Remove legacy __sync_fetch_and_add with __atomic_fetch_add #5
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…SEQ_CST was picked to enforce total ordering with all other __ATOMIC_SEQ_CST operations.) Signed-off-by: Chun-Hung Tseng <henrybear327@gmail.com>
|
Thank @henrybear327 for contributing! I have amended the git commit message. |
|
Thank you @jserv ! |
linD026
added a commit
to linD026/concurrent-programs
that referenced
this pull request
Apr 22, 2023
Currently, the reader_func is counting the number of grace periods based on the value of dut, which is updated by mthpc_rcu_replace_pointer. However, the value of dut actually represents the time we update the value, not the number of grace periods. Also, the original method might result in incorrect counting if someone tried to update the gp_idx while others who saw the same dut value with prev_count still depend on the old gp_idx to increase the counter. To fix the problem, instead of relying on the dut value to increase the gp_idx, we manually increase gp_idx on write side. Then, we can easily determine the gp on read side. For dut value, we simply check the old count value is not greater than the newest one. Additionally, since synchronize_rcu is quite slow, readers generally will pass through the critical section during the first grace period. To generate more realistic output, we add a delay on read side before entering the critical section. Before: 100 reader(s), 5 update run(s), 6 grace period(s) [grace period #0] 100 reader(s) [grace period sysprog21#1] 0 reader(s) [grace period sysprog21#2] 0 reader(s) [grace period sysprog21#3] 0 reader(s) [grace period sysprog21#4] 0 reader(s) [grace period sysprog21#5] 0 reader(s) After, we added a delay: 100 reader(s), 5 update run(s), 6 grace period(s) [grace period #0] 76 reader(s) [grace period sysprog21#1] 0 reader(s) [grace period sysprog21#2] 1 reader(s) [grace period sysprog21#3] 0 reader(s) [grace period sysprog21#4] 3 reader(s) [grace period sysprog21#5] 20 reader(s)
linD026
added a commit
to linD026/concurrent-programs
that referenced
this pull request
Apr 22, 2023
Currently, the reader_func is counting the number of grace periods based on the value of dut, which is updated by rcu_assign_pointer. However, the value of dut actually represents the time we update the value, not the number of grace periods. Also, the original method might result in incorrect counting if someone tried to update the gp_idx while others who saw the same dut value with prev_count still depend on the old gp_idx to increase the counter. To fix the problem, instead of relying on the dut value to increase the gp_idx, we manually increase gp_idx on write side. Then, we can easily determine the gp on read side. For dut value, we simply check the old count value is not greater than the newest one. Additionally, since synchronize_rcu is quite slow, readers generally will pass through the critical section during the first grace period. To generate more realistic output, we add a delay on read side before entering the critical section. Before: 100 reader(s), 5 update run(s), 6 grace period(s) [grace period #0] 100 reader(s) [grace period sysprog21#1] 0 reader(s) [grace period sysprog21#2] 0 reader(s) [grace period sysprog21#3] 0 reader(s) [grace period sysprog21#4] 0 reader(s) [grace period sysprog21#5] 0 reader(s) After, we added a delay: 100 reader(s), 5 update run(s), 6 grace period(s) [grace period #0] 76 reader(s) [grace period sysprog21#1] 0 reader(s) [grace period sysprog21#2] 1 reader(s) [grace period sysprog21#3] 0 reader(s) [grace period sysprog21#4] 3 reader(s) [grace period sysprog21#5] 20 reader(s)
linD026
added a commit
to linD026/concurrent-programs
that referenced
this pull request
Apr 22, 2023
Currently, the reader_func is counting the number of grace periods based on the value of dut, which is updated by rcu_assign_pointer. However, the value of dut actually represents the time we update the value, not the number of grace periods. Also, the original method might result in incorrect counting if someone tried to update the gp_idx while others who saw the same dut value with prev_count still depend on the old gp_idx to increase the counter. To fix the problem, instead of relying on the dut value to increase the gp_idx, we manually increase gp_idx on write side. Then, we can easily determine the gp on read side. For dut value, we simply check the old count value is not greater than the newest one. Additionally, since synchronize_rcu is quite slow, readers generally will pass through the critical section during the first grace period. To generate more realistic output, we add a delay on read side before entering the critical section. Before: 100 reader(s), 5 update run(s), 6 grace period(s) [grace period #0] 100 reader(s) [grace period sysprog21#1] 0 reader(s) [grace period sysprog21#2] 0 reader(s) [grace period sysprog21#3] 0 reader(s) [grace period sysprog21#4] 0 reader(s) [grace period sysprog21#5] 0 reader(s) After, we added a delay: 100 reader(s), 5 update run(s), 6 grace period(s) [grace period #0] 76 reader(s) [grace period sysprog21#1] 0 reader(s) [grace period sysprog21#2] 1 reader(s) [grace period sysprog21#3] 0 reader(s) [grace period sysprog21#4] 3 reader(s) [grace period sysprog21#5] 20 reader(s)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
__ATOMIC_SEQ_CST was picked to enforce total ordering with all other __ATOMIC_SEQ_CST operations.)