Re: 0x444ba45b in prof_mark () from /usr/lib/perl5/5.6.1/i686-linux/auto/Devel/DProf/DProf.so

2002-12-09 Thread Ged Haywood
Hi Stas,

On Sun, 8 Dec 2002, Stas Bekman wrote:

 I see no reason why we should mislead readers, saying do not upgrade to 
 5.8.0, but to 5.7.2 instead.

I'm sorry if I misled anyone.  I thought it was clear that my own
personal recommendation was at least 5.7.2 and that there are those
who would recommend 5.8.0 but I'm not one of them.  That's because I
have no personal experience of that release.

Anyway, if they've read all this I think they have the idea now... :)

73,
Ged.




Re: 0x444ba45b in prof_mark () from /usr/lib/perl5/5.6.1/i686-linux/auto/Devel/DProf/DProf.so

2002-12-08 Thread Ged Haywood
Hi there,

On Sat, 7 Dec 2002, Richard Clarke wrote:

 I seem to have hit a slight stumbling block in my mod_perl development.
[snip]
 As you can see from the topic, the segv happens in the DProf library.

Yup, that'll do it.

Try upgrading to Perl 5.7.2 at least.  It worked for me.

There are those who would recommend 5.8.0 minimum, but I'm not one.
(Yet:).

73,
Ged.




Re: 0x444ba45b in prof_mark () from /usr/lib/perl5/5.6.1/i686-linux/auto/Devel/DProf/DProf.so

2002-12-08 Thread Stas Bekman
Ged Haywood wrote:
[...]

Try upgrading to Perl 5.7.2 at least.  It worked for me.

There are those who would recommend 5.8.0 minimum, but I'm not one.
(Yet:).


Eh? 5.8.0 is almost the same as 5.7.2, with mucho bugs fixed. I don't 
understand why do you suggest to install an unstable developers version, 
when there is a stable and better version out there.

Whether you are running 5.7.2 or 5.8.0, do not compile with ithreads 
support, unless you need it, as it makes the interpreter slower.

__
Stas BekmanJAm_pH -- Just Another mod_perl Hacker
http://stason.org/ mod_perl Guide --- http://perl.apache.org
mailto:[EMAIL PROTECTED] http://use.perl.org http://apacheweek.com
http://modperlbook.org http://apache.org   http://ticketmaster.com



Re: 0x444ba45b in prof_mark () from /usr/lib/perl5/5.6.1/i686-linux/auto/Devel/DProf/DProf.so

2002-12-08 Thread Ged Haywood
Hi Stas,

On Sun, 8 Dec 2002, Stas Bekman wrote:

 Ged Haywood wrote:
 [...]
  Try upgrading to Perl 5.7.2 at least.  It worked for me.
  
  There are those who would recommend 5.8.0 minimum, but I'm not one.
  (Yet:).
 
 Eh? 5.8.0 is almost the same as 5.7.2, with mucho bugs fixed. I don't 
 understand why do you suggest to install an unstable developers version, 
 when there is a stable and better version out there.

Simply that I've never upgraded to 5.8.0 myself and I've seen a few
people with some troubles when they did.  I've had no trouble at all
with 5.7.2.

You may remember I had segfault problems with dprof and 5.7.1 which
went away when I upgraded to 5.7.2 (it was you who suggested it:).

73,
Ged.





Re: 0x444ba45b in prof_mark () from /usr/lib/perl5/5.6.1/i686-linux/auto/Devel/DProf/DProf.so

2002-12-08 Thread Stas Bekman
Ged Haywood wrote:

Hi Stas,

On Sun, 8 Dec 2002, Stas Bekman wrote:



Ged Haywood wrote:
[...]


Try upgrading to Perl 5.7.2 at least.  It worked for me.

There are those who would recommend 5.8.0 minimum, but I'm not one.
(Yet:).


Eh? 5.8.0 is almost the same as 5.7.2, with mucho bugs fixed. I don't 
understand why do you suggest to install an unstable developers version, 
when there is a stable and better version out there.


Simply that I've never upgraded to 5.8.0 myself and I've seen a few
people with some troubles when they did.  I've had no trouble at all
with 5.7.2.


If there are any troubles, they should be reported to p5p. Personally I 
didn't have any. Mandrake 8.0 uses 5.8.0 for its all tools and it works, 
so does the new RH.

You may remember I had segfault problems with dprof and 5.7.1 which
went away when I upgraded to 5.7.2 (it was you who suggested it:).


At that time there was no 5.8.0 available.

I see no reason why we should mislead readers, saying do not upgrade to 
5.8.0, but to 5.7.2 instead.

__
Stas BekmanJAm_pH -- Just Another mod_perl Hacker
http://stason.org/ mod_perl Guide --- http://perl.apache.org
mailto:[EMAIL PROTECTED] http://use.perl.org http://apacheweek.com
http://modperlbook.org http://apache.org   http://ticketmaster.com



0x444ba45b in prof_mark () from /usr/lib/perl5/5.6.1/i686-linux/auto/Devel/DProf/DProf.so

2002-12-07 Thread Richard Clarke
List,
I seem to have hit a slight stumbling block in my mod_perl development.
Without wanting to write a whole essay on the exact combination of events
that causes this segv to happen I'll just say for now that it happens after
my custom Template::Provider returns a status indicating a template could
not be found. My apache error log indicates that the segv happens
immediately after my contenthandler returns SERVER_ERROR.
As you can see from the topic, the segv happens in the DProf library. I have
listed the gdb output at the end of my email.
My question is simply, is this a mod_perl problem, a template toolkit
problem or a problem somewhere in the dprof library?.
If I really need to use Apache::DProf then I'll just avoid requesting
templates that don't exist. I'm still curious however, as to whether there
is a more sinister problem waiting to cause my problems in the future.
Any advice? Has anyone else ever experienced a code path which causes a
similar segv in dprof?

Richard.

(gdb) c
Continuing.

Program received signal SIGSEGV, Segmentation fault.
0x4477145b in prof_mark () from
/usr/lib/perl5/5.6.1/i686-linux/auto/Devel/DProf/DProf.so
(gdb) bt
#0  0x4477145b in prof_mark () from
/usr/lib/perl5/5.6.1/i686-linux/auto/Devel/DProf/DProf.so
#1  0x44771b5d in XS_DB_sub () from
/usr/lib/perl5/5.6.1/i686-linux/auto/Devel/DProf/DProf.so
#2  0x444e2732 in Perl_pp_entersub () from /usr/lib/libperl.so
#3  0x444dc7e0 in Perl_runops_standard () from /usr/lib/libperl.so
#4  0x444900cc in S_call_body () from /usr/lib/libperl.so
#5  0x4448c28e in perl_call_sv () from /usr/lib/libperl.so
#6  0x4448fcbe in perl_call_method () from /usr/lib/libperl.so
#7  0x0806fbe1 in perl_call_handler ()
#8  0x0806f472 in perl_run_stacked_handlers ()
#9  0x0806dfc4 in perl_handler ()
#10 0x0808e1b9 in ap_invoke_handler ()
#11 0x080a3f7f in ap_some_auth_required ()
#12 0x080a3fea in ap_process_request ()
#13 0x0809a876 in ap_child_terminate ()
#14 0x0809aa35 in ap_child_terminate ()
#15 0x0809abb6 in ap_child_terminate ()
#16 0x0809b24d in ap_child_terminate ()
#17 0x0809babc in main ()
#18 0x445743bd in __libc_start_main () from /lib/libc.so.6
(gdb)




Re: 0x444ba45b in prof_mark () from/usr/lib/perl5/5.6.1/i686-linux/auto/Devel/DProf/DProf.so

2002-12-07 Thread Ilya Martynov
 On Sat, 7 Dec 2002 19:58:40 -, Richard Clarke [EMAIL PROTECTED] said:

RC List,
RC I seem to have hit a slight stumbling block in my mod_perl development.
RC Without wanting to write a whole essay on the exact combination of events
RC that causes this segv to happen I'll just say for now that it happens after
RC my custom Template::Provider returns a status indicating a template could
RC not be found. My apache error log indicates that the segv happens
RC immediately after my contenthandler returns SERVER_ERROR.
RC As you can see from the topic, the segv happens in the DProf library. I have
RC listed the gdb output at the end of my email.
RC My question is simply, is this a mod_perl problem, a template toolkit
RC problem or a problem somewhere in the dprof library?.

Likely it is Dprof's problem. 

RC If I really need to use Apache::DProf then I'll just avoid requesting
RC templates that don't exist. I'm still curious however, as to whether there
RC is a more sinister problem waiting to cause my problems in the future.
RC Any advice? Has anyone else ever experienced a code path which causes a
RC similar segv in dprof?

Many times. Apache::DProf happily segfaults on any non-trivial
program.

-- 
Ilya Martynov,  [EMAIL PROTECTED]
CTO IPonWEB (UK) Ltd
Quality Perl Programming and Unix Support
UK managed @ offshore prices - http://www.iponweb.net
Personal website - http://martynov.org