report file sizes in Eden's debugInodeStatus() call

Summary: Remove file size stat() call from cli and move logic into server-side code

Reviewed By: chadaustin

Differential Revision: D17080186

fbshipit-source-id: 882d1c01db841e7b13b7659f47091e6b5c46ab57

eden/cli/debug.py

@@ -404,14 +404,11 @@ class FileStatsCMD(Subcmd):
         parser.add_argument("--sizes", action="store_true", help="Compute file sizes")
 
     def make_file_entries(
-        self, base_path: str, paths: List[str], get_sizes: bool
+        self, base_path: str, paths: List[Tuple[str, Optional[int]]], get_sizes: bool
     ) -> List[Dict[str, Any]]:
         if get_sizes:
-            return [
-                {"path": path, "size": os.path.getsize(os.path.join(base_path, path))}
-                for path in paths
-            ]
-        return [{"path": path} for path in paths]
+            return [{"path": path, "size": file_size} for (path, file_size) in paths]
+        return [{"path": path} for (path, _) in paths]
 
     def run(self, args: argparse.Namespace) -> int:
         request_root = args.path

eden/cli/test/util_test.py

@@ -41,6 +41,7 @@ class UtilTest(unittest.TestCase):
                     loaded=True,
                     materialized=False,
                     hash=b"1abc",
+                    fileSize=300,
                 ),
                 TreeInodeEntryDebugInfo(
                     name=b"written_file",
@@ -48,6 +49,7 @@ class UtilTest(unittest.TestCase):
                     mode=stat.S_IFREG,
                     loaded=True,
                     materialized=True,
+                    fileSize=400,
                 ),
             ],
             refcount=0,
@@ -58,8 +60,8 @@ class UtilTest(unittest.TestCase):
         read_files, written_files = util.split_inodes_by_operation_type(
             self.INODE_RESULTS_0
         )
-        self.assertListEqual(read_files, ["some_path/d1/read_file"])
-        self.assertListEqual(written_files, ["some_path/d1/written_file"])
+        self.assertListEqual(read_files, [("some_path/d1/read_file", 300)])
+        self.assertListEqual(written_files, [("some_path/d1/written_file", 400)])
 
     INODE_RESULTS_1 = [
         TreeInodeDebugInfo(
@@ -75,6 +77,7 @@ class UtilTest(unittest.TestCase):
                     loaded=True,
                     materialized=False,
                     hash=b"1abc",
+                    fileSize=300,
                 ),
                 TreeInodeEntryDebugInfo(
                     name=b"written_file",
@@ -82,6 +85,7 @@ class UtilTest(unittest.TestCase):
                     mode=stat.S_IFDIR,
                     loaded=True,
                     materialized=True,
+                    fileSize=400,
                 ),
             ],
             refcount=0,
@@ -109,6 +113,7 @@ class UtilTest(unittest.TestCase):
                     loaded=True,
                     materialized=False,
                     hash=b"1abc",
+                    fileSize=300,
                 )
             ],
             refcount=0,
@@ -125,6 +130,7 @@ class UtilTest(unittest.TestCase):
                     mode=stat.S_IFREG,
                     loaded=True,
                     materialized=True,
+                    fileSize=400,
                 )
             ],
             refcount=0,
@@ -135,8 +141,8 @@ class UtilTest(unittest.TestCase):
         read_files, written_files = util.split_inodes_by_operation_type(
             self.INODE_RESULTS_2
         )
-        self.assertListEqual(read_files, ["some_path/d1/read_file"])
-        self.assertListEqual(written_files, ["some_path/d1/written_file"])
+        self.assertListEqual(read_files, [("some_path/d1/read_file", 300)])
+        self.assertListEqual(written_files, [("some_path/d1/written_file", 400)])
 
     INODE_RESULTS_3 = [
         TreeInodeDebugInfo(
@@ -152,6 +158,7 @@ class UtilTest(unittest.TestCase):
                     loaded=False,
                     materialized=False,
                     hash=b"1abc",
+                    fileSize=300,
                 ),
                 TreeInodeEntryDebugInfo(
                     name=b"written_file",
@@ -159,6 +166,7 @@ class UtilTest(unittest.TestCase):
                     mode=stat.S_IFREG,
                     loaded=False,
                     materialized=True,
+                    fileSize=400,
                 ),
             ],
             refcount=0,

eden/cli/util.py

@@ -511,21 +511,26 @@ def get_username() -> str:
 class LoadedNode(typing.NamedTuple):
     path: str
     is_write: bool
+    file_size: Optional[int]
 
 
 def split_inodes_by_operation_type(
     inode_results: typing.Sequence[TreeInodeDebugInfo]
-) -> typing.Tuple[typing.List[str], typing.List[str]]:
+) -> typing.Tuple[
+    typing.List[typing.Tuple[str, Optional[int]]],
+    typing.List[typing.Tuple[str, Optional[int]]],
+]:
     loaded_node_info = [
         LoadedNode(
             path=os.path.join(os.fsdecode(tree.path), os.fsdecode(n.name)),
             is_write=n.materialized or not n.hash,
+            file_size=n.fileSize,
         )
         for tree in inode_results
         for n in tree.entries
         if n.loaded and stat.S_IFMT(n.mode) == stat.S_IFREG
     ]
 
-    read_files = [o.path for o in loaded_node_info if not o.is_write]
-    written_files = [o.path for o in loaded_node_info if o.is_write]
+    read_files = [(o.path, o.file_size) for o in loaded_node_info if not o.is_write]
+    written_files = [(o.path, o.file_size) for o in loaded_node_info if o.is_write]
     return read_files, written_files

eden/fs/inodes/FileInode.h

@@ -223,6 +223,8 @@ class FileInode final : public InodeBaseMetadata<FileInodeState> {
 
   void fsync(bool datasync);
 
+  folly::Future<struct stat> stat();
+
  private:
   using State = FileInodeState;
   class LockedState;
@@ -367,8 +369,6 @@ class FileInode final : public InodeBaseMetadata<FileInodeState> {
       size_t numIovecs,
       off_t off);
 
-  folly::Future<struct stat> stat();
-
   /**
    * Update the st_blocks field in a stat structure based on the st_size value.
    */

eden/fs/inodes/TreeInode.cpp

@@ -3274,6 +3274,7 @@ void TreeInode::getDebugStatus(vector<TreeInodeDebugInfo>& results) const {
     }
   }
 
+  std::vector<folly::Future<std::pair<size_t, uint64_t>>> futures;
   for (const auto& childData : childInodes) {
     info.entries.emplace_back();
     auto& infoEntry = info.entries.back();
@@ -3299,8 +3300,23 @@ void TreeInode::getDebugStatus(vector<TreeInodeDebugInfo>& results) const {
       auto blobHash = childFile->getBlobHash();
       infoEntry.materialized = !blobHash.has_value();
       infoEntry.hash = thriftHash(blobHash);
+      futures.push_back(
+          childFile->stat().thenValue([i = info.entries.size() - 1](auto st) {
+            auto fileSize = st.st_size;
+            return std::make_pair(i, fileSize);
+          }));
     }
   }
+  auto fileSizeMappings = folly::collectAll(futures).get();
+  for (const auto& future : fileSizeMappings) {
+    auto [i, fileSize] = future.value();
+
+    auto& infoEntry = info.entries[i];
+    // We must use set_size here because size is
+    // optional and if it is set directly then it will
+    // not get serialized correctly.
+    infoEntry.set_fileSize(fileSize);
+  }
   results.push_back(info);
 
   // Recurse into all children directories after we finish building our own

eden/fs/service/eden.thrift

@@ -390,6 +390,10 @@ struct TreeInodeEntryDebugInfo {
    * control Blob or Tree.
    */
   6: BinaryHash hash
+  /**
+   * Size of the file in bytes, won't be set for directories
+  */
+  7: optional i64 fileSize
 }
 
 struct WorkingDirectoryParents {