Use boost::thread locking instead of interprocess #1292

pull sipa wants to merge 1 commits into bitcoin:master from sipa:threadsync changing 3 files +54 −59

sipa commented at 10:53 PM on May 13, 2012: member

Luke noticed there was a boost/thread/* set of synchronisation primitives, separate from those in boost/interprocess/sync/*. Given that they do not need to cope with locks being in shared memory, they are different and simpler on some systems.

This commit replaces all sync.{h,cpp} infrastructure by systems provided by boost/thread/*.

So far, it seems to work well on Linux, OSX and BSD (the latter two required a workaround before, as the interprocess semaphores and condition variables used spinlocks - it seems the thread ones don't).
jgarzik commented at 11:29 PM on May 13, 2012: contributor

ACK
laanwj commented at 5:24 AM on May 14, 2012: member

ACK
gmaxwell commented at 4:42 PM on May 14, 2012: contributor

boost::interprocess::interprocess_mutex dd_mutex in DEBUG_LOCKORDER? (Edit: and, indeed, DEBUG_LOCKORDER is currently broken)
sipa commented at 9:56 PM on May 14, 2012: member

Fixed DEBUG_LOCKORDER.
Use boost::thread locking instead of interprocess 660ff174f2

in src/makefile.unix:None in 1e5b74b72c outdated

   3 | @@ -4,7 +4,7 @@
   4 |  
   5 |  USE_UPNP:=0
   6 |  
   7 | -DEFS=-DUSE_IPV6
   8 | +DEFS=-DUSE_IPV6 -DDEBUG_LOCKORDER

laanwj commented at 11:32 AM on May 17, 2012:

I don't think this should sneak into the makefile :)

gavinandresen merged this on May 18, 2012
gavinandresen closed this on May 18, 2012
gavinandresen commented at 3:11 PM on May 18, 2012: contributor

ACK, good CPU usage on my mac.
suprnurd referenced this in commit cfc8954b42 on Dec 5, 2017
lateminer referenced this in commit 5ad9c81acc on May 6, 2020
DrahtBot locked this on Sep 8, 2021

Contributors