[Devel] [PATCH] c/r: coordinator to report correct error on restart failure

[Oren Laadan]

If restart fails it is usually due to an error for a restoring task,
which is place in ctx->errno. Then the coordinator wakes up and sees
an -EINTR.

This patch changes the coordinator's behavior to report the error
value placed in ctx->errno (if an error occurred) rather than report
a confusing -EINTR.

Signed-off-by: Oren Laadan <orenl at cs.columbia.edu>
---
 checkpoint/restart.c |   15 +++++++++++++--
 1 files changed, 13 insertions(+), 2 deletions(-)

diff --git a/checkpoint/restart.c b/checkpoint/restart.c
index 5daadc4..9b75de8 100644
--- a/checkpoint/restart.c
+++ b/checkpoint/restart.c
@@ -711,7 +711,7 @@ static inline int is_task_active(struct ckpt_ctx *ctx, pid_t pid)
 }
 
 /* should not be called under write_lock_irq(&tasklist_lock) */
-static inline void _restore_notify_error(struct ckpt_ctx *ctx, int errno)
+static void _restore_notify_error(struct ckpt_ctx *ctx, int errno)
 {
 	/* first to fail: notify everyone (racy but harmless) */
 	if (!ckpt_test_ctx_error(ctx)) {
@@ -1263,9 +1263,20 @@ static int do_restore_coord(struct ckpt_ctx *ctx, pid_t pid)
 		post_restore_task();
 
 	restore_debug_error(ctx, ret);
-	if (ret < 0) {
+
+	if (ret < 0)
 		ckpt_set_ctx_error(ctx, ret);
+
+	if (ckpt_test_ctx_error(ctx)) {
 		destroy_descendants(ctx);
+		/*
+		 * If a restaring task (or we) reported an error, that set
+		 * out return value to that error. (Need the unlikely loop
+		 * because the error is recorded after the flag is set).
+		 */
+		while (!ctx->errno)
+			yield();
+		ret = ctx->errno;
 	} else {
 		ckpt_set_ctx_success(ctx);
 		wake_up_all(&ctx->waitq);
-- 
1.6.0.4

_______________________________________________
Containers mailing list
Containers at lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers