webnet/tailscale
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

ipn/ipnlocal: replace the LockedOnEntry pattern with conventional lock/unlock discipline (#16925)

Browse Source 
There are several methods within the LocalBackend that used an unusual and
error-prone lock discipline whereby they require the caller to hold the backend
mutex on entry, but release it on the way out.

In #11650 we added some support code to make this pattern more visible.
Now it is time to eliminate the pattern (at least within this package).
This is intended to produce no semantic changes, though I am relying on
integration tests and careful inspection to achieve that.

To the extent possible I preserved the existing control flow. In a few places,
however, I replaced this with an unlock/lock closure. This means we will
sometimes reacquire a lock only to release it again one frame up the stack, but
these operations are not performance sensitive and the legibility gain seems
worthwhile.

We can probably also pull some of these out into separate methods, but I did
not do that here so as to avoid other variable scope changes that might be hard
to see. I would like to do some more cleanup separately.

As a follow-up, we could also remove the unlockOnce helper, but I did not do
that here either.

Updates #11649

Change-Id: I4c92d4536eca629cfcd6187528381c33f4d64e20
Signed-off-by: M. J. Fromberger <fromberger@tailscale.com>
This commit is contained in:
M. J. Fromberger
2025-08-23 02:07:22 -07:00
committed by GitHub
parent fa0e83ab4f
commit 6c8fef961e
3 changed files with 172 additions and 159 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ipn/ipnlocal/profiles.go
+7 -8
View File
@@ -180,7 +180,7 @@ func (pm *profileManager) SwitchToProfile(profile ipn.LoginProfileView) (cp ipn.
f(pm.currentProfile, pm.prefs, false)
}
// Do not call pm.extHost.NotifyProfileChange here; it is invoked in
// [LocalBackend.resetForProfileChangeLockedOnEntry] after the netmap reset.
// [LocalBackend.resetForProfileChangeLocked] after the netmap reset.
// TODO(nickkhyl): Consider moving it here (or into the stateChangeCb handler
// in [LocalBackend]) once the profile/node state, including the netmap,
// is actually tied to the current profile.
@@ -359,9 +359,9 @@ func (pm *profileManager) SetPrefs(prefsIn ipn.PrefsView, np ipn.NetworkProfile)
// where prefsIn is the previous profile's prefs with an updated Persist, LoggedOut,
// WantRunning and possibly other fields. This may not be the desired behavior.
//
// Additionally, LocalBackend doesn't treat it as a proper profile switch, meaning that
// [LocalBackend.resetForProfileChangeLockedOnEntry] is not called and certain
// node/profile-specific state may not be reset as expected.
// Additionally, LocalBackend doesn't treat it as a proper profile switch,
// meaning that [LocalBackend.resetForProfileChangeLocked] is not called and
// certain node/profile-specific state may not be reset as expected.
//
// However, [profileManager] notifies [ipnext.Extension]s about the profile change,
// so features migrated from LocalBackend to external packages should not be affected.
@@ -494,10 +494,9 @@ func (pm *profileManager) setProfilePrefsNoPermCheck(profile ipn.LoginProfileVie
oldPrefs := pm.prefs
pm.prefs = clonedPrefs
// Sadly, profile prefs can be changed in multiple ways.
// It's pretty chaotic, and in many cases callers use
// unexported methods of the profile manager instead of
// going through [LocalBackend.setPrefsLockedOnEntry]
// Sadly, profile prefs can be changed in multiple ways. It's pretty
// chaotic, and in many cases callers use unexported methods of the
// profile manager instead of going through [LocalBackend.setPrefsLocked]
// or at least using [profileManager.SetPrefs].
//
// While we should definitely clean this up to improve