autopatch/edk2
6
0
Fork 0
Code Issues Pull requests Releases Activity

Rebase AlmaLinux X64 ISA-baseline pin onto edk2 20260821 (GCC48 -> GCCNOLTO) #1

Merged
eabdullin merged 1 commit from triage/fix-edk2-manual-86018 into a10s 2026-09-24 07:46:28 +00:00
Conversation 0 Commits 1 Files changed 1 +10 -9
almalinux-automation commented 2026-09-23 20:56:26 +00:00
Member
Copy link

Problem

The AlmaLinux-only autopatch 0001-BaseTools-pin-the-X64-ISA-baseline-to-x86-64-v2.patch no longer applied after the edk2 rebase from 20260221-7 to 20260821-2, failing %prep on every arch:

error: patch failed: BaseTools/Conf/tools_def.template:865
Rejected hunk #1.

What drifted upstream

Upstream reorganized BaseTools/Conf/tools_def.template and removed the deprecated GCC48/GCC49/GCC5 toolchains. The GCC X64 flag DEFINE block moved from line ~865 to line ~618, and the macro was renamed:

  • old: DEFINE GCC48_X64_CC_FLAGS = DEF(GCC48_ALL_CC_FLAGS) DEF(GCC_IA32_X64_CC_FLAGS) -m64 ...
  • new: DEFINE GCCNOLTO_X64_CC_FLAGS = DEF(GCC_ALL_CC_FLAGS) DEF(GCC_IA32_X64_CC_FLAGS) -m64 ...

GCCNOLTO_X64_CC_FLAGS is the successor base: GCC_X64_CC_FLAGS = DEF(GCCNOLTO_X64_CC_FLAGS), and the LTO GCC toolchain used by the spec (%define TOOLCHAIN GCC) derives its X64 CC flags from GCC_X64_CC_FLAGS. So pinning GCCNOLTO_X64_CC_FLAGS propagates the ISA baseline to exactly the toolchain that builds OVMF, preserving the original patch's intent.

Fix

Re-anchored the single one-line hunk onto the renamed/relocated GCCNOLTO_X64_CC_FLAGS line, inserting -march=x86-64-v2 immediately after -m64 (identical placement to the original). The diff body was regenerated with git so the hunk header/counts are correct by construction. The commit message/body is unchanged.

Verification

patch -p1 --fuzz=0 --dry-run and a real patch -p1 --fuzz=0 against a pristine copy of src/ both apply cleanly (exit=0) and yield the intended -march=x86-64-v2 pin.

## Problem The AlmaLinux-only autopatch `0001-BaseTools-pin-the-X64-ISA-baseline-to-x86-64-v2.patch` no longer applied after the edk2 rebase from 20260221-7 to 20260821-2, failing `%prep` on every arch: ``` error: patch failed: BaseTools/Conf/tools_def.template:865 Rejected hunk #1. ``` ## What drifted upstream Upstream reorganized `BaseTools/Conf/tools_def.template` and removed the deprecated `GCC48`/`GCC49`/`GCC5` toolchains. The GCC X64 flag DEFINE block moved from line ~865 to line ~618, and the macro was renamed: - old: `DEFINE GCC48_X64_CC_FLAGS = DEF(GCC48_ALL_CC_FLAGS) DEF(GCC_IA32_X64_CC_FLAGS) -m64 ...` - new: `DEFINE GCCNOLTO_X64_CC_FLAGS = DEF(GCC_ALL_CC_FLAGS) DEF(GCC_IA32_X64_CC_FLAGS) -m64 ...` `GCCNOLTO_X64_CC_FLAGS` is the successor base: `GCC_X64_CC_FLAGS = DEF(GCCNOLTO_X64_CC_FLAGS)`, and the LTO GCC toolchain used by the spec (`%define TOOLCHAIN GCC`) derives its X64 CC flags from `GCC_X64_CC_FLAGS`. So pinning `GCCNOLTO_X64_CC_FLAGS` propagates the ISA baseline to exactly the toolchain that builds OVMF, preserving the original patch's intent. ## Fix Re-anchored the single one-line hunk onto the renamed/relocated `GCCNOLTO_X64_CC_FLAGS` line, inserting `-march=x86-64-v2` immediately after `-m64` (identical placement to the original). The diff body was regenerated with git so the hunk header/counts are correct by construction. The commit message/body is unchanged. ## Verification `patch -p1 --fuzz=0 --dry-run` and a real `patch -p1 --fuzz=0` against a pristine copy of `src/` both apply cleanly (`exit=0`) and yield the intended `-march=x86-64-v2` pin.
eabdullin deleted branch triage/fix-edk2-manual-86018 2026-09-24 07:46:28 +00:00
Sign in to join this conversation.
Reviewers
No reviewers
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
autopatch/edk2!1
No description provided.