aboutsummaryrefslogtreecommitdiff
path: root/src
diff options
context:
space:
mode:
authorChristian Grothoff <christian@grothoff.org>2022-10-22 00:27:32 +0200
committerChristian Grothoff <christian@grothoff.org>2022-10-22 00:27:32 +0200
commit4524180ada7d49e2f3ea4356df4706f2c9b0841d (patch)
treeac44d30d13e60e52517886625e7f01268c605080 /src
parent3b90f9cae240623799917a839c018039dae506b0 (diff)
-document why re-locking here is safe -- at least right now
Diffstat (limited to 'src')
-rw-r--r--src/bank-lib/fakebank.c4
1 files changed, 4 insertions, 0 deletions
diff --git a/src/bank-lib/fakebank.c b/src/bank-lib/fakebank.c
index fa1475b7a..34b87be57 100644
--- a/src/bank-lib/fakebank.c
+++ b/src/bank-lib/fakebank.c
@@ -3630,6 +3630,10 @@ access_withdrawals_confirm (struct TALER_FAKEBANK_Handle *h,
TALER_EC_BANK_DUPLICATE_RESERVE_PUB_SUBJECT,
NULL);
}
+ /* Re-acquiring the lock and continuing to operate on 'wo'
+ is currently (!) acceptable because we NEVER free 'wo'
+ until shutdown. We may want to revise this if keeping
+ all withdraw operations in RAM becomes an issue... */
GNUNET_assert (0 ==
pthread_mutex_lock (&h->big_lock));
wo->confirmation_done = true;