|
1 | | -$Header: /cvsroot/pgsql/src/backend/storage/lmgr/README,v 1.2 1998/01/28 02:29:26 momjian Exp $ |
| 1 | +$Header: /cvsroot/pgsql/src/backend/storage/lmgr/README,v 1.3 1998/07/06 18:16:07 momjian Exp $ |
2 | 2 |
|
3 | | -This file is an attempt to save me (and future code maintainers) some |
4 | | -time and a lot of headaches. The existing lock manager code at the time |
5 | | -of this writing (June 16 1992) can best be described as confusing. The |
6 | | -complexity seems inherent in lock manager functionality, but variable |
7 | | -names chosen in the current implementation really confuse me everytime |
8 | | -I have to track down a bug. Also, what gets done where and by whom isn't |
9 | | -always clear.... |
| 3 | +There are two fundemental lock structures. Lock methods describe the |
| 4 | +locking behavior. We currently only support multi-level locking. Lock |
| 5 | +modes describe the mode of the lock(read/write or shared/exclusive). |
| 6 | +See src/tools/backend/index.html and src/include/storage/lock.h for more |
| 7 | +details. |
10 | 8 |
|
11 | | -Starting with the data structures the lock manager relies upon... |
12 | | - |
13 | | -(NOTE - these will undoubtedly change over time and it is likely |
14 | | -that this file won't always be updated along with the structs.) |
| 9 | +--------------------------------------------------------------------------- |
15 | 10 |
|
16 | 11 | The lock manager's LOCK: |
17 | 12 |
|
@@ -90,10 +85,3 @@ activeHolders - |
90 | 85 |
|
91 | 86 | --------------------------------------------------------------------------- |
92 | 87 |
|
93 | | -Locks are accessed in two ways. Each PROC structure has a lockQueue, |
94 | | -that is a circular linked list of LOCK pointers that this process holds |
95 | | -or is waiting on. |
96 | | - |
97 | | -Second, there is a hash table that can do a lookup by combined LOCK |
98 | | -address and transaction id(xid) which allows a process to see what |
99 | | -type of locks it holds on that table. |