summaryrefslogtreecommitdiffstats
path: root/dom/ipc/tests/browser_bug1646088.js
diff options
context:
space:
mode:
authorDaniel Baumann <daniel.baumann@progress-linux.org>2024-04-19 00:47:55 +0000
committerDaniel Baumann <daniel.baumann@progress-linux.org>2024-04-19 00:47:55 +0000
commit26a029d407be480d791972afb5975cf62c9360a6 (patch)
treef435a8308119effd964b339f76abb83a57c29483 /dom/ipc/tests/browser_bug1646088.js
parentInitial commit. (diff)
downloadfirefox-upstream/124.0.1.tar.xz
firefox-upstream/124.0.1.zip
Adding upstream version 124.0.1.upstream/124.0.1
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'dom/ipc/tests/browser_bug1646088.js')
-rw-r--r--dom/ipc/tests/browser_bug1646088.js71
1 files changed, 71 insertions, 0 deletions
diff --git a/dom/ipc/tests/browser_bug1646088.js b/dom/ipc/tests/browser_bug1646088.js
new file mode 100644
index 0000000000..18e9afb49e
--- /dev/null
+++ b/dom/ipc/tests/browser_bug1646088.js
@@ -0,0 +1,71 @@
+let dir = getChromeDir(getResolvedURI(gTestPath));
+dir.append("file_dummy.html");
+const uriString = Services.io.newFileURI(dir).spec;
+
+add_task(async function () {
+ await BrowserTestUtils.withNewTab(
+ "https://example.com",
+ async function (browser) {
+ // Override the browser's `prepareToChangeRemoteness` so that we can delay
+ // the process switch for an indefinite amount of time. This will allow us
+ // to control the timing of the resolve call to trigger the bug.
+ let prepareToChangeCalled = Promise.withResolvers();
+ let finishSwitch = Promise.withResolvers();
+ let oldPrepare = browser.prepareToChangeRemoteness;
+ browser.prepareToChangeRemoteness = async () => {
+ prepareToChangeCalled.resolve();
+ await oldPrepare.call(browser);
+ await finishSwitch.promise;
+ };
+
+ // Begin a process switch, which should cause `prepareToChangeRemoteness` to
+ // be called.
+ // NOTE: This used to avoid BrowserTestUtils.loadURI, as that call would
+ // previously eagerly perform a process switch meaning that the interesting
+ // codepath wouldn't be triggered. Nowadays the process switch codepath
+ // always happens during navigation as required by this test.
+ info("Beginning process switch into file URI process");
+ let browserLoaded = BrowserTestUtils.browserLoaded(browser);
+ BrowserTestUtils.startLoadingURIString(browser, uriString);
+ await prepareToChangeCalled.promise;
+
+ // The tab we opened is now midway through process switching. Open another
+ // browser within the same tab, and immediately close it after the load
+ // finishes.
+ info("Creating new tab loaded in file URI process");
+ let fileProcess;
+ let browserParentDestroyed = Promise.withResolvers();
+ await BrowserTestUtils.withNewTab(
+ uriString,
+ async function (otherBrowser) {
+ let remoteTab = otherBrowser.frameLoader.remoteTab;
+ fileProcess = remoteTab.contentProcessId;
+ info("Loaded test URI in pid: " + fileProcess);
+
+ browserParentDestroyed.resolve(
+ TestUtils.topicObserved(
+ "ipc:browser-destroyed",
+ subject => subject === remoteTab
+ )
+ );
+ }
+ );
+ await browserParentDestroyed.promise;
+
+ // This browser has now been closed, which could cause the file content
+ // process to begin shutting down, despite us process switching into it.
+ // We can now allow the process switch to finish, and wait for the load to
+ // finish as well.
+ info("BrowserParent has been destroyed, finishing process switch");
+ finishSwitch.resolve();
+ await browserLoaded;
+
+ info("Load complete");
+ is(
+ browser.frameLoader.remoteTab.contentProcessId,
+ fileProcess,
+ "Should have loaded in the same file URI process"
+ );
+ }
+ );
+});