Hello Francois, Jan, Christian, and all
Sorry for the late reply, I was waiting for the administrator of the Boot Architecture mailing list to accept my subscription request, but it seems it will take a bit more time. I will send this reply and hope it will not be blocked. I have also added the u-boot mailing list to Cc, as Tom suggested (although I'm not a member), the CIP mailing list, Jan Kiszka (one of the main developers of Efibootguard) and Christian (an expert in software updates).
Background: during the last Linaro connect in Bangkok I was told that Linaro Edge (LEDGE) were working on a secure software update mechanism based on UEFI capsules that would flash firmware updates from a UEFI application, instead of using a Linux agent such as SWUpdate. Then, I had an online meeting with Francois, director of LEDGE. I explained to Francois that in CIP we are using the Linux agent approach right now, and we are also considering the use of a UEFI application (Efibootguard) to arm a watchdog and deal with the state-machine variables (installed, testing, ok, failed..) needed for A/B software updates. Efibootguard sounds like an excellent place to collaborate with Linaro (particularly on the watchdog drivers front) because it does not strictly depend on where the firmware is flashed (UEFI capsule or Linux agent).
On Fri, Apr 19, 2019 at 12:48:51PM +0200, Francois Ozog wrote:
Hi Daniel,
We will be conducting a UEFI gap analysis to support EFIBootGuard in U-Boot.
As we are working on UEFI SecureBoot implementation in U-Boot, how do you expect the boot process to be secured? Would U-Boot UEFI SecureBoot verify EFIBootGuard signature and in turn EFIBootGuard will check either grub or Linux signature?
Please elaborate on your vision of a secured boot process.
Efibootguard is composed of two parts. - A UEFI application that can arm a watchdog and decide what environment (kernel, boot args, etc.) to use next depending on a set of variables (update status, highest revision, etc.) stored in FAT16 partitions. - A Linux application that can read and set those variables from Linux (similar to u-boot's fw_setenv). This functionality is also available in the form of a library.
As far as I know, there is no concept of "Secure Booting" in Efibootguard at the moment. Adding signature checks before booting into the selected kernel would be a possible solution.
Thanks, Daniel
[prolonging the CCs with the efibootguard mailing list]
On 24.04.19 03:23, daniel.sangorrin@toshiba.co.jp wrote:
Hello Francois, Jan, Christian, and all EFI Boot Guard is now shipped in quite a few devices, to my knowledge not only at Sorry for the late reply, I was waiting for the administrator of the Boot Architecture mailing list to accept my subscription request, but it seems it will take a bit more time. I will send this reply and hope it will not be blocked. I have also added the u-boot mailing list to Cc, as Tom suggested (although I'm not a member), the CIP mailing list, Jan Kiszka (one of the main developers of Efibootguard) and Christian (an expert in software updates).
Background: during the last Linaro connect in Bangkok I was told that Linaro Edge (LEDGE) were working on a secure software update mechanism based on UEFI capsules that would flash firmware updates from a UEFI application, instead of using a Linux agent such as SWUpdate.
How would capsules help with writing to arbitrary storage, updating only files on filesystem, reducing the update size (binary diffs), or talking to the cloud?
Then, I had an online meeting with Francois, director of LEDGE. I explained to Francois that in CIP we are using the Linux agent approach right now, and we are also considering the use of a UEFI application (Efibootguard) to arm a watchdog and deal with the state-machine variables (installed, testing, ok, failed..) needed for A/B software updates. Efibootguard sounds like an excellent place to collaborate with Linaro (particularly on the watchdog drivers front) because it does not strictly depend on where the firmware is flashed (UEFI capsule or Linux agent).
On Fri, Apr 19, 2019 at 12:48:51PM +0200, Francois Ozog wrote:
Hi Daniel,
We will be conducting a UEFI gap analysis to support EFIBootGuard in U-Boot.
As we are working on UEFI SecureBoot implementation in U-Boot, how do you expect the boot process to be secured? Would U-Boot UEFI SecureBoot verify EFIBootGuard signature and in turn EFIBootGuard will check either grub or Linux signature?
Please elaborate on your vision of a secured boot process.
Efibootguard is composed of two parts.
- A UEFI application that can arm a watchdog and decide what environment (kernel, boot args, etc.) to use next depending on a set of variables (update status, highest revision, etc.) stored in FAT16 partitions.
- A Linux application that can read and set those variables from Linux (similar to u-boot's fw_setenv). This functionality is also available in the form of a library.
As far as I know, there is no concept of "Secure Booting" in Efibootguard at the moment. Adding signature checks before booting into the selected kernel would be a possible solution.
Secure boot is a pending feature on our to-do list. It's a bit more complicated than that, like secure boot is "a bit" more complicated than you think once you actually try to implement it. Once we do that, it's really about adding signature checks or relying on UEFI validating the payloads we boot for us PLUS ensuring the our config sections can either be validated (despite being volatile) or split the security-wise critical parts (specifically EFI payload parameters) from the less critical ones (update states) and remove the latter from the validation.
BTW, what we do in EFI Board Guard could also be done in any other UEFI bootloader, may it be grub (if you like to use that complex and fragile beast in production), systemd-boot or even TianoCore. But for now, it was easier - and more robust - to add our requirements in form of this tiny bootloader to the ecosystem. EFI Boot Guard is now shipped in quite a few devices, to my best knowledge not only at Siemens.
Jan
Hi Jan, Francois:
Grant: thanks!
From: Jan Kiszka jan.kiszka@siemens.com On 24.04.19 03:23, daniel.sangorrin@toshiba.co.jp wrote:
Hello Francois, Jan, Christian, and all EFI Boot Guard is now shipped in quite a few devices, to my knowledge not only at Sorry for the late reply, I was waiting for the administrator of the Boot Architecture mailing list to accept my
subscription request, but it seems it will take a bit more time. I will send this reply and hope it will not be blocked. I have also added the u-boot mailing list to Cc, as Tom suggested (although I'm not a member), the CIP mailing list, Jan Kiszka (one of the main developers of Efibootguard) and Christian (an expert in software updates).
Background: during the last Linaro connect in Bangkok I was told that Linaro Edge (LEDGE) were working on
a secure software update mechanism based on UEFI capsules that would flash firmware updates from a UEFI application, instead of using a Linux agent such as SWUpdate.
How would capsules help with writing to arbitrary storage, updating only files on filesystem, reducing the update size (binary diffs), or talking to the cloud?
- arbitrary storage: I guess they can only write to what is supported by the machine's UEFI implementation. - updating only files on filesystem: I assume this is out of scope in their architecture (Francois: do you want to support file-based updates or only block-based ones?) - reducing the update size (binary diffs), or talking to the cloud: they will do that from non-secure Linux. It would be dangerous to use a fragile network stack from an UEFI application or the secure world. In that sense, they also need a Linux agent.
I believe that LEDGE is looking for a software update method that works the same on any machine (that supports UEFI). To do that they want to use the UEFI interfaces/services. They also want the ability to update the TrustZone secure world (you can't do that unless you have enough privileges).
As far as I know, there is no concept of "Secure Booting" in Efibootguard at the moment. Adding signature
checks before booting into the selected kernel would be a possible solution.
Secure boot is a pending feature on our to-do list. It's a bit more complicated than that, like secure boot is "a bit" more complicated than you think once you actually try to implement it. Once we do that, it's really about adding signature checks or relying on UEFI validating the payloads we boot for us PLUS ensuring the our config sections can either be validated (despite being volatile) or split the security-wise critical parts (specifically EFI payload parameters) from the less critical ones (update states) and remove the latter from the validation.
I suppose that those "bits" are hard to predict until you start the implementation. From an architectural point of view, I guess that the "revision" variables will need to be secured to avoid downgrades (e.g. an attacker causing a rollback to a previous revision of the OS image).
BTW, what we do in EFI Board Guard could also be done in any other UEFI bootloader, may it be grub (if you like to use that complex and fragile beast in production), systemd-boot or even TianoCore. But for now, it was easier - and more robust - to add our requirements in form of this tiny bootloader to the ecosystem. EFI Boot Guard is now shipped in quite a few devices, to my best knowledge not only at Siemens.
Jan: do you have a schedule or a list of tasks that need to be done? Francois: what direction should we take from here?
Thanks, Daniel
Hi,
Background: during the last Linaro connect in Bangkok I was told that Linaro Edge (LEDGE) were working on a secure software update mechanism based on UEFI capsules that would flash firmware updates from a UEFI application, instead of using a Linux agent such as SWUpdate.
How would capsules help with writing to arbitrary storage, updating only files on filesystem, reducing the update size (binary diffs), or talking to the cloud?
arbitrary storage: I guess they can only write to what is supported by the machine's UEFI implementation.
updating only files on filesystem: I assume this is out of scope in their architecture (Francois: do you want to support file-based updates or only block-based ones?)
- reducing the update size (binary diffs), or talking to the cloud: they will do that from non-secure Linux. It would be dangerous to use a fragile network stack from an UEFI application or the secure world. In that sense, they also need a Linux agent.
If I understand the idea correctly, there's then a Linux companion agent that does, e.g., the cloud communication and sets up the update payload to be consumed and applied by the UEFI firmware update application. The actual update action will take place on reboot, e.g., flashing the inactive partition in an A/B setup and thereafter booting into the newly flashed, then active partition.
If so, this would prolong the time the system is unavailable due to disk flashing I/O in the UEFI firmware update application compared to flashing the inactive partition from within Linux and have the bootloader just switch the to be booted partitions.
Generally, but in particular regarding binary diffs, a binary diff application mechanism would have to be implemented and maintained in the UEFI realm as opposed to "just" using the respective shared libraries (e.g. librsync, casync, ...) in the Linux realm. Granted, one could come up with a unified format to which the output of, e.g, librsync's rdiff or a casync run is converted but nonetheless, this creates some porting/ implementation and maintenance effort.
Kind regards, Christian
On Fri, 26 Apr 2019 at 10:30, Christian Storm christian.storm@siemens.com wrote:
Hi,
Background: during the last Linaro connect in Bangkok I was told that Linaro Edge (LEDGE) were working on a secure software update mechanism based on UEFI capsules that would flash firmware updates from a UEFI application, instead of using a Linux agent such as SWUpdate.
How would capsules help with writing to arbitrary storage, updating only files on filesystem, reducing the update size (binary diffs), or talking to the cloud?
See below
- arbitrary storage: I guess they can only write to what is supported by the machine's UEFI implementation.
Yes, see below
- updating only files on filesystem: I assume this is out of scope in their architecture (Francois: do you want to support file-based updates or only block-based ones?)
Both, see below
- reducing the update size (binary diffs), or talking to the cloud: they will do that from non-secure Linux. It would be dangerous to use a fragile network stack from an UEFI application or the secure world. In that sense, they also need a Linux agent.
If I understand the idea correctly, there's then a Linux companion agent that does, e.g., the cloud communication and sets up the update payload to be consumed and applied by the UEFI firmware update application. The actual update action will take place on reboot, e.g., flashing the inactive partition in an A/B setup and thereafter booting into the newly flashed, then active partition.
If so, this would prolong the time the system is unavailable due to disk flashing I/O in the UEFI firmware update application compared to flashing the inactive partition from within Linux and have the bootloader just switch the to be booted partitions.
Generally, but in particular regarding binary diffs, a binary diff application mechanism would have to be implemented and maintained in the UEFI realm as opposed to "just" using the respective shared libraries (e.g. librsync, casync, ...) in the Linux realm. Granted, one could come up with a unified format to which the output of, e.g, librsync's rdiff or a casync run is converted but nonetheless, this creates some porting/ implementation and maintenance effort.
Here are the guiding principles of our efforts : 0) we want a cross architecture (x86/Arm/...), cross vendor and cross processor model update solution 1) untrusted world cannot update trusted world 2) what the UEFI implementation does with the capsule is platform specific 3) the update capsule payload is opaque
1) is a "philosophy" decision. When you have a root of trust down to the operating system. So in theory everything should be fine. But the attack surface is such that we can't rule out hacking (and history prove this is unfortunately a safe assumption). In addition, there may be liability aspects related to the who does the update: the hardware platform administrator may not be the legal entity than the operating system administrator. For instance: - on Packet.net could, a customer can flash up to the BL33 untrusted firmware but that is all. - A surveillance camera can be operated by city personnel but only law enforcement agency can see raw video (un-blurred faces and licence plates). This can be implemented by a derivative of OPTEE SecureMedia Path but if you allow untrusted world to update the trusted one, city personnel can overcome the legal restriction. With 1) this means that even U-Boot code shall not be able/allowed to update S-EL3 firmware (be it Trusted Firmware A or something else), Secure EL1/0 software (OPTEE and its applications or something else). If possible, allowing the operating system administrator to selectively (BL33 is OK but not S-EL3) update firmware is at least platform dependent. Hence defeats 0)
With 2) we do not impose reboot to update. Some platform may impose reboot or some designers will prefer reboot. We say that there is a chain of responsibility for updates. So it is perfectly OK to have a Linux software agent receive an update by any mean (network, USB, serial...). The agent will pack this (or those) into a capsule and push it to UEFI implementation. The UEFI implementation (U-Boot or Tianocore) will then do whatever it pleases for the update providing it complies with 1) So the UEFI implementation can live update up to BL33 firmware. Should the update be targeted to secure world, then the UEFI implementation can pass it to S-EL3 for update (platform specific) which means the update can also be live. It is a designer decision.
With 3) we have flexibility but sometimes too much flexibility is a problem. Alexander Graf suggested we can use a FIT object to pass diverse objects. It is "half" opaque but I really like the idea. The contents of individual FIT components can be blocks to be placed at a fix location in a NOR flash or a file, no importance.
What do everyone think about those design principles ?
Kind regards, Christian
-- Dr. Christian Storm Siemens AG, Corporate Technology, CT RDA IOT SES-DE Otto-Hahn-Ring 6, 81739 München, Germany
On 26.04.19 11:07, Francois Ozog wrote:
On Fri, 26 Apr 2019 at 10:30, Christian Storm christian.storm@siemens.com wrote:
Hi,
Background: during the last Linaro connect in Bangkok I was told that Linaro Edge (LEDGE) were working on a secure software update mechanism based on UEFI capsules that would flash firmware updates from a UEFI application, instead of using a Linux agent such as SWUpdate.
How would capsules help with writing to arbitrary storage, updating only files on filesystem, reducing the update size (binary diffs), or talking to the cloud?
See below
- arbitrary storage: I guess they can only write to what is supported by the machine's UEFI implementation.
Yes, see below
- updating only files on filesystem: I assume this is out of scope in their architecture (Francois: do you want to support file-based updates or only block-based ones?)
Both, see below
- reducing the update size (binary diffs), or talking to the cloud: they will do that from non-secure Linux. It would be dangerous to use a fragile network stack from an UEFI application or the secure world. In that sense, they also need a Linux agent.
If I understand the idea correctly, there's then a Linux companion agent that does, e.g., the cloud communication and sets up the update payload to be consumed and applied by the UEFI firmware update application. The actual update action will take place on reboot, e.g., flashing the inactive partition in an A/B setup and thereafter booting into the newly flashed, then active partition.
If so, this would prolong the time the system is unavailable due to disk flashing I/O in the UEFI firmware update application compared to flashing the inactive partition from within Linux and have the bootloader just switch the to be booted partitions.
Generally, but in particular regarding binary diffs, a binary diff application mechanism would have to be implemented and maintained in the UEFI realm as opposed to "just" using the respective shared libraries (e.g. librsync, casync, ...) in the Linux realm. Granted, one could come up with a unified format to which the output of, e.g, librsync's rdiff or a casync run is converted but nonetheless, this creates some porting/ implementation and maintenance effort.
Here are the guiding principles of our efforts : 0) we want a cross architecture (x86/Arm/...), cross vendor and cross processor model update solution
- untrusted world cannot update trusted world
Conceptually, it can. It's a matter of validating the update by the trusted world before using it. A trusted instance can allow an untrusted one to write version 2, validate that before switching to it, and stick with version 1 if that fails.
what the UEFI implementation does with the capsule is platform specific
the update capsule payload is opaque
is a "philosophy" decision. When you have a root of trust down to
the operating system. So in theory everything should be fine. But the attack surface is such that we can't rule out hacking (and history prove this is unfortunately a safe assumption). In addition, there may be liability aspects related to the who does the update: the hardware platform administrator may not be the legal entity than the operating system administrator. For instance:
- on Packet.net could, a customer can flash up to the BL33 untrusted
firmware but that is all.
- A surveillance camera can be operated by city personnel but only law
enforcement agency can see raw video (un-blurred faces and licence plates). This can be implemented by a derivative of OPTEE SecureMedia Path but if you allow untrusted world to update the trusted one, city personnel can overcome the legal restriction. With 1) this means that even U-Boot code shall not be able/allowed to update S-EL3 firmware (be it Trusted Firmware A or something else), Secure EL1/0 software (OPTEE and its applications or something else). If possible, allowing the operating system administrator to selectively (BL33 is OK but not S-EL3) update firmware is at least platform dependent. Hence defeats 0)
With 2) we do not impose reboot to update. Some platform may impose reboot or some designers will prefer reboot. We say that there is a chain of responsibility for updates. So it is perfectly OK to have a Linux software agent receive an update by any mean (network, USB, serial...). The agent will pack this (or those) into a capsule and push it to UEFI implementation. The UEFI implementation (U-Boot or Tianocore) will then do whatever it pleases for the update providing it complies with 1) So the UEFI implementation can live update up to BL33 firmware. Should the update be targeted to secure world, then the UEFI implementation can pass it to S-EL3 for update (platform specific) which means the update can also be live. It is a designer decision.
With 3) we have flexibility but sometimes too much flexibility is a problem. Alexander Graf suggested we can use a FIT object to pass diverse objects. It is "half" opaque but I really like the idea. The contents of individual FIT components can be blocks to be placed at a fix location in a NOR flash or a file, no importance.
What do everyone think about those design principles ?
UEFI and capsules can be fine for those platform that support it (and it's still a rare feature) and for stuff like boot and peripheral firmware. I don't think it's a wise, future-proof idea to use it for more.
UEFI is not a very healthy ecosystem yet, and I'm personally skeptical it will evolve towards that (looking at that as both a user as well as an OEM). It's not even present in quite a few of our use cases. In some it will never be - think of safety-critical system: not affordable with such a complex approach like UEFI. It should not expand further into OS domains. Updating complete filesystems and their content is beyond its duties. We have all the required, tested, matured means for the OS in the OS. For firmware, it can be an option - if that firmware is UEFI compliant, with all needed options.
So let's use UEFI and capsules as a one possible building block in the design, but not as the cornerstone. Just like we do mandate that all updates must be served and managed by our IoT cloud ;).
Jan
On 26/04/2019 10:49, Jan Kiszka wrote:
On 26.04.19 11:07, Francois Ozog wrote:
[...]
Here are the guiding principles of our efforts : 0) we want a cross architecture (x86/Arm/...), cross vendor and cross processor model update solution
- untrusted world cannot update trusted world
Conceptually, it can. It's a matter of validating the update by the trusted world before using it. A trusted instance can allow an untrusted one to write version 2, validate that before switching to it, and stick with version 1 if that fails.
- what the UEFI implementation does with the capsule is platform
specific 3) the update capsule payload is opaque
- is a "philosophy" decision. When you have a root of trust down to
the operating system. So in theory everything should be fine. But the attack surface is such that we can't rule out hacking (and history prove this is unfortunately a safe assumption). In addition, there may be liability aspects related to the who does the update: the hardware platform administrator may not be the legal entity than the operating system administrator. For instance: - on Packet.net could, a customer can flash up to the BL33 untrusted firmware but that is all.
- A surveillance camera can be operated by city personnel but only law
enforcement agency can see raw video (un-blurred faces and licence plates). This can be implemented by a derivative of OPTEE SecureMedia Path but if you allow untrusted world to update the trusted one, city personnel can overcome the legal restriction. With 1) this means that even U-Boot code shall not be able/allowed to update S-EL3 firmware (be it Trusted Firmware A or something else), Secure EL1/0 software (OPTEE and its applications or something else). If possible, allowing the operating system administrator to selectively (BL33 is OK but not S-EL3) update firmware is at least platform dependent. Hence defeats 0)
With 2) we do not impose reboot to update. Some platform may impose reboot or some designers will prefer reboot. We say that there is a chain of responsibility for updates. So it is perfectly OK to have a Linux software agent receive an update by any mean (network, USB, serial...). The agent will pack this (or those) into a capsule and push it to UEFI implementation. The UEFI implementation (U-Boot or Tianocore) will then do whatever it pleases for the update providing it complies with 1) So the UEFI implementation can live update up to BL33 firmware. Should the update be targeted to secure world, then the UEFI implementation can pass it to S-EL3 for update (platform specific) which means the update can also be live. It is a designer decision.
With 3) we have flexibility but sometimes too much flexibility is a problem. Alexander Graf suggested we can use a FIT object to pass diverse objects. It is "half" opaque but I really like the idea. The contents of individual FIT components can be blocks to be placed at a fix location in a NOR flash or a file, no importance.
What do everyone think about those design principles ?
UEFI and capsules can be fine for those platform that support it (and it's still a rare feature) and for stuff like boot and peripheral firmware. I don't think it's a wise, future-proof idea to use it for more.
UEFI is not a very healthy ecosystem yet, and I'm personally skeptical it will evolve towards that (looking at that as both a user as well as an OEM). It's not even present in quite a few of our use cases. In some it will never be - think of safety-critical system: not affordable with such a complex approach like UEFI.
Can I challenge that view a bit? On the Tianocore side I agree that the ecosystem isn't very healthy. That project in particular struggles with what to do with board support, having decided early on that board support generally doesn't need to be in the main repository.
However, U-Boot support for the UEFI ABI is improving in leaps and bounds. SUSE and Fedora both depend on U-Boot UEFI for booting on all the Arm SBCs that they support, and enabling UEFI in U-Boot is just a config option or two. There is a fair bit of encouragement from within the project to enable UEFI by default.
UEFI gets a bad rap at being complicated, but I think the U-Boot work has shown that implementing the core UEFI ABI doesn't require much code and isn't the complicated mess they everyone fears it to be.
I don't see the conflict with safety critical for boot services. I can however see the arguement against UEFI runtime services as a poor implementation could result in unbounded execution times. There's been recent movement on the UEFI spec to make runtime services optional, and in U-Boot they are mostly empty stubs.
It should not expand further into OS domains. Updating complete filesystems and their content is beyond its duties. We have all the required, tested, matured means for the OS in the OS. For firmware, it can be an option - if that firmware is UEFI compliant, with all needed options.
Here I agree 100%. Capsule update is a useful ABI for the OS to pass firmware update blobs without needing to know specifics about the platform, but everything from the kernel on up is out of scope.
So let's use UEFI and capsules as a one possible building block in the design, but not as the cornerstone. Just like we do mandate that all updates must be served and managed by our IoT cloud ;).
What do the firmware stacks in your projects look like? What percentage are based on U-Boot? Littlekernel? Others?
UEFI has the advantage that it has a defined ABI, and only a small amount of it needs to be implemented to execute UEFI OS loaders (ie. the Linux UEFI stub). For the reference platform, I think it makes sense to make the design decision that firmware provides UEFI, and the update infrastructure is based on that. However, if UEFI doesn't work for a particular stack/project then it can be swapped out. There will need to be work done on the boot sequence, but all the rest of the OS stack is still relevant.
g.
- team-ledge@linaro.org as it is now a public discussion
On Fri, 26 Apr 2019 at 12:21, Grant Likely Grant.Likely@arm.com wrote:
On 26/04/2019 10:49, Jan Kiszka wrote:
On 26.04.19 11:07, Francois Ozog wrote:
[...]
Here are the guiding principles of our efforts : 0) we want a cross architecture (x86/Arm/...), cross vendor and cross processor model update solution
- untrusted world cannot update trusted world
Conceptually, it can. It's a matter of validating the update by the trusted world before using it. A trusted instance can allow an untrusted one to write version 2, validate that before switching to it, and stick with version 1 if that fails.
- what the UEFI implementation does with the capsule is platform
specific 3) the update capsule payload is opaque
- is a "philosophy" decision. When you have a root of trust down to
the operating system. So in theory everything should be fine. But the attack surface is such that we can't rule out hacking (and history prove this is unfortunately a safe assumption). In addition, there may be liability aspects related to the who does the update: the hardware platform administrator may not be the legal entity than the operating system administrator. For instance:
- on Packet.net could, a customer can flash up to the BL33 untrusted
firmware but that is all.
- A surveillance camera can be operated by city personnel but only law
enforcement agency can see raw video (un-blurred faces and licence plates). This can be implemented by a derivative of OPTEE SecureMedia Path but if you allow untrusted world to update the trusted one, city personnel can overcome the legal restriction. With 1) this means that even U-Boot code shall not be able/allowed to update S-EL3 firmware (be it Trusted Firmware A or something else), Secure EL1/0 software (OPTEE and its applications or something else). If possible, allowing the operating system administrator to selectively (BL33 is OK but not S-EL3) update firmware is at least platform dependent. Hence defeats 0)
With 2) we do not impose reboot to update. Some platform may impose reboot or some designers will prefer reboot. We say that there is a chain of responsibility for updates. So it is perfectly OK to have a Linux software agent receive an update by any mean (network, USB, serial...). The agent will pack this (or those) into a capsule and push it to UEFI implementation. The UEFI implementation (U-Boot or Tianocore) will then do whatever it pleases for the update providing it complies with 1) So the UEFI implementation can live update up to BL33 firmware. Should the update be targeted to secure world, then the UEFI implementation can pass it to S-EL3 for update (platform specific) which means the update can also be live. It is a designer decision.
With 3) we have flexibility but sometimes too much flexibility is a problem. Alexander Graf suggested we can use a FIT object to pass diverse objects. It is "half" opaque but I really like the idea. The contents of individual FIT components can be blocks to be placed at a fix location in a NOR flash or a file, no importance.
What do everyone think about those design principles ?
UEFI and capsules can be fine for those platform that support it (and it's still a rare feature) and for stuff like boot and peripheral firmware. I don't think it's a wise, future-proof idea to use it for
more.
UEFI is not a very healthy ecosystem yet, and I'm personally skeptical it will evolve towards that (looking at that as both a user as well as an OEM). It's not even present in quite a few of our use cases. In some it will never be - think of safety-critical system: not affordable with such a complex approach like UEFI.
Can I challenge that view a bit? On the Tianocore side I agree that the ecosystem isn't very healthy. That project in particular struggles with what to do with board support, having decided early on that board support generally doesn't need to be in the main repository.
However, U-Boot support for the UEFI ABI is improving in leaps and bounds. SUSE and Fedora both depend on U-Boot UEFI for booting on all the Arm SBCs that they support, and enabling UEFI in U-Boot is just a config option or two. There is a fair bit of encouragement from within the project to enable UEFI by default.
UEFI gets a bad rap at being complicated, but I think the U-Boot work has shown that implementing the core UEFI ABI doesn't require much code and isn't the complicated mess they everyone fears it to be.
I don't see the conflict with safety critical for boot services. I can however see the arguement against UEFI runtime services as a poor implementation could result in unbounded execution times. There's been recent movement on the UEFI spec to make runtime services optional, and in U-Boot they are mostly empty stubs.
It should not expand further into OS domains. Updating complete filesystems and their content is beyond its duties. We have all the required, tested, matured means for the OS in the OS. For firmware, it can be an option - if that firmware is UEFI compliant, with all needed options.
Here I agree 100%. Capsule update is a useful ABI for the OS to pass firmware update blobs without needing to know specifics about the platform, but everything from the kernel on up is out of scope.
So let's use UEFI and capsules as a one possible building block in the design, but not as the cornerstone. Just like we do mandate that all updates must be served and managed by our IoT cloud ;).
What do the firmware stacks in your projects look like? What percentage are based on U-Boot? Littlekernel? Others?
UEFI has the advantage that it has a defined ABI, and only a small amount of it needs to be implemented to execute UEFI OS loaders (ie. the Linux UEFI stub). For the reference platform, I think it makes sense to make the design decision that firmware provides UEFI, and the update infrastructure is based on that. However, if UEFI doesn't work for a particular stack/project then it can be swapped out. There will need to be work done on the boot sequence, but all the rest of the OS stack is still relevant.
g.
On 26.04.19 12:21, Grant Likely wrote:
On 26/04/2019 10:49, Jan Kiszka wrote:
On 26.04.19 11:07, Francois Ozog wrote:
[...]
Here are the guiding principles of our efforts : 0) we want a cross architecture (x86/Arm/...), cross vendor and cross processor model update solution
- untrusted world cannot update trusted world
Conceptually, it can. It's a matter of validating the update by the trusted world before using it. A trusted instance can allow an untrusted one to write version 2, validate that before switching to it, and stick with version 1 if that fails.
- what the UEFI implementation does with the capsule is platform
specific 3) the update capsule payload is opaque
- is a "philosophy" decision. When you have a root of trust down to
the operating system. So in theory everything should be fine. But the attack surface is such that we can't rule out hacking (and history prove this is unfortunately a safe assumption). In addition, there may be liability aspects related to the who does the update: the hardware platform administrator may not be the legal entity than the operating system administrator. For instance: - on Packet.net could, a customer can flash up to the BL33 untrusted firmware but that is all.
- A surveillance camera can be operated by city personnel but only law
enforcement agency can see raw video (un-blurred faces and licence plates). This can be implemented by a derivative of OPTEE SecureMedia Path but if you allow untrusted world to update the trusted one, city personnel can overcome the legal restriction. With 1) this means that even U-Boot code shall not be able/allowed to update S-EL3 firmware (be it Trusted Firmware A or something else), Secure EL1/0 software (OPTEE and its applications or something else). If possible, allowing the operating system administrator to selectively (BL33 is OK but not S-EL3) update firmware is at least platform dependent. Hence defeats 0)
With 2) we do not impose reboot to update. Some platform may impose reboot or some designers will prefer reboot. We say that there is a chain of responsibility for updates. So it is perfectly OK to have a Linux software agent receive an update by any mean (network, USB, serial...). The agent will pack this (or those) into a capsule and push it to UEFI implementation. The UEFI implementation (U-Boot or Tianocore) will then do whatever it pleases for the update providing it complies with 1) So the UEFI implementation can live update up to BL33 firmware. Should the update be targeted to secure world, then the UEFI implementation can pass it to S-EL3 for update (platform specific) which means the update can also be live. It is a designer decision.
With 3) we have flexibility but sometimes too much flexibility is a problem. Alexander Graf suggested we can use a FIT object to pass diverse objects. It is "half" opaque but I really like the idea. The contents of individual FIT components can be blocks to be placed at a fix location in a NOR flash or a file, no importance.
What do everyone think about those design principles ?
UEFI and capsules can be fine for those platform that support it (and it's still a rare feature) and for stuff like boot and peripheral firmware. I don't think it's a wise, future-proof idea to use it for more.
UEFI is not a very healthy ecosystem yet, and I'm personally skeptical it will evolve towards that (looking at that as both a user as well as an OEM). It's not even present in quite a few of our use cases. In some it will never be - think of safety-critical system: not affordable with such a complex approach like UEFI.
Can I challenge that view a bit? On the Tianocore side I agree that the ecosystem isn't very healthy. That project in particular struggles with what to do with board support, having decided early on that board support generally doesn't need to be in the main repository.
However, U-Boot support for the UEFI ABI is improving in leaps and bounds. SUSE and Fedora both depend on U-Boot UEFI for booting on all the Arm SBCs that they support, and enabling UEFI in U-Boot is just a config option or two. There is a fair bit of encouragement from within the project to enable UEFI by default.
I don't disagree that this aspect is a step forward (though pulling in things redundant code via grub & Co is not really progress). But a Unix community would have probably designed a technically more elegant solution on a green field than the clumsy, legacy-based UEFI interfaces.
UEFI gets a bad rap at being complicated, but I think the U-Boot work has shown that implementing the core UEFI ABI doesn't require much code and isn't the complicated mess they everyone fears it to be.
Depends on how much you start to rely on UEFI features.
I don't see the conflict with safety critical for boot services. I can
Everything is certifiable - with infinite time and money. So you will likely reduce the boot process to the very essential of your concrete system, removing then unneeded abstractions and stages. Or you will find ways to eliminate that from your argumentation, which would definitely include removing runtime services.
however see the arguement against UEFI runtime services as a poor implementation could result in unbounded execution times. There's been recent movement on the UEFI spec to make runtime services optional, and in U-Boot they are mostly empty stubs.
Of course, it also depends on how much you need from UEFI to boot or to use certain services like capsule updates (and which would be gone when you remove runtime services).
It should not expand further into OS domains. Updating complete filesystems and their content is beyond its duties. We have all the required, tested, matured means for the OS in the OS. For firmware, it can be an option - if that firmware is UEFI compliant, with all needed options.
Here I agree 100%. Capsule update is a useful ABI for the OS to pass firmware update blobs without needing to know specifics about the platform, but everything from the kernel on up is out of scope.
OK, that's good.
So let's use UEFI and capsules as a one possible building block in the design, but not as the cornerstone. Just like we do mandate that all updates must be served and managed by our IoT cloud ;).
What do the firmware stacks in your projects look like? What percentage are based on U-Boot? Littlekernel? Others?
I don't have absolute numbers. We have a lot of "standard" x86 UEFI systems, many U-boot devices (of which, to my knowledge, none is on UEFI yet - no surprise, that feature is still too "new"), and quite a few Coreboot designs (so far only x86) that just need to load Linux and bypass a lot therefore. There will likely be more exotic things but they will likely also do not deserve attention.
UEFI has the advantage that it has a defined ABI, and only a small amount of it needs to be implemented to execute UEFI OS loaders (ie. the Linux UEFI stub). For the reference platform, I think it makes sense to make the design decision that firmware provides UEFI, and the update infrastructure is based on that. However, if UEFI doesn't work for a particular stack/project then it can be swapped out. There will need to be work done on the boot sequence, but all the rest of the OS stack is still relevant.
Yes, I would design it in a way that a perfect UEFI implementation can take a lot of boot-related tasks like safe boot configuration switching, watchdog management, firmware and peripheral updating (possibly in lock-step with the OS) etc. But I would always ensure that incomplete or immature UEFI implementations will not complicate partial or complete replacements. Because the latter will remain the default for quite a while, if not longer, I'm sure. UEFI should not become the systemd of unattended device updates (or worse, due to closed implementations).
Jan
On 4/26/19 1:21 PM, Jan Kiszka wrote:
On 26.04.19 12:21, Grant Likely wrote:
On 26/04/2019 10:49, Jan Kiszka wrote:
On 26.04.19 11:07, Francois Ozog wrote:
[...]
Here are the guiding principles of our efforts : 0) we want a cross architecture (x86/Arm/...), cross vendor and cross processor model update solution
- untrusted world cannot update trusted world
Conceptually, it can. It's a matter of validating the update by the trusted world before using it. A trusted instance can allow an untrusted one to write version 2, validate that before switching to it, and stick with version 1 if that fails.
- what the UEFI implementation does with the capsule is platform
specific 3) the update capsule payload is opaque
- is a "philosophy" decision. When you have a root of trust down to
the operating system. So in theory everything should be fine. But the attack surface is such that we can't rule out hacking (and history prove this is unfortunately a safe assumption). In addition, there may be liability aspects related to the who does the update: the hardware platform administrator may not be the legal entity than the operating system administrator. For instance: - on Packet.net could, a customer can flash up to the BL33 untrusted firmware but that is all.
- A surveillance camera can be operated by city personnel but only law
enforcement agency can see raw video (un-blurred faces and licence plates). This can be implemented by a derivative of OPTEE SecureMedia Path but if you allow untrusted world to update the trusted one, city personnel can overcome the legal restriction. With 1) this means that even U-Boot code shall not be able/allowed to update S-EL3 firmware (be it Trusted Firmware A or something else), Secure EL1/0 software (OPTEE and its applications or something else). If possible, allowing the operating system administrator to selectively (BL33 is OK but not S-EL3) update firmware is at least platform dependent. Hence defeats 0)
With 2) we do not impose reboot to update. Some platform may impose reboot or some designers will prefer reboot. We say that there is a chain of responsibility for updates. So it is perfectly OK to have a Linux software agent receive an update by any mean (network, USB, serial...). The agent will pack this (or those) into a capsule and push it to UEFI implementation. The UEFI implementation (U-Boot or Tianocore) will then do whatever it pleases for the update providing it complies with 1) So the UEFI implementation can live update up to BL33 firmware. Should the update be targeted to secure world, then the UEFI implementation can pass it to S-EL3 for update (platform specific) which means the update can also be live. It is a designer decision.
With 3) we have flexibility but sometimes too much flexibility is a problem. Alexander Graf suggested we can use a FIT object to pass diverse objects. It is "half" opaque but I really like the idea. The contents of individual FIT components can be blocks to be placed at a fix location in a NOR flash or a file, no importance.
What do everyone think about those design principles ?
UEFI and capsules can be fine for those platform that support it (and it's still a rare feature) and for stuff like boot and peripheral firmware. I don't think it's a wise, future-proof idea to use it for more.
UEFI is not a very healthy ecosystem yet, and I'm personally skeptical it will evolve towards that (looking at that as both a user as well as an OEM). It's not even present in quite a few of our use cases. In some it will never be - think of safety-critical system: not affordable with such a complex approach like UEFI.
Can I challenge that view a bit? On the Tianocore side I agree that the ecosystem isn't very healthy. That project in particular struggles with what to do with board support, having decided early on that board support generally doesn't need to be in the main repository.
However, U-Boot support for the UEFI ABI is improving in leaps and bounds. SUSE and Fedora both depend on U-Boot UEFI for booting on all the Arm SBCs that they support, and enabling UEFI in U-Boot is just a config option or two. There is a fair bit of encouragement from within the project to enable UEFI by default.
I don't disagree that this aspect is a step forward (though pulling in things redundant code via grub & Co is not really progress). But a Unix community would have probably designed a technically more elegant solution on a green field than the clumsy, legacy-based UEFI interfaces.
UEFI gets a bad rap at being complicated, but I think the U-Boot work has shown that implementing the core UEFI ABI doesn't require much code and isn't the complicated mess they everyone fears it to be.
Depends on how much you start to rely on UEFI features.
The format for a UEFI capsule is described in the EFI_FIRMWARE_MANAGEMENT_PROTOCOL chapter of the UEFI spec. Essentially it is a file containing multiple UEFI binaries which are individually signed and can be loaded as UEFI boottime drivers. Further payloads are passed to the SetImage() method of the EFI_FIRMWARE_MANAGEMENT_PROTOCOL.
Two ways for the delivery of a capsule are defined. Capsules can be delivered by placing a file in the \EFI\UpdateCapsule directory or by calling the UpdateCapsule() boot service. The UpdateCapsule() boot service can either be implemented as available at boottime only or as a full runtime service.
The development target that we have defined for U-Boot is to reach conformance to the Embedded Base Boot Requirements (EBBR) specification. EBBR 1.0 does not mandate implementing update via UEFI capsules.
In U-Boot we have severe size limitations. On many boards U-Boot has to fit into 512 kB. So I think we should not link the capsule update functionality into the main U-Boot binary.
The implementation of the EFI_FIRMWARE_MANAGEMENT_PROTOCOL will be device specific. So if you are interested in adding capsule support I think it should be provided as a device specific UEFI driver that can be loaded by U-Boot.
Currently loading and unloading of drivers is not fully implemented in U-Boot. But it is on my agenda.
Best regards
Heinrich
I don't see the conflict with safety critical for boot services. I can
Everything is certifiable - with infinite time and money. So you will likely reduce the boot process to the very essential of your concrete system, removing then unneeded abstractions and stages. Or you will find ways to eliminate that from your argumentation, which would definitely include removing runtime services.
however see the arguement against UEFI runtime services as a poor implementation could result in unbounded execution times. There's been recent movement on the UEFI spec to make runtime services optional, and in U-Boot they are mostly empty stubs.
Of course, it also depends on how much you need from UEFI to boot or to use certain services like capsule updates (and which would be gone when you remove runtime services).
It should not expand further into OS domains. Updating complete filesystems and their content is beyond its duties. We have all the required, tested, matured means for the OS in the OS. For firmware, it can be an option - if that firmware is UEFI compliant, with all needed options.
Here I agree 100%. Capsule update is a useful ABI for the OS to pass firmware update blobs without needing to know specifics about the platform, but everything from the kernel on up is out of scope.
OK, that's good.
So let's use UEFI and capsules as a one possible building block in the design, but not as the cornerstone. Just like we do mandate that all updates must be served and managed by our IoT cloud ;).
What do the firmware stacks in your projects look like? What percentage are based on U-Boot? Littlekernel? Others?
I don't have absolute numbers. We have a lot of "standard" x86 UEFI systems, many U-boot devices (of which, to my knowledge, none is on UEFI yet - no surprise, that feature is still too "new"), and quite a few Coreboot designs (so far only x86) that just need to load Linux and bypass a lot therefore. There will likely be more exotic things but they will likely also do not deserve attention.
UEFI has the advantage that it has a defined ABI, and only a small amount of it needs to be implemented to execute UEFI OS loaders (ie. the Linux UEFI stub). For the reference platform, I think it makes sense to make the design decision that firmware provides UEFI, and the update infrastructure is based on that. However, if UEFI doesn't work for a particular stack/project then it can be swapped out. There will need to be work done on the boot sequence, but all the rest of the OS stack is still relevant.
Yes, I would design it in a way that a perfect UEFI implementation can take a lot of boot-related tasks like safe boot configuration switching, watchdog management, firmware and peripheral updating (possibly in lock-step with the OS) etc. But I would always ensure that incomplete or immature UEFI implementations will not complicate partial or complete replacements. Because the latter will remain the default for quite a while, if not longer, I'm sure. UEFI should not become the systemd of unattended device updates (or worse, due to closed implementations).
Jan
On 26.04.19 15:46, Heinrich Schuchardt wrote:
On 4/26/19 1:21 PM, Jan Kiszka wrote:
On 26.04.19 12:21, Grant Likely wrote:
On 26/04/2019 10:49, Jan Kiszka wrote:
On 26.04.19 11:07, Francois Ozog wrote:
[...]
Here are the guiding principles of our efforts : 0) we want a cross architecture (x86/Arm/...), cross vendor and cross processor model update solution
- untrusted world cannot update trusted world
Conceptually, it can. It's a matter of validating the update by the trusted world before using it. A trusted instance can allow an untrusted one to write version 2, validate that before switching to it, and stick with version 1 if that fails.
- what the UEFI implementation does with the capsule is platform
specific 3) the update capsule payload is opaque
- is a "philosophy" decision. When you have a root of trust down to
the operating system. So in theory everything should be fine. But the attack surface is such that we can't rule out hacking (and history prove this is unfortunately a safe assumption). In addition, there may be liability aspects related to the who does the update: the hardware platform administrator may not be the legal entity than the operating system administrator. For instance: - on Packet.net could, a customer can flash up to the BL33 untrusted firmware but that is all.
- A surveillance camera can be operated by city personnel but only law
enforcement agency can see raw video (un-blurred faces and licence plates). This can be implemented by a derivative of OPTEE SecureMedia Path but if you allow untrusted world to update the trusted one, city personnel can overcome the legal restriction. With 1) this means that even U-Boot code shall not be able/allowed to update S-EL3 firmware (be it Trusted Firmware A or something else), Secure EL1/0 software (OPTEE and its applications or something else). If possible, allowing the operating system administrator to selectively (BL33 is OK but not S-EL3) update firmware is at least platform dependent. Hence defeats 0)
With 2) we do not impose reboot to update. Some platform may impose reboot or some designers will prefer reboot. We say that there is a chain of responsibility for updates. So it is perfectly OK to have a Linux software agent receive an update by any mean (network, USB, serial...). The agent will pack this (or those) into a capsule and push it to UEFI implementation. The UEFI implementation (U-Boot or Tianocore) will then do whatever it pleases for the update providing it complies with 1) So the UEFI implementation can live update up to BL33 firmware. Should the update be targeted to secure world, then the UEFI implementation can pass it to S-EL3 for update (platform specific) which means the update can also be live. It is a designer decision.
With 3) we have flexibility but sometimes too much flexibility is a problem. Alexander Graf suggested we can use a FIT object to pass diverse objects. It is "half" opaque but I really like the idea. The contents of individual FIT components can be blocks to be placed at a fix location in a NOR flash or a file, no importance.
What do everyone think about those design principles ?
UEFI and capsules can be fine for those platform that support it (and it's still a rare feature) and for stuff like boot and peripheral firmware. I don't think it's a wise, future-proof idea to use it for more.
UEFI is not a very healthy ecosystem yet, and I'm personally skeptical it will evolve towards that (looking at that as both a user as well as an OEM). It's not even present in quite a few of our use cases. In some it will never be - think of safety-critical system: not affordable with such a complex approach like UEFI.
Can I challenge that view a bit? On the Tianocore side I agree that the ecosystem isn't very healthy. That project in particular struggles with what to do with board support, having decided early on that board support generally doesn't need to be in the main repository.
However, U-Boot support for the UEFI ABI is improving in leaps and bounds. SUSE and Fedora both depend on U-Boot UEFI for booting on all the Arm SBCs that they support, and enabling UEFI in U-Boot is just a config option or two. There is a fair bit of encouragement from within the project to enable UEFI by default.
I don't disagree that this aspect is a step forward (though pulling in things redundant code via grub & Co is not really progress). But a Unix community would have probably designed a technically more elegant solution on a green field than the clumsy, legacy-based UEFI interfaces.
UEFI gets a bad rap at being complicated, but I think the U-Boot work has shown that implementing the core UEFI ABI doesn't require much code and isn't the complicated mess they everyone fears it to be.
Depends on how much you start to rely on UEFI features.
The format for a UEFI capsule is described in the EFI_FIRMWARE_MANAGEMENT_PROTOCOL chapter of the UEFI spec. Essentially
Are you sure? I thought that protocol was about other devices in the system, not the main firmware. It also seems to be optional - so we could just have a board specific function to implement CapsuleUpdate, but no cruft for all the other bits in the spec.
That's where the idea came from to just put a fit image into the capsule body (set CapsuleGUID to a newly defined FIT-GUID). With that, we could share a lot of code inside of U-Boot, as DT parsing is already there. We could then have individual segments, that can either be data or command payloads and thus a capsule update could basically just be a few data segments with a U-Boot script.
it is a file containing multiple UEFI binaries which are individually signed and can be loaded as UEFI boottime drivers. Further payloads are passed to the SetImage() method of the EFI_FIRMWARE_MANAGEMENT_PROTOCOL.
Two ways for the delivery of a capsule are defined. Capsules can be delivered by placing a file in the \EFI\UpdateCapsule directory or by calling the UpdateCapsule() boot service. The UpdateCapsule() boot service can either be implemented as available at boottime only or as a full runtime service.
The development target that we have defined for U-Boot is to reach conformance to the Embedded Base Boot Requirements (EBBR) specification. EBBR 1.0 does not mandate implementing update via UEFI capsules.
Correct, this is about post-1.0 :).
In U-Boot we have severe size limitations. On many boards U-Boot has to fit into 512 kB. So I think we should not link the capsule update functionality into the main U-Boot binary.
The implementation of the EFI_FIRMWARE_MANAGEMENT_PROTOCOL will be device specific. So if you are interested in adding capsule support I think it should be provided as a device specific UEFI driver that can be loaded by U-Boot.
Currently loading and unloading of drivers is not fully implemented in U-Boot. But it is on my agenda.
One of the things I love about U-Boot is that it's monolithic :). As an end user, it means that I get something and it just contains everything I need. Imagine a world where you first have to load your driver to update your firmware from some arbitrary storage. It would create a huge mess.
Alex
On Sat, Apr 27, 2019 at 09:56:08AM +0200, Alexander Graf wrote:
UEFI gets a bad rap at being complicated, but I think the U-Boot work has shown that implementing the core UEFI ABI doesn't require much code and isn't the complicated mess they everyone fears it to be.
Depends on how much you start to rely on UEFI features.
The format for a UEFI capsule is described in the EFI_FIRMWARE_MANAGEMENT_PROTOCOL chapter of the UEFI spec. Essentially
Are you sure? I thought that protocol was about other devices in the system, not the main firmware. It also seems to be optional - so we could just have a board specific function to implement CapsuleUpdate, but no cruft for all the other bits in the spec.
There's no requirement to use that format in order to use capsule updates, but some vendors have found it convenient. There's also no requirement to implement the EFI_FIRMWARE_MANAGEMENT_PROTOCOL at all - and the only place it's distinctly useful from an application point of view (i.e. while trying to drive a firmware update) is if you want to independently update two of the same kind of peripheral device's firmware from within a UEFI application. It's not clear to me that's a very interesting use case from an EBBR point of view - I'm haven't seen much demand for it on server, laptop, or desktop systems either.
But it's also how option ROMs register their firmware to be updated via UpdateCapsule(). That's interesting on machines where you regularly expect peripherals to be in PCIe slots, but I'm still not sure it's that useful for EBBR style devices. I guess it could be useful in the thunderbolt case, but I think in order to avoid thwarting DMA attack mitigations, that would best be done from the OS just like for USB devices.
That's where the idea came from to just put a fit image into the capsule body (set CapsuleGUID to a newly defined FIT-GUID). With that, we could share a lot of code inside of U-Boot, as DT parsing is already there. We could then have individual segments, that can either be data or command payloads and thus a capsule update could basically just be a few data segments with a U-Boot script.
This makes a lot of sense to me, though there have been some pain points about this approach in the past. The biggest one is because the EFI_FIRMWARE_MANAGEMENT_PROTOCOL capsule format has a header on it that's the same as the EFI_CAPSULE_HEADER structure, which we have to add in fwupd if we can't find it. If we plan on recommending that vendors implement these ad-hoc firmware blobs, we should also require that when they do so, they already include the EFI_CAPSULE_HEADER on the image, with a GUID that matches what's in the ESRT, and is specifically not EFI_FIRMWARE_MANAGEMENT_CAPSULE_ID_GUID.
it is a file containing multiple UEFI binaries which are individually signed and can be loaded as UEFI boottime drivers. Further payloads are passed to the SetImage() method of the EFI_FIRMWARE_MANAGEMENT_PROTOCOL.
Two ways for the delivery of a capsule are defined.
Three, I think? I'm not seeing why an application can't also call FMP->SetImage() to do this when it's present. I'm not sure it's worth supporting at all, though.
Capsules can be delivered by placing a file in the \EFI\UpdateCapsule directory or by calling the UpdateCapsule() boot service. The UpdateCapsule() boot service can either be implemented as available at boottime only or as a full runtime service.
Right now in fwupd we support only boot-service calls to do this, because of the history of bugs with SetVirtualAddressMap() and ExitBootServices(), and because it's the only thing required to be tested for Windows logo cert. I suspect MS has made that the case because of the exact same bugs.
The path-based mechanism makes a lot of sense with EBBR, but runtime UpdateCapsule() is a lot more viable here than on the desktop or servers as well. The reason I say that is because with these devices we're more in a situation where the hardware vendor is likely to be producing the kernel as well, so there's a better integration testing story here.
On Fri, 26 Apr 2019 at 11:49, Jan Kiszka jan.kiszka@siemens.com wrote:
On 26.04.19 11:07, Francois Ozog wrote:
On Fri, 26 Apr 2019 at 10:30, Christian Storm christian.storm@siemens.com wrote:
Hi,
Background: during the last Linaro connect in Bangkok I was told that Linaro Edge (LEDGE) were working on a secure software update mechanism based on UEFI capsules that would flash firmware updates from a UEFI application, instead of using a Linux agent such as SWUpdate.
How would capsules help with writing to arbitrary storage, updating only files on filesystem, reducing the update size (binary diffs), or talking to the cloud?
See below
- arbitrary storage: I guess they can only write to what is supported by the machine's UEFI implementation.
Yes, see below
- updating only files on filesystem: I assume this is out of scope in their architecture (Francois: do you want to support file-based updates or only block-based ones?)
Both, see below
- reducing the update size (binary diffs), or talking to the cloud: they will do that from non-secure Linux. It would be dangerous to
use
a fragile network stack from an UEFI application or the secure
world.
In that sense, they also need a Linux agent.
If I understand the idea correctly, there's then a Linux companion agent that does, e.g., the cloud communication and sets up the update payload to be consumed and applied by the UEFI firmware update application. The actual update action will take place on reboot, e.g., flashing the inactive partition in an A/B setup and thereafter booting into the newly flashed, then active partition.
If so, this would prolong the time the system is unavailable due to disk flashing I/O in the UEFI firmware update application compared to
flashing
the inactive partition from within Linux and have the bootloader just switch the to be booted partitions.
Generally, but in particular regarding binary diffs, a binary diff application mechanism would have to be implemented and maintained in the UEFI realm as opposed to "just" using the respective shared libraries (e.g. librsync, casync, ...) in the Linux realm. Granted, one could come up with a unified format to which the output of, e.g, librsync's
rdiff
or a casync run is converted but nonetheless, this creates some porting/ implementation and maintenance effort.
Here are the guiding principles of our efforts : 0) we want a cross architecture (x86/Arm/...), cross vendor and cross processor model update solution
- untrusted world cannot update trusted world
Conceptually, it can. It's a matter of validating the update by the trusted world before using it. A trusted instance can allow an untrusted one to write version 2, validate that before switching to it, and stick with version 1 if that fails.
Agreed. Then is a better principle: untrusted world cannot directly
update trusted world objects that are part of the current chain of trust This would allow changing B partition while A is in force. There may be another aspect to take into account, not every use case need the same security "level". May be a principle to define security levels and guidelines to achieve each level is more relevant than the "untrusted world cannot directly update trusted world objects that are part of the current chain of trust" which implies a certain security level goal.
- what the UEFI implementation does with the capsule is platform specific
the update capsule payload is opaque
is a "philosophy" decision. When you have a root of trust down to
the operating system. So in theory everything should be fine. But the attack surface is such that we can't rule out hacking (and history prove this is unfortunately a safe assumption). In addition, there may be liability aspects related to the who does the update: the hardware platform administrator may not be the legal entity than the operating system administrator. For instance:
- on Packet.net could, a customer can flash up to the BL33 untrusted
firmware but that is all.
- A surveillance camera can be operated by city personnel but only law
enforcement agency can see raw video (un-blurred faces and licence plates). This can be implemented by a derivative of OPTEE SecureMedia Path but if you allow untrusted world to update the trusted one, city personnel can overcome the legal restriction. With 1) this means that even U-Boot code shall not be able/allowed to update S-EL3 firmware (be it Trusted Firmware A or something else), Secure EL1/0 software (OPTEE and its applications or something else). If possible, allowing the operating system administrator to selectively (BL33 is OK but not S-EL3) update firmware is at least platform dependent. Hence defeats 0)
With 2) we do not impose reboot to update. Some platform may impose reboot or some designers will prefer reboot. We say that there is a chain of responsibility for updates. So it is perfectly OK to have a Linux software agent receive an update by any mean (network, USB, serial...). The agent will pack this (or those) into a capsule and push it to UEFI implementation. The UEFI implementation (U-Boot or Tianocore) will then do whatever it pleases for the update providing it complies with 1) So the UEFI implementation can live update up to BL33 firmware. Should the update be targeted to secure world, then the UEFI implementation can pass it to S-EL3 for update (platform specific) which means the update can also be live. It is a designer decision.
With 3) we have flexibility but sometimes too much flexibility is a problem. Alexander Graf suggested we can use a FIT object to pass diverse objects. It is "half" opaque but I really like the idea. The contents of individual FIT components can be blocks to be placed at a fix location in a NOR flash or a file, no importance.
What do everyone think about those design principles ?
UEFI and capsules can be fine for those platform that support it (and it's still a rare feature) and for stuff like boot and peripheral firmware. I don't think it's a wise, future-proof idea to use it for more.
Agreed. if we say that {shim, grub , kernel, initrd, commandline} is to be
named the "BL33 payload" then they are out of scope of the UEFI update capsules, as well as any untrusted apps.= (on top of OS). Update capsules can be used to update firmwares of any sort, keys, certificates, secrets, blobs that are for pre-BL33 execution environment.
UEFI is not a very healthy ecosystem yet, and I'm personally skeptical it
will evolve towards that (looking at that as both a user as well as an OEM). It's not even present in quite a few of our use cases. In some it will never be - think of safety-critical system: not affordable with such a complex approach like UEFI. It should not expand further into OS domains. Updating complete filesystems and their content is beyond its duties. We have all the required, tested, matured means for the OS in the OS. For firmware, it can be an option - if that firmware is UEFI compliant, with all needed options.
So let's use UEFI and capsules as a one possible building block in the design, but not as the cornerstone. Just like we do mandate that all updates must be served and managed by our IoT cloud ;).
Agreed. Out of all the edge and fog deployments some will have very
specific requirements. This effort is not to rule them out as rule out proprietary mechanisms. It is to say that 90% of the cases can be handled by a generic method and we should define this one.
Jan
-- Siemens AG, Corporate Technology, CT RDA IOT SES-DE Corporate Competence Center Embedded Linux
On 24/04/2019 02:23, daniel.sangorrin@toshiba.co.jp wrote:
Hello Francois, Jan, Christian, and all
Sorry for the late reply, I was waiting for the administrator of the Boot Architecture mailing list to accept my subscription request, but it seems it will take a bit more time. I will send this reply and hope it will not be blocked. I have also added the u-boot mailing list to Cc, as Tom suggested (although I'm not a member), the CIP mailing list, Jan Kiszka (one of the main developers of Efibootguard) and Christian (an expert in software updates).
Hmmm, that's weird. The boot-architecture list doesn't require approval to subscribe. It just requires confirmation of email address. I think there is a problem with lists.linaro.org. I'll send an IT support request to Linaro.
Regardless, I've approved this post and added you to the allowed senders list. You can post freely. :-)
g.
Background: during the last Linaro connect in Bangkok I was told that Linaro Edge (LEDGE) were working on a secure software update mechanism based on UEFI capsules that would flash firmware updates from a UEFI application, instead of using a Linux agent such as SWUpdate. Then, I had an online meeting with Francois, director of LEDGE. I explained to Francois that in CIP we are using the Linux agent approach right now, and we are also considering the use of a UEFI application (Efibootguard) to arm a watchdog and deal with the state-machine variables (installed, testing, ok, failed..) needed for A/B software updates. Efibootguard sounds like an excellent place to collaborate with Linaro (particularly on the watchdog drivers front) because it does not strictly depend on where the firmware is flashed (UEFI capsule or Linux agent).
On Fri, Apr 19, 2019 at 12:48:51PM +0200, Francois Ozog wrote:
Hi Daniel,
We will be conducting a UEFI gap analysis to support EFIBootGuard in U-Boot.
As we are working on UEFI SecureBoot implementation in U-Boot, how do you expect the boot process to be secured? Would U-Boot UEFI SecureBoot verify EFIBootGuard signature and in turn EFIBootGuard will check either grub or Linux signature?
Please elaborate on your vision of a secured boot process.
Efibootguard is composed of two parts.
- A UEFI application that can arm a watchdog and decide what environment (kernel, boot args, etc.) to use next depending on a set of variables (update status, highest revision, etc.) stored in FAT16 partitions.
- A Linux application that can read and set those variables from Linux (similar to u-boot's fw_setenv). This functionality is also available in the form of a library.
As far as I know, there is no concept of "Secure Booting" in Efibootguard at the moment. Adding signature checks before booting into the selected kernel would be a possible solution.
Thanks, Daniel
boot-architecture mailing list boot-architecture@lists.linaro.org https://lists.linaro.org/mailman/listinfo/boot-architecture
boot-architecture@lists.linaro.org
-
Alexander Graf -
Christian Storm -
daniel.sangorrin＠toshiba.co.jp -
Francois Ozog -
Grant Likely -
Heinrich Schuchardt -
Jan Kiszka -
Peter Jones