barebox/barebox-tools: upgrade 2025.09.0 -> 2025.09.3
PATCH 1/1 barebox/barebox-tools: upgrade 2025.09.0 -> 2025.09.3
From: Ankur Tyagi <[email protected]>
2025.09.3
---------
Fixed FIT image vulnerability
https://lore.barebox.org/barebox/[email protected]/
Changelog:
https://github.com/barebox/barebox/compare/v2025.09.2...v2025.09.3
2025.09.2
---------
Changelog:
https://github.com/barebox/barebox/compare/v2025.09.1...v2025.09.2
2025.09.1
---------
This stable release is specifically targeted at whinlatter which is currently at v2025.09.0
https://lore.barebox.org/barebox/[email protected]/
Changelog:
https://github.com/barebox/barebox/compare/v2025.09.0...v2025.09.1
--- meta/recipes-bsp/barebox/barebox-common.inc | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
meta/recipes-bsp/barebox/barebox-common.inc +2 -2
@@ -3,6 +3,6 @@ SECTION = "bootloaders" LIC_FILES_CHKSUM = "file://COPYING;md5=f5125d13e000b9ca1f0d3364286c4192" -PV = "2025.09.0" +PV = "2025.09.3" SRC_URI = "https://barebox.org/download/barebox-${PV}.tar.bz2" -SRC_URI[sha256sum] = "7df1aa47bb7bf1763a729137ac773e69a4052812af094475d739fc63a9295f0d" +SRC_URI[sha256sum] = "e87eb863cbe45e4f5af8930825c8f6e20c02b82451e9e1125ea1c73c1fb49a87"
Quoted 21 lines · Ankur Tyagi via lists.openembedded.org
From: Ankur Tyagi <[email protected]>
2025.09.3
---------
Fixed FIT image vulnerability
https://lore.barebox.org/barebox/[email protected]/Changelog:
https://github.com/barebox/barebox/compare/v2025.09.2...v2025.09.32025.09.2
---------
Changelog:
https://github.com/barebox/barebox/compare/v2025.09.1...v2025.09.22025.09.1
---------
This stable release is specifically targeted at whinlatter which is currently at v2025.09.0
https://lore.barebox.org/barebox/[email protected]/
Quoted 4 lines
Changelog:
https://github.com/barebox/barebox/compare/v2025.09.0...v2025.09.1Signed-off-by Ankur Tyagi <[email protected]>
I'm a bit torn on this one. Because on one hand, this looks like a
proper stable branch (and I appreciate the effort), but on the other
hand, there are changes that I find too risky:
For example:
* API Change: https://github.com/barebox/barebox/commit/5e9f709b2b56bd9689e174572ef2af59b9fed5d2
* New feature: https://github.com/barebox/barebox/compare/v2025.09.0...v2025.09.3#diff-2dd6d3e4c91931c982f091155a47dd8d03e26731f8d59b112e938508fe12255aR1
Thoughts?
In the meantime, I will keep it in my branch for awareness.
Quoted 17 lines
--- meta/recipes-bsp/barebox/barebox-common.inc | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)meta/recipes-bsp/barebox/barebox-common.inc +2
-2@@ -3,6 +3,6 @@ SECTION = "bootloaders" LIC_FILES_CHKSUM = "file://COPYING;md5=f5125d13e000b9ca1f0d3364286c4192" -PV = "2025.09.0" +PV = "2025.09.3" SRC_URI = "https://barebox.org/download/barebox-${PV}.tar.bz2" -SRC_URI[sha256sum] = "7df1aa47bb7bf1763a729137ac773e69a4052812af094475d739fc63a9295f0d" +SRC_URI[sha256sum] = "e87eb863cbe45e4f5af8930825c8f6e20c02b82451e9e1125ea1c73c1fb49a87"
Signature · 2 lines
Yoann Congal Smile ECS
Hello Yoann!
Quoted 22 lines · Yoann Congal via lists.openembedded.org
Quoted 21 lines · Ankur Tyagi via lists.openembedded.org
From: Ankur Tyagi <[email protected]>
2025.09.3
---------
Fixed FIT image vulnerability
https://lore.barebox.org/barebox/[email protected]/Changelog:
https://github.com/barebox/barebox/compare/v2025.09.2...v2025.09.32025.09.2
---------
Changelog:
https://github.com/barebox/barebox/compare/v2025.09.1...v2025.09.22025.09.1
---------
This stable release is specifically targeted at whinlatter which is currently at v2025.09.0
https://lore.barebox.org/barebox/[email protected]/
Quoted 7 lines
Quoted 4 lines
Changelog:
https://github.com/barebox/barebox/compare/v2025.09.0...v2025.09.1Signed-off-by Ankur Tyagi <[email protected]>I'm a bit torn on this one. Because on one hand, this looks like a proper stable branch (and I appreciate the effort),
I had prepared the v2025.09.y releases, so we (barebox) can hone a workflow to be able to support the release that makes it into Wrynose for a longer time than the usual 1 month it takes between the monthly releases of barebox.
Quoted 4 lines
but on the other
hand, there are changes that I find too risky:
For example:
* API Change: https://github.com/barebox/barebox/commit/5e9f709b2b56bd9689e174572ef2af59b9fed5d2
barebox is a bare metal bootloader. Its API surface is:
- the data formats it parses
- the data it passes along, e.g. fixed up device trees
- the boot/runtime services it provides to an OS (UEFI/PSCI/TEE-supplicant).
The change to bootm_set_overrides is not a change that affects user-visible API for any of that; it is completely internal to barebox. Revisiting it, I still think it was appropriate to backport it.
Quoted 1 line
Which features do you see there?
I read through the 50 commit titles on the first page and nearly all of them are fixes. The very few that aren't direct fixes are relatively benign preparatory commits followed by a fix and they were then backported together.
Quoted 1 line
Thoughts?
For the record, v2026.09.3 and v2026.03.1 were released together and that's the number of commits they have compared to v2025.09.0:
$ git log v2025.09.0..v2025.09.3 --oneline | wc -l 111
$ git log v2025.09.0..v2026.03.1 --oneline | wc -l 1023
So roughly 10% of commits to master since v2025.09.0 were hand selected to be backported as fixes. FWIW, I got curious and compared v6.19.12 to v7.0 and there it's 21%, so there's still some untapped potential (for squashing bugs!).
Quoted 1 line
In the meantime, I will keep it in my branch for awareness.
I appreciate that the stable branch was picked up, so many thanks for that!
Cheers,
Ahmad
Quoted 30 lines
Quoted 10 lines
--- meta/recipes-bsp/barebox/barebox-common.inc | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)meta/recipes-bsp/barebox/barebox-common.inc +0
-0meta/recipes-bsp/barebox/barebox-common.inc SECTION = "bootloaders" · line 8
Quoted 8 lines
meta/recipes-bsp/barebox/barebox-common.inc +2
-2@@ -3,6 +3,6 @@ SECTION = "bootloaders" LIC_FILES_CHKSUM = "file://COPYING;md5=f5125d13e000b9ca1f0d3364286c4192" -PV = "2025.09.0" +PV = "2025.09.3" SRC_URI = "https://barebox.org/download/barebox-${PV}.tar.bz2" -SRC_URI[sha256sum] = "7df1aa47bb7bf1763a729137ac773e69a4052812af094475d739fc63a9295f0d" +SRC_URI[sha256sum] = "e87eb863cbe45e4f5af8930825c8f6e20c02b82451e9e1125ea1c73c1fb49a87"-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#235329): https://lists.openembedded.org/g/openembedded-core/message/235329
Mute This Topic: https://lists.openembedded.org/mt/118717996/4830399
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
Signature · 4 lines
Pengutronix e.K. | | Steuerwalder Str. 21 | http://www.pengutronix.de/ | 31137 Hildesheim, Germany | Phone: +49-5121-206917-0 | Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |
Quoted 45 lines · Ahmad Fatoum
Hello Yoann!
Quoted 22 lines · Yoann Congal via lists.openembedded.org
Quoted 21 lines · Ankur Tyagi via lists.openembedded.org
From: Ankur Tyagi <[email protected]>
2025.09.3
---------
Fixed FIT image vulnerability
https://lore.barebox.org/barebox/[email protected]/Changelog:
https://github.com/barebox/barebox/compare/v2025.09.2...v2025.09.32025.09.2
---------
Changelog:
https://github.com/barebox/barebox/compare/v2025.09.1...v2025.09.22025.09.1
---------
This stable release is specifically targeted at whinlatter which is currently at v2025.09.0
https://lore.barebox.org/barebox/[email protected]/Quoted 7 lines
Quoted 4 lines
Changelog:
https://github.com/barebox/barebox/compare/v2025.09.0...v2025.09.1Signed-off-by Ankur Tyagi <[email protected]>I'm a bit torn on this one. Because on one hand, this looks like a proper stable branch (and I appreciate the effort),
I had prepared the v2025.09.y releases, so we (barebox) can hone a workflow to be able to support the release that makes it into Wrynose for a longer time than the usual 1 month it takes between the monthly releases of barebox.
Quoted 4 lines
but on the other
hand, there are changes that I find too risky:
For example:
* API Change: https://github.com/barebox/barebox/commit/5e9f709b2b56bd9689e174572ef2af59b9fed5d2barebox is a bare metal bootloader. Its API surface is:
Quoted 7 lines
- the data formats it parses
- the data it passes along, e.g. fixed up device trees
- the boot/runtime services it provides to an OS (UEFI/PSCI/TEE-supplicant).The change to
bootm_set_overridesis not a change that affects user-visible API for any of that; it is completely internal to barebox. Revisiting it, I still think it was appropriate to backport it.
I kind of agreeing with you I just remember my BSP dev days where this is the kind of function I would call for my custom boot flow (But in this case, fix looks easy enough)
Quoted 4 lines
Quoted 2 lines
Which features do you see there?
Sorry the URL is not as direct as I thought. This should link to the new "scripts/imx/pread.c" file. Also, looks like imx-image gained a new option :
"-a authenticate additionaly the DCD block..." But those look like dependencies of a bugfix.
Quoted 16 lines
I read through the 50 commit titles on the first page and nearly all of them are fixes. The very few that aren't direct fixes are relatively benign preparatory commits followed by a fix and they were then backported together.
Quoted 1 line
Thoughts?
For the record, v2026.09.3 and v2026.03.1 were released together and that's the number of commits they have compared to v2025.09.0:
$ git log v2025.09.0..v2025.09.3 --oneline | wc -l 111$ git log v2025.09.0..v2026.03.1 --oneline | wc -l 1023
Quoted 8 lines
So roughly 10% of commits to master since v2025.09.0 were hand selected to be backported as fixes. FWIW, I got curious and compared v6.19.12 to v7.0 and there it's 21%, so there's still some untapped potential (for squashing bugs!).
Quoted 1 line
In the meantime, I will keep it in my branch for awareness.
I appreciate that the stable branch was picked up, so many thanks for that!
Globally, I lean towards taking the upgrade. We will see if there are other point of views.
And by the way, thank you for creating and maintaining a stable branch :)
Quoted 34 lines
Cheers,
AhmadQuoted 30 lines
Quoted 10 lines
--- meta/recipes-bsp/barebox/barebox-common.inc | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)meta/recipes-bsp/barebox/barebox-common.inc +0
-0meta/recipes-bsp/barebox/barebox-common.inc SECTION = "bootloaders" · line 8
Quoted 8 lines
meta/recipes-bsp/barebox/barebox-common.inc +2
-2@@ -3,6 +3,6 @@ SECTION = "bootloaders" LIC_FILES_CHKSUM = "file://COPYING;md5=f5125d13e000b9ca1f0d3364286c4192" -PV = "2025.09.0" +PV = "2025.09.3" SRC_URI = "https://barebox.org/download/barebox-${PV}.tar.bz2" -SRC_URI[sha256sum] = "7df1aa47bb7bf1763a729137ac773e69a4052812af094475d739fc63a9295f0d" +SRC_URI[sha256sum] = "e87eb863cbe45e4f5af8930825c8f6e20c02b82451e9e1125ea1c73c1fb49a87"-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#235329): https://lists.openembedded.org/g/openembedded-core/message/235329
Mute This Topic: https://lists.openembedded.org/mt/118717996/4830399
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
Signature · 2 lines
Yoann Congal Smile ECS
Hello Yoann,
Cc += [email protected] who are listed as recipe maintainers.
Quoted 13 lines · Yoann Congal
Quoted 12 lines · Ahmad Fatoum
Quoted 1 line · Yoann Congal via lists.openembedded.org
On Wed Apr 8, 2026 at 2:09 AM CEST, Ankur Tyagi via lists.openembedded.org wrote:
I had prepared the v2025.09.y releases, so we (barebox) can hone a workflow to be able to support the release that makes it into Wrynose for a longer time than the usual 1 month it takes between the monthly releases of barebox.
Quoted 4 lines
but on the other
hand, there are changes that I find too risky:
For example:
* API Change: https://github.com/barebox/barebox/commit/5e9f709b2b56bd9689e174572ef2af59b9fed5d2
Quoted 5 lines
Quoted 5 lines
barebox is a bare metal bootloader. Its API surface is:
- the data formats it parses
- the data it passes along, e.g. fixed up device trees
- the boot/runtime services it provides to an OS (UEFI/PSCI/TEE-supplicant).
Rereading this, it's probably worth to spell out also:
- The protocols that barebox speaks
- The build integration (e.g. Kconfig options)
- The shell environment
Such changes are also not suitable for a point release.
Quoted 7 lines
Quoted 3 lines
The change to
bootm_set_overridesis not a change that affects user-visible API for any of that; it is completely internal to barebox. Revisiting it, I still think it was appropriate to backport it.I kind of agreeing with you I just remember my BSP dev days where this is the kind of function I would call for my custom boot flow (But in this case, fix looks easy enough)
We had rather heated discussions not so long ago about the extent we should accommodate out-of-tree board code. :-)
I believe a build-time failure is as good as it gets if an API is just not salvageable like the one you pointed out; It didn't take recursive calls into account, which occurred in real world scenarios.
For compatibility breakage that is not immediately obvious at compile-time, we are maintaining migration notes:
https://www.barebox.org/doc/latest/migration-guides/index.html
But nothing that deserves an entry there should be something that would be ok to include in a point release - save grave security bugs that can't be fixed any other way.
Most recently, both barebox v2026.03.1 and U-Boot v2026.04.0 will refuse to boot a FIT with a verified signed configuration when some of the images were left unsigned:
https://lore.kernel.org/barebox/[email protected]/T/#u
Migration guide entry is the best we can do here.
Quoted 6 lines
Quoted 3 lines
Quoted 1 line
Which features do you see there?
Sorry the URL is not as direct as I thought. This should link to the new "
scripts/imx/pread.c" file.
We don't have mingw build in CI, so this fixes a build regression even if doesn't sound this way.
Quoted 2 lines
Also, looks like imx-image gained a new option :
"-a authenticate additionaly the DCD block..."
So this was the second commit in a series of two patches. The first is a fix and the second fixes an incompatibility with the external uuu utility. I concede that you can decide either way whether this patch is backport material or not.
The impact on a user not affected by this issue is nonetheless minimal.
Quoted 1 line
But those look like dependencies of a bugfix.
Fixes are occasionally split for master (next release) and next (release after next), when the change is too involved.
Recent example is that TF-A v2.14 broke backwards compatibility for its consumers:
https://lore.barebox.org/barebox/[email protected]/
https://lore.barebox.org/barebox/[email protected]/
No one size fits all.
Quoted 7 lines
Quoted 1 line
I appreciate that the stable branch was picked up, so many thanks for that!
Globally, I lean towards taking the upgrade. We will see if there are other point of views.
And by the way, thank you for creating and maintaining a stable branch :)
This has been a pain point I heard at conferences in the past and I am happy to get feedback on how we can improve the process.
The LTS and the migration guides are inspired by Yocto and we see it as an important part of making sure that barebox is a reliable bootloader in secure boot contexts.
I think there's generally a lot to improve[1] the status quo of embedded bootloader security and I see making bootloader updates easier as important as shiny new features like barebox security policies (coming hopefully soon to an OE-core near you!), even if the former is admittedly not as fun..
Thanks,
Ahmad
Quoted 37 lines
Quoted 35 lines
Cheers,
AhmadQuoted 30 lines
Quoted 10 lines
--- meta/recipes-bsp/barebox/barebox-common.inc | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)meta/recipes-bsp/barebox/barebox-common.inc +0
-0meta/recipes-bsp/barebox/barebox-common.inc SECTION = "bootloaders" · line 8
Quoted 8 lines
meta/recipes-bsp/barebox/barebox-common.inc +2
-2@@ -3,6 +3,6 @@ SECTION = "bootloaders" LIC_FILES_CHKSUM = "file://COPYING;md5=f5125d13e000b9ca1f0d3364286c4192" -PV = "2025.09.0" +PV = "2025.09.3" SRC_URI = "https://barebox.org/download/barebox-${PV}.tar.bz2" -SRC_URI[sha256sum] = "7df1aa47bb7bf1763a729137ac773e69a4052812af094475d739fc63a9295f0d" +SRC_URI[sha256sum] = "e87eb863cbe45e4f5af8930825c8f6e20c02b82451e9e1125ea1c73c1fb49a87"-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#235329): https://lists.openembedded.org/g/openembedded-core/message/235329
Mute This Topic: https://lists.openembedded.org/mt/118717996/4830399
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
Signature · 4 lines
Pengutronix e.K. | | Steuerwalder Str. 21 | http://www.pengutronix.de/ | 31137 Hildesheim, Germany | Phone: +49-5121-206917-0 | Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |