From fe9538746cb1c974c528df5f978243907dbfae1d Mon Sep 17 00:00:00 2001
From: "Kyle J. McKay" <mackyle@gmail.com>
Date: Wed, 2 Sep 2020 14:51:10 -0700
Subject: [PATCH] taskd/clone.sh: clean up more crud on exceeds limit failures

Refactor part of the cleanup_failed_clone function into a new,
internal _cleanup_failed_clone_bloat function that can be reused
and which is responsible for removing any "crud" files that
might potentially use a non-trivial amount of space.

Call the new _cleanup_failed_clone_bloat from the cleanup_failed_clone
to make it retain exactly the same semantics as before.

Add a call to the new _cleanup_failed_clone_bloat function from the
clear_all_objects_and_packs function so that any exceeds-limit
clones immediately remove potentially bloating files at the same
time they remove their objects and refs.

Signed-off-by: Kyle J. McKay <mackyle@gmail.com>
---
 taskd/clone.sh | 70 +++++++++++++++++++++++++++++++++++++++++-----------------
 1 file changed, 50 insertions(+), 20 deletions(-)

diff --git a/taskd/clone.sh b/taskd/clone.sh
index a9aad6b..94e543b 100755
--- a/taskd/clone.sh
+++ b/taskd/clone.sh
@@ -73,6 +73,52 @@ git_bzr_fetch() (
 	return $?
 )
 
+# removes any git-svn leftovers
+cleanup_git_svn_leftovers() {
+
+	# Remove any stale git-svn temp files
+	# The git-svn process creates temp files with random 10 character names
+	# in the root of $GIT_DIR.  Unfortunately they do not have a recognizable
+	# prefix, so we just have to kill any files with a 10-character name.
+	# All characters are chosen from
+	# [A-Za-z0-9_] so we can at least check that and fortunately the only
+	# collision is 'FETCH_HEAD' but that doesn't matter.
+	# There may also be temp files with a Git_ prefix as well.
+	_randchar='[A-Za-z0-9_]'
+	_randchar2="$_randchar$_randchar"
+	_randchar4="$_randchar2$_randchar2"
+	_randchar10="$_randchar4$_randchar4$_randchar2"
+	find -L . -maxdepth 1 -type f -name "$_randchar10" -exec rm -f '{}' + || :
+	find -L . -maxdepth 1 -type f -name "Git_*" -exec rm -f '{}' + || :
+}
+
+# removes all crud leftover from a previous failed clone attempt
+# only files that could potentially consume a non-trivial amount
+# of space are removed by this function
+# refs and objects are left unchanged by this function
+_cleanup_failed_clone_bloat() {
+
+	# Any pre-existing FETCH_HEAD from a previous clone failed or not is
+	# now garbage to be removed
+	rm -f FETCH_HEAD
+
+	# Remove any left-over svn dir from a previous failed attempt
+	rm -rf svn
+
+	# Remove any left-over .darcs dirs from a previous failed attempt
+	rm -rf *.darcs
+
+	# Remove any left-over repo.hg dir from a previous failed attempt
+	rm -rf repo.hg
+
+	# Remove any left-over import/export/temp files from a previous failed attempt
+	rm -f bfe-marks dfe-marks hg2git-heads hg2git-mapping hg2git-marks* hg2git-state \
+		gfi-marks gfi-packs .pkts-temp .refs-temp
+
+	# Remove any git-svn junk
+	cleanup_git_svn_leftovers
+}
+
 clear_all_objects_and_packs() {
 	if [ -d objects ]; then
 		# make sure the repository is not left broken
@@ -81,6 +127,7 @@ clear_all_objects_and_packs() {
 		find -H refs objects -type f -exec rm -f '{}' + >/dev/null 2>&1 || :
 		! [ -d htmlcache ] || { >htmlcache/changed; } 2>/dev/null || :
 	fi
+	_cleanup_failed_clone_bloat
 }
 
 exit_err=0
@@ -204,6 +251,9 @@ cleanup_git_svn_leftovers() {
 # removes all leftovers from a previous failed clone attempt
 cleanup_failed_clone() {
 
+	# Remove any left-over clone bloat
+	_cleanup_failed_clone_bloat
+
 	# Remove any left-over svn-remote.svn or remote.origin config
 	git config --remove-section svn-remote.svn 2>/dev/null || :
 	git config --remove-section remote.origin 2>/dev/null || :
@@ -217,29 +267,9 @@ cleanup_failed_clone() {
 		fi
 	done
 
-	# Any pre-existing FETCH_HEAD from a previous clone failed or not is
-	# now garbage to be removed
-	rm -f FETCH_HEAD
-
 	# Remove any stale ref locks
 	clear_stale_ref_locks
 
-	# Remove any left-over svn dir from a previous failed attempt
-	rm -rf svn
-
-	# Remove any left-over .darcs dirs from a previous failed attempt
-	rm -rf *.darcs
-
-	# Remove any left-over repo.hg dir from a previous failed attempt
-	rm -rf repo.hg
-
-	# Remove any left-over import/export/temp files from a previous failed attempt
-	rm -f bfe-marks dfe-marks hg2git-heads hg2git-mapping hg2git-marks* hg2git-state \
-		gfi-marks gfi-packs .pkts-temp .refs-temp
-
-	# Remove any git-svn junk
-	cleanup_git_svn_leftovers
-
 	# We want a gc right after the clone, so re-enable that just in case.
 	# There's a potential race where we could add it and gc.sh could remove
 	# it, but we'll reunset lastgc just before we remove .delaygc at the end.
-- 
2.11.4.GIT