[GHC] #3874: GHC.Exts.traceEvent segfaults

[GHC]

#3874: GHC.Exts.traceEvent segfaults
-------------------------------+--------------------------------------------
    Reporter:  cjs             |       Owner:                 
        Type:  bug             |      Status:  new            
    Priority:  normal          |   Component:  Runtime System 
     Version:  6.12.1          |    Keywords:  event, eventlog
          Os:  Linux           |    Testcase:                 
Architecture:  x86_64 (amd64)  |     Failure:  Runtime crash  
-------------------------------+--------------------------------------------

Comment(by cjs):

 My earlier "postSafeuserMsg" patch was wrong in several ways; instead use
 the attachment just added now, 3874-1.patch, which is tested and fixes
 more bugs. With a test program {{{main = traceEvent "hello"}}}, this fixes
 the following cases that were previously segfaulted:
 {{{
 ghc        -eventlog -o test test.hs && ./test +RTS -ls
 ghc -debug -eventlog -o test test.hs && ./test +RTS -ls -v
 }}}
 However, the following test case still segfaults:
 {{{
 ghc        -eventlog -o test test.hs && ./test +RTS -ls -v
 }}}
 The backtrace for this case is quite different:
 {{{
 #0  0x000000000043dcef in postSchedEvent ()
 #1  0x00000000004358a6 in createThread ()
 #2  0x000000000044e759 in createIOThread ()
 #3  0x000000000044e812 in rts_evalLazyIO ()
 #4  0x0000000000430d11 in real_main ()
 #5  0x0000000000430e1f in hs_main ()
 #6  0x00007f1df3e245a6 in __libc_start_main () from /lib/libc.so.6
 #7  0x0000000000402f09 in _start () at ../sysdeps/x86_64/elf/start.S:113
 }}}
 I had a quick look here, and the problem does not appear to be the same
 va_list issue.

-- 
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/3874#comment:4>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler