Re: [PATCH v3 5/7] MIPS: perf: Allocate per-core counters on demand

2018-05-17 Thread Matt Redfearn
Hi James, On 16/05/18 19:05, James Hogan wrote: On Fri, Apr 20, 2018 at 11:23:07AM +0100, Matt Redfearn wrote: Previously when performance counters are per-core, rather than per-thread, the number available were divided by 2 on detection, and the counters used by each thread in a core were

Re: [PATCH v3 5/7] MIPS: perf: Allocate per-core counters on demand

2018-05-17 Thread Matt Redfearn
Hi James, On 16/05/18 19:05, James Hogan wrote: On Fri, Apr 20, 2018 at 11:23:07AM +0100, Matt Redfearn wrote: Previously when performance counters are per-core, rather than per-thread, the number available were divided by 2 on detection, and the counters used by each thread in a core were

Re: [PATCH v3 5/7] MIPS: perf: Allocate per-core counters on demand

2018-05-16 Thread James Hogan
On Fri, Apr 20, 2018 at 11:23:07AM +0100, Matt Redfearn wrote: > Previously when performance counters are per-core, rather than > per-thread, the number available were divided by 2 on detection, and the > counters used by each thread in a core were "swizzled" to ensure > separation. However, this

Re: [PATCH v3 5/7] MIPS: perf: Allocate per-core counters on demand

2018-05-16 Thread James Hogan
On Fri, Apr 20, 2018 at 11:23:07AM +0100, Matt Redfearn wrote: > Previously when performance counters are per-core, rather than > per-thread, the number available were divided by 2 on detection, and the > counters used by each thread in a core were "swizzled" to ensure > separation. However, this

[PATCH v3 5/7] MIPS: perf: Allocate per-core counters on demand

2018-04-20 Thread Matt Redfearn
Previously when performance counters are per-core, rather than per-thread, the number available were divided by 2 on detection, and the counters used by each thread in a core were "swizzled" to ensure separation. However, this solution is suboptimal since it relies on a couple of assumptions: a)

[PATCH v3 5/7] MIPS: perf: Allocate per-core counters on demand

2018-04-20 Thread Matt Redfearn
Previously when performance counters are per-core, rather than per-thread, the number available were divided by 2 on detection, and the counters used by each thread in a core were "swizzled" to ensure separation. However, this solution is suboptimal since it relies on a couple of assumptions: a)