www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - druntime investigation troubles

reply Denis Shelomovskij <verylonglogin.reg gmail.com> writes:
Looks like `_STI_monitor_staticctor` is called by C runtime on Windows. 
And C runtime isn't open-source.

It creates troubles for investigation druntime (for me at least). Why is 
it so? Why not to call everything in C `main`, what's the reason?

If there is no list of stuff done before C `main` druntime isn't 
open-source for me because I can't understand what is taking place just 
with druntime sources.

-- 
Денис В. Шеломовский
Denis V. Shelomovskij
May 20 2012
parent reply =?UTF-8?B?QWxleCBSw7hubmUgUGV0ZXJzZW4=?= <alex lycus.org> writes:
On 20-05-2012 10:41, Denis Shelomovskij wrote:

 Looks like `_STI_monitor_staticctor` is called by C runtime on Windows.
 And C runtime isn't open-source.

 It creates troubles for investigation druntime (for me at least). Why is
 it so? Why not to call everything in C `main`, what's the reason?

 If there is no list of stuff done before C `main` druntime isn't
 open-source for me because I can't understand what is taking place just
 with druntime sources.



Huh? It's perfectly open source. Read src/rt/monitor.c.

-- 
Alex Rønne Petersen
alex lycus.org
http://lycus.org
May 20 2012
next sibling parent reply =?UTF-8?B?QWxleCBSw7hubmUgUGV0ZXJzZW4=?= <alex lycus.org> writes:
On 20-05-2012 18:20, Alex Rønne Petersen wrote:

 On 20-05-2012 10:41, Denis Shelomovskij wrote:

 Looks like `_STI_monitor_staticctor` is called by C runtime on Windows.
 And C runtime isn't open-source.

 It creates troubles for investigation druntime (for me at least). Why is
 it so? Why not to call everything in C `main`, what's the reason?

 If there is no list of stuff done before C `main` druntime isn't
 open-source for me because I can't understand what is taking place just
 with druntime sources.



 Huh? It's perfectly open source. Read src/rt/monitor.c.



Seems like I misunderstood what you were saying. Right, the C runtime on 
*Windows* is closed source. But, I don't know why you think that 
function is called by the C runtime; see src/rt/dmain2.d.

-- 
Alex Rønne Petersen
alex lycus.org
http://lycus.org
May 20 2012
parent reply Jacob Carlborg <doob me.com> writes:
On 2012-05-20 18:25, Alex Rønne Petersen wrote:


 Seems like I misunderstood what you were saying. Right, the C runtime on
 *Windows* is closed source. But, I don't know why you think that
 function is called by the C runtime; see src/rt/dmain2.d.



Have a look again. It's only called on Posix:

https://github.com/D-Programming-Language/druntime/blob/7d663821d39cfe8874cb95b0df46b5065a770cef/src/rt/dmain2.d#L364

-- 
/Jacob Carlborg
May 20 2012
parent reply =?UTF-8?B?QWxleCBSw7hubmUgUGV0ZXJzZW4=?= <alex lycus.org> writes:
On 20-05-2012 22:13, Jacob Carlborg wrote:

 On 2012-05-20 18:25, Alex Rønne Petersen wrote:


 Seems like I misunderstood what you were saying. Right, the C runtime on
 *Windows* is closed source. But, I don't know why you think that
 function is called by the C runtime; see src/rt/dmain2.d.



 Have a look again. It's only called on Posix:

 https://github.com/D-Programming-Language/druntime/blob/7d663821d39cfe8874cb95b0df46b5065a770cef/src/rt/dmain2.d#L364



I stand corrected. I had no idea about the magic involved here!

The proprietary Windows tool chain is seriously problematic...

-- 
Alex Rønne Petersen
alex lycus.org
http://lycus.org
May 20 2012
parent reply Denis Shelomovskij <verylonglogin.reg gmail.com> writes:
21.05.2012 2:13, Alex Rønne Petersen написал:

 On 20-05-2012 22:13, Jacob Carlborg wrote:

 On 2012-05-20 18:25, Alex Rønne Petersen wrote:


 Seems like I misunderstood what you were saying. Right, the C runtime on
 *Windows* is closed source. But, I don't know why you think that
 function is called by the C runtime; see src/rt/dmain2.d.



 Have a look again. It's only called on Posix:

 https://github.com/D-Programming-Language/druntime/blob/7d663821d39cfe8874cb95b0df46b5065a770cef/src/rt/dmain2.d#L364



 I stand corrected. I had no idea about the magic involved here!

 The proprietary Windows tool chain is seriously problematic...



So can anybody do something with it? At least document a bit what does 
proprietary part do.

-- 
Денис В. Шеломовский
Denis V. Shelomovskij
May 22 2012
next sibling parent Jacob Carlborg <doob me.com> writes:
On 2012-05-22 23:01, Steven Schveighoffer wrote:


 It looks like code that is not called on Windows. Which doesn't make
 sense. It would seem that you must initialize a critical section in
 order to use it.

 I can't find any reference to STI_monitor in dmd, dmc, or druntime
 source code, except those calls that are done for Posix only. This isn't
 some closed-source mystery, I think it is just unused code.

 Sean, does this make sense? Are we using uninitialized critical sections?

 -Steve



"_STI_critical_init" is defined both in critical.c and critical_.d, but 
I can't find a place where it's called on Windows.

-- 
/Jacob Carlborg
May 22 2012
prev sibling parent reply Jacob Carlborg <doob me.com> writes:
On 2012-05-22 23:01, Steven Schveighoffer wrote:


 It looks like code that is not called on Windows. Which doesn't make
 sense. It would seem that you must initialize a critical section in
 order to use it.

 I can't find any reference to STI_monitor in dmd, dmc, or druntime
 source code, except those calls that are done for Posix only. This isn't
 some closed-source mystery, I think it is just unused code.

 Sean, does this make sense? Are we using uninitialized critical sections?

 -Steve



I found a call now, it's in "_d_criticalenter", both in critical.c and 
critical_.d.

-- 
/Jacob Carlborg
May 22 2012
next sibling parent Jacob Carlborg <doob me.com> writes:
On 2012-05-23 13:31, Steven Schveighoffer wrote:

 On Wed, 23 May 2012 02:21:14 -0400, Jacob Carlborg <doob me.com> wrote:


 On 2012-05-22 23:01, Steven Schveighoffer wrote:


 It looks like code that is not called on Windows. Which doesn't make
 sense. It would seem that you must initialize a critical section in
 order to use it.

 I can't find any reference to STI_monitor in dmd, dmc, or druntime
 source code, except those calls that are done for Posix only. This isn't
 some closed-source mystery, I think it is just unused code.

 Sean, does this make sense? Are we using uninitialized critical
 sections?

 -Steve



 I found a call now, it's in "_d_criticalenter", both in critical.c and
 critical_.d.



 OK, so does this mean monitor is only used on Posix, and critical is
 used only on windows? This seems a little weird. Especially since we are
 still compiling monitor.c.

 -Steve



No, there's a "_d_criticalenter" for Posix in critical(_).d/c which 
calls "_STI_critical_init". It seems there's a duplicate call for 
"_STI_critical_init" on Posix, in dmain2.d and in critical(_).d/c.

Also this actually looks like a big mess:

* critical_.d and critical.c contains the same code. critical_.d is used 
in the posix makefile. critical.c _and_ critical_.d is used in the win32 
makefile

* "_STI_critical_init" is implemented on both Posix and Windows, but 
it's only called on Posix

* "_STI_critical_init" is called on multiple places on Posix. In 
dmain2.d and in "_d_criticalenter".

-- 
/Jacob Carlborg
May 23 2012
prev sibling parent reply Denis Shelomovskij <verylonglogin.reg gmail.com> writes:
23.05.2012 10:21, Jacob Carlborg написал:

 On 2012-05-22 23:01, Steven Schveighoffer wrote:


 It looks like code that is not called on Windows. Which doesn't make
 sense. It would seem that you must initialize a critical section in
 order to use it.

 I can't find any reference to STI_monitor in dmd, dmc, or druntime
 source code, except those calls that are done for Posix only. This isn't
 some closed-source mystery, I think it is just unused code.

 Sean, does this make sense? Are we using uninitialized critical sections?

 -Steve



 I found a call now, it's in "_d_criticalenter", both in critical.c and
 critical_.d.



What call have you found in "_d_criticalenter"?

By the way, `_STI_critical_init` is called before C main (uncomment 
printf's and check), so it is definitely called not by druntime.

-- 
Денис В. Шеломовский
Denis V. Shelomovskij
May 23 2012
parent reply Jacob Carlborg <doob me.com> writes:
On 2012-05-23 19:32, Denis Shelomovskij wrote:


 What call have you found in "_d_criticalenter"?

 By the way, `_STI_critical_init` is called before C main (uncomment
 printf's and check), so it is definitely called not by druntime.



"_STI_critical_init" is called here:

* 
https://github.com/D-Programming-Language/druntime/blob/master/src/rt/critical_.d#L149

* 
https://github.com/D-Programming-Language/druntime/blob/master/src/rt/critical.c#L124

* 
https://github.com/D-Programming-Language/druntime/blob/master/src/rt/dmain2.d#L364

Found another one:

* 
https://github.com/D-Programming-Language/druntime/blob/master/src/rt/dmain2.d#L255

I assume, and hope, that only one of these are actually called. That's 
why I said all this is a big mess.

The last one is only called if the runtime is manually initialized, i.e. 
via C using the "rt_init" function. Which is stupid, "rt_init" should be 
called from the C main function as well. There's unnecessary code 
duplication here.

-- 
/Jacob Carlborg
May 23 2012
next sibling parent Jacob Carlborg <doob me.com> writes:
On 2012-05-23 21:58, Steven Schveighoffer wrote:


 +1

 dmain2 has that "magic runtime" feel where you look at it and are so
 confused that you are afraid to touch anything :P

 -Steve



I previously made a pull request and tried to fix code duplication in 
dmain2.d. Unfortunately some test case failed, I'm going to look into it 
again.

-- 
/Jacob Carlborg
May 23 2012
prev sibling parent Jacob Carlborg <doob me.com> writes:
On 2012-05-24 21:49, Sean Kelly wrote:


 There's a difference in how unittest execution is handled.  This could
probably be sorted out though.



Yeah, I've noticed that. Does the unit test really need to run when you 
call rt_init from C?

-- 
/Jacob Carlborg
May 24 2012
prev sibling next sibling parent "Steven Schveighoffer" <schveiguy yahoo.com> writes:
On Tue, 22 May 2012 15:16:30 -0400, Denis Shelomovskij  =

<verylonglogin.reg gmail.com> wrote:


 21.05.2012 2:13, Alex R=C3=B8nne Petersen =D0=BD=D0=B0=D0=BF=D0=B8=D1=81=




 On 20-05-2012 22:13, Jacob Carlborg wrote:

 On 2012-05-20 18:25, Alex R=C3=B8nne Petersen wrote:


 Seems like I misunderstood what you were saying. Right, the C runti=














 on
 *Windows* is closed source. But, I don't know why you think that
 function is called by the C runtime; see src/rt/dmain2.d.



 Have a look again. It's only called on Posix:

 https://github.com/D-Programming-Language/druntime/blob/7d663821d39c=












 I stand corrected. I had no idea about the magic involved here!

 The proprietary Windows tool chain is seriously problematic...



 So can anybody do something with it? At least document a bit what does=





 proprietary part do.



It looks like code that is not called on Windows.  Which doesn't make  =

sense.  It would seem that you must initialize a critical section in ord=
er  =

to use it.

I can't find any reference to STI_monitor in dmd, dmc, or druntime sourc=
e  =

code, except those calls that are done for Posix only.  This isn't some =
 =

closed-source mystery, I think it is just unused code.

Sean, does this make sense?  Are we using uninitialized critical section=
s?

-Steve



 May 22 2012



prev sibling next sibling parent  "Steven Schveighoffer" <schveiguy yahoo.com>  writes:


On Wed, 23 May 2012 02:21:14 -0400, Jacob Carlborg <doob me.com> wrote:


 On 2012-05-22 23:01, Steven Schveighoffer wrote:


 It looks like code that is not called on Windows. Which doesn't make
 sense. It would seem that you must initialize a critical section in
 order to use it.

 I can't find any reference to STI_monitor in dmd, dmc, or druntime
 source code, except those calls that are done for Posix only. This isn't
 some closed-source mystery, I think it is just unused code.

 Sean, does this make sense? Are we using uninitialized critical  
 sections?

 -Steve



 I found a call now, it's in "_d_criticalenter", both in critical.c and  
 critical_.d.



OK, so does this mean monitor is only used on Posix, and critical is used  
only on windows?  This seems a little weird.  Especially since we are  
still compiling monitor.c.

-Steve



 May 23 2012



prev sibling next sibling parent  "Steven Schveighoffer" <schveiguy yahoo.com>  writes:


On Wed, 23 May 2012 15:46:16 -0400, Jacob Carlborg <doob me.com> wrote:


 On 2012-05-23 19:32, Denis Shelomovskij wrote:


 What call have you found in "_d_criticalenter"?

 By the way, `_STI_critical_init` is called before C main (uncomment
 printf's and check), so it is definitely called not by druntime.



 "_STI_critical_init" is called here:

 *  
 https://github.com/D-Programming-Language/druntime/blob/master/src/rt/critical_.d#L149

 *  
 https://github.com/D-Programming-Language/druntime/blob/master/src/rt/critical.c#L124

 *  
 https://github.com/D-Programming-Language/druntime/blob/master/src/rt/dmain2.d#L364

 Found another one:

 *  
 https://github.com/D-Programming-Language/druntime/blob/master/src/rt/dmain2.d#L255

 I assume, and hope, that only one of these are actually called. That's  
 why I said all this is a big mess.

 The last one is only called if the runtime is manually initialized, i.e.  
 via C using the "rt_init" function. Which is stupid, "rt_init" should be  
 called from the C main function as well. There's unnecessary code  
 duplication here.



+1

dmain2 has that "magic runtime" feel where you look at it and are so  
confused that you are afraid to touch anything :P

-Steve



 May 23 2012



prev sibling next sibling parent  Sean Kelly <sean invisibleduck.org>  writes:


On May 23, 2012, at 7:49 AM, Jacob Carlborg wrote:


 On 2012-05-23 13:31, Steven Schveighoffer wrote:

 On Wed, 23 May 2012 02:21:14 -0400, Jacob Carlborg <doob me.com> =







=20

 On 2012-05-22 23:01, Steven Schveighoffer wrote:
=20

 It looks like code that is not called on Windows. Which doesn't =













 sense. It would seem that you must initialize a critical section in
 order to use it.
=20
 I can't find any reference to STI_monitor in dmd, dmc, or druntime
 source code, except those calls that are done for Posix only. This =













 some closed-source mystery, I think it is just unused code.
=20
 Sean, does this make sense? Are we using uninitialized critical
 sections?
=20
 -Steve



 I found a call now, it's in "_d_criticalenter", both in critical.c =










 critical_.d.
=20



 OK, so does this mean monitor is only used on Posix, and critical is
 used only on windows? This seems a little weird. Especially since we =







 still compiling monitor.c.
=20
 -Steve



 No, there's a "_d_criticalenter" for Posix in critical(_).d/c which =



"_STI_critical_init" on Posix, in dmain2.d and in critical(_).d/c.

=20
 Also this actually looks like a big mess:
=20
 * critical_.d and critical.c contains the same code. critical_.d is =



win32 makefile

=20
 * "_STI_critical_init" is implemented on both Posix and Windows, but =




=20
 * "_STI_critical_init" is called on multiple places on Posix. In =




If I remember correctly, Brad Roberts tried porting the old C critical =
section code to D a while back and something didn't work right.  Both =
versions of the code still exist, and at some point the D code needs to =
be sorted and the C code should be thrown away.  I suppose I should make =
this my top priority, when I find the time.=



 May 24 2012



prev sibling next sibling parent  Sean Kelly <sean invisibleduck.org>  writes:


On May 22, 2012, at 2:01 PM, Steven Schveighoffer wrote:


 On Tue, 22 May 2012 15:16:30 -0400, Denis Shelomovskij =




=20

 21.05.2012 2:13, Alex R=C3=B8nne Petersen =D0=BD=D0=B0=D0=BF=D0=B8=D1=81=







 On 20-05-2012 22:13, Jacob Carlborg wrote:

 On 2012-05-20 18:25, Alex R=C3=B8nne Petersen wrote:
=20

 Seems like I misunderstood what you were saying. Right, the C =
















 *Windows* is closed source. But, I don't know why you think that
 function is called by the C runtime; see src/rt/dmain2.d.



 Have a look again. It's only called on Posix:
=20
 =












cb95b0df46b5065a770cef/src/rt/dmain2.d#L364

=20
=20
=20



 I stand corrected. I had no idea about the magic involved here!
=20
 The proprietary Windows tool chain is seriously problematic...
=20



 So can anybody do something with it? At least document a bit what =







=20



 It looks like code that is not called on Windows.  Which doesn't make =



order to use it.

=20
 I can't find any reference to STI_monitor in dmd, dmc, or druntime =



isn't some closed-source mystery, I think it is just unused code.

=20
 Sean, does this make sense?  Are we using uninitialized critical =




This code is before my time, but I believe that DMC implicitly treats =
STI functions as module ctors, and DMD inherits this behavior because it =
shares a C runtime with DMC.  It's been a while since I've looked at all =
of this, but the full C runtime source is shipped with DMC--I have a =
copy.=



 May 24 2012



prev sibling next sibling parent  Sean Kelly <sean invisibleduck.org>  writes:


On May 23, 2012, at 12:46 PM, Jacob Carlborg wrote:


 On 2012-05-23 19:32, Denis Shelomovskij wrote:
=20

 What call have you found in "_d_criticalenter"?
=20
 By the way, `_STI_critical_init` is called before C main (uncomment
 printf's and check), so it is definitely called not by druntime.
=20



 "_STI_critical_init" is called here:
=20
 * =



ical_.d#L149

=20
 * =



ical.c#L124

=20
 * =



n2.d#L364

=20
 Found another one:
=20
 * =



n2.d#L255

=20
 I assume, and hope, that only one of these are actually called. That's =




=20
 The last one is only called if the runtime is manually initialized, =



should be called from the C main function as well. There's unnecessary =
code duplication here.

There's a difference in how unittest execution is handled.  This could =
probably be sorted out though.=



 May 24 2012



prev sibling next sibling parent  "Steven Schveighoffer" <schveiguy yahoo.com>  writes:


On Thu, 24 May 2012 15:44:28 -0400, Sean Kelly <sean invisibleduck.org> =
 =

wrote:


 On May 22, 2012, at 2:01 PM, Steven Schveighoffer wrote:


 On Tue, 22 May 2012 15:16:30 -0400, Denis Shelomovskij  =







 <verylonglogin.reg gmail.com> wrote:


 21.05.2012 2:13, Alex R=C3=B8nne Petersen =D0=BD=D0=B0=D0=BF=D0=B8=D1=










 On 20-05-2012 22:13, Jacob Carlborg wrote:

 On 2012-05-20 18:25, Alex R=C3=B8nne Petersen wrote:


 Seems like I misunderstood what you were saying. Right, the C  =



















 runtime on
 *Windows* is closed source. But, I don't know why you think that
 function is called by the C runtime; see src/rt/dmain2.d.



 Have a look again. It's only called on Posix:

 https://github.com/D-Programming-Language/druntime/blob/7d663821d3=


















 I stand corrected. I had no idea about the magic involved here!

 The proprietary Windows tool chain is seriously problematic...



 So can anybody do something with it? At least document a bit what do=











 proprietary part do.



 It looks like code that is not called on Windows.  Which doesn't make=








 sense.  It would seem that you must initialize a critical section in =








 order to use it.

 I can't find any reference to STI_monitor in dmd, dmc, or druntime  =







 source code, except those calls that are done for Posix only.  This  =







 isn't some closed-source mystery, I think it is just unused code.

 Sean, does this make sense?  Are we using uninitialized critical  =







 sections?



 This code is before my time, but I believe that DMC implicitly treats =





 STI functions as module ctors, and DMD inherits this behavior because =





 shares a C runtime with DMC.  It's been a while since I've looked at a=





 of this, but the full C runtime source is shipped with DMC--I have a  =




 copy.



Yeah, I do too.  I did a full text search for STI_monitor and found  =

nothing.

That's why I'm confused...

-Steve



 May 24 2012



prev sibling next sibling parent  Sean Kelly <sean invisibleduck.org>  writes:


On May 24, 2012, at 1:27 PM, "Steven Schveighoffer" <schveiguy yahoo.com> wr=
ote:


 On Thu, 24 May 2012 15:44:28 -0400, Sean Kelly <sean invisibleduck.org> wr=




=20

 On May 22, 2012, at 2:01 PM, Steven Schveighoffer wrote:
=20

 On Tue, 22 May 2012 15:16:30 -0400, Denis Shelomovskij <verylonglogin.re=










=20

 21.05.2012 2:13, Alex R=C3=B8nne Petersen =D0=BD=D0=B0=D0=BF=D0=B8=D1=81=













 On 20-05-2012 22:13, Jacob Carlborg wrote:

 On 2012-05-20 18:25, Alex R=C3=B8nne Petersen wrote:
=20

 Seems like I misunderstood what you were saying. Right, the C runtim=






















 *Windows* is closed source. But, I don't know why you think that
 function is called by the C runtime; see src/rt/dmain2.d.



 Have a look again. It's only called on Posix:
=20
 https://github.com/D-Programming-Language/druntime/blob/7d663821d39cf=



















=20
=20
=20



 I stand corrected. I had no idea about the magic involved here!
=20
 The proprietary Windows tool chain is seriously problematic...
=20



 So can anybody do something with it? At least document a bit what does p=













=20



 It looks like code that is not called on Windows.  Which doesn't make se=









se it.

=20
 I can't find any reference to STI_monitor in dmd, dmc, or druntime sourc=









osed-source mystery, I think it is just unused code.

=20
 Sean, does this make sense?  Are we using uninitialized critical section=










=20
 This code is before my time, but I believe that DMC implicitly treats STI=






 a C runtime with DMC.  It's been a while since I've looked at all of this, b=
ut the full C runtime source is shipped with DMC--I have a copy.

=20
 Yeah, I do too.  I did a full text search for STI_monitor and found nothin=




=20
 That's why I'm confused...



It's not STI_monitor specifically. STI_anything is a module ctor and STD_any=
thing is a module dtor. Or at least that's my recollection. There are a few p=
laces in the DMC runtime that have initializers like this. I think one is so=
mewhere in the floating-point support code.=20=



 May 24 2012



prev sibling next sibling parent  "Steven Schveighoffer" <schveiguy yahoo.com>  writes:


On Thu, 24 May 2012 16:40:46 -0400, Sean Kelly <sean invisibleduck.org>  
wrote:


 It's not STI_monitor specifically. STI_anything is a module ctor and  
 STD_anything is a module dtor. Or at least that's my recollection. There  
 are a few places in the DMC runtime that have initializers like this. I  
 think one is somewhere in the floating-point support code.



oh.  That's.... horrible.

Yeah, we should eliminate this ASAP.  Is there any reason it has to be  
done before C's main?

-Steve



 May 25 2012



prev sibling next sibling parent  Sean Kelly <sean invisibleduck.org>  writes:


On May 25, 2012, at 7:31 AM, "Steven Schveighoffer" <schveiguy yahoo.com> wr=
ote:


 On Thu, 24 May 2012 16:40:46 -0400, Sean Kelly <sean invisibleduck.org> wr=




=20

 It's not STI_monitor specifically. STI_anything is a module ctor and STD_=






ew places in the DMC runtime that have initializers like this. I think one i=
s somewhere in the floating-point support code.

=20
 oh.  That's.... horrible.
=20
 Yeah, we should eliminate this ASAP.  Is there any reason it has to be don=




None that I can think of. I bet it's just really old code.=20=



 May 25 2012



prev sibling  parent  Sean Kelly <sean invisibleduck.org>  writes:


On May 24, 2012, at 11:25 PM, Jacob Carlborg <doob me.com> wrote:


 On 2012-05-24 21:49, Sean Kelly wrote:
=20

 There's a difference in how unittest execution is handled.  This could pr=







=20
 Yeah, I've noticed that. Does the unit test really need to run when you ca=




Not really. I wanted initialization to be the same either way, but unittests=
 seem optional.=20=



 May 25 2012