TF-M March 2023

tf-m@lists.trustedfirmware.org

20 participants
14 discussions

by Bohdan.Hunko＠infineon.com

Hi everyone, Why investigation some interrupts related stuff I have noticed a problem/bug in the scheduler. PendSV_Handler has following code: " movs r0, #"M2S(EXC_RETURN_SECURE_STACK)" \n" " mov r1, lr \n" " tst r0, r1 \n" /* NS interrupted */ " beq v8b_pendsv_exit \n" /* No schedule */ Which means that if NSPE was interrupted then no scheduling will be done. This created a problem in following scenario: 1. NSPE is executing 2. Secure interrupts arrives. 3. NSPE is interrupted, secure interrupt is handled. 4. Secure interrupts triggers scheduling of some partition. 5. Secure interrupt returns 6. Scheduling is triggered 7. PendSV returns (because /* NS interrupted */ /* No schedule */ ) and does not perform scheduling even thought pending partition is present. This means that pending partition will only be scheduled when next PSA call will be done be NSPE. Even if there would be subsequent Secure interrupts that preempt NSPE, partition will not be scheduled because PendSV will always return without scheduling because NSPE was interrupted. I was not able to find any information about this in TFM docs so I assume this is a bug. Please let me know if I am wrong. And if this is real problem then is there a plan to fix this? Regards, Bohdan Hunko Cypress Semiconductor Ukraine Engineer CSUKR CSS ICW SW FW Mobile: +38099 50 19 714 Bohdan.Hunko(a)infineon.com<mailto:Bohdan.Hunko@infineon.com>

6 months, 1 week

Support mbedTLS 3.3.0

by S Krishnan, Archanaa

Hi, TF-M v1.7 utilizes mbedTLS v3.2.1. Are there plans to update to mbedTLS v3.3.0? Regards, Archanaa

6 months, 2 weeks

Partitions without manifest files are not linked correctly in L3

by Bohdan.Hunko＠infineon.com

In isolation level 3 partitions code/data in linker script are gathered together and aligned using information from manifest files. Currently there are 2 partitions that are not using manifest files, and instead have hand written load_info.c files. These partitions are: NS agent trust zone and idle partition. When partition does not have manifest file then its code/data is not gathered together (as there is no manifest to provide needed information). This results in partition code/data being linked directly to SPM. Also code/data may be not correctly aligned (if platform requires special alignment for PSA/APP RoT partitions). For example if platform define custom TFM_LINKER_PSA_ROT_LINKER_DATA_ALIGNMENT, NS agent TZ and idle partitions stacks will not be aligned properly. This is a problem because resulting alignment is not sufficient for the platform, which means that functions that apply protections fail. I see several solutions to this problem: 1. Add alignment to stack of these special partitions. Both the start and the size of the stack should be aligned to satisfy alignment requirements. This is fairly easy fix with small amount of changes. The problem is that code/data of these partitions will still be located in SPM code/data sections which is not ideal solution. I would say this is bare minimum solution, just to make things work. 2. Better solution might be to move these special partitions to now use manifest files. The problem I see is that these partition use special priorities values which are not supported by manifest tool. Also NS Agent TZ uses special PID = 0, which I believe is also not supported by manifest tool. I think this is more time consuming fix but overall this should result in better and easier to understand code. Would be glad to hear a feedback on this topic. Regards, Bohdan Hunko Cypress Semiconductor Ukraine Engineer CSUKR CSS ICW SW FW Mobile: +38099 50 19 714 Bohdan.Hunko(a)infineon.com<mailto:Bohdan.Hunko@infineon.com>

6 months, 2 weeks

tfm_hal_memory_symbols

by Chris.Brand＠infineon.com

Hi, I'm looking at https://git.trustedfirmware.org/TF-M/trusted-firmware-m.git/commit/?id=df87… which added tfm_hal_memory_symbols.{c|h}, containing tfm_hal_sp_meta_start and tfm_hal_sp_meta_end. The former is used in backend_ipc.c, while the latter is unused. The commit message says "this is the first example of using defined symbols to get memory info" but that commit is over two years old now and there doesn't seem to be a second example. Is there still a plan to move in this direction? If so, can somebody outline what that involves? Thanks, Chris Brand Cypress Semiconductor (Canada), Inc. An Infineon Technologies Company Sr Prin Software Engr CSCA CSS ICW SW PSW 1 Chris.Brand(a)infineon.com<mailto:Chris.Brand@infineon.com> International Place 13700 V6V 2X8 Richmond Canada www.infineon.com<www.cypress.com> www.cypress.com<http://www.cypress.com> Discoveries<http://www.infineon.com/discoveries> Facebook<http://www.facebook.com/infineon> Twitter<http://www.twitter.com/Infineon> LinkedIn<http://www.linkedin.com/company/infineon-technologies> Part of your life. Part of tomorrow. NOTICE: The information transmitted is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material of Infineon Technologies AG and its affiliated entities which is for the exclusive use of the individual designated above as the recipient. Any review, retransmission, dissemination or other use of, or taking of any action in reliance upon, this information by persons or entities other than the intended recipient is prohibited. If you received this in error, please contact immediately the sender by returning e-mail and delete the material from any computer. If you are not the specified recipient, you are hereby notified that all disclosure, reproduction, distribution or action taken on the basis of this message is prohibited.

6 months, 3 weeks

2023

2022

2021

2020

2019

2018

TF-M March 2023