Hello, I'm fine with reverting for now. But could you give me a way to reproduce this error? I've not seen crashes on either linux and windows in various configs. Cheers, Andreas Ben Gamari schrieb:

Simon Peyton Jones via ghc-devs writes:

...

OK I have verified that

* This split_marker crash happens on a clean HEAD build * Reverting "NCG: New code layout algorithm.", 575515b4909f3d77b3d31f2f6c22d14a92d8b8e0, solves the problem. Andreas: I propose to revert in HEAD unless you have a rapid fix, or believe that is somehow my fault. (Or maybe someone else can revert.)

Simon, are you using split-objs by any chance?

Regardless, yes, let's revert until we work out what is going on here.

Cheers,

- Ben

I might already have found the specific cause. It seems with split sections we generate a dummy CmmProc, which has as entry label

error (split_sections_entry)

My patch likely introduces strictness on that field where it was lazy before. If this is the cause I expect to have a patch up in a hour or two and will merge it after it validates. Cheers, Andreas Ben Gamari schrieb:

Andreas Klebinger writes:

...

Hello,

I'm fine with reverting for now. But could you give me a way to reproduce this error?

I've not seen crashes on either linux and windows in various configs.

I suspect that Simon is building with SplitObjects enabled. To be honest, I would really like to remove this feature; SplitSections is better in nearly every regard. However, we have been stalled since SplitSections doesn't quite work yet on Windows (or, IIRC, the toolchain is prohibitively slow when it's used). I believe Tamar was working on fixing this.

Cheers,

- Ben

I do indeed have SplitObs=YES in my validate.mk. In the past at least, that made binaries quite a bit smaller. Happy to change it to SplitSections=YES if that is more kosher these days? Simon From: Andreas Klebinger Sent: 19 November 2018 20:38 To: Ben Gamari Cc: Phyx ; Simon Peyton Jones ; ghc-devs@haskell.org Subject: Re: split_marker crash I might already have found the specific cause. It seems with split sections we generate a dummy CmmProc, which has as entry label

error (split_sections_entry)

My patch likely introduces strictness on that field where it was lazy before. If this is the cause I expect to have a patch up in a hour or two and will merge it after it validates. Cheers, Andreas Ben Gamari schrieb: Andreas Klebinger mailto:klebinger.andreas@gmx.at writes: Hello, I'm fine with reverting for now. But could you give me a way to reproduce this error? I've not seen crashes on either linux and windows in various configs. I suspect that Simon is building with SplitObjects enabled. To be honest, I would really like to remove this feature; SplitSections is better in nearly every regard. However, we have been stalled since SplitSections doesn't quite work yet on Windows (or, IIRC, the toolchain is prohibitively slow when it's used). I believe Tamar was working on fixing this. Cheers, - Ben

Simon Peyton Jones via ghc-devs writes:

I do indeed have SplitObs=YES in my validate.mk.

Ahh, so perhaps Andreas's finding isn't the culprit afterall. Looking at the source it looks like `split_marker_entry` is indeed emitted for split objects support.c

In the past at least, that made binaries quite a bit smaller. Happy to change it to SplitSections=YES if that is more kosher these days?

We generally recommend SplitSections at this point. Not only is it simpler, but it may reduce build times, result in smaller binaries, and generally stays closer to what C compilers do, avoiding toolchain bugs. Cheers, - Ben

Phyx writes:

On Mon, Nov 19, 2018 at 8:33 PM Ben Gamari wrote:

...

Andreas Klebinger writes:

...

Hello,

I'm fine with reverting for now. But could you give me a way to reproduce this error?

I've not seen crashes on either linux and windows in various configs.

I suspect that Simon is building with SplitObjects enabled. To be honest, I would really like to remove this feature; SplitSections is better in nearly every regard. However, we have been stalled since SplitSections doesn't quite work yet on Windows (or, IIRC, the toolchain is prohibitively slow when it's used). I believe Tamar was working on fixing this.

SplitObjects has been off in favor of SplitSections on Windows since Aug. https://github.com/ghc/ghc/commit/23774c98f1368b41515cbd5223b87ea6dbf644e1

Ahh, I had forgotten about this. I'll need to check this when I'm back home but I suspect this means that we can finally rip out (or at least deprecate) split object support for 8.8. Cheers, - Ben

Steven Shaw writes:

Curious if the GHC libraries _are_ still built with the `-split-objs` flag, as pointed out in the latest documentation.

Good catch; the build system uses split sections by default if supported. If not it instead uses split objects, if supported. I do not believe there are any cases where we support split objects but not split sections. Consequently I believe we should always use split sections. Cheers, - Ben