T8089 failing with ghci, threaded1, threaded2, profthreaded

=====> T8089(ghci) 17 of 17 [2, 16, 0]
cd "/run/user/1001/ghctest-m_wu2f4i/test   spaces/../../libraries/base/tests/T8089.run" && "/home/alp/WT/ghc-slow-validate/bindisttest/install   dir/bin/ghc" T8089.hs -dcore-lint -dcmm-lint -no-user-package-db -rtsopts -fno-warn-missed-specialisations -fshow-warning-groups -fdiagnostics-color=never -fno-diagnostics-show-caret -dno-debug-output  --interactive -v0 -ignore-dot-ghci -fno-ghci-history +RTS -I0.1 -RTS < T8089.genscript
Wrong exit code for T8089(ghci) (expected 99 , actual 0 )
*** unexpected failure for T8089(ghci)
=====> T8089(threaded1) 17 of 17 [2, 17, 0]
cd "/run/user/1001/ghctest-m_wu2f4i/test   spaces/../../libraries/base/tests/T8089.run" &&  "/home/alp/WT/ghc-slow-validate/bindisttest/install   dir/bin/ghc" -o T8089 T8089.hs -dcore-lint -dcmm-lint -no-user-package-db -rtsopts -fno-warn-missed-specialisations -fshow-warning-groups -fdiagnostics-color=never -fno-diagnostics-show-caret -dno-debug-output  -threaded -debug 
cd "/run/user/1001/ghctest-m_wu2f4i/test   spaces/../../libraries/base/tests/T8089.run" && ./T8089  
Wrong exit code for T8089(threaded1)(expected 99 , actual 0 )
*** unexpected failure for T8089(threaded1)
=====> T8089(threaded2) 17 of 17 [2, 18, 0]
cd "/run/user/1001/ghctest-m_wu2f4i/test   spaces/../../libraries/base/tests/T8089.run" &&  "/home/alp/WT/ghc-slow-validate/bindisttest/install   dir/bin/ghc" -o T8089 T8089.hs -dcore-lint -dcmm-lint -no-user-package-db -rtsopts -fno-warn-missed-specialisations -fshow-warning-groups -fdiagnostics-color=never -fno-diagnostics-show-caret -dno-debug-output  -O -threaded -eventlog 
cd "/run/user/1001/ghctest-m_wu2f4i/test   spaces/../../libraries/base/tests/T8089.run" && ./T8089 +RTS -N2 -ls -RTS 
Wrong exit code for T8089(threaded2)(expected 99 , actual 0 )
*** unexpected failure for T8089(threaded2)  
=====> T8089(profthreaded) 17 of 17 [2, 19, 0]
cd "/run/user/1001/ghctest-m_wu2f4i/test   spaces/../../libraries/base/tests/T8089.run" &&  "/home/alp/WT/ghc-slow-validate/bindisttest/install   dir/bin/ghc" -o T8089 T8089.hs -dcore-lint -dcmm-lint -no-user-package-db -rtsopts -fno-warn-missed-specialisations -fshow-warning-groups -fdiagnostics-color=never -fno-diagnostics-show-caret -dno-debug-output  -O -prof -static -fprof-auto -threaded 
cd "/run/user/1001/ghctest-m_wu2f4i/test   spaces/../../libraries/base/tests/T8089.run" && ./T8089 +RTS -p -RTS 
Wrong exit code for T8089(profthreaded)(expected 99 , actual 0 )
*** unexpected failure for T8089(profthreaded)

I'm going to write a patch to expect the test to be broken in these ways, but it seems really wrong that the exit code we get is not the same accross all ways.

Trac metadata

Trac field	Value
Version	8.5
Type	Bug
TypeOfFailure	OtherFailure
Priority	normal
Resolution	Unresolved
Component	libraries/base
Test case
Differential revisions
BlockedBy
Related	#8089 (closed)
Blocking
CC
Operating system
Architecture

mentioned in issue #8089 (closed)

changed weight to 5

added Tbug Trac import core libraries labels

#7325 (closed) and #15064 (closed) are also relevant.

Trac metadata

Trac field	Value
Related	#8089 (closed) → #7325 (closed), #8089 (closed), #15064 (closed)

It looks to me like there is a bug in the event manager. Specifically, it apparently makes no attempt to handle overflow when computing expiration times. Namely, GHC.Event.TimerManager.registerTimeout has:

now <- getMonotonicTimeNSec                
let expTime = fromIntegral us * 1000 + now 

Even worse, we currently throw away the delay requested by the user, meaning it is no longer possible to account for overflow after registration.

This can be fixed, but will require a bit more allocation in timeout registration.

changed milestone to %8.6.1

Trac metadata

Trac field	Value
Differential revisions	- → D4700

mentioned in commit 21a9fb5f

closed

Trac metadata

Trac field	Value
Resolution	Unresolved → ResolvedFixed

When will the fix be released? This bug will likely cause some fairly widespread failures.

https://www.reddit.com/r/haskell/comments/97zmpa/psa_forever_threaddelay_maxbound_idiom_broken_on/

EDIT: OK, not sure about "widespread". But clearly there is live code out there that will break. So, it's good to know when the fix will land. In our case, we implemented a fairly ugly work-around, so we'd like to know when we can remove it. Thanks for the prompt fix!

EDIT2: And answering my own question, I see that the fix made it into 8.6.1. Thanks! Sorry for the noise.

Another question - this code seems to be mixing up milliseconds and microseconds. What am I missing?

EDIT: Never mind, I got it. now is in **nanoseconds**.

Indeed this will be fixed in 8.6.1.

added incorrect runtime result label

added Pnormal label