Transactional locking with read-write locks in transactional memory systems

• US 8,973,004 B2
• Filed: 06/26/2009
• Issued: 03/03/2015
• Est. Priority Date: 06/26/2009
• Status: Active Grant

• Current Assignee: Sun Microsystems Incorporated (Oracle Corporation)
• Original Assignee: Oracle America, Inc. (Oracle Corporation)
• Inventors: Dice, David, Shavit, Nir N.
• Primary Examiner(s): Kawsar, Abdullah Al
• Application Number:
  US12/492,654
• Publication Number:
  US 20100333096A1
• Time in Patent Office: 2,076 Days
• Field of Search: 718/102
• US Class Current: 718/102
• CPC Class Codes: G06F 12/00 Accessing, addressing or allocating within memory systems or architectures digital input from, or digital output to record carriers, e.g. to disk storage units, G06F3/06
• CPC Class Codes: G06F 12/14 Protection against unauthorised use of memory or access to memory
• CPC Class Codes: G06F 2209/523 Mode
• CPC Class Codes: G06F 9/467 Transactional memory G06F9/528 takes precedence

Abstract

A system and method for transactional memory using read-write locks is disclosed. Each of a plurality of shared memory areas is associated with a respective read-write lock, which includes a read-lock portion indicating whether any thread has a read-lock for read-only access to the memory area and a write-lock portion indicating whether any thread has a write-lock for write access to the memory area. A thread executing a group of memory access operations as an atomic transaction acquires the proper read or write permissions before performing a memory operation. To perform a read access, the thread attempts to obtain the corresponding read-lock and succeeds if no other thread holds a write-lock for the memory area. To perform a write-access, the thread attempts to obtain the corresponding write-lock and succeeds if no other thread holds a write-lock or read-lock for the memory area.

Claims

1. A computer-implemented method, comprising:

associating each of a plurality of shared memory areas with a respective read-write lock, wherein for each shared memory area the read-write lock includes a read-lock portion indicating whether any thread has a read-lock for read-only access to the associated shared memory area and a write-lock portion indicating whether any thread has a write-lock for write access to the associated shared memory area, wherein for each respective read-write lock, the read-lock portion comprises a plurality of slots, wherein for each slot;

a value of the slot is modifiable only by a respective thread and the value indicates whether the respective thread holds a read-lock for read-only access to the associated shared memory area;

a thread executing a set of program instructions including a plurality of memory access operations to shared memory to be completed as an atomic transaction, wherein for each memory access operation of the atomic transaction said executing comprises;

if the memory access is a read access;

attempting to obtain the read-lock portion of the read-write lock for the corresponding shared memory area, wherein the read-lock portion is obtained only if no other thread holds the write-lock portion of the read-write lock; and

performing the read access only after successfully obtaining the read-lock portion of the read-write lock; and

if the memory access is a write access;

attempting to obtain the write-lock portion of the read-write lock for the corresponding shared memory area, wherein the write-lock portion is obtained only if no other thread holds the write-lock portion of the read-write lock, wherein the write-lock portion is obtained if no other thread holds the write-lock portion while one or more other threads hold the read-lock portion of the read-write lock; and

performing the write access only after successfully obtaining the write-lock portion of the read-write lock and no other thread holds the read-lock portion of the read-write lock.

2. The computer-implemented method of claim 1, wherein for each of the plurality of shared memory areas, according to the respective associated read-write lock multiple threads can concurrently have read access but only one thread can have write access at any given time.

3. The computer-implemented method of claim 1, wherein said executing further comprises committing the transaction, wherein said committing comprises releasing all read-write locks held by the thread for the transaction.

4. The computer-implemented method of claim 1, wherein said executing further comprises aborting the transaction if said attempting to obtain the read-lock portion or said attempting to obtain the write-lock portion is unsuccessful after a time-out criterion.

5. The computer-implemented method of claim 4, wherein said executing further comprises maintaining a read log for each read access performed for the transaction and maintaining a write log for each write access performed for the transaction, wherein said aborting comprises using the write log to undo all write accesses to shared memory performed for the transaction.

6. The computer-implemented method of claim 1, further comprising specifying the transaction as an irrevocable transaction, wherein all other transactions accessing the same shared memory areas as the irrevocable transaction are revocable so that any deadlock between the irrevocable transaction and one or more other transactions in obtaining one of the read-write locks will be resolved in favor of the irrevocable transaction.

7. The computer-implemented method of claim 6, wherein said specifying the transaction as an irrevocable transaction is performed in response to detecting a lack of progress by the transaction due to the transaction having been aborted one or more times.

8. The computer-implemented method of claim 1, wherein said executing further comprises privatizing a node of a data structure in the shared memory for exclusive use by the thread, wherein said privatizing does not require any additional operations other than, for each shared memory area of the node, obtaining the write-lock portion of the respective associated read-write lock and waiting until no other thread holds the read-lock portion of the read-write lock before performing any write access to that shared memory area.

9. A non-transitory computer-readable storage medium storing program instructions executable by one or more processors to implement:

associating each of a plurality of shared memory areas with a respective read-write lock, wherein for each shared memory area the read-write lock includes a read-lock portion indicating whether any thread has a read-lock for read-only access to the associated shared memory area and a write-lock portion indicating whether any thread has a write-lock for write access to the associated shared memory area, wherein for each respective read-write lock, the read-lock portion comprises a plurality of slots, wherein for each slot;

a value of the slot is modifiable only by a respective thread and the value indicates whether the respective thread holds a read-lock for read-only access to the associated shared memory area;

a thread executing a set of program instructions including a plurality of memory access operations to shared memory to be completed as an atomic transaction, wherein for each memory access operation of the atomic transaction said executing comprises;

if the memory access is a read access;

attempting to obtain the read-lock portion of the read-write lock for the corresponding shared memory area, wherein the read-lock portion is obtained only if no other thread holds the write-lock portion of the read-write lock; and

performing the read access only after successfully obtaining the read-lock portion of the read-write lock; and

if the memory access is a write access;

attempting to obtain the write-lock portion of the read-write lock for the corresponding shared memory area, wherein the write-lock portion is obtained only if no other thread holds the write-lock portion of the read-write lock even if one or more other threads hold the read-lock portion of the read-write lock; and

performing the write access only after successfully obtaining the write-lock portion of the read-write lock and no other thread holds the read-lock portion of the read-write lock.

10. The non-transitory computer-readable storage medium of claim 9, wherein for each of the plurality of shared memory areas, according to the respective associated read-write lock multiple threads can concurrently have read access but only one thread can have write access at any given time.

11. The non-transitory computer-readable storage medium of claim 9, wherein said executing further comprises aborting the transaction if said attempting to obtain the read-lock portion or said attempting to obtain the write-lock portion is unsuccessful after a time-out criterion.

12. The non-transitory computer-readable storage medium of claim 11, wherein said executing further comprises maintaining a read log for each read access performed for the transaction and maintaining a write log for each write access performed for the transaction, wherein said aborting comprises using the write log to undo all write accesses to shared memory performed for the transaction.

13. The non-transitory computer-readable storage medium of claim 9, wherein the program instructions are further executable to implement:

specifying the transaction as an irrevocable transaction, wherein all other transactions accessing the same shared memory areas as the irrevocable transaction are revocable so that any deadlock between the irrevocable transaction and one or more other transactions in obtaining one of the read-write locks will be resolved in favor of the irrevocable transaction.

14. The non-transitory computer-readable storage medium of claim 13, wherein said specifying the transaction as an irrevocable transaction is performed in response to detecting a lack of progress by the transaction due to the transaction having been aborted one or more times.

15. The non-transitory computer-readable storage medium of claim 9, wherein said executing further comprises privatizing a node of a data structure in the shared memory for exclusive use by the thread, wherein said privatizing does not require any additional operations other than, for each shared memory area of the node, obtaining the write-lock portion of the respective associated read-write lock and waiting until no other thread holds the read-lock portion of the read-write lock before performing any write access to that shared memory area.

16. A computer system, comprising:

one or more processors;

a memory coupled to the one or more processors storing program instructions executable by the one or more processors to implement;

associating each of a plurality of shared memory areas with a respective read-write lock, wherein for each shared memory area the read-write lock includes a read-lock portion indicating whether any thread has a read-lock for read-only access to the associated shared memory area and a write-lock portion indicating whether any thread has a write-lock for write access to the associated shared memory area, wherein for each respective read-write lock, the read-lock portion comprises a plurality of slots, wherein for each slot;

a value of the slot is modifiable only by a respective thread and the value indicates whether the respective thread holds a read-lock for read-only access to the associated shared memory area;

a thread executing a set of program instructions including a plurality of memory access operations to shared memory to be completed as an atomic transaction, wherein for each memory access operation of the atomic transaction said executing comprises;

if the memory access is a read access;

attempting to obtain the read-lock portion of the read-write lock for the corresponding shared memory area, wherein the read-lock portion is obtained only if no other thread holds the write-lock portion of the read-write lock; and

performing the read access only after successfully obtaining the read-lock portion of the read-write lock; and

if the memory access is a write access;

attempting to obtain the write-lock portion of the read-write lock for the corresponding shared memory area, wherein the write-lock portion is obtained only if no other thread holds the write-lock portion of the read-write lock even if one or more other threads hold the read-lock portion of the read-write lock; and

performing the write access only after successfully obtaining the write-lock portion of the read-write lock and no other thread holds the read-lock portion of the read-write lock.

17. The system of claim 16, wherein said executing further comprises:

maintaining a read log for each read access performed for the transaction and maintaining a write log for each write access performed for the transaction;

aborting the transaction if said attempting to obtain the read-lock portion or said attempting to obtain the write-lock portion is unsuccessful after a time-out criterion, wherein said aborting comprises using the write log to undo all write accesses to shared memory performed for the transaction.

18. The system of claim 16, wherein the program instructions are further executable to implement:

specifying the transaction as an irrevocable transaction, wherein all other transactions accessing the same shared memory areas as the irrevocable transaction are revocable so that any deadlock between the irrevocable transaction and one or more other transactions in obtaining one of the read-write locks will be resolved in favor of the irrevocable transaction.

19. The system of claim 16, wherein said executing further comprises privatizing a node of a data structure in the shared memory for exclusive use by the thread, wherein said privatizing does not require any additional operations other than, for each shared memory area of the node, obtaining the write-lock portion of the respective associated read-write lock and waiting until no other thread holds the read-lock portion of the read-write lock before performing any write access to that shared memory area.