RFR (S): 8228857: Refactor PlatformMonitor into PlatformMutex and PlatformMonitor

Daniel D. Daugherty daniel.daugherty at oracle.com
Sun Aug 11 12:57:34 UTC 2019


On 8/9/19 2:25 AM, David Holmes wrote:
> Okay here is v2 with Kim's non-copyable adjustments and an alternate 
> approach to the "impl" class so that the PlatformMutex doesn't have to 
> pay for the unused pthread_cond_t. Trade-off in this scheme is that 
> PlatformMonitor now has two "impl" pointers so consumes more space.
>
> Also made comment changes suggested by Dan.
>
> http://cr.openjdk.java.net/~dholmes/8228857/webrev.v2/

src/hotspot/os/posix/os_posix.cpp
     No comments.

src/hotspot/os/posix/os_posix.hpp
     L237: // former pthread_cond_t, by using a freelist of 
mutex/condvar pairs.
         You aren't allocating pairs anymore right?

     L239: #ifdef __APPLE__
     L240: #define PLATFORM_MONITOR_IMPL_INDIRECT 1
     L241: #else
     L242: #define PLATFORM_MONITOR_IMPL_INDIRECT 1
     L243: #endif
         Leftover from compile testing on a non Apple platform?
         Update: Same comment as Dean.

     L268:   PlatformMutex();            // Use freelist allocation of impl.
         nit - comment was previously aligned with L271.

src/hotspot/os/posix/os_posix.inline.hpp
     No comments.

src/hotspot/os/solaris/os_solaris.cpp
     No comments.

src/hotspot/os/solaris/os_solaris.hpp
     No comments.

src/hotspot/os/windows/os_windows.hpp
     No comments.

src/hotspot/os/windows/os_windows.inline.hpp
     No comments.

Thumbs up.

Dan


>
> Thanks,
> David
> -----



More information about the hotspot-dev mailing list