Re: Correct place to catch SIGXCPU

--===============0820131594==
Content-Type: multipart/alternative; boundary=90e6ba181850fc650504a00a6eec

--90e6ba181850fc650504a00a6eec
Content-Type: text/plain; charset=ISO-8859-1

SIGXCPU is used by both mono GC's to restart a stopped thread.

The pattern is the following:

//stop a thread, collect its register and wait for restart signal
pthread_kill (thread, SIGPWR)

//restart a thread
pthread_kill (thread, SIGXCPU)

The signal handler for sigpwr uses sigsuspend to block until it is signaled
to resume.
You can see this in mono/metadata/sgen-gc.c suspend_handler and
libgc/pthread_stop_world.c _GC_suspend_handler.

Both use the same pattern to restart.

If sigsuspend is no good on your host, it can be  easily changed to use a
semaphore instead.

Cheers,
Rodrigo


On Sun, Apr 3, 2011 at 7:03 PM, Maik Greubel <[email protected]> wrote:

> Hello list,
>
> I've got a problem regarding the SIGXCPU signal raised obviously by garbage
> collector, which is described here:
> http://www.mono-project.com/Embedding_Mono
>
> First let me explain what I have did (I have no X11 or X.org installed):
>
> - Compile & install DirectFB (git version)
> - Compile and install gtk+ 2.24 against DirectFB
> - Compile several libraries needed by mono & gtk-sharp (both latest stable)
> - Compile and install webkit
> - Compile and install webkit-sharp
> - Create a simple C# application which makes use of the webkit-sharp
> - Run it using mono
>
> The application exits using SIGXCPU and writes "CPU time limit exceeded."
> to
> console. gdb does not really help here. I can provide all traces needed to
> find a solution for that.
>
> I strongly believe, it has something to do with my architecture and has
> nothing to do with mono itself or any of the used components. A test with
> GtkHTML patched into gtk-sharp gives the same result.
>
> Where is the correct place to catch that SIGXCPU and provide the signal
> back
> to app so my application is able to run?
>
> When the time has come and my application runs well I want to contribute
> all
> of my patches to the several projects.
>
> Thanks and regards
>
> Maik
>
> _______________________________________________
> Mono-gc-list maillist  -  [email protected]
> http://lists.ximian.com/mailman/listinfo/mono-gc-list
>

--90e6ba181850fc650504a00a6eec
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<meta charset=3D"utf-8">SIGXCPU is used by both mono GC&#39;s to restart a =
stopped thread.<div><br></div><div>The pattern is the following:</div><div>=
<br></div><div>//stop a thread, collect its register and wait for restart s=
ignal</div>
<div>pthread_kill (thread,=A0SIGPWR)</div><div><br></div><div>//restart a t=
hread</div><div><meta charset=3D"utf-8"><div>pthread_kill (thread,=A0<meta =
charset=3D"utf-8">SIGXCPU)</div><div><br></div><div>The signal handler for =
sigpwr uses=A0sigsuspend to block until it is signaled to resume.</div>
<div>You can see this in mono/metadata/sgen-gc.c suspend_handler and</div><=
div>libgc/pthread_stop_world.c=A0_GC_suspend_handler.</div><div><br></div><=
div>Both use the same pattern to restart.</div><div><br></div><div>If=A0sig=
suspend is no good on your host, it can be=A0=A0easily=A0changed to use a s=
emaphore instead.</div>
<meta charset=3D"utf-8"><meta charset=3D"utf-8"><div><br></div><div>Cheers,=
</div><div>Rodrigo</div><div><br></div><br><div class=3D"gmail_quote">On Su=
n, Apr 3, 2011 at 7:03 PM, Maik Greubel <span dir=3D"ltr">&lt;<a href=3D"ma=
ilto:[email protected]">[email protected]</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex;">Hello list,<br>
<br>
I&#39;ve got a problem regarding the SIGXCPU signal raised obviously by gar=
bage<br>
collector, which is described here:<br>
<a href=3D"http://www.mono-project.com/Embedding_Mono" target=3D"_blank">ht=
tp://www.mono-project.com/Embedding_Mono</a><br>
<br>
First let me explain what I have did (I have no X11 or X.org installed):<br=
>
<br>
- Compile &amp; install DirectFB (git version)<br>
- Compile and install gtk+ 2.24 against DirectFB<br>
- Compile several libraries needed by mono &amp; gtk-sharp (both latest sta=
ble)<br>
- Compile and install webkit<br>
- Compile and install webkit-sharp<br>
- Create a simple C# application which makes use of the webkit-sharp<br>
- Run it using mono<br>
<br>
The application exits using SIGXCPU and writes &quot;CPU time limit exceede=
d.&quot; to<br>
console. gdb does not really help here. I can provide all traces needed to<=
br>
find a solution for that.<br>
<br>
I strongly believe, it has something to do with my architecture and has<br>
nothing to do with mono itself or any of the used components. A test with<b=
r>
GtkHTML patched into gtk-sharp gives the same result.<br>
<br>
Where is the correct place to catch that SIGXCPU and provide the signal bac=
k<br>
to app so my application is able to run?<br>
<br>
When the time has come and my application runs well I want to contribute al=
l<br>
of my patches to the several projects.<br>
<br>
Thanks and regards<br>
<br>
Maik<br>
<br>
_______________________________________________<br>
Mono-gc-list maillist =A0- =A0<a href=3D"mailto:[email protected]=
om">[email protected]</a><br>
<a href=3D"http://lists.ximian.com/mailman/listinfo/mono-gc-list" target=3D=
"_blank">http://lists.ximian.com/mailman/listinfo/mono-gc-list</a><br>
</blockquote></div><br></div>

--90e6ba181850fc650504a00a6eec--

--===============0820131594==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Mono-gc-list maillist  -  [email protected]
http://lists.ximian.com/mailman/listinfo/mono-gc-list

--===============0820131594==--