );ut_assert_nextlinen ("----" );ut_assert_nextline (" 6 ethernet eth_bootdev" ); -ut_assert_skip_to_line ("(total hunters: 9)" ); +ut_assert_skip_to_line ("(total hunters: 10)" );ut_assert_console_end ();/* Use the MMC hunter and see that it updates */
@@ -358,7 +359,7 @@ static int bootdev_test_cmd_hunt(struct unit_test_state *uts)ut_assertok (run_command ("bootdev hunt -l" ,0 ));ut_assert_skip_to_line (" 5 ide ide_bootdev" );ut_assert_nextline (" 2 * mmc mmc_bootdev" ); -ut_assert_skip_to_line ("(total hunters: 9)" ); +ut_assert_skip_to_line ("(total hunters: 10)" );ut_assert_console_end ();/* Scan all hunters */
@@ -376,6 +377,7 @@ static int bootdev_test_cmd_hunt(struct unit_test_state *uts)/* mmc hunter has already been used so should not run again */ ut_assert_nextline ("Hunting with: nvme" ); +ut_assert_nextline ("Hunting with: qfw" );ut_assert_nextline ("Hunting with: scsi" );ut_assert_nextline ("scanning bus for devices..." );ut_assert_skip_to_line ("Hunting with: spi_flash" );
@@ -394,11 +396,12 @@ static int bootdev_test_cmd_hunt(struct unit_test_state *uts)ut_assert_nextline (" 5 * ide ide_bootdev" );ut_assert_nextline (" 2 * mmc mmc_bootdev" );ut_assert_nextline (" 4 * nvme nvme_bootdev" ); +ut_assert_nextline (" 4 * qfw qfw_bootdev" );ut_assert_nextline (" 4 * scsi scsi_bootdev" );ut_assert_nextline (" 4 * spi_flash sf_bootdev" );ut_assert_nextline (" 5 * usb usb_bootdev" );ut_assert_nextline (" 4 * virtio virtio_bootdev" ); -ut_assert_nextline ("(total hunters: 9)" ); +ut_assert_nextline ("(total hunters: 10)" );ut_assert_console_end ();ut_asserteq (GENMASK (MAX_HUNTER ,0 ), std->hunters_used);
@@ -585,8 +588,8 @@ static int bootdev_test_next_label(struct unit_test_state *uts)ut_asserteq_str ("scsi.id0lun0.bootdev" , dev->name);ut_asserteq (BOOTFLOW_METHF_SINGLE_UCLASS , mflags); -/* SCSI is sixth in the list, so bit 5 */ -ut_asserteq (BIT (MMC_HUNTER ) |BIT (5 ), std->hunters_used); +/* SCSI is 7th in the list, so bit 6 */ +ut_asserteq (BIT (MMC_HUNTER ) |BIT (6 ), std->hunters_used);ut_assertok (bootdev_next_label (&iter, &dev, &mflags));ut_assert_console_end ();
@@ -596,7 +599,7 @@ static int bootdev_test_next_label(struct unit_test_state *uts) mflags);/* dhcp: Ethernet is first so bit 0 */ -ut_asserteq (BIT (MMC_HUNTER ) |BIT (5 ) |BIT (0 ), std->hunters_used); +ut_asserteq (BIT (MMC_HUNTER ) |BIT (6 ) |BIT (0 ), std->hunters_used);ut_assertok (bootdev_next_label (&iter, &dev, &mflags));ut_assert_console_end ();
@@ -606,7 +609,7 @@ static int bootdev_test_next_label(struct unit_test_state *uts) mflags);/* pxe: Ethernet is first so bit 0 */ -ut_asserteq (BIT (MMC_HUNTER ) |BIT (5 ) |BIT (0 ), std->hunters_used); +ut_asserteq (BIT (MMC_HUNTER ) |BIT (6 ) |BIT (0 ), std->hunters_used); mflags =123 ;ut_asserteq (-ENODEV ,bootdev_next_label (&iter, &dev, &mflags));
@@ -614,7 +617,7 @@ static int bootdev_test_next_label(struct unit_test_state *uts)ut_assert_console_end ();/* no change */ -ut_asserteq (BIT (MMC_HUNTER ) |BIT (5 ) |BIT (0 ), std->hunters_used); +ut_asserteq (BIT (MMC_HUNTER ) |BIT (6 ) |BIT (0 ), std->hunters_used);return 0 ; }
test/boot/bootstd_common.h +1 -1
@@ -21,7 +21,7 @@#define TEST_VERNUM 0x00010002enum { -MAX_HUNTER =8 , +MAX_HUNTER =9 ,MMC_HUNTER =3 ,/* ID of MMC hunter */ };
Signature · 1 line
2.39.1.456.gfc5497dd1b-goog
Re: [PATCH 13/13] arm: qemu: Move GUIDs to the C file
Quoted 3 lines · Simon Glass
These are only used in one place, so move them there.
Signed-off-by Simon Glass <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom
Quoted 9 lines · Simon Glass
This series fixes a few reported problems with virtio block-device handling.
It also moves QEMU over to use standard boot on ARM, by adding a suitable bootdev and bootmeth for qfw.
Finally, it moves the boards to use a text-based environment. The only remaining item in the
config.hheader file isCFG_SYS_SDRAM_BASEso it would be nice to move that toKconfig, if possible.
Figuring out what to do about CFG_SYS_SDRAM_BASE is a good question. For this series, have you been able to perform other tests on top of it, such as verifying off the shelf distributions still boot / install as expected?
Signature · 1 line
Tom
Hi Tom,
Quoted 8 lines · Tom Rini
Quoted 6 lines · Simon Glass
This series fixes a few reported problems with virtio block-device handling.
It also moves QEMU over to use standard boot on ARM, by adding a suitable bootdev and bootmeth for qfw.
Quoted 8 lines
Quoted 3 lines
Finally, it moves the boards to use a text-based environment. The only remaining item in the
config.hheader file isCFG_SYS_SDRAM_BASEso it would be nice to move that toKconfig, if possible.Figuring out what to do about
CFG_SYS_SDRAM_BASEis a good question. For this series, have you been able to perform other tests on top of it, such as verifying off the shelf distributions still boot / install as expected?
I did not do a new install, but did check that it booting Debian OK. Should we try to get some of this into CI or would it be too painful?
I have done a lot of checking with RockPro64 which of course uses the same logic (except for qfw which is new and which I manually tested). Of course we also have a lot of unit tests for bootstd, to prevent regressionsl,
Regards,
Simon