ShakeLooseVCaches needs to avoid CVInit vcaches

the code refactoring missed this case. sadly, it means we
can panic under heavy vcache contention.

FIXES 127645

Change-Id: I41929df066b8e3d447ba72e77be6777a306b8bf8
Reviewed-on: http://gerrit.openafs.org/2449
Reviewed-by: Simon Wilkinson <sxw@inf.ed.ac.uk>
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: Derrick Brashear <shadow@dementia.org>

diff --git a/src/afs/afs_vcache.c b/src/afs/afs_vcache.c
index af829510e7eebb58b75a6a633461c1599116085f..117c919fa46a5de4674a0a07742e7c4966886092 100644 (file)
--- a/src/afs/afs_vcache.c
+++ b/src/afs/afs_vcache.c
@@ -666,6 +666,8 @@ afs_ShakeLooseVCaches(afs_int32 anumber)
            refpanic("Exceeded pool of AFS vnodes(VLRU cycle?)");
        } else if (QNext(uq) != tq) {
            refpanic("VLRU inconsistent");
+       } else if (tvc->f.states & CVInit) {
+           continue;
        }
 
        fv_slept = 0;