RTS options broken for dynamic libraries
There are several problems here. Firstly, GHC doesn't seem to respect -rtsopts when linking dynamic libraries. In !DriverPipeline, linkDynLib doesn't care about the flag, only linkBinary does. This causes hs_init to fail if called with some RTS options. This is a blocker for us.
Secondly, here is a snipped from hs_init in !RtsStartup.c:
setFullProgArgv(*argc,*argv);
setupRtsFlags(argc, *argv, &rts_argc, rts_argv);
setProgArgv(*argc,*argv);
Of these, setFullProgArgv copies the things pointed to by argc and argv (good), setProgArgv doesn't copy (bad) and setupRtsFlags doesn't seem to, either (it calls setProgName which doesn't copy, at least). We probably should have setupRtsFlags and setProgArgv create its own copy of the args as well and document that the caller doesn't have to hang on to the memory after the call. Alternatively, the docs should make it very clear that the memory must remain allocated until the RTS is shut down.
Finally, hs_init simply aborts if it doesn't like the RTS arguments which is quite unhelpful for dynamic libraries. I took me a day to find out that an application crash was caused by a failing hs_init (because of the -rtsopts problem). I would like to add a check for this to our code but there doesn't seem to be a way to do this. It would be much nicer if hs_init returned a failure/success code, at least for dynamic libraries.
Trac metadata
| Trac field | Value |
|---|---|
| Version | 7.1 |
| Type | Bug |
| TypeOfFailure | OtherFailure |
| Priority | normal |
| Resolution | Unresolved |
| Component | Compiler |
| Test case | |
| Differential revisions | |
| BlockedBy | |
| Related | |
| Blocking | |
| CC | |
| Operating system | |
| Architecture |