Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix retry race condition that can lead to double decrementing inFlightSubStreams and so miss calling closed #11026

Merged
merged 3 commits into from
Mar 22, 2024
Merged

Fix retry race condition that can lead to double decrementing inFlightSubStreams and so miss calling closed #11026

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
736edcb
Fix retry race condition that can lead to double decrementing inFligh…
larry-safran Mar 21, 2024
ba5e4e0
Handle case where the race condition with cancel causes the decrement…
larry-safran Mar 21, 2024
6000630
Update comment about using parent instead of child locks to be on the…
larry-safran Mar 21, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
52 changes: 35 additions & 17 deletions core/src/main/java/io/grpc/internal/RetriableStream.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -149,11 +149,10 @@
this.throttle = throttle;
}

@SuppressWarnings("GuardedBy")
@SuppressWarnings("GuardedBy") // TODO(b/145386688) this.lock==ScheduledCancellor.lock so ok
@Nullable // null if already committed
@CheckReturnValue
private Runnable commit(final Substream winningSubstream) {

synchronized (lock) {
if (state.winningSubstream != null) {
return null;
Expand All @@ -165,10 +164,9 @@
// subtract the share of this RPC from channelBufferUsed.
channelBufferUsed.addAndGet(-perRpcBufferUsed);

final boolean wasCancelled = (scheduledRetry != null) ? scheduledRetry.isCancelled() : false;
final Future<?> retryFuture;
if (scheduledRetry != null) {
// TODO(b/145386688): This access should be guarded by 'this.scheduledRetry.lock'; instead
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why remove these? These look disabled by the @SuppressWarnings("GuardedBy")

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The way that the comments were done was intrusive and ugly. Changed them all to be on the SuppressWarnings annotation.

// found: 'this.lock'
retryFuture = scheduledRetry.markCancelled();
scheduledRetry = null;
} else {
Expand All @@ -177,8 +175,6 @@
// cancel the scheduled hedging if it is scheduled prior to the commitment
final Future<?> hedgingFuture;
if (scheduledHedging != null) {
// TODO(b/145386688): This access should be guarded by 'this.scheduledHedging.lock'; instead
// found: 'this.lock'
hedgingFuture = scheduledHedging.markCancelled();
scheduledHedging = null;
} else {
Expand All @@ -196,7 +192,21 @@
}
if (retryFuture != null) {
retryFuture.cancel(false);
if (!wasCancelled && inFlightSubStreams.decrementAndGet() == Integer.MIN_VALUE) {
assert savedCloseMasterListenerReason != null;
listenerSerializeExecutor.execute(
new Runnable() {

Check warning on line 198 in core/src/main/java/io/grpc/internal/RetriableStream.java

View check run for this annotation

Codecov / codecov/patch

core/src/main/java/io/grpc/internal/RetriableStream.java#L197-L198 

Added lines #L197 - L198 were not covered by tests
@Override
public void run() {
isClosed = true;
masterListener.closed(savedCloseMasterListenerReason.status,
savedCloseMasterListenerReason.progress,
savedCloseMasterListenerReason.metadata);
}

Check warning on line 205 in core/src/main/java/io/grpc/internal/RetriableStream.java

View check run for this annotation

Codecov / codecov/patch

core/src/main/java/io/grpc/internal/RetriableStream.java#L201-L205 

Added lines #L201 - L205 were not covered by tests
});
}
}

if (hedgingFuture != null) {
hedgingFuture.cancel(false);
}
Expand Down Expand Up @@ -415,7 +425,7 @@
drain(substream);
}

@SuppressWarnings("GuardedBy")
@SuppressWarnings("GuardedBy") // TODO(b/145386688) this.lock==ScheduledCancellor.lock so ok
private void pushbackHedging(@Nullable Integer delayMillis) {
if (delayMillis == null) {
return;
Expand All @@ -434,8 +444,6 @@
return;
}

// TODO(b/145386688): This access should be guarded by 'this.scheduledHedging.lock'; instead
// found: 'this.lock'
futureToBeCancelled = scheduledHedging.markCancelled();
scheduledHedging = future = new FutureCanceller(lock);
}
Expand Down Expand Up @@ -469,16 +477,13 @@
}
callExecutor.execute(
new Runnable() {
@SuppressWarnings("GuardedBy")
@SuppressWarnings("GuardedBy") //TODO(b/145386688) lock==ScheduledCancellor.lock so ok
@Override
public void run() {
boolean cancelled = false;
FutureCanceller future = null;

synchronized (lock) {
// TODO(b/145386688): This access should be guarded by
// 'HedgingRunnable.this.scheduledHedgingRef.lock'; instead found:
// 'RetriableStream.this.lock'
if (scheduledHedgingRef.isCancelled()) {
cancelled = true;
} else {
Expand Down Expand Up @@ -810,13 +815,11 @@
&& !state.hedgingFrozen;
}

@SuppressWarnings("GuardedBy")
@SuppressWarnings("GuardedBy") // TODO(b/145386688) this.lock==ScheduledCancellor.lock so ok
private void freezeHedging() {
Future<?> futureToBeCancelled = null;
synchronized (lock) {
if (scheduledHedging != null) {
// TODO(b/145386688): This access should be guarded by 'this.scheduledHedging.lock'; instead
// found: 'this.lock'
futureToBeCancelled = scheduledHedging.markCancelled();
scheduledHedging = null;
}
Expand Down Expand Up @@ -999,9 +1002,19 @@
synchronized (lock) {
scheduledRetry = scheduledRetryCopy = new FutureCanceller(lock);
}

class RetryBackoffRunnable implements Runnable {
@Override
@SuppressWarnings("FutureReturnValueIgnored")
public void run() {
synchronized (scheduledRetryCopy.lock) {
if (scheduledRetryCopy.isCancelled()) {
return;

Check warning on line 1012 in core/src/main/java/io/grpc/internal/RetriableStream.java

View check run for this annotation

Codecov / codecov/patch

core/src/main/java/io/grpc/internal/RetriableStream.java#L1012 

Added line #L1012 was not covered by tests
} else {
scheduledRetryCopy.markCancelled();
}
}

callExecutor.execute(
new Runnable() {
@Override
Expand Down Expand Up @@ -1563,11 +1576,16 @@
}

void setFuture(Future<?> future) {
boolean wasCancelled;
synchronized (lock) {
if (!cancelled) {
wasCancelled = cancelled;
if (!wasCancelled) {
this.future = future;
}
}
if (wasCancelled) {
future.cancel(false);

Check warning on line 1587 in core/src/main/java/io/grpc/internal/RetriableStream.java

View check run for this annotation

Codecov / codecov/patch

core/src/main/java/io/grpc/internal/RetriableStream.java#L1587 

Added line #L1587 was not covered by tests
}
}

@GuardedBy("lock")
Expand Down
2 changes: 2 additions & 0 deletions core/src/test/java/io/grpc/internal/RetriableStreamTest.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -705,6 +705,7 @@ public void retry_unretriableClosed_cancel() {
// cancel
retriableStream.cancel(Status.CANCELLED);
inOrder.verify(retriableStreamRecorder, never()).postCommit();
verify(masterListener, times(1)).closed(any(), any(), any());
}

@Test
Expand Down Expand Up @@ -733,6 +734,7 @@ public void retry_cancelWhileBackoff() {

verifyNoMoreInteractions(mockStream1);
verifyNoMoreInteractions(mockStream2);
verify(masterListener, times(1)).closed(any(), any(), any());
}

@Test
Expand Down
Loading