Bug 451877

Summary:	unconfined_execmem_exec_t needed for several GHC-built Haskell binaries
Product:	[Fedora] Fedora	Reporter:	Bryan O'Sullivan <bos>
Component:	selinux-policy	Assignee:	Daniel Walsh <dwalsh>
Status:	CLOSED RAWHIDE	QA Contact:	Fedora Extras Quality Assurance <extras-qa>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	rawhide	CC:	drepper, haskell-devel, notting, petersen, rvokal
Target Milestone:	---
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2009-04-13 14:06:17 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:
Bug Blocks:	452440

Description Bryan O'Sullivan 2008-06-17 22:30:45 UTC

The following Haskell programs, built by GHC, need unconfined_execmem_exec_t
added to the SELinux policies:

/usr/bin/haddock
/usr/bin/haddock-0.9
/usr/bin/hasktags
/usr/bin/runghc
/usr/bin/runhaskell
/usr/libexec/ghc-*/ghc-6.8.[0-9]
/usr/libexec/ghc-*/ghc-pkg.bin
/usr/libexec/ghc-*/hsc2hs-bin

Comment 1 Bryan O'Sullivan 2008-06-17 22:33:35 UTC

Tibbs suggests that Haddock doesn't need to be blocked on this.

Comment 2 Daniel Walsh 2008-06-22 10:55:05 UTC

Why doesn't haddock fix their code to not need execmem?

Is this java or mono?

What does haddock need both executable and writeable memory at the same time?

Comment 3 Jens Petersen 2008-06-22 22:59:27 UTC

(In reply to comment #2)
> Is this java or mono?

Haskell ;)

Comment 4 Bryan O'Sullivan 2008-06-23 02:07:40 UTC

The runtime system for programs compiled by GHC generates code dynamically and
executes it.

The interaction with SELinux's enforcing mode is a known problem, which was
previously addressed with a hack: the %post scripts for Haskell programs were
using chcon to add unconfined_exec_mem_t.  This obviously didn't work in lots of
circumstances, hence wanting to apply the policy properly.

The underlying problem, namely the way GHC allocates memory that it intends to
execute dynamically, should be fixed within the next six months or so.  See
http://hackage.haskell.org/trac/ghc/ticket/738 for details.

Comment 5 Daniel Walsh 2008-07-02 17:46:12 UTC

Fixed in selinux-policy-3.3.1-74.fc9.noarch

Comment 6 Bill Nottingham 2008-10-24 20:08:23 UTC

Can Haskell users verify this and close the bug if it's fixed?

Comment 7 Jens Petersen 2009-01-21 00:37:35 UTC

I started a comment long ago (which was then lost by a browser crash or restart)...

Basic summary is most (all) haskell programs shipped can run now in enforcing.
However there have been a number of path changes that should probably be updated
in selinux-policy.  I can help to do that.

Currently the changes needed are being done for that at install time.

Comment 8 Daniel Walsh 2009-01-26 16:46:20 UTC

What are the paths?

Comment 9 Jens Petersen 2009-06-02 03:57:39 UTC

I tested in F11 and it seems AFAICT the %post stuff we have is no longer needed for ghc executables so I am removing it for f12.