Bug 16511 – Suspected Win64 release-mode code-gen bug

Status
NEW
Severity
critical
Priority
P2
Component
dmd
Product
D
Version
D2
Platform
x86_64
OS
Linux
Creation time
2016-09-19T17:31:30Z
Last change time
2024-12-13T18:50:09Z
Assigned to
No Owner
Creator
Nick Sabalausky
Moved to GitHub: dmd#17775 →

Comments

Comment #0 by bus_dbugzilla — 2016-09-19T17:31:30Z
I've reveived a bug report on a lib of mine that I'm suspecting may be a 64-bit release-mode code-gen bug, because it involves what, at least *appears*, to be a fiber overflowing its stack *ONLY* on 64-bit in release mode. I could understand a stack overlflow on 32-bit-only, but this seems backwards. And a release-mode-only crash seems to suggest a code-gen bug as well. Sorry I dont have a reduced case right now, but you can reproduce like this: On a Win64 machine with DUB 1.0.0 <http://code.dlang.org/files/dub-1.0.0-setup.exe>: - Download this Gist to a fresh directory: https://gist.github.com/MrSmith33/c2f435937aefbc3eead25aa873363d84 - From that directory, run: > dub --arch=x86_64 --build=release --verbose --compiler=dmd It will build, run, and then crash. Omitting *either* of --arch=x86_64 or --build=release will eliminate the crash.
Comment #1 by bus_dbugzilla — 2016-09-19T17:44:43Z
Just for the sake of reference, the original report/conversation is here: https://github.com/Abscissa/libInputVisitor/issues/1
Comment #2 by bus_dbugzilla — 2016-09-19T18:05:50Z
Got a report of the crash happening with both --build=debug (dub's default) and --build=release. It only crashes with --build=release for me. So not really sure what's going on. In my lib, I'm going to be working around it, but wanted to post this here in case there is an underlying codegen thing.
Comment #3 by robert.schadek — 2024-12-13T18:50:09Z
THIS ISSUE HAS BEEN MOVED TO GITHUB https://github.com/dlang/dmd/issues/17775 DO NOT COMMENT HERE ANYMORE, NOBODY WILL SEE IT, THIS ISSUE HAS BEEN MOVED TO GITHUB