Bug #79487: Provide non-Windows implementations for LF_BACKOFF #40
Conversation
|
Hi, thank you for your contribution. Please confirm this code is submitted under the terms of the OCA (Oracle's Contribution Agreement) you have previously signed by cutting and pasting the following text as a comment: |
|
See also: http://bugs.mysql.com/bug.php?id=72755 My comment would be that perhaps the UT_LOW_PRIORITY_CPU() and UT_RESUME_PRIORITY_CPU() macros should be called in my_yield_processor() rather than trusting the caller to also call them. Without these, the code generated for POWER in lf_hash and lf_alloc-pin will not do exactly what is intended - it'll delay, but it won't relax the CPU thread to give more core resources to other threads executing in the core. Then there's the issue of the rest of the code in lf_hash... which I'm not convinced is correct. |
Add LF_BACKOFF implementations for GCC and Sun Studio + some minor code unification with InnoDB.
|
Thanks, @stewart-ibm. I moved UT_LOW_PRIORITY_CPU() and UT_RESUME_PRIORITY_CPU() code from InnoDB to my_atomic.h and used it in my_yield_processor(). |
|
I confirm the code being submitted is offered under the terms of the OCA, and that I am authorized to contribute it. |
|
Hi, thank you for your contribution. Your code has been assigned to an internal queue. Please follow |
…te-system-table-metrics Setup explicit separate list for system table metrics
Add LF_BACKOFF implementations for GCC and Sun Studio + some minor code
unification with InnoDB.
This obviously needs further unification with InnoDB code (in particular, ut_delay()) and the contributed patch for http://bugs.mysql.com/bug.php?id=74832.
But I understand this patch will not be merged as is anyway, so I tried to make it as isolated as possible, but avoid code duplication with InnoDB at the same time.