This started in b34, probably with the putback for this:
     6178789 Support big page heap on x86 and amd64 platforms

In the below, note the warning about MPSS. This warning occurs with b30 - b33
but the slowdown does not. With b34, the warning disappears
and the slowdown appeares.

These commands were run on machine vm-v20z-10.
I do not see this slowdown running on mho:

time $prom/b30/binaries/solaris-amd64/fastdebug/bin/java -version
Java HotSpot(TM) Server VM warning: Attempt to use MPSS failed.
java version "1.6.0-ea"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.6.0-ea-b30)
Java HotSpot(TM) Server VM (build 1.6.0-ea-b30-debug, mixed mode)
0.0u 0.0s 0:00e 0%- swps=0 pf=0 in=0 out=0 usd=0k uss=0k mm=0


time $prom/b40/binaries/solaris-amd64/fastdebug/bin/java -version
java version "1.6.0-ea-fastdebug"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.6.0-ea-fastdebug-b40)
Java HotSpot(TM) Server VM (build 1.6.0-ea-b40-debug, mixed mode)
0.0u 93.0s 1:34e 98%- swps=0 pf=0 in=0 out=0 usd=0k uss=0k mm=0


time $prom/b50/binaries/solaris-amd64/fastdebug/bin/java -version
java version "1.6.0-ea-fastdebug"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.6.0-ea-fastdebug-b50)
Java HotSpot(TM) Server VM (build 1.6.0-ea-fastdebug-b50-debug, mixed mode)
0.0u 93.0s 1:35e 97%- swps=0 pf=0 in=0 out=0 usd=0k uss=0k mm=0


time $prom/b58/binaries/solaris-amd64/fastdebug/bin/java -version
java version "1.6.0-ea-fastdebug"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.6.0-ea-fastdebug-b58)
Java HotSpot(TM) Server VM (build 1.6.0-ea-fastdebug-b58-debug, mixed mode)
0.0u 95.0s 1:37e 97%- swps=0 pf=0 in=0 out=0 usd=0k uss=0k mm=0


When I run with -server -XX:-UseLargePages, the problem does not occur.

I believe this bug makes these tests timeout:
     nsk/serial/Unsafe/allocateMemory/allocmem001
     nsk/serial/Unsafe/allocateMemory/reallocmem001

but see also bug
   6305280 Fastdebug GC spending mostt time in MutableSpace::mangle_unused_area