[GHC] #13883: T5435_dyn_asm fails with ld.gold
GHC
ghc-devs at haskell.org
Tue Jun 27 16:25:42 UTC 2017
#13883: T5435_dyn_asm fails with ld.gold
-------------------------------------+-------------------------------------
Reporter: bgamari | Owner: (none)
Type: bug | Status: new
Priority: normal | Milestone:
Component: Compiler | Version: 8.0.1
Keywords: | Operating System: Unknown/Multiple
Architecture: | Type of failure: None/Unknown
Unknown/Multiple |
Test Case: | Blocked By:
Blocking: | Related Tickets:
Differential Rev(s): | Wiki Page:
-------------------------------------+-------------------------------------
I'm seeing `T5435_dyn_asm` fail with Phab:D3449 using the gold linker.
Specifically,
{{{
=====> T5435_dyn_asm(normal) 1 of 1 [0, 0, 0]
cd "./rts/T5435_dyn_asm.run" && $MAKE -s --no-print-directory
T5435_dyn_asm
T5435_dyn_asm failed with ['initArray1', 'initArray2', 'ctors2', 'ctors1',
'success'], see all.T for details
*** unexpected failure for T5435_dyn_asm(normal)
}}}
It seems that the ctor list in `T5435_asm.c` is in order `ctors2,ctors1`
as it expects that constructors are run in reverse order, yet somehow with
gold the constructors are being run in forward order.
I've searched high and low for a document specifying this behavior, but
the best I can come up with is GCC's
[[https://gcc.gnu.org/onlinedocs/gccint/Initialization.html|internals
manual]]. I currently don't have a great answer for why gold flips the
order. I suspected that the (enabled by default) `--ctors-in-init-array`
flag might have something to do with it, but alas that doesn't appear to
be the case.
Anyways, given that this most certainly isn't GHC's fault and the order
requirements aren't well specified anyways, I'm just going to accept the
new ordering.
--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13883>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
More information about the ghc-tickets
mailing list