There seems to be new breakage in threadDelay on Windows. I noticed conc070 failing in a validate:

=====> conc070(ghci) 191 of 3265 [0, 0, 0]

cd ./concurrent/should_run && 'c:/simonmar/ghc-validate/bindisttest/install   dir/bin/ghc.exe' -fforce-recomp -dcore-lint -dcmm-lint -dno-debug-output -no-user-package-conf -rtsopts  -fno-ghci-history conc070.hs --interactive -v0 -ignore-dot-ghci +RTS -I0.1 -RTS   <conc070.genscript 1>conc070.interp.stdout 2>conc070.interp.stderr

Actual stdout output differs from expected:

--- ./concurrent/should_run/conc070.stdout	2011-08-02 14:43:53 +0100
+++ ./concurrent/should_run/conc070.run.stdout	2012-04-27 10:54:04 +0100
@@ -1 +1 @@
-[ThreadBlocked BlockedOnMVar,ThreadBlocked BlockedOnMVar,ThreadRunning,ThreadFinished]
+[ThreadFinished,ThreadBlocked BlockedOnMVar,ThreadRunning,ThreadFinished]

*** unexpected failure for conc070(ghci)

and strangely if you try threadDelay 1000000 in GHCi it returns immediately.

This test is also failing:

=====> ThreadDelay001(threaded1) 7 of 7 [0, 0, 0]
cd . && 'c:/simonmar/ghc-validate/inplace/bin/ghc-stage2.exe' -fforce-recomp -dcore-lint -dcmm-lint -dno-debug-output -no-user-package-conf -rtsopts  -fno-ghci-history -o ThreadDelay001 ThreadDelay001.hs -threaded -debug   >ThreadDelay001.comp.stderr 2>&1
cd . && ./ThreadDelay001    </dev/null >ThreadDelay001.run.stdout 2>ThreadDelay001.run.stderr
Actual stdout output differs from expected:
--- /dev/null   2012-04-30 11:17:41 +0100
+++ ./ThreadDelay001.run.stdout 2012-04-30 11:17:41 +0100
@@ -0,0 +1,2 @@
+(Mon Apr 30 11:17:29 GMT Daylight Time 2012,Mon Apr 30 11:17:29 GMT Daylight Time 2012,1000000,0,-1000000,-1.0e-6)
+(Mon Apr 30 11:17:29 GMT Daylight Time 2012,Mon Apr 30 11:17:29 GMT Daylight Time 2012,5000000,0,-5000000,-5.0e-6)
*** unexpected failure for ThreadDelay001(threaded1)

Possibly related to the fix for #5865 (closed)