Re: [webkit-dev] Webkit JavaScript interpreter crash (v. 153112/ppc64le).

2015-10-27 Thread Atul Sowani 
I tried with JSC_useJIT=0 as well as with JSC_useLLInt=0, however that
did not help. The crash is still happening. First there was an error
about page alignment, which I bypassed. Now the program hangs in
dtoa() --> mult(). One doubt: wouldn't dtoa code (in dtoa.cpp/.h) need
specific porting for ppc64le?

On Tue, Oct 27, 2015 at 10:48 AM, Atul Sowani <sowa...@us.ibm.com> wrote:
> Hi Ryosuke,
>
> Thanks for the JSC_useJIT option, I was not aware of it. I will try to use
> webkit with JSC_useJIT=0 option. However, I am doubtful if I can replace
> existing webkit used in PhantomJS with the latest version, since it will
> require integration with the PhantomJS code. I am going to check both these
> options now (env. var and feasibility of using new webkit version) and will
> update this forum with my findings.
>
> Thanks,
> Atul.
>
>
>
> From:Ryosuke Niwa <rn...@webkit.org>
> To:Atul Sowani/Beaverton/Contr/IBM@IBMUS
> Cc:"webkit-dev@lists.webkit.org" <webkit-dev@lists.webkit.org>
> Date:    10/26/2015 07:15 PM
> Subject:Re: [webkit-dev] Webkit JavaScript interpreter crash (v.
> 153112/ppc64le).
> Sent by:ryosuke.n...@gmail.com
> 
>
>
>
>
> On Monday, October 26, 2015, Atul Sowani <sowa...@us.ibm.com> wrote:
> I am trying to port PhantomJS 2.0.1 to IBM PPC64LE platform. PhantomJS uses
> qtwebkit version 153112 and I am facing a few crashes in the JavaScript
> interpreter code. I was directed to this forum by PhantomJS developers to
> get help/insight about the interpreter.
>
> That WebKit revision is more than two years old:
> http://trac.webkit.org/changeset/153112
>
> Requesting this forum admins/contributors to confirm if this is the right
> forum to discuss JS interpreter issues, and is it mandatory to open a bug
> report in order to discuss it on this mailing list.
>
> Can you reproduce the same interpreter crash on Mac or GTK+ if you built
> trunk WebKit?  (You can export JSC_useJIT=0 in env to disable JIT)
>
> If not, can reproduce the crash on Mac/GTK+ ports at r153112?  If you can,
> you can probably binary search the revision that fixed the crash.
>
> - R. Niwa
>
>
>
> --
> - R. Niwa
>
>
> ___
> webkit-dev mailing list
> webkit-dev@lists.webkit.org
> https://lists.webkit.org/mailman/listinfo/webkit-dev
>
___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev

Re: [webkit-dev] Webkit JavaScript interpreter crash (v. 153112/ppc64le).

2015-10-27 Thread Filip Pizlo 
Usually, making JSC work on a new platform requires a non-trivial engineering 
investment. I wouldn't expect any version of JSC to work on PPC out of the box. 

-Filip

> On Oct 27, 2015, at 4:15 AM, Atul Sowani <sow...@gmail.com> wrote:
> 
> I tried with JSC_useJIT=0 as well as with JSC_useLLInt=0, however that
> did not help. The crash is still happening. First there was an error
> about page alignment, which I bypassed. Now the program hangs in
> dtoa() --> mult(). One doubt: wouldn't dtoa code (in dtoa.cpp/.h) need
> specific porting for ppc64le?
> 
>> On Tue, Oct 27, 2015 at 10:48 AM, Atul Sowani <sowa...@us.ibm.com> wrote:
>> Hi Ryosuke,
>> 
>> Thanks for the JSC_useJIT option, I was not aware of it. I will try to use
>> webkit with JSC_useJIT=0 option. However, I am doubtful if I can replace
>> existing webkit used in PhantomJS with the latest version, since it will
>> require integration with the PhantomJS code. I am going to check both these
>> options now (env. var and feasibility of using new webkit version) and will
>> update this forum with my findings.
>> 
>> Thanks,
>> Atul.
>> 
>> 
>> 
>> From:Ryosuke Niwa <rn...@webkit.org>
>> To:Atul Sowani/Beaverton/Contr/IBM@IBMUS
>> Cc:    "webkit-dev@lists.webkit.org" <webkit-dev@lists.webkit.org>
>> Date:10/26/2015 07:15 PM
>> Subject:Re: [webkit-dev] Webkit JavaScript interpreter crash (v.
>> 153112/ppc64le).
>> Sent by:ryosuke.n...@gmail.com
>> 
>> 
>> 
>> 
>> 
>> On Monday, October 26, 2015, Atul Sowani <sowa...@us.ibm.com> wrote:
>> I am trying to port PhantomJS 2.0.1 to IBM PPC64LE platform. PhantomJS uses
>> qtwebkit version 153112 and I am facing a few crashes in the JavaScript
>> interpreter code. I was directed to this forum by PhantomJS developers to
>> get help/insight about the interpreter.
>> 
>> That WebKit revision is more than two years old:
>> http://trac.webkit.org/changeset/153112
>> 
>> Requesting this forum admins/contributors to confirm if this is the right
>> forum to discuss JS interpreter issues, and is it mandatory to open a bug
>> report in order to discuss it on this mailing list.
>> 
>> Can you reproduce the same interpreter crash on Mac or GTK+ if you built
>> trunk WebKit?  (You can export JSC_useJIT=0 in env to disable JIT)
>> 
>> If not, can reproduce the crash on Mac/GTK+ ports at r153112?  If you can,
>> you can probably binary search the revision that fixed the crash.
>> 
>> - R. Niwa
>> 
>> 
>> 
>> --
>> - R. Niwa
>> 
>> 
>> ___
>> webkit-dev mailing list
>> webkit-dev@lists.webkit.org
>> https://lists.webkit.org/mailman/listinfo/webkit-dev
> ___
> webkit-dev mailing list
> webkit-dev@lists.webkit.org
> https://lists.webkit.org/mailman/listinfo/webkit-dev
___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev

Re: [webkit-dev] Webkit JavaScript interpreter crash (v. 153112/ppc64le).

2015-10-26 Thread Atul Sowani 
Hi Ryosuke,

Thanks for the JSC_useJIT option, I was not aware of it. I will try to use 
webkit with JSC_useJIT=0 option. However, I am doubtful if I can replace 
existing webkit used in PhantomJS with the latest version, since it will 
require integration with the PhantomJS code. I am going to check both 
these options now (env. var and feasibility of using new webkit version) 
and will update this forum with my findings.

Thanks,
Atul.



From:   Ryosuke Niwa <rn...@webkit.org>
To: Atul Sowani/Beaverton/Contr/IBM@IBMUS
Cc: "webkit-dev@lists.webkit.org" <webkit-dev@lists.webkit.org>
Date:   10/26/2015 07:15 PM
Subject:    Re: [webkit-dev] Webkit JavaScript interpreter crash (v. 
153112/ppc64le).
Sent by:ryosuke.n...@gmail.com




On Monday, October 26, 2015, Atul Sowani <sowa...@us.ibm.com> wrote:
I am trying to port PhantomJS 2.0.1 to IBM PPC64LE platform. PhantomJS 
uses qtwebkit version 153112 and I am facing a few crashes in the 
JavaScript interpreter code. I was directed to this forum by PhantomJS 
developers to get help/insight about the interpreter.

That WebKit revision is more than two years old: 
http://trac.webkit.org/changeset/153112

Requesting this forum admins/contributors to confirm if this is the right 
forum to discuss JS interpreter issues, and is it mandatory to open a bug 
report in order to discuss it on this mailing list.

Can you reproduce the same interpreter crash on Mac or GTK+ if you built 
trunk WebKit?  (You can export JSC_useJIT=0 in env to disable JIT)

If not, can reproduce the crash on Mac/GTK+ ports at r153112?  If you can, 
you can probably binary search the revision that fixed the crash.

- R. Niwa
 


-- 
- R. Niwa


___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev

Re: [webkit-dev] Webkit JavaScript interpreter crash (v. 153112/ppc64le).

2015-10-26 Thread Ryosuke Niwa 
On Monday, October 26, 2015, Atul Sowani  wrote:

> I am trying to port PhantomJS 2.0.1 to IBM PPC64LE platform. PhantomJS
> uses qtwebkit version 153112 and I am facing a few crashes in the
> JavaScript interpreter code. I was directed to this forum by PhantomJS
> developers to get help/insight about the interpreter.


That WebKit revision is more than two years old:
http://trac.webkit.org/changeset/153112

Requesting this forum admins/contributors to confirm if this is the right
> forum to discuss JS interpreter issues, and is it mandatory to open a bug
> report in order to discuss it on this mailing list.
>

Can you reproduce the same interpreter crash on Mac or GTK+ if you built
trunk WebKit?  (You can export JSC_useJIT=0 in env to disable JIT)

If not, can reproduce the crash on Mac/GTK+ ports at r153112?  If you can,
you can probably binary search the revision that fixed the crash.

- R. Niwa



-- 
- R. Niwa
___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev