Since I'm cross-compiling rather than building natively on a target I think 
that :

*snapshot=off*

is the right choice also according to what I have read over the internet. I 
have re-built both the v8 package and node.js. V8 is built now with the 
default options for mips architecture as we discussed earlier while node is 
configured with just :

*./configure --prefix=/usr --without-snapshot --shared-v8 
--shared-v8-includes="$(MYDIR)/include/" 
--shared-v8-libpath="$(MYDIR)/usr/lib";*

A little progress has been made, by which I mean that time I don't get the 
floating point expcetion :

<https://lh3.googleusercontent.com/-1ti7OfvfV2E/VNsghzVA0bI/AAAAAAAAAvQ/LylFO8DLKrQ/s1600/no_floating.png>

however, the node interpreter is stuck every time I launch its shell, or 
npm or a simple test program such as the JS Hello World.

I am a owner of a *Dog Hunter Linino ONE* :

http://shop.linino.org/products/linino-one

and of a *Arduino YUN* :

http://shop.linino.org/products/arduino-yun-dogstick

running both OpenWRT 12.09 (with linux kernel 3.3.8). The first one is 
basically the same of the Yun except for a reduced form factor (the 
hardware is the same : Atheros AR9331 as linux cpu and Atmel ATmega32u4 as 
MCU)

Thank you in advance for everything you can find about this issue
 
   Regards, Arturo    

Il giorno mercoledì 11 febbraio 2015 06:24:37 UTC+1, paul lind ha scritto:
>
> Hi Arturo - If you are building natively on your target, then you should 
> be able to build with snapshot=on (the default). That makes startup a lot 
> faster.
>
> If your board is in a commercially available product, please let me know 
> what it is. We might give one a try here.
>
> paul
>
>
> On Feb 10, 2015, at 5:31 PM, Arturo Rinaldi <arty...@gmail.com 
> <javascript:>> wrote:
>
> Hello Paul, you're guessing right. I always build the v0.10.x series of 
> the node tarballs with a *shared* build of V8 from your v8m-rb repo.
>
> I am well aware that v8 3.20 and on have total support for MIPS 
> architecture and better performances on it. For this very reason I was 
> trying to build an updated version of both V8 and node itself for my board.
>
> Please consider also the fact that *OpenWRT* (the OS running on my board) 
> has a embedded* SDK* of its own and and a build with QEMU would be really 
> useless to me. The *MAKE* arguments are the following for *V8* :
>
> *$(MAKE) mips.release werror=no library=shared snapshot=off 
> i18nsupport=off*
>
> and for *node.js* :
>
> *$(MAKE) snapshot=off werror=no* (I think that i18nsupport=off is not 
> necessary here because of the previous build of V8 without i18n support)
>
> I'll try tomorrow to force the build with the default FPU options enabled 
> and report here in the list. I hope to get my Atheros9300 CPU based board 
> working with this build...
>
>    Regards, Arturo
>
> Il giorno mercoledì 11 febbraio 2015 00:06:10 UTC+1, paul lind ha scritto:
>>
>> Sorry, but that is not going to work. From your earlier screenshot I am 
>> guessing that you previously used node 10, with v8 3.14 from our old github 
>> repo.
>>
>> The good news is that the MIPS big-endian support is all upstream in v8 
>> repo since 3.26, which of course includes the 3.28 version in node 12.
>>
>> The not-so-good news is that after 3.17, v8 does not support soft-float 
>> builds, for either ARM or MIPS.
>>
>> However, if you build v8 with FP instructions, it will work fine using 
>> the kernel FP emulator that is in any more-or-less recent kernel. I am 
>> running big-endian boards here with 3.4.27 kernel. It's slow on 
>> FP-intensive work, like some benchmarks, but overall it works very well for 
>> most apps.
>>
>> You can build for FP by using the GYP defaults, or setting them 
>> explicitly like this:
>>
>> export GYPFLAGS="-Dv8_use_mips_abi_hardfloat=true 
>> -Dv8_can_use_fpu_instructions=true"
>>
>> You should try a build without snapshots, and big-endian does not 
>> currently support i18n:
>> make mips.release -j12 snapshot=off i18nsupport=off
>>
>> If you want to use snapshots, you can either build on your native mips 
>> board, or use qemu as a helper to cross-compile the snapshot image:
>> https://github.com/paul99/v8m-rb/wiki/Building-v8-for-BE-with-QEMU
>>
>> Hope this helps...
>>
>> On Feb 10, 2015, at 2:44 PM, Arturo Rinaldi <arty...@gmail.com> wrote:
>>
>> Luckily, it is just a typo in my mail. I double checked and the settings 
>> in the makefiles are right. I have also performed a deep "sweep" into the 
>> *gyp* files and found the environment variables to build the V8 
>> libraries with.
>>
>> I am now able (maybe) to track down my issues to the node source. I have 
>> always built the sources until now with these CLI options :
>>
>> *--dest-cpu=mips --dest-os=linux --with-mips-float-abi=soft;*
>>
>> However, I noticed time ago (since the unstable release of the node 
>> v0.11.x series) that this option has been removed. Just before discovering 
>> this mips-switch in the configure help, I used to build node with the :
>>
>> *--with-arm-float-abi=soft*
>>
>> option, which did the trick very well without any sort of issues. Do you 
>> think that reinstating this last option could solve my issue ? (i.e. the 
>> floating point will work in the right way on MIPS arch as well).
>>
>> It could be a great achievement for me running the latest version of node 
>> on my embedded platform since I need for work purposes....
>>
>> Il giorno martedì 10 febbraio 2015 22:55:49 UTC+1, Jakob Kummerow ha 
>> scritto:
>>>
>>> What Ben said. Most flags are defined in build/toolchain.gypi and 
>>> build/features.gypi.
>>>
>>> In the case at hand:
>>>
>>> On Tue, Feb 10, 2015 at 5:51 PM, Arturo Rinaldi <arty...@gmail.com> 
>>>> wrote:
>>>> > GYPFLAGS="-Dv8_use_mips_abi_hardfloat=false -Dv8_can_use_*fpiu*
>>>> _instructions=false"
>>>>
>>>
>>> If "fpiu" is not just a typo in your mail, then that's your problem. It 
>>> should be "fpu".
>>>  
>>>
>>>> >
>>>> > I have recently built the latest tarball of node, v0.12.0, for my 
>>>> platform but get this "Floatgin Point error" exception
>>>> >
>>>> > any hint on this issue ?
>>>> >
>>>> >    Regards, Arturo
>>>>
>>>> You can find out what GYP flags are supported by checking the
>>>> 'variables' sections in the various *.gyp and *.gypi files that are
>>>> scattered across the source tree.
>>>>
>>>> --
>>>> --
>>>> v8-users mailing list
>>>> v8-u...@googlegroups.com
>>>> http://groups.google.com/group/v8-users
>>>> ---
>>>> You received this message because you are subscribed to the Google 
>>>> Groups "v8-users" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send 
>>>> an email to v8-users+u...@googlegroups.com.
>>>> For more options, visit https://groups.google.com/d/optout.
>>>>
>>>
>>>
>> -- 
>> -- 
>> v8-users mailing list
>> v8-u...@googlegroups.com
>> http://groups.google.com/group/v8-users
>> --- 
>> You received this message because you are subscribed to the Google Groups 
>> "v8-users" group.
>> To unsubscribe from this group and stop receiving emails from it, send an 
>> email to v8-users+u...@googlegroups.com.
>> For more options, visit https://groups.google.com/d/optout.
>>
>>
>>
> -- 
> -- 
> v8-users mailing list
> v8-u...@googlegroups.com <javascript:>
> http://groups.google.com/group/v8-users
> --- 
> You received this message because you are subscribed to the Google Groups 
> "v8-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to v8-users+u...@googlegroups.com <javascript:>.
> For more options, visit https://groups.google.com/d/optout.
>
>
>

-- 
-- 
v8-users mailing list
v8-users@googlegroups.com
http://groups.google.com/group/v8-users
--- 
You received this message because you are subscribed to the Google Groups 
"v8-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to v8-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to