bootconfig: Make size and checksum fields le32
Subsystems (2)
Masami Hiramatsu
maintainercover bootconfig: Make size and checksum fields le32
Hello,
This is a series of patches to make the size and the checksum fields in the footer le32 instead of u32.
In the thread for alignment series[1], Steve pointed that the current footer format didn't specify the endianness thus it is hard to apply the bootconfig for cross-build initrd if the target endianness is different from the host machine.
I've proposed that the hexadecimal ASCII string in the previous series
[2] but Linus pointed that making it le32 was enough.
So this just make those fields le32.
Thank you,
[1] https://lore.kernel.org/lkml/[email protected]/
[2] https://lore.kernel.org/linux-doc/CAHk-=wi9RedSQoGF06dVs2mp7tBp4QoiW8+XZzNcDFJr3Zo5gg@mail.gmail.com/
---
Masami Hiramatsu (3):
bootconfig: Load size and checksum in the footer as le32 tools/bootconfig: Store size and checksum in footer as le32
docs: bootconfig: Add the endianness of fields
Documentation/admin-guide/bootconfig.rst | 4 +++- init/main.c | 4 ++-- tools/bootconfig/main.c | 7 +++++-- 3 files changed, 10 insertions(+), 5 deletions(-)
--
Masami Hiramatsu (Linaro) <[email protected]>
Masami Hiramatsu
maintainerPATCH 1/3 bootconfig: Load size and checksum in the footer as le32
Load the size and the checksum fields in the footer as le32 instead of u32. This will allow us to apply bootconfig to the cross build initrd without caring the endianness.
--- init/main.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
init/main.c +2 -2
@@ -288,8 +288,8 @@ static void * __init get_boot_config_from_initrd(u32 *_size, u32 *_csum) found: hdr = (u32 *)(data -8 ); - size = hdr[0 ]; - csum = hdr[1 ]; + size =le32_to_cpu (hdr[0 ]); + csum =le32_to_cpu (hdr[1 ]); data = ((void *)hdr) - size;if ((unsigned long )data < initrd_start) {
Masami Hiramatsu
maintainerPATCH 2/3 tools/bootconfig: Store size and checksum in footer as le32
In reply to Masami HiramatsuStore the size and the checksum fields in the footer as le32 instead of u32. This will allow us to apply bootconfig to the cross build initrd without caring the endianness.
--- tools/bootconfig/main.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-)
tools/bootconfig/main.c +5 -2
@@ -10,6 +10,7 @@#include <unistd.h> #include <string.h> #include <errno.h> +#include <endian.h> #include <linux/kernel.h> #include <linux/bootconfig.h>
@@ -183,9 +184,11 @@ static int load_xbc_from_initrd(int fd, char **buf)if (read (fd, &size,sizeof (u32)) <0 )return pr_errno ("Failed to read size" , -errno); + size =le32toh (size);if (read (fd, &csum,sizeof (u32)) <0 )return pr_errno ("Failed to read checksum" , -errno); + csum =le32toh (csum);/* Wrong size error */ if (stat.st_size < size +8 +BOOTCONFIG_MAGIC_LEN ) {
@@ -407,10 +410,10 @@ static int apply_xbc(const char *path, const char *xbc_path)/* Add a footer */ p = data + size; - *(u32 *)p = size; + *(u32 *)p =htole32 (size); p +=sizeof (u32); - *(u32 *)p = csum; + *(u32 *)p =htole32 (csum); p +=sizeof (u32);memcpy (p,BOOTCONFIG_MAGIC ,BOOTCONFIG_MAGIC_LEN );
Masami Hiramatsu
maintainerPATCH 3/3 docs: bootconfig: Add the endianness of fields
In reply to Masami HiramatsuAdd a description about the endianness of the size and the checksum fields. Those must be stored as le32 instead of u32. This will allow us to apply bootconfig to the cross build initrd without caring the endianness.
--- Documentation/admin-guide/bootconfig.rst | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
Documentation/admin-guide/bootconfig.rst +3 -1
@@ -140,7 +140,9 @@ Since the boot configuration file is loaded with initrd, it will be added to the end of the initrd (initramfs) image file with padding, size, checksum and 12-byte magic word as below. -[initrd][bootconfig][padding][size(u32)][checksum(u32)][#BOOTCONFIG\n] +[initrd][bootconfig][padding][size(le32)][checksum(le32)][#BOOTCONFIG\n] + +The size and checksum fields are unsigned 32bit little endian value. When the boot configuration is added to the initrd image, the total file size is aligned to 4 bytes. To fill the gap, null characters
Quoted 13 lines · On Fri, 20 Nov 2020 11:28:55 +0900 Masami Hiramatsu <[email protected]> wrote:
Hello,
This is a series of patches to make the size and the checksum fields in the footer le32 instead of u32.
In the thread for alignment series[1], Steve pointed that the current footer format didn't specify the endianness thus it is hard to apply the bootconfig for cross-build initrd if the target endianness is different from the host machine.
I've proposed that the hexadecimal ASCII string in the previous series
[2] but Linus pointed that making it le32 was enough.
Quoted 9 lines
So this just make those fields le32.
Thank you,
[1] https://lore.kernel.org/lkml/[email protected]/
[2] https://lore.kernel.org/linux-doc/CAHk-=wi9RedSQoGF06dVs2mp7tBp4QoiW8+XZzNcDFJr3Zo5gg@mail.gmail.com/---
I have this all tested. Is this something that should go into the current -rc release, or wait for the next merge window?
-- Steve
Masami Hiramatsu
maintainerQuoted 20 lines · On Tue, 1 Dec 2020 10:48:18 -0500 Steven Rostedt <[email protected]> wrote:
Quoted 17 lines · On Fri, 20 Nov 2020 11:28:55 +0900 Masami Hiramatsu <[email protected]> wrote:
Hello,
This is a series of patches to make the size and the checksum fields in the footer le32 instead of u32.
In the thread for alignment series[1], Steve pointed that the current footer format didn't specify the endianness thus it is hard to apply the bootconfig for cross-build initrd if the target endianness is different from the host machine.
I've proposed that the hexadecimal ASCII string in the previous series
[2] but Linus pointed that making it le32 was enough.So this just make those fields le32.
Thank you,
Quoted 8 lines
Quoted 5 lines
[1] https://lore.kernel.org/lkml/[email protected]/
[2] https://lore.kernel.org/linux-doc/CAHk-=wi9RedSQoGF06dVs2mp7tBp4QoiW8+XZzNcDFJr3Zo5gg@mail.gmail.com/---
I have this all tested. Is this something that should go into the current -rc release, or wait for the next merge window?
Hrm, this kind of things usually good for the next merge window because it is only for the different endianess cross-build case. But the next release will be LTS (6 years life cycle), so it is possible someone hits this issue before EOL and (I guess) finally they need this series.
So if possible, could you push this in -rc release? I think this doesn't change so much.
Thank you,
Signature · 1 line
Masami Hiramatsu <[email protected]>