This patch enables AutoFDO build support for Rust code within the Linux kernel. This allows Rust code to be profiled and optimized based on the profile. The RUSTFLAGS variable was suffixed with *_AUTOFDO_CLANG to match the naming of the config option, which is called CONFIG_AUTOFDO_CLANG. This implementation has been verified in Android, first by inspecting the object files and confirming that they look correct. After that, it was verified as below: 1. Running the binderAddInts benchmark [1] with Rust Binder built as rust_binder.ko module, using a Pixel 9 Pro. 2. Collecting a profile on a Pixel 10 Pro XL using the app-launch benchmark, which starts different apps many times, on a device with Rust Binder as a built-in kernel module. (C Binder was not present on the device.) 3. Using the collected profile, run the binderAddInts benchmark again with Rust Binder built both as a rust_binder.ko module, and as a built-in kernel module. 4. In both cases, Rust Binder without AutoFDO was approximately 13% slower than the AutoFDO optimized version. Built-in vs .ko did not make a measurable performance difference. All of the above was verified in conjunction with my helpers inlining series [2], which confirmed that this worked correctly for helpers too once [3] was fixed in the helpers inlining series.
Unified · Split
| @@ -3,14 +3,18 @@ · 2 unmodified lines | ||
| 3 | 3 | # Enable available and selected Clang AutoFDO features. |
| 4 | 4 | |
| 5 | 5 | CFLAGS_AUTOFDO_CLANG := -fdebug-info-for-profiling -mllvm -enable-fs-discriminator=true -mllvm -improved-fs-discriminator=true |
| 6 | +RUSTFLAGS_AUTOFDO_CLANG := -Zdebug-info-for-profiling -Cllvm-args=-enable-fs-discriminator=true -Cllvm-args=-improved-fs-discriminator=true | |
| 6 | 7 | |
| 7 | 8 | ifndef CONFIG_DEBUG_INFO |
| 8 | 9 | CFLAGS_AUTOFDO_CLANG += -gmlt |
| 10 | + RUSTFLAGS_AUTOFDO_CLANG += -Cdebuginfo=line-tables-only | |
| 9 | 11 | endif |
| 10 | 12 | |
| 11 | 13 | ifdef CLANG_AUTOFDO_PROFILE |
| 12 | 14 | CFLAGS_AUTOFDO_CLANG += -fprofile-sample-use=$(CLANG_AUTOFDO_PROFILE) -ffunction-sections |
| 13 | 15 | CFLAGS_AUTOFDO_CLANG += -fsplit-machine-functions |
| 16 | + RUSTFLAGS_AUTOFDO_CLANG += -Zprofile-sample-use=$(CLANG_AUTOFDO_PROFILE) -Zfunction-sections=y | |
| 17 | + RUSTFLAGS_AUTOFDO_CLANG += -Cllvm-args=-split-machine-functions | |
| 14 | 18 | endif |
| 15 | 19 | |
| 16 | 20 | ifdef CONFIG_LTO_CLANG_THIN |
| @@ -21,4 +25,4 @@ ifdef CONFIG_LTO_CLANG_THIN · 4 unmodified lines | ||
| 21 | 25 | KBUILD_LDFLAGS += -plugin-opt=-split-machine-functions |
| 22 | 26 | endif |
| 23 | 27 | |
| 24 | −export CFLAGS_AUTOFDO_CLANG | |
| 28 | +export CFLAGS_AUTOFDO_CLANG RUSTFLAGS_AUTOFDO_CLANG | |
| @@ -123,6 +123,9 @@ ifeq ($(CONFIG_AUTOFDO_CLANG),y) · 122 unmodified lines | ||
| 123 | 123 | _c_flags += $(if $(patsubst n%,, \ |
| 124 | 124 | $(AUTOFDO_PROFILE_$(target-stem).o)$(AUTOFDO_PROFILE)$(is-kernel-object)), \ |
| 125 | 125 | $(CFLAGS_AUTOFDO_CLANG)) |
| 126 | +_rust_flags += $(if $(patsubst n%,, \ | |
| 127 | + $(AUTOFDO_PROFILE_$(target-stem).o)$(AUTOFDO_PROFILE)$(is-kernel-object)), \ | |
| 128 | + $(RUSTFLAGS_AUTOFDO_CLANG)) | |
| 126 | 129 | endif |
| 127 | 130 | |
| 128 | 131 | # |