Merge lp:~stewart/percona-server/5.5.27-plus-bugfixes into lp:percona-server/5.5
Status: | Work in progress |
---|---|
Proposed branch: | lp:~stewart/percona-server/5.5.27-plus-bugfixes |
Merge into: | lp:percona-server/5.5 |
Prerequisite: | lp:~stewart/percona-server/5.5.27 |
Diff against target: |
55 lines (+15/-6) 1 file modified
Percona-Server/storage/innobase/buf/buf0lru.c (+15/-6) |
To merge this branch: | bzr merge lp:~stewart/percona-server/5.5.27-plus-bugfixes |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Alexey Kopytov (community) | Needs Information on 2012-08-15 | ||
Laurynas Biveinis (community) | 2012-08-10 | Needs Information on 2012-08-10 | |
Review via email:
|
Description of the change
Include critical bugfixes based on Alexey's feedback to initial 5.5.27 branch. This merge req has that branch as a prerequisite.
http://
Stewart Smith (stewart) wrote : | # |
Laurynas Biveinis <email address hidden> writes:
> Review: Needs Information
>
> Does this apply to 5.1? (If it does, should be fixed there first and
> merged up...)
Haven't checked to be honest :)
>
> Why did you add the "impossible case" if (!block_mutex) ... ?
It was there in original code that was missed during merge.... it is a
bit odd though.
--
Stewart Smith
> > Does this apply to 5.1? (If it does, should be fixed there first and
> > merged up...)
>
> Haven't checked to be honest :)
OK if the fixes do not apply for 5.1. If they do, then 5.1 has to be fixed and upmerged.
Stewart Smith (stewart) wrote : | # |
> > > Does this apply to 5.1? (If it does, should be fixed there first and
> > > merged up...)
> >
> > Haven't checked to be honest :)
>
> OK if the fixes do not apply for 5.1. If they do, then 5.1 has to be fixed and
> upmerged.
FYI they don't seem to apply to 5.1 from code inspection.
Alexey Kopytov (akopytov) wrote : | # |
I don't the "It may be impossible case" code in the original fix. The upstream code is:
---
} else {
block_mutex = buf_page_
mutex_
if (bpage-
mutex_
/* We cannot remove this page during
this scan yet; maybe the system is
currently reading it in, or flushing
the modifications to the file */
all_freed = FALSE;
goto next_page;
}
}
---
Which appears to be rather different than what is now in Percona Server. Why?
Does this apply to 5.1? (If it does, should be fixed there first and merged up...)
Why did you add the "impossible case" if (!block_mutex) ... ?