[LibOS] Spark meets error because of missing syscall fcntl(F_SETLK/F_SETLKW/F_GETLK)

[llly]

Description of the problem

Trying to run spark using Graphene-SGX and caught Exception Function not implemented.

Steps to reproduce

Run spark executor using Graphene-SGX.

Expected results

Can work.

Actual results

Got following exception:

20/09/23 12:52:12 ERROR Executor: Exception in task 0.0 in stage 0.0 (TID 0)
java.io.IOException: Function not implemented
        at java.base/sun.nio.ch.FileDispatcherImpl.lock0(Native Method)
        at java.base/sun.nio.ch.FileDispatcherImpl.lock(FileDispatcherImpl.java:96)
        at java.base/sun.nio.ch.FileChannelImpl.lock(FileChannelImpl.java:1119)
        at java.base/java.nio.channels.FileChannel.lock(FileChannel.java:1063)
        at org.apache.spark.util.Utils$.fetchFile(Utils.scala:492)
        at org.apache.spark.executor.Executor$$anonfun$org$apache$spark$executor$Executor$$updateDependencies$5.apply(Executor.scala:811)
        at org.apache.spark.executor.Executor$$anonfun$org$apache$spark$executor$Executor$$updateDependencies$5.apply(Executor.scala:803)
        at scala.collection.TraversableLike$WithFilter$$anonfun$foreach$1.apply(TraversableLike.scala:733)
        at scala.collection.mutable.HashMap$$anonfun$foreach$1.apply(HashMap.scala:130)
        at scala.collection.mutable.HashMap$$anonfun$foreach$1.apply(HashMap.scala:130)
        at scala.collection.mutable.HashTable$class.foreachEntry(HashTable.scala:236)
        at scala.collection.mutable.HashMap.foreachEntry(HashMap.scala:40)
        at scala.collection.mutable.HashMap.foreach(HashMap.scala:130)
        at scala.collection.TraversableLike$WithFilter.foreach(TraversableLike.scala:732)
        at org.apache.spark.executor.Executor.org$apache$spark$executor$Executor$$updateDependencies(Executor.scala:803)
        at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:375)
        at java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1128)
        at java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:628)
        at java.base/java.lang.Thread.run(Thread.java:834)

The Java code is https://github.com/openjdk/jdk/blob/d7a0fb9ebc898e76207c27166b81630e837a064a/src/java.base/unix/classes/sun/nio/ch/FileDispatcherImpl.java#L96
I searched the log and found failed syscall

shim_fcntl(278,F_SETLKW,0x9a8462db0) = -38

because fcntl(F_SETLK/F_SETLKW/F_GETLK) are not implemented.
https://github.com/oscarlab/graphene/blob/f94c72b8f55918aff6e7b0b45b1487f216039245/LibOS/shim/src/sys/shim_fcntl.c#L173-L175

[mkow]

This is a duplicate of #437, closing. If you want to highlight the fact that this specific workload needs this functionality please comment under #437.

[llly]

It blocks Spark usage in Graphene. I'm glad to see that we have put it into production TODO list.

[debin-yang]

Suggest give those TODO tasks higher priority and start addressing it earlier. This is a blocking issue of Graphene to be used by the targeted usage which uses Spark and Graphene. Workaround is identified and can work temporarily for POC of the usage but the workaround will not work for production of the usage.

[mkow]

@llly @debin-yang Please move the discussion to #437, so that we have everything about file locking in one place.