Re: [EXTERNAL] [PATCH 11/20] mach-k3: r5: common: add helper functions needed in LPM resume sequence

[Kumar, Abhash]

On 4/30/2026 2:14 PM, Richard Genoud (TI) wrote:
From: Prasanth Babu Mantena <p-mantena@ ti. com> Add helper functions 
that are used by respective SoCs in LPM resume flow. - lpm_process() 
is called at boot time to: - retrieve the LPM memory region from DTS - 
save ATF/OPTEE certificates
ZjQcmQRYFpfptBannerStart
This message was sent from outside of Texas Instruments.
Do not click links or open attachments unless you recognize the source 
of this email and know the content is safe.
Report Suspicious
<https://us-phishalarm-ewt.proofpoint.com/EWT/v1/G3vK!undqXRfPtm2bSgZlPFMD_PaDMAG16nyrFWtHumWcmHVJHF2x9T6ju4MVod9Xz5kkhLg8jvqD8Vb2RXV8TKm6Nx3q-RI8sEAX0Ue59g$> 

ZjQcmQRYFpfptBannerEnd
From: Prasanth Babu Mantena <p-mant...@ti.com>

Add helper functions that are used by respective SoCs in LPM resume flow.

- lpm_process() is called at boot time to:
   - retrieve the LPM memory region from DTS
   - save ATF/OPTEE certificates information and DM code in this memory
     region
   - Forward the LPM address to TIFS via TISCI_MSG_LPM_SAVE_ADDR
TIFS will use this address to save TFA context and its own minimal
context just before suspend.

- do_resume() is called at resume, just after bringing the DDR out of
   retention to:
   - retrieve the LPM memory region from DTS
   - authenticate certificates from LPM memory region and apply firewalls
   - ask TIFS to restore TFA and its own minimal context
   - start TFA on remote proc
   - load and jump to DM

https://software-dl.ti.com/tisci/esd/latest/2_tisci_msgs/pm/lpm.html#lpm-msg-lpm-save-addr

Signed-off-by: Prasanth Babu Mantena <p-mant...@ti.com>
Co-developed-by: Richard Genoud (TI) <richard.gen...@bootlin.com>
Signed-off-by: Richard Genoud (TI) <richard.gen...@bootlin.com>
---
  arch/arm/mach-k3/common.h        |  16 +++
  arch/arm/mach-k3/lpm-common.h    |  15 ++
  arch/arm/mach-k3/r5/Kconfig      |   4 +
  arch/arm/mach-k3/r5/Makefile     |   1 +
  arch/arm/mach-k3/r5/common.c     |  17 +--
  arch/arm/mach-k3/r5/lpm-common.c | 228 +++++++++++++++++++++++++++++++
  6 files changed, 269 insertions(+), 12 deletions(-)
  create mode 100644 arch/arm/mach-k3/lpm-common.h
  create mode 100644 arch/arm/mach-k3/r5/lpm-common.c

diff --git a/arch/arm/mach-k3/common.h b/arch/arm/mach-k3/common.h
index e970076d08ec..50e670f845e3 100644
--- a/arch/arm/mach-k3/common.h
+++ b/arch/arm/mach-k3/common.h
@@ -8,6 +8,7 @@
  
  #include <asm/armv7_mpu.h>
  #include <asm/hardware.h>
+#include <image.h>
  #include <mach/security.h>
  
+static int extract_lpm_region(void)
+{
+       ofnode node;
+       fdt_addr_t lpm_reg_addr;
+       fdt_size_t lpm_reg_size;
+
+       node = ofnode_path("/reserved-memory/lpm-memory");
+       if (!ofnode_valid(node)) {
+               printf("lpm will not be functional\n");
+               return -ENODEV;
+       }
+
+       lpm_reg_addr = ofnode_get_addr(node);
+       if (lpm_reg_addr == FDT_ADDR_T_NONE) {
+               printf("Can't find a valid reserved node!\n");
+               return -ENODEV;
+       }
+
+       lpm_reg_size = ofnode_get_size(node);
+       if (lpm_reg_size == FDT_ADDR_T_NONE) {
+               printf("Can't find a valid reserved node!\n");
+               return -ENODEV;
+       }
+
+       mem_addr_lpm.context_save_addr = (u32 *)lpm_reg_addr;
+       mem_addr_lpm.atf_cert_addr = mem_addr_lpm.context_save_addr + 
FW_IMAGE_SIZE;
+       mem_addr_lpm.optee_cert_addr = mem_addr_lpm.atf_cert_addr + 
FW_IMAGE_SIZE;

Hi Richard,

Adding a u32 pointer to an integer "FW_IMAGE_SIZE" would scale 
FW_IMAGE_SIZE by sizeof(u32).

This would cause the address values to come out different than intended.

+       mem_addr_lpm.dm_save_addr = mem_addr_lpm.optee_cert_addr + (2 * 
FW_IMAGE_SIZE);
+       mem_addr_lpm.size = lpm_reg_size;
+
+       return 0;
+}
+

Thanks,

Abhash