merge: prefetch merging file contents

Summary: Batch prefetch contents of all files' contents involved in a merge. This might avoid serial content fetches in certain cases during rebase (with lots of files to be merged).

Reviewed By: quark-zju

Differential Revision: D65402133

fbshipit-source-id: 3f8a095ac89f73b71532daa51736304a37eec231

eden/scm/sapling/ext/remotefilelog/remotefilectx.py

@@ -16,7 +16,6 @@
 from ..extlib.phabricator import graphql
 from . import shallowutil
 
-
 propertycache = util.propertycache
 conduit = None
 FASTLOG_TIMEOUT_IN_SECS = 0.5

eden/scm/sapling/filemerge.py

@@ -1042,7 +1042,7 @@ def _makebackup(repo, ui, wctx, fcd, premerge):
         # so we don't disturb the working directory.
         relpath = back[len(repo.wvfs.base) + 1 :]
         if premerge:
-            wctx[relpath].write(fcd.data(), fcd.flags())
+            wctx[relpath].write(fcd, fcd.flags())
         return wctx[relpath]
     else:
         if premerge:

eden/scm/sapling/merge.py

@@ -1457,6 +1457,31 @@ def applyupdates(repo, actions, wctx, mctx, overwrite, labels=None, ancestors=No
     for m, l in actions.items():
         l.sort()
 
+    # Prefetch content for files to be merged to avoid serial lookups.
+    merge_prefetch = []
+    for f, args, msg in (
+        actions[ACTION_CHANGED_DELETED]
+        + actions[ACTION_DELETED_CHANGED]
+        + actions[ACTION_MERGE]
+    ):
+        f1, f2, fa, move, anc = args
+        if f1 is not None:
+            merge_prefetch.append(wctx[f1])
+        if f2 is not None:
+            merge_prefetch.append(mctx[f2])
+        actx = repo[anc]
+        if fa in actx:
+            merge_prefetch.append(actx[fa])
+    if merge_prefetch and hasattr(repo, "fileservice"):
+        repo.fileservice.prefetch(
+            [
+                (fc.path(), fc.filenode())
+                for fc in merge_prefetch
+                if fc.filenode() not in (None, nullid)
+            ],
+            fetchhistory=False,
+        )
+
     # These are m(erge) actions that aren't actually conflicts, such as remote copying a
     # file. We don't want to expose them to merge drivers since merge drivers might get
     # confused.

eden/scm/tests/test-rebase-file-fetches.t

@@ -46,7 +46,7 @@ Make sure we batch tree fetches well:
   TRACE tree_fetches: attrs=["content"] keys=["a/b@105dfd91"]
   TRACE file_fetches: attrs=["history"] length=Some(1) keys=["a/b/c2/file"]
 
-FIXME Make sure we batch fetch content for files needing merge:
+Make sure we batch fetch content for files needing merge:
   $ newserver server3
   $ drawdag <<EOS
   >      # C/bar = 2\n2\n3\n
@@ -60,6 +60,7 @@ FIXME Make sure we batch fetch content for files needing merge:
   $ newclientrepo client3 test:server3
   $ LOG=file_fetches=trace,tree_fetches=trace hg rebase -q -r $B -d $C
   TRACE tree_fetches: attrs=["content"] keys=["@52df54bd", "@9e73d36f", "@c0749e87"]
+  TRACE file_fetches: attrs=["content", "header", "aux"] keys=["bar", "bar", "foo", "foo"]
   TRACE file_fetches: attrs=["content", "header"] keys=["bar"]
   TRACE file_fetches: attrs=["content", "header"] keys=["bar"]
   TRACE file_fetches: attrs=["history"] length=Some(1) keys=["bar"]