From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from www.dynamicdevices.co.uk (www.dynamicdevices.co.uk [89.200.136.37]) by yocto-www.yoctoproject.org (Postfix) with ESMTP id 0369AE01757 for ; Tue, 5 Nov 2013 01:47:11 -0800 (PST) Received: from [127.0.0.1] (cpc16-live22-2-0-cust180.know.cable.virginm.net [82.42.220.181]) (using TLSv1 with cipher ECDHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by www.dynamicdevices.co.uk (Postfix) with ESMTPSA id A63EE860A9; Tue, 5 Nov 2013 09:47:08 +0000 (UTC) Message-ID: <5278BE99.7050105@dynamicdevices.co.uk> Date: Tue, 05 Nov 2013 09:47:05 +0000 From: Alex J Lennon User-Agent: Mozilla/5.0 (Windows NT 6.3; WOW64; rv:24.0) Gecko/20100101 Thunderbird/24.1.0 MIME-Version: 1.0 To: yocto@yoctoproject.org References: <20131029110034.GF3697@jama> <20131029134236.GI3697@jama> In-Reply-To: Cc: Paul Eggleton Subject: [meta-mono] Mono 3.2.3 support X-BeenThere: yocto@yoctoproject.org X-Mailman-Version: 2.1.13 Precedence: list List-Id: Discussion of all things Yocto Project List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 05 Nov 2013 09:47:14 -0000 Content-Type: multipart/alternative; boundary="------------090506060302030506040208" --------------090506060302030506040208 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Hi all, I've successfully built and tested a Mono 3.2.3 recipe with Yocto. The code is in a fork I took of current meta-mono head, which is here: git@git.assembla.com:ciseco-eve.meta-mono.git Web access:https://www.assembla.com/code/ciseco-eve/git-2/nodes I have tested against a qemux86 target and also against an i.MX6 target built out of the meta-arm-fsl layer (MACHINE = 'imx6qsabresd', fsl-image-gui) Bitbake layer versions and major environment settings are here http://pastebin.com/SQhB4JgL The testing I have done thus far is basic, simply running a console "Hello World" application and a corresponding version showing a Windows Form with a button. Floating Point ============ At this time Mono does not support arm hardfp although there may be some patches floating around. There are some notes on upcoming support for Mono in the the near future but nothing I can see in the repository yet, ref: http://blog.alexrp.com/2013/07/21/recent-improvements-to-mono-on-arm/ As such any images which are tuned by default to use hardfp need this overridden (e.g. fsl-image-gui) To build fsl-image-gui I added the following to my local.conf DEFAULTTUNE_mx6 = "cortexa9-neon" (I'm not sure if this is the best way - any advice on this would be appreciated) Build Issues ========== Lastly I had a lot of trouble building. From comments seen on the inter-web I suspect there is an issue with bitbake tracking dependencies across differing tunings. "I tried this on imx6qsabresd and have no issues. Right before release we did have problems with disabling hwfp correctly but it was enabled in the final update. Also I've seen multiple times that switching between hwfp and swfp can corrupt build states. I always keep these completely separate builds both with cache and tmp directories." ref: https://community.freescale.com/thread/310233 I was seeing that recipe builds failed, and in one case when I investigated the RPM .spec file the spec file had been created with dependencies on packages with -hardfp in the naming. I had to create a completely new build directory for my softfp build and with this done the build then worked. I think this may have been something touched upon by commentators at ELCE the other week. --- Lastly, if anybody has time to review the recipes I've modified here and advise me on any non-compliances with Yocto policies I'd be pleased to make any needed changes to bring them into compliance with a view to committal to the official git repo. Best Regards, Alex Lennon --------------090506060302030506040208 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Hi all,

I've successfully built and tested a Mono 3.2.3 recipe with Yocto.

The code is in a fork I took of current meta-mono head, which is here:

git@git.assembla.com:ciseco-eve.meta-mono.git

Web access: https://www.assembla.com/code/ciseco-eve/git-2/nodes

I have tested against a qemux86 target and also against an i.MX6 target
built out of the meta-arm-fsl layer (MACHINE = 'imx6qsabresd', fsl-image-gui)

Bitbake layer versions and major environment settings are here http://pastebin.com/SQhB4JgL

The testing I have done thus far is basic, simply running a console "Hello World" application
and a corresponding version showing a Windows Form with a button.

Floating Point
============

At this time Mono does not support arm hardfp although there may be some patches
floating around. There are some notes on upcoming support for Mono in the the near
future but nothing I can see in the repository yet,

ref:  http://blog.alexrp.com/2013/07/21/recent-improvements-to-mono-on-arm/

As such any images which are tuned by default to use hardfp need this overridden (e.g. fsl-image-gui)

To build fsl-image-gui I added the following to my local.conf

DEFAULTTUNE_mx6 = "cortexa9-neon"

(I'm not sure if this is the best way - any advice on this would be appreciated)

Build Issues
==========

Lastly I had a lot of trouble building. From comments seen on the inter-web I suspect
there is an issue with bitbake tracking dependencies across differing tunings.

"I tried this on imx6qsabresd and have no issues.   Right before release we did have problems with
disabling hwfp correctly but it was enabled in the final update. Also I've seen multiple times that switching
between hwfp and swfp can corrupt build states.  I always keep these completely separate builds both
with cache and tmp directories."

ref: https://community.freescale.com/thread/310233

I was seeing that recipe builds failed, and in one case when I investigated the RPM .spec file
the spec file had been created with dependencies on packages with -hardfp in the naming.

I had to create a completely new build directory for my softfp build and with this done
the build then worked.

I think this may have been something touched upon by commentators at ELCE the other week.

---

Lastly, if anybody has time to review the recipes I've modified here and advise me on any non-compliances
with Yocto policies I'd be pleased to make any needed changes to bring them into compliance with a view
to committal to the official git repo.

Best Regards,

Alex Lennon

--------------090506060302030506040208--