Differences

This shows you the differences between two versions of the page.

--- fourier:chromium_fourier [2026/04/28 22:06] – fourier: chromium_fourier: add 2026-04-28 superseded notice pointing at fourier:start markus_fritsche
+++ fourier:chromium_fourier [2026/04/29 04:16] (current) – fix HTML-entity over-escape (' -> ' ; " -> "); restores monospace markup markus_fritsche
@@ Line 4: / Line 4: @@
 > campaign has since grown into 4 sibling patch series + 1 kernel RFC;
 > the current 4-patch chromium-fourier shape and the end-to-end ohm
-> validation (~3.8× CPU reduction on &#039;&#039;bbb_1080p30_h264.mp4&#039;&#039;) are tracked
+> validation (~3.8× CPU reduction on ''bbb_1080p30_h264.mp4'') are tracked
-> in &#039;&#039;[[fourier:start|fourier:start]]&#039;&#039; and &#039;&#039;[[fourier:patch_series]]&#039;&#039;.
+> in ''[[fourier:start|fourier:start]]'' and ''[[fourier:patch_series]]''.
 ===== (a) What chromium-fourier gets us =====
-When the cross-compile build on **CT220** (&#039;&#039;chromium-builder-x86&#039;&#039; on data) finishes, we get a stock Chromium 147 aarch64 binary at &#039;&#039;out/Default/chrome&#039;&#039; plus support files (&#039;&#039;*.pak&#039;&#039;, &#039;&#039;locales/&#039;&#039;, &#039;&#039;icudtl.dat&#039;&#039;, &#039;&#039;v8_context_snapshot.bin&#039;&#039;, &#039;&#039;chromedriver&#039;&#039;, &#039;&#039;chrome_sandbox&#039;&#039;). Packaged as &#039;&#039;chromium-fourier-147.0.7727.116-1-aarch64.pkg.tar.zst&#039;&#039; via the marfrit PKGBUILD — total install ~250-400 MB.
+When the cross-compile build on **CT220** (''chromium-builder-x86'' on data) finishes, we get a stock Chromium 147 aarch64 binary at ''out/Default/chrome'' plus support files (''*.pak'', ''locales/'', ''icudtl.dat'', ''v8_context_snapshot.bin'', ''chromedriver'', ''chrome_sandbox''). Packaged as ''chromium-fourier-147.0.7727.116-1-aarch64.pkg.tar.zst'' via the marfrit PKGBUILD — total install ~250-400 MB.
-Built with &#039;&#039;use_v4l2_codec=true&#039;&#039;, &#039;&#039;use_v4lplugin=true&#039;&#039;, &#039;&#039;use_linux_v4l2_only=true&#039;&#039; baked in at compile time. **That&#039;s the part &#039;&#039;brave-bin&#039;&#039; does not have.** Whether those flags actually flip the right factory for our hantro stateless decoder is the test we run post-install on ohm; if not, the next iteration adds a small source patch to the factory check (original chromium-fourier &#039;&#039;NEXT.md&#039;&#039; plan B — much smaller scope than full libva-multiplanar).
+Built with ''use_v4l2_codec=true'', ''use_v4lplugin=true'', ''use_linux_v4l2_only=true'' baked in at compile time. **That's the part ''brave-bin'' does not have.** Whether those flags actually flip the right factory for our hantro stateless decoder is the test we run post-install on ohm; if not, the next iteration adds a small source patch to the factory check (original chromium-fourier ''NEXT.md'' plan B — much smaller scope than full libva-multiplanar).
-What we lose vs &#039;&#039;brave-bin&#039;&#039;: Brave Shields (ad-block), Brave Wallet, Brave-specific sync, Brave Search integration, Talk, Rewards. UX downgrade for power users; for &quot;video plays, browser browses&quot; it is a wash.
+What we lose vs ''brave-bin'': Brave Shields (ad-block), Brave Wallet, Brave-specific sync, Brave Search integration, Talk, Rewards. UX downgrade for power users; for "video plays, browser browses" it is a wash.
 ===== (b) The rug-pull on Brave =====
@@ Line 19: / Line 19: @@
 **Rug-pull does not reach the code that needs to change.**
-&#039;&#039;brave-bin&#039;&#039; links some things as shared libraries — those **CAN** be swapped:
+''brave-bin'' links some things as shared libraries — those **CAN** be swapped:
-  * &#039;&#039;libva&#039;&#039; — already done with &#039;&#039;LIBVA_DRIVER_NAME=v4l2_request&#039;&#039; pointing at our &#039;&#039;libva-v4l2-request-fourier&#039;&#039;
+  * ''libva'' — already done with ''LIBVA_DRIVER_NAME=v4l2_request'' pointing at our ''libva-v4l2-request-fourier''
-  * &#039;&#039;/opt/brave-bin/lib/libffmpeg.so&#039;&#039; — our &#039;&#039;ffmpeg-v4l2-request-git&#039;&#039; libavcodec could be dropped in there
+  * ''/opt/brave-bin/lib/libffmpeg.so'' — our ''ffmpeg-v4l2-request-git'' libavcodec could be dropped in there
-But the decoder factory gating is **statically baked** into the brave binary as compile-time &#039;&#039;BUILDFLAG(IS_CHROMEOS)&#039;&#039; guards, not a runtime function call. There is no symbol to &#039;&#039;LD_PRELOAD&#039;&#039; over. The &#039;&#039;V4L2VideoDecoder&#039;&#039; class exists in &#039;&#039;brave-bin&#039;&#039; (verified via &#039;&#039;strings&#039;&#039;), but the factory that decides &quot;should I instantiate it?&quot; is hard-coded IS_CHROMEOS-only at compile time. **A library swap cannot undo a &#039;&#039;#ifdef&#039;&#039;.**
+But the decoder factory gating is **statically baked** into the brave binary as compile-time ''BUILDFLAG(IS_CHROMEOS)'' guards, not a runtime function call. There is no symbol to ''LD_PRELOAD'' over. The ''V4L2VideoDecoder'' class exists in ''brave-bin'' (verified via ''strings''), but the factory that decides "should I instantiate it?" is hard-coded IS_CHROMEOS-only at compile time. **A library swap cannot undo a ''#ifdef''.**
-Library swap also does not help with the chromeos pipeline &#039;&#039;ImageProcessor&#039;&#039; failure that we hit on stock &#039;&#039;brave-bin&#039;&#039; — that is also static.
+Library swap also does not help with the chromeos pipeline ''ImageProcessor'' failure that we hit on stock ''brave-bin'' — that is also static.
 ===== Viable Brave paths (ascending effort) =====
-  - **Use &#039;&#039;chromium-fourier&#039;&#039;** — ship the result of today&#039;s build, accept the loss of Brave-specific features. Quickest.
+  - **Use ''chromium-fourier''** — ship the result of today's build, accept the loss of Brave-specific features. Quickest.
-  - **Rebuild Brave from source** with the same &#039;&#039;use_v4l2_codec&#039;&#039; gn args we use for &#039;&#039;chromium-fourier&#039;&#039;. Brave&#039;s source = chromium + a &#039;&#039;brave-core&#039;&#039; overlay; same gn build, same cross-compile environment on CT220, same chromium-fourier patches on top of brave-core&#039;s tree. Another 6-10 h build. Ships a &#039;&#039;brave-fourier&#039;&#039; next to &#039;&#039;chromium-fourier&#039;&#039;. Real Brave with V4L2 unlocked.
+  - **Rebuild Brave from source** with the same ''use_v4l2_codec'' gn args we use for ''chromium-fourier''. Brave's source = chromium + a ''brave-core'' overlay; same gn build, same cross-compile environment on CT220, same chromium-fourier patches on top of brave-core's tree. Another 6-10 h build. Ships a ''brave-fourier'' next to ''chromium-fourier''. Real Brave with V4L2 unlocked.
-  - **Binary-patch &#039;&#039;brave-bin&#039;&#039;** to flip the IS_CHROMEOS check at the call sites. Find the conditional jump in the &#039;&#039;V4L2VideoDecoder&#039;&#039; factory, NOP-out the chromeos gate. Fragile, has to be redone every Brave version, and the chromium guards are usually inlined past easy patching.
+  - **Binary-patch ''brave-bin''** to flip the IS_CHROMEOS check at the call sites. Find the conditional jump in the ''V4L2VideoDecoder'' factory, NOP-out the chromeos gate. Fragile, has to be redone every Brave version, and the chromium guards are usually inlined past easy patching.
   - **Wait for upstream Brave/Chromium** to enable V4L2 on Linux non-ChromeOS. Vaporware horizon.
-**Recommended:** (1) for ohm right now, (2) as the proper followup once &#039;&#039;chromium-fourier&#039;&#039; proves the V4L2 unlock actually engages decode. The real validation is &quot;does it decode?&quot; — once confirmed on &#039;&#039;chromium-fourier&#039;&#039;, the same patches are guaranteed to work on Brave&#039;s tree because Brave is chromium + UI/feature additions; the decoder code is unchanged.
+**Recommended:** (1) for ohm right now, (2) as the proper followup once ''chromium-fourier'' proves the V4L2 unlock actually engages decode. The real validation is "does it decode?" — once confirmed on ''chromium-fourier'', the same patches are guaranteed to work on Brave's tree because Brave is chromium + UI/feature additions; the decoder code is unchanged.
 ----