Commit message (Collapse) | Author | Age | |
---|---|---|---|
* | Stage #1 of core breakup: move everything under lib | 2016-03-25 | |
| | |||
* | Fix formatting | 2016-03-04 | |
| | |||
* | Fix race in poll() based pollset | 2016-03-04 | |
| | | | | | | It was possible for entries in watchers[] to be deleted by another thread before we called begin_poll. I'd like to do something nicer than this eventually... but that'll be easier once we've got the polling loop cleaned up (which is currently WIP) | ||
* | Revert "Revert "Add an implementation firewall against pollset_set"" | 2016-02-25 | |
| | |||
* | Revert "Add an implementation firewall against pollset_set" | 2016-02-25 | |
| | |||
* | Change pollset mutex ownership | 2016-02-25 | |
| | |||
* | Provide an interface firewall between pollset and its implementations | 2016-02-19 | |
| | | | | | | | Starting to allow for >1 implementation of pollset within a binary. Do so without requiring an extra allocation for completion queues (which we could not tolerate). | ||
* | Updated copyrights | 2016-01-12 | |
| | |||
* | Spam cleanup | 2016-01-06 | |
| | |||
* | Remove pollset fd deletion path: its unused | 2015-12-11 | |
| | |||
* | stream_op cleanup: pollset changes | 2015-11-02 | |
| | |||
* | clang-format | 2015-10-02 | |
| | |||
* | Stabilize mac build | 2015-10-02 | |
| | |||
* | Completing wakeup story | 2015-10-01 | |
| | |||
* | Merge github.com:grpc/grpc into pfft | 2015-09-30 | |
|\ | |||
* | | Fix alarms | 2015-09-28 | |
| | | | | | | | | | | | | - make kick_poller() do something on POSIX - fix some conditions whereby alarms are held in a pollset exec context for too long - make channel_connectivity tests dependent on the correct behavior | ||
| * | Add some TODO comments for possible poll optimization | 2015-09-25 | |
|/ | |||
* | Merge pull request #3476 from vjpai/block_annotate | 2015-09-25 | |
|\ | | | | | Annotate blocking points | ||
| * | Move block_annotate from iomgr to support since it's used in other | 2015-09-24 | |
| | | | | | | | | core places besides iomgr | ||
| * | Put in blocking point annotations at places in the code where we may block ↵ | 2015-09-24 | |
| | | | | | | | | for reasons other than synchronization | ||
* | | clang-format all core files | 2015-09-22 | |
| | | |||
* | | Move argument passing to start of list | 2015-09-22 | |
| | | |||
* | | Move arguments to the start of lists | 2015-09-22 | |
| | | |||
* | | indent pass to get logical source lines on one physical line | 2015-09-22 | |
| | | |||
* | | Rename call_list to closure_list | 2015-09-22 | |
| | | |||
* | | Call list progress | 2015-09-18 | |
|/ | |||
* | Mac build fixes | 2015-09-10 | |
| | |||
* | Enable -Wconversion | 2015-09-10 | |
| | |||
* | Merge github.com:grpc/grpc into y12kdm3 | 2015-08-18 | |
|\ | |||
| * | clang-format all source | 2015-08-18 | |
| | | |||
* | | Working towards a non-blocking API test | 2015-08-06 | |
|/ | |||
* | Save some iterations | 2015-07-31 | |
| | |||
* | Allow specific pollers to be woken | 2015-07-29 | |
| | | | | | | | | | | | | | | | | | | | Currently, if two threads call grpc_completion_queue_pluck on the same completion queue for different tags, there is a 50% chance that we deliver the completion wakeup to the wrong poller - forcing the correct poller to wait until its polling times out before it can return an event up to the application. This change tweaks our polling interfaces so that we can indeed wake a specific poller. Nothing has been performance tuned yet. It's definitely sub-optimal in a number of places. Wakeup file-descriptors should be recycled. We should have a path that avoids calling poll() followed by epoll(). We can probably live without it right at the second though. This code will fail on Windows at least (I'll do that port when I'm in the office and have a Windows machine). | ||
* | Fix a leaked lock, and make a debug-only detection for this class of bug | 2015-07-16 | |
| | |||
* | Hoist epoll_ctl outside of pollset lock | 2015-07-14 | |
| | |||
* | Fix a bug in the multipoll on poll path where we use a file descriptor thats ↵ | 2015-07-01 | |
| | | | | orphaned by mistake | ||
* | Remove return status from maybe_work. | 2015-06-22 | |
| | | | | | | | Since alarm checks may mutate work deadlines for pollsets, the value returned from maybe_work is meaningless. Instead, maybe pollset_work always return 1 if maybe_work is invoked, and then redo the deadline check _on the next call_ to pollset_work. | ||
* | Docs | 2015-06-18 | |
| | |||
* | Be better about timeout rounding | 2015-06-15 | |
| | |||
* | clang-format | 2015-06-05 | |
| | |||
* | Release all fd refs before signalling shutdown | 2015-06-01 | |
| | |||
* | fd refcount debugging | 2015-05-31 | |
| | |||
* | Merge branch 'but-maybe-i-want-to-poll' into we-dont-need-no-backup | 2015-05-29 | |
|\ | |||
| * | clang-format | 2015-05-29 | |
| | | |||
* | | Merge branch 'but-maybe-i-want-to-poll' into we-dont-need-no-backup | 2015-05-29 | |
|\| | |||
| * | Make it possible to run with poll() instead of epoll() | 2015-05-29 | |
| | | | | | | | | | | Mostly to facilitate testing, but maybe in the future we want to work on Linux systems with broken epoll support? | ||
* | | Refactor pollset_kick | 2015-05-28 | |
| | | | | | | | | | | Allow multiple threads to be polling Remove unnecessary windows stubs | ||
* | | Remove condition variable from pollset | 2015-05-28 | |
|/ | |||
* | More scalable unary polling | 2015-05-14 | |
| | | | | | | - admit only one poller for read and one for write at a time (poll is level triggered, so this avoids a thundering herd on each event) - wake only one poller when more pollers are needed, again avoiding a thundering herd | ||
* | Stop kicking ourselves | 2015-04-27 | |
| |