www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - dmd 2.064 beta 4 badly broken: dirEntries segfaults + other

reply Timothee Cour <thelastmammoth gmail.com> writes:
see my recent regressions I posted on bugzilla.
Oct 30 2013
next sibling parent Lionello Lunesu <lionello lunesu.remove.com> writes:
On 10/31/13, 2:19, Timothee Cour wrote:

 see my recent regressions I posted on bugzilla.


Ugh, you don't want to know how much time I wasted on this. I was 
convinced it was something I had done, some old phobos somewhere.
Oct 31 2013
prev sibling parent reply "monarch_dodra" <monarchdodra gmail.com> writes:
On Thursday, 31 October 2013 at 01:19:59 UTC, Timothee Cour wrote:

 see my recent regressions I posted on bugzilla.


Adding a link always helps:
http://d.puremagic.com/issues/show_bug.cgi?id=11392

Do you have any idea if this worked in beta 1-3 ? It would help 
locate the regression. I *pray* it's not in emplace :/
Oct 31 2013
next sibling parent reply "Gary Willoughby" <dev nomad.so> writes:
On Thursday, 31 October 2013 at 08:42:49 UTC, monarch_dodra wrote:

 On Thursday, 31 October 2013 at 01:19:59 UTC, Timothee Cour 
 wrote:

 see my recent regressions I posted on bugzilla.


 Adding a link always helps:
 http://d.puremagic.com/issues/show_bug.cgi?id=11392

 Do you have any idea if this worked in beta 1-3 ? It would help 
 locate the regression. I *pray* it's not in emplace :/


This bug also seems to affect GDC and LDC.

http://forum.dlang.org/thread/wzzfsnshgdjdoypanamp forum.dlang.org#post-qrftnizdsbuimxicolov:40forum.dlang.org
Oct 31 2013
next sibling parent Iain Buclaw <ibuclaw ubuntu.com> writes:
On 31 October 2013 08:56, Gary Willoughby <dev nomad.so> wrote:


 On Thursday, 31 October 2013 at 08:42:49 UTC, monarch_dodra wrote:


 On Thursday, 31 October 2013 at 01:19:59 UTC, Timothee Cour wrote:


 see my recent regressions I posted on bugzilla.


 Adding a link always helps:
 http://d.puremagic.com/issues/**show_bug.cgi?id=11392<http://d.puremagic.com/issues/show_bug.cgi?id=11392>

 Do you have any idea if this worked in beta 1-3 ? It would help locate
 the regression. I *pray* it's not in emplace :/


 This bug also seems to affect GDC and LDC.

 http://forum.dlang.org/thread/**wzzfsnshgdjdoypanamp forum.**
 dlang.org#post-**qrftnizdsbuimxicolov:40forum.**dlang.org<http://forum.dlang.org/thread/wzzfsnshgdjdoypanamp forum.dlang.org#post-qrftnizdsbuimxicolov:40forum.dlang.org>



I can't reproduce on GDC development.

-- 
Iain Buclaw

*(p < e ? p++ : p) = (c & 0x0f) + '0';
Oct 31 2013
prev sibling parent reply "David Nadlinger" <code klickverbot.at> writes:
On Thursday, 31 October 2013 at 08:56:12 UTC, Gary Willoughby 
wrote:

 This bug also seems to affect GDC and LDC.

 http://forum.dlang.org/thread/wzzfsnshgdjdoypanamp forum.dlang.org#post-qrftnizdsbuimxicolov:40forum.dlang.org


How is this related to the (since fixed) LDC dirEntries issue?

The latter was a linking problem due to LDC trying to 
second-guess DMD's (up to 2.063.2) naive template emission 
strategy.

David
Oct 31 2013
parent "Gary Willoughby" <dev nomad.so> writes:
On Thursday, 31 October 2013 at 12:00:33 UTC, David Nadlinger 
wrote:

 On Thursday, 31 October 2013 at 08:56:12 UTC, Gary Willoughby 
 wrote:

 This bug also seems to affect GDC and LDC.

 http://forum.dlang.org/thread/wzzfsnshgdjdoypanamp forum.dlang.org#post-qrftnizdsbuimxicolov:40forum.dlang.org


 How is this related to the (since fixed) LDC dirEntries issue?

 The latter was a linking problem due to LDC trying to 
 second-guess DMD's (up to 2.063.2) naive template emission 
 strategy.

 David


My mistake.
Oct 31 2013
prev sibling next sibling parent "Maxim Fomin" <maxim maxim-fomin.ru> writes:
On Thursday, 31 October 2013 at 08:42:49 UTC, monarch_dodra wrote:

 On Thursday, 31 October 2013 at 01:19:59 UTC, Timothee Cour 
 wrote:

 see my recent regressions I posted on bugzilla.


 Adding a link always helps:
 http://d.puremagic.com/issues/show_bug.cgi?id=11392

 Do you have any idea if this worked in beta 1-3 ? It would help 
 locate the regression. I *pray* it's not in emplace :/


 From what I know, direntries/ByLine bugs may be related to 
temporary structs dtor bugs in cases of exception. In some cases 
dmd incorrectly generates code, so if you have within function 
scope struct dtor invocation and call to some function which 
throws exception, there are may be segfaults. These bugs are 
sensitive to g and O switches and typically involve memory 
corruption in Refcounted struct dtor.

Probably related

http://d.puremagic.com/issues/show_bug.cgi?id=10723

http://d.puremagic.com/issues/show_bug.cgi?id=9438

I haven't debugged new issues so don't take this for sure.

What is more important that there is issue in bugzilla which 
likely is root of these dtor/RefCounted/dirEntries/ByLine 
problems which I was debugging and which is not still fixed, but 
I cannot find it right now.
Oct 31 2013
prev sibling parent Jacob Carlborg <doob me.com> writes:
On 2013-10-31 09:42, monarch_dodra wrote:


 Adding a link always helps:
 http://d.puremagic.com/issues/show_bug.cgi?id=11392

 Do you have any idea if this worked in beta 1-3 ? It would help locate
 the regression. I *pray* it's not in emplace :/


Just run a git bisect with an automated test.

-- 
/Jacob Carlborg
Oct 31 2013