afs-analyze-avoid-potential-loop-20030423
authorDerrick Brashear <shadow@dementia.org>
Wed, 23 Apr 2003 05:31:14 +0000 (05:31 +0000)
committerDerrick Brashear <shadow@dementia.org>
Wed, 23 Apr 2003 05:31:14 +0000 (05:31 +0000)
commit42a1981fff5c9b3f4855e4f9200fb6a9de53642b
tree31de4f5f4f660d42b5ef321175f4aefe57bece76
parentea410c1c4d78c74419a300ab1ee1aee62ea5db17
afs-analyze-avoid-potential-loop-20030423

avoid potentially looping forever in e.g. afs_FetchStatus because
we exit afs_Analyze too early to increment busyCount, and despite having
areq->busyCount non-zero we have servers still not_busy which are already down
that we'll never talk to again. that combined with no conn means we never
touch most of afs_Analyze, exiting early with shouldRetry set.

incrementing busyCount whenever we exit with shouldRetry set should
ensure we can never end up looping forever.
src/afs/afs_analyze.c