| Bug #72520 | os_event_wait_time_low(): wait time calculation is messed up | ||
|---|---|---|---|
| Submitted: | 3 May 2014 4:53 | Modified: | 5 May 2014 5:08 |
| Reporter: | Inaam Rana (OCA) | Email Updates: | |
| Status: | Verified | Impact on me: | |
| Category: | MySQL Server: InnoDB storage engine | Severity: | S3 (Non-critical) |
| Version: | 5.5, 5.6, 5.7 | OS: | Any |
| Assigned to: | CPU Architecture: | Any | |
[5 May 2014 5:08]
MySQL Verification Team
Hello Inaam, Thank you for the bug report and contribution. Thanks, Umesh
[5 May 2014 5:41]
Sunny Bains
This was IIRC found and fixed by Yasufumi recently. I will check the latest 5.5 and 5.6 code too:
------------------------------------------------------------
revno: 7694
revision-id: yasufumi.kinoshita@oracle.com-20140313074852-0u31sml409l9llbg
parent: tatjana.nuernberg@oracle.com-20140313062138-e4hnkrn6dziemou2
committer: Yasufumi Kinoshita <yasufumi.kinoshita@oracle.com>
branch nick: mysql-trunk
timestamp: Thu 2014-03-13 08:48:52 +0100
message:
Bug #18386498 OS_EVENT_WAIT_TIME_LOW() RETURNS OS_SYNC_TIME_EXCEEDED BEFORE THE TIME ELAPSED
approved by Sunny over IM.
sunny@linux innobase> bzr diffp -c 7694
=== modified file 'storage/innobase/os/os0event.cc'
--- storage/innobase/os/os0event.cc revid:tatjana.nuernberg@oracle.com-20140313062138-e4hnkrn6dziemou2
+++ storage/innobase/os/os0event.cc revid:yasufumi.kinoshita@oracle.com-20140313074852-0u31sml409l9llbg
@@ -514,7 +514,7 @@ os_event::wait_time_low(
tv.tv_usec += time_in_usec;
if ((ulint) tv.tv_usec >= MICROSECS_IN_A_SECOND) {
- tv.tv_sec += time_in_usec / MICROSECS_IN_A_SECOND;
+ tv.tv_sec += tv.tv_usec / MICROSECS_IN_A_SECOND;
tv.tv_usec %= MICROSECS_IN_A_SECOND;
}
[8 Apr 2016 9:05]
zhai weixiang
This bug has been fixed, so it should be closed ?
$git show 5529e0f96d82419def08da8a37ed1cee04ca1114
commit 5529e0f96d82419def08da8a37ed1cee04ca1114
Author: Yasufumi Kinoshita <yasufumi.kinoshita@oracle.com>
Date: Thu Mar 13 08:48:52 2014 +0100
Bug #18386498 OS_EVENT_WAIT_TIME_LOW() RETURNS OS_SYNC_TIME_EXCEEDED BEFORE THE TIME ELAPSED
approved by Sunny over IM.
Description: os_event_wait_time_low() is used for timed wait. The calculation of absolute time for timeout is broken. 514 tv.tv_usec += time_in_usec; 515 516 if ((ulint) tv.tv_usec >= MICROSECS_IN_A_SECOND) { 517 tv.tv_sec += time_in_usec / MICROSECS_IN_A_SECOND; 518 tv.tv_usec %= MICROSECS_IN_A_SECOND; 519 } time_in_usec is time in microsecond that caller wants to set the timeout value to. Line 517 should not use time_in_usec, instead it should use tv.tv_usec. This has been present since 5.5 atleast but it didn't cause any grief because all caller pass multiples of seconds as wait time. In 5.7, however, we have page_cleaner thread which passes sub-second timeout values. When time_in_usec is less than MICROSECS_IN_A_SECOND, we'll never add to tv_tv_sec and we'll likely return much before than actual intended timeout. This can cause seriously erratic behavior of page_cleaner. How to repeat: see code. Suggested fix: === modified file 'storage/innobase/os/os0event.cc' --- storage/innobase/os/os0event.cc 2013-12-12 16:28:03 +0000 +++ storage/innobase/os/os0event.cc 2014-05-03 04:51:44 +0000 @@ -514,7 +514,7 @@ tv.tv_usec += time_in_usec; if ((ulint) tv.tv_usec >= MICROSECS_IN_A_SECOND) { - tv.tv_sec += time_in_usec / MICROSECS_IN_A_SECOND; + tv.tv_sec += tv.tv_usec / MICROSECS_IN_A_SECOND; tv.tv_usec %= MICROSECS_IN_A_SECOND; }