electron/patches/chromium/support_mixed_sandbox_with_zygote.patch
electron-roller[bot] 5513e66982
chore: bump chromium to 95.0.4612.5 (main) (#30503)
* chore: bump chromium in DEPS to 94.0.4604.0

* build: 3-way merge of chromium patches

* chore: bump chromium in DEPS to 94.0.4605.0

* build: 3-way merge of chromium patches

* 3076040: Reland Remove delete_children RemoveAllChildViews arg

Ref: https://chromium-review.googlesource.com/c/chromium/src/+/3076040

* 3069287: Remove the remaining uses and delete the deprecated API

Ref: https://chromium-review.googlesource.com/c/chromium/src/+/3069287

* 2297212: Replace RemoveWithoutPathExpansion(.*, nullptr) with Value::RemoveKey()

Refs: https://chromium-review.googlesource.com/c/chromium/src/+/2297212

Also: https://chromium-review.googlesource.com/c/chromium/src/+/3060296

* 3082756: Change transport_security_persister_path param to be a path to a file.

Refs: https://chromium-review.googlesource.com/c/chromium/src/+/3082756

> this CL intentionally changes the name of the parameter
> in the network context parameters and the order of the constructor
> parameters to ensure all callers update their code to pass a full
> file path rather than a path to a directory.

The 'path' in this diff is already an absolute path, coming from
`CHECK(base::PathService::Get(chrome::DIR_USER_DATA, &path_));` at
08ff1c2cbf/shell/browser/electron_browser_context.cc (L126)

* iwyu: network::mojom::HttpRawHeaderPair

* fixup! 3076040: Reland Remove delete_children RemoveAllChildViews arg

Missed one.

* 2999884: CodeHealth: Remove DictionaryValue::GetStringWithoutPathExpansion

Refs: https://chromium-review.googlesource.com/c/chromium/src/+/2999884
(example of replacing GetStringWithoutPathExpansion() w/FindStringKey())

Also: https://chromium-review.googlesource.com/c/chromium/src/+/3060296
(removal of DictionaryValue::GetStringWithoutPathExpansion)

* 3059260: Remove kSameSiteByDefaultCookies and kCookiesWithoutSameSiteMustBeSecure

Refs: https://chromium-review.googlesource.com/c/chromium/src/+/3059260

We had both of these in a 'disable_features' list. Since these feature have
been removed upstream, remove them from our disable list, too.

IMPORTANT: this commit should not be backported to older branches that
still have these features, because doing so would un-disable them.

* 2920890: Load reroute_info from download in-progress and history db back into DownloadItem.

Refs: https://chromium-review.googlesource.com/c/chromium/src/+/2920890

* 3039323: [Clipboard API] Clipboard Custom Formats implementation Part 5.

Refs: https://chromium-review.googlesource.com/c/chromium/src/+/3039323

* chore: bump chromium in DEPS to 94.0.4606.0

* 3084502: Add a new PrintRasterizePdfDpi policy.

Refs: https://chromium-review.googlesource.com/c/chromium/src/+/3084502

* chore: update patches

* chore: bump chromium in DEPS to 94.0.4606.3

* chore: bump chromium in DEPS to 95.0.4608.0

* chore: bump chromium in DEPS to 95.0.4609.0

* [DevTools] Remove report_raw_headers from network::ResourceRequest

https://chromium-review.googlesource.com/c/chromium/src/+/2856099

* Remove content::WebContentsObserver::OnInterfaceRequestFromFrame

https://chromium-review.googlesource.com/c/chromium/src/+/3092665

* Disable kDesktopCaptureMacV2

https://chromium-review.googlesource.com/c/chromium/src/+/3069272

* Add a new PrintRasterizePdfDpi policy.

https://chromium-review.googlesource.com/c/chromium/src/+/3084502

* chore: update patches

* chore: bump chromium in DEPS to 95.0.4609.3

* disable `use_lld` for macos

* chore: update patches

* Linux: use chrome_crashpad_handler instead of crashpad_handler

https://chromium-review.googlesource.com/c/chromium/src/+/3054290

* chore: fix lint

* Revert "[DevTools] Remove report_raw_headers from network::ResourceRequest"

This reverts commit 28f4da1582d046e96cb58f3cbb590503e89dfd0d.

* [DevTools] Remove report_raw_headers from network::ResourceRequest (Attempt #2)

https://chromium-review.googlesource.com/c/chromium/src/+/2856099

* DCHECK that predictor always has a non-empty NetworkIsolationKey.

https://chromium-review.googlesource.com/c/chromium/src/+/3067698

* Remove --no-untrusted-code-mitigations from //content and //gin

https://chromium-review.googlesource.com/c/chromium/src/+/3096585

* fixup! Remove kSameSiteByDefaultCookies and kCookiesWithoutSameSiteMustBeSecure

https://chromium-review.googlesource.com/c/chromium/src/+/3059260

* fixup! Remove kSameSiteByDefaultCookies and kCookiesWithoutSameSiteMustBeSecure

* Convert PrintManager to RenderFrameHostReceiverSet.

https://chromium-review.googlesource.com/c/chromium/src/+/3072019

* chore: bump chromium in DEPS to 95.0.4612.5

* chore: disable v8 oilpan

* [Compiler] Remove untrusted code mitigations.

https://chromium-review.googlesource.com/c/v8/v8/+/3045704

* Remove most FTP logic from services/network.

https://chromium-review.googlesource.com/c/chromium/src/+/3076119

* Rename scale_factor.h -> resource_scale_factor.h

https://chromium-review.googlesource.com/c/chromium/src/+/3057113

* [GURL -> SiteForCookies] extensions/

https://chromium-review.googlesource.com/c/chromium/src/+/3100825

* breadcrumbs: add desktop entry point

https://chromium-review.googlesource.com/c/chromium/src/+/3021746

* Move args_ to private in ExtensionFunction

https://chromium-review.googlesource.com/c/chromium/src/+/3076261

* chore: iwyu

* fixup! Remove kSameSiteByDefaultCookies and kCookiesWithoutSameSiteMustBeSecure

* Disable kDesktopCaptureMacV2

https://chromium-review.googlesource.com/c/chromium/src/+/3069272

* fixup! [Compiler] Remove untrusted code mitigations.

* fixup! Disable kDesktopCaptureMacV2

* Revert "chore: disable v8 oilpan"

This reverts commit 5d255cf1d8e8efbb906047937a713279e5f800d0.

* Reland "chore: disable v8 oilpan"

This reverts commit 1c252765b07a205560e7b5eed06de2605336e2d8.

The previous revert was to test on which platforms did the
heapsnapshot test actually fail.

* [Clipboard API] Clipboard Custom Formats implementation Part 5.

https://chromium-review.googlesource.com/c/chromium/src/+/3039323

* Convert ExtensionFrameHost to RenderFrameHostReceiverSet.

https://chromium-review.googlesource.com/c/chromium/src/+/3063358

* Convert PDFWebContentsHelper to RenderFrameHostReceiverSet.

https://chromium-review.googlesource.com/c/chromium/src/+/3049751

* [Underscore Migration] Migrate ui/legacy

https://chromium-review.googlesource.com/c/devtools/devtools-frontend/+/3093160

* chore: remove unknown permission error

* chore: fix lint

* chore: ignore -Wunreachable-code-return for node deps/

* fixup! chore: ignore -Wunreachable-code-return for node deps/

* fix: windows build

* fix: build dependency

Dependency was missed in cbeae20438

* 3108669: arm,dsp: Fix 8bpp Dct64_NEON().

https://chromium-review.googlesource.com/c/codecs/libgav1/+/3108669

* chore: revert libgav1 roll

* Revert "3108669: arm,dsp: Fix 8bpp Dct64_NEON()."

This reverts commit 7ed31323127aac8ba2eaff9cae6c9be9a4954f33.

* Revert "chore: revert libgav1 roll"

This reverts commit 084a490d298811267316c786762fe7aa91b6318d.

* chore: revert clang roll

* chore:  Fix -Wunreachable-code-aggressive warnings in arm and arm64 code

Co-authored-by: electron-roller[bot] <84116207+electron-roller[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
Co-authored-by: deepak1556 <hop2deep@gmail.com>
Co-authored-by: mlaurencin <mlaurencin@electronjs.org>
Co-authored-by: John Kleinschmidt <jkleinsc@electronjs.org>
2021-08-23 20:52:17 -04:00

84 lines
3.6 KiB
Diff

From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Jeremy Apthorp <nornagon@nornagon.net>
Date: Wed, 28 Nov 2018 13:20:27 -0800
Subject: support_mixed_sandbox_with_zygote.patch
On Linux, Chromium launches all new renderer processes via a "zygote"
process which has the sandbox pre-initialized (see
//docs/linux_zygote.md). In order to support mixed-sandbox mode, in
which some renderers are launched with the sandbox engaged and others
without it, we need the option to launch non-sandboxed renderers without
going through the zygote.
Chromium already supports a `--no-zygote` flag, but it turns off the
zygote completely, and thus also disables sandboxing. This patch allows
the `--no-zygote` flag to affect renderer processes on a case-by-case
basis, checking immediately prior to launch whether to go through the
zygote or not based on the command-line of the to-be-launched renderer.
This patch could conceivably be upstreamed, as it does not affect
production Chromium (which does not use the `--no-zygote` flag).
However, the patch would need to be reviewed by the security team, as it
does touch a security-sensitive class.
diff --git a/content/browser/renderer_host/render_process_host_impl.cc b/content/browser/renderer_host/render_process_host_impl.cc
index 80826440781926cb6a244c62117f8e2964c701d0..8c61ce5f2035a63675d17fda429b839c1e1e8e3b 100644
--- a/content/browser/renderer_host/render_process_host_impl.cc
+++ b/content/browser/renderer_host/render_process_host_impl.cc
@@ -431,10 +431,18 @@ class RendererSandboxedProcessLauncherDelegate
public:
RendererSandboxedProcessLauncherDelegate() = default;
+#if BUILDFLAG(USE_ZYGOTE_HANDLE)
+ RendererSandboxedProcessLauncherDelegate(bool use_zygote):
+ use_zygote_(use_zygote) {}
+#endif
+
~RendererSandboxedProcessLauncherDelegate() override = default;
#if BUILDFLAG(USE_ZYGOTE_HANDLE)
ZygoteHandle GetZygote() override {
+ if (!use_zygote_) {
+ return nullptr;
+ }
const base::CommandLine& browser_command_line =
*base::CommandLine::ForCurrentProcess();
base::CommandLine::StringType renderer_prefix =
@@ -452,6 +460,11 @@ class RendererSandboxedProcessLauncherDelegate
sandbox::policy::SandboxType GetSandboxType() override {
return sandbox::policy::SandboxType::kRenderer;
}
+
+ private:
+#if BUILDFLAG(USE_ZYGOTE_HANDLE)
+ bool use_zygote_ = true;
+#endif
};
#if defined(OS_WIN)
@@ -463,6 +476,9 @@ class RendererSandboxedProcessLauncherDelegateWin
bool is_jit_disabled)
: renderer_code_integrity_enabled_(
GetContentClient()->browser()->IsRendererCodeIntegrityEnabled()) {
+#if BUILDFLAG(USE_ZYGOTE_HANDLE)
+ use_zygote_ = !cmd_line->HasSwitch(switches::kNoZygote);
+#endif
if (is_jit_disabled) {
dynamic_code_can_be_disabled_ = true;
return;
@@ -2014,9 +2030,15 @@ bool RenderProcessHostImpl::Init() {
std::unique_ptr<SandboxedProcessLauncherDelegate> sandbox_delegate =
std::make_unique<RendererSandboxedProcessLauncherDelegateWin>(
cmd_line.get(), IsJitDisabled());
+#else
+#if BUILDFLAG(USE_ZYGOTE_HANDLE)
+ bool use_zygote = !cmd_line->HasSwitch(switches::kNoZygote);
+ std::unique_ptr<SandboxedProcessLauncherDelegate> sandbox_delegate =
+ std::make_unique<RendererSandboxedProcessLauncherDelegate>(use_zygote);
#else
std::unique_ptr<SandboxedProcessLauncherDelegate> sandbox_delegate =
std::make_unique<RendererSandboxedProcessLauncherDelegate>();
+#endif
#endif
// Spawn the child process asynchronously to avoid blocking the UI thread.
// As long as there's no renderer prefix, we can use the zygote process