specification: updating for latest linux changes by dpsmith · Pull Request #35 · TrenchBoot/documentation

dpsmith · 2026-04-08T23:43:23Z

Update the specification to align with the changes made for the Linux maintainers.

Update the specification to align with the changes made for the Linux maintainers. Signed-off-by: Daniel P. Smith <dpsmith@apertussolutions.com>

krystian-hebel · 2026-04-09T06:55:47Z

-        u64 boot_params_addr;
        u64 slrt;
-        u64 txt_info;
+        u64 txt_info; 


Remove trailing space.

krystian-hebel · 2026-04-09T08:04:14Z

+:uuid0: Chunk 0 of UUID, 9082AC5A
+:uuid1: Chunk 1 of UUID, 74A7476F
+:uuid2: Chunk 2 of UUID, A2555C0F
+:uuid3: Chunk 3 of UUID, 42B651CB


Should it be unconditionally included in non-TXT binaries as well? It probably makes sense for x86 as vendor can be easily detected, but I'm not so sure about other architectures.

Yes, while it originates from Intel TXT, we want to standardize the dl-stub across architectures and this provides the cleanest, IOW maximize reuse, method to communicate the necessary information to the dl-stub

SergiiDmytruk · 2026-04-09T13:18:51Z

+:uuid1: Chunk 1 of UUID, 74A7476F
+:uuid2: Chunk 2 of UUID, A2555C0F
+:uuid3: Chunk 3 of UUID, 42B651CB
+:header_len: Length of the header.


To be unambiguous since this is not always the case:

Suggested change

:header_len: Length of the header.

:header_len: Length of the header in bytes.

SergiiDmytruk · 2026-04-09T13:20:12Z

+:header_len: Length of the header.
+:version: Header version.
+:entry_point: Address of the DLME entry point.
+:first_valid_page: Address to first valid page of DLME.


Suggested change

:first_valid_page: Address to first valid page of DLME.

:first_valid_page: Address of the first valid 4KiB-page of DLME.

I suppose it also needs to have zeroes in its lowest bits.

I would not specify 4K, it should be kept platform agnostic. I am good with the other wording change.

SergiiDmytruk · 2026-04-09T13:22:21Z

+:start: Offset to first byte of DLME within page space.
+:end: Offset to last byte + 1 of DLME within page space.


The base for the offsets is unspecified.

This is how it is defined in MLE guide,

Note: linear addresses start at the beginning of the executable region of the binary,
not the beginning of the file. The offset between the beginning of the binary file and
beginning of the linear address space is ABI-dependent.

I am going to have to think of a good way to say this in relation to this spec. Suggestions are welcome.

SergiiDmytruk · 2026-04-09T13:22:56Z

+Handler to facilitate it in launching the kernel.
+
+
+DLME Header Structure


Are there any alignment requirements?

The only requirement from TXT spec is that it must be within the MLE range. If we want to impose alignment, I'm open to that discussion. It may require a slight change to the Linux series, though I don't think it would be difficult.

SergiiDmytruk · 2026-04-09T13:24:23Z

-| **1.3** - All other registers besides EDI/RDI are not guarenteed
-| **1.4** - The invoking code **SHALL** use a long jump to the DLE Handler
-| **1.5** - The DLE Handler **SHALL NOT** return control on error
+| **2.3** - All other registers besides EDI/RDI are not guarenteed


A typo:

Suggested change

| **2.3** - All other registers besides EDI/RDI are not guarenteed

| **2.3** - All other registers besides EDI/RDI are not guaranteed

SergiiDmytruk · 2026-04-09T13:25:47Z

+    typedef struct tdEFI_SLAUNCH_PROTOCOL {
+      EFI_SLAUNCH_SLRT_UPDATE_REGISTER_DLME RegisterDlme;
+      EFI_SLAUNCH_LAUNCH Launch;
+    } EFI_SLAUNCH_PROTOCOL


Suggested change

} EFI_SLAUNCH_PROTOCOL

} EFI_SLAUNCH_PROTOCOL;

SergiiDmytruk · 2026-04-09T13:26:12Z

+                        IN EFI_PHYSICAL_ADDRESS *DlmeBase
+                        IN EFI_PHYSICAL_ADDRESS *DlmeHeader


Suggested change

IN EFI_PHYSICAL_ADDRESS *DlmeBase

IN EFI_PHYSICAL_ADDRESS *DlmeHeader

IN EFI_PHYSICAL_ADDRESS *DlmeBase,

IN EFI_PHYSICAL_ADDRESS *DlmeHeader,

SergiiDmytruk · 2026-04-09T13:32:00Z

+:dlme_header_offset: The offest into the DLME of the DLME Header.
+:dlme_config_table: The offsee into the DLME of the DLME Config Table.


Suggested change

:dlme_header_offset: The offest into the DLME of the DLME Header.

:dlme_config_table: The offsee into the DLME of the DLME Config Table.

:dlme_header_offset: The offset into the DLME of the DLME Header.

:dlme_config_table: The offset into the DLME of the DLME Config Table.

specification: updating for latest linux changes

5297853

Update the specification to align with the changes made for the Linux maintainers. Signed-off-by: Daniel P. Smith <dpsmith@apertussolutions.com>

pietrushnic requested a review from SergiiDmytruk April 9, 2026 04:47

krystian-hebel reviewed Apr 9, 2026

View reviewed changes

SergiiDmytruk reviewed Apr 9, 2026

View reviewed changes

	:header_len: Length of the header.
	:header_len: Length of the header in bytes.

	:first_valid_page: Address to first valid page of DLME.
	:first_valid_page: Address of the first valid 4KiB-page of DLME.

		:start: Offset to first byte of DLME within page space.
		:end: Offset to last byte + 1 of DLME within page space.

		Handler to facilitate it in launching the kernel.


		DLME Header Structure

	\| 2.3 - All other registers besides EDI/RDI are not guarenteed
	\| 2.3 - All other registers besides EDI/RDI are not guaranteed

		IN EFI_PHYSICAL_ADDRESS *DlmeBase
		IN EFI_PHYSICAL_ADDRESS *DlmeHeader

		:dlme_header_offset: The offest into the DLME of the DLME Header.
		:dlme_config_table: The offsee into the DLME of the DLME Config Table.

Conversation

dpsmith commented Apr 8, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants