aCC: error 1922: "/opt/aCC/lbin/ctcom" terminated with signal 10

[Srikrishna Erra]

Hi,
I am trying to compiler a CPP program and got below error.

aCC: error 1922: "/vob/hpux-intel/opt/aCC/lbin/ctcom" terminated with signal 10

I have HPUX-IA machine.
$ uname -a
HP-UX eshpx800 B.11.23 U ia64 0091977010 unlimited-user license

This time, the cpp file has undergone some changes. Earlier without latest changes there were no issues when compiled with same version of compiler.

And i don't think the changes done this time will generate a signal 10 error.

Could you help in resolving this issue/understanding resons why signal 10 is generated.

**********************************************
Here is the log.

/vob/hpux-intel/opt/aCC/bin/aCC -DU_COMMON_IMPLEMENTATION -DHAVE_CONFIG_H -I../common -I/vob/icu/src/source/common -I/vob/icu/src/source/i18n "-DDEFAULT_ICU_PLUGINS=\"/vob/icu/hpux-ia64.32/build/lib/icu\" " +O0 +u1 -D_REENTRANT +DD32 -I/vob/hpux-intel/opt/aCC/include -Wc,-ansi_for_scope,on +W740 +W749 +W823 +W4232 -I/vob/hpux-pa/opt/aCC/include_std -c +Z -o brkiter.o /vob/icu/src/source/common/brkiter.cpp
CXXOPTS=-v
/vob/hpux-intel/opt/aCC/lbin/ctcom -align one -diags 740 -diags 749 -diags 823 -diags 4232 -ia64abi all -architecture 32 -inst compiletime -diags 523 -target_os 11.23 -test namespaces -koenig_lookup
on -inline_power 1 -link_type dynamic -fpeval float -tls_dyn on -o brkiter.o -ansi_for_scope on -I../common -I/vob/icu/src/source/common -I/vob/icu/src/source/i18n -I/vob/hpux-intel/opt/aCC/include -I
/vob/hpux-pa/opt/aCC/include_std -I/vob/hpux-intel/opt/aCC/include_std -I/vob/hpux-intel/opt/aCC/include_std/iostream_compat -I/usr/include -I/usr -DU_COMMON_IMPLEMENTATION -DHAVE_CONFIG_H -DDEFAULT_ICU_PLUGINS="/vob/icu/hpux-ia64.32/build/lib/icu" -D_REENTRANT -D_HP_IA64ABI -D__hpux -D__unix -D__ia64=1 -D_BIG_ENDIAN=1 -D__STDCPP__ -D_HP_NAMESPACE_STD -D_ILP32 -D__HP_aCC=55000 -D_INLINE_ASM -D_BIND_LIBCALLS -D_Math_errhandling=MATH_ERREXCEPT -D_FLT_EVAL_METHOD=0 -plusolistoption -Ol04all! -plusolistoption -Ol11aggressive! -plusooption -Oq00,al,ag,cn,sz,ic,vo,Mf,Po,es,rs,Rf,Pr,sp,in,cl,om,vc,pi,fa,Pe,rr,pa,pv,nf,cp,lx,Pg,ug,lu,lb,uj,dn,sg,pt,kt,em,np,ar,rp,fs,bp,wp,pc,mp,lr,cx,cr,pi,so,Rc,fa,ft,fe,ap,st,lc,Bl,ib,sd,ll,rl,Lt,ol,fl,lm,ts,rd,dp,If! /vob/icu/src/source/common/brkiter.cpp
Signal 10
aCC: error 1922: "/vob/hpux-intel/opt/aCC/lbin/ctcom" terminated with signal 10
gmake: *** [brkiter.o] Error 1
$
**********************************************

Here are the details of compiler.
$ /vob/hpux-intel/opt/aCC/bin/aCC -V
aCC: HP aC++/ANSI C B3910B A.05.50 [Jan 06 2003]

$ what /vob/hpux-intel/opt/ansic/lbin/ctcom
/vob/hpux-intel/opt/ansic/lbin/ctcom:
REV: HP SESD Support Code
Ucode-2 Version 2-9
linkstamp Wed Jan 15 01:02:11 2003 PST
HP aC++/C B3910B A.05.50 [Jan 06 2003] IPF
HP Low-level optimizer B.11.23.134 (01/13/03)
Ucode2 PBO Annotator - Rev 1.1 Ucode2 2.9 - 02-Dec-02.22:41
HP IPF libflowdb - Rev 1.2 date: 12-Nov-02.15:13 [51099 544]

Thanks in advance.

regards.
Srikrishna Erra.

[Dennis Handly · Accepted solution]

>Could you help in resolving this issue/understanding reasons why signal 10 is generated.

As I mentioned to you before, aCC5 is NOT supported. The latest version is A.06.25.

Also you should not be using +O0 on Integrity.
Nor should you be using the -AP include paths:
-I/vob/hpux-intel/opt/aCC/include

If you are happy with the answers you were given, please read the following on how to assign points:
http://forums.itrc.hp.com/service/forums/helptips.do?#33

Your unassigned threads:
http://forums.itrc.hp.com/service/forums/pageList.do?userId=WW2137&listType=unassigned

[Srikrishna Erra]

Hi,
Thank you for the reply.

This is the first time i got this issue and sent a mail.

Upgrading to latest compiler is the only solution? because this issue is with only one .cpp file.

I am not sure i can move to latest compiler.

But before that i just want to know why "signal 10" i am getting? can Modifying code of .cpp file (part of the code responsible for this signal 10 error) resolve this issue.

For this i need to know first, what could be the scenarios for getting signal 10 errors.

Could you please help me resolve this issue.

Thanks

regards,
Srikrishna Erra.

[Dennis Handly]

>Upgrading to latest compiler is the only solution? I am not sure I can move to latest compiler.

That's how you solve nearly all compiler aborts.
Any particular reason you can't upgrade?

>I just want to know why signal 10 I am getting?

It's a compiler bug in an obsolete product.

>can modifying code of .cpp file resolve this issue.

Possibly, though it would be much easier if you just installed a supported compiler and if you still had problems, you could contact the Response Center.

>what could be the scenarios for getting signal 10 errors.

You would have to look at all bug reports to see what matched your case. Though without a stack trace, it would be hard to match it up.
You should at least patch to PHSS_29655, A.05.52.

[Srikrishna Erra]

Hi,
Thank you once again.

Compiler decision is not in my hands.
There were other applications uses the same compiler and a high level team exists which will look into and make decisions.

What i can do is just propose the "need/requirement".

Anyhow i will do the same.

At the same time, i will also search for patches that suites this error and will try them apart from the one you had given.

I need one more help.

I am getting just getting error without ant useful info.

aCC: error 1922: "/vob/hpux-intel/opt/aCC/lbin/ctcom" terminated with signal 10

Is there any compiler option/the way by which we can get detailed information like "where/what is" something which can be helpful?

Could you please let me know.


Thanks

regards,
Srikrishna Erra.

[Dennis Handly]

>I am getting just getting error without any useful info.
aCC: error 1922: ctcom terminated with signal 10

The compiler got another signal 10 trying to print a stack trace.

>will try them apart from the one you had given.

There won't be any except for possibly PHSS_40541, which may make the stack trace work?

>Is there any compiler option which we can get detailed information which can be helpful?

No, the compiler aborted trying to give you that. Did it produce a core file?
If so, use gdb to get a stack trace and include it here.

Otherwise you could create a .i file with "-E -.i" and chop up that file until the problem goes away.

[Srikrishna Erra]

Hi Dennis,

Core file is created.
Here is the stack trace...

********************************************
$ gdb -c core
HP gdb 5.9 for HP Itanium (32 or 64 bit) and target HP-UX 11.2x.
Copyright 1986 - 2001 Free Software Foundation, Inc.
Hewlett-Packard Wildebeest 5.9 (based on GDB) is covered by the
GNU General Public License. Type "show copying" to see the conditions to
change it and/or distribute copies. Type "show warranty" for warranty/support.


warning: Unknown symbols for 'core'; use the 'symbol-file' command.
Core was generated by `ctcom'.
Program terminated with signal 10, Bus error.
BUS_ADRALN - Invalid address alignment. Please refer to the following link that
helps in handling unaligned data: http://docs.hp.com/en/7730/newhelp0610/pragmas
.htm#pragma-pack-ex3
No object file symbols.
#0 0x60000000c1e77db0 in ()
(gdb) bt
#0 0x60000000c1e77db0 in ()
warning: Attempting to unwind past bad PC 0x60000000c1e77db0
#1 0x60000000c1e873a0 in ()
********************************************

I have done strings on core and i got...
$ strings core
yzU
04:Unable to find library '/usr/lib/nls/loc/hpux32/locales.1/C'.
$ ls /usr/lib/nls/loc/hpux32/locales.1/C
/usr/lib/nls/loc/hpux32/locales.1/C not found
$
$ ls /usr/lib/nls/loc/hpux32/locales.1/C*
/usr/lib/nls/loc/hpux32/locales.1/C.iso88591
/usr/lib/nls/loc/hpux32/locales.1/C.iso885915
/usr/lib/nls/loc/hpux32/locales.1/C.utf8

Meanwhile i will also try to create .i file.

regards,
Srikrishna Erra.

[Dennis Handly]

>$ gdb -c core

You forgot to specify the executable:
$ gdb /vob/hpux-intel/opt/aCC/lbin/ctcom core

>I have done strings on core and I got.

This is particularly useless.

If you want any more help, you need to assign points to every reply in these threads:
http://forums.itrc.hp.com/service/forums/pageList.do?userId=WW2137&listType=unassigned

[Srikrishna Erra]

Hi Dennis,
Sorry. Here is the gdb info

$ gdb /vob/hpux-intel/opt/aCC/lbin/ctcom -c core
HP gdb 5.9 for HP Itanium (32 or 64 bit) and target HP-UX 11.2x.
Copyright 1986 - 2001 Free Software Foundation, Inc.
Hewlett-Packard Wildebeest 5.9 (based on GDB) is covered by the
GNU General Public License. Type "show copying" to see the conditions to
change it and/or distribute copies. Type "show warranty" for warranty/support.
..
Core was generated by `ctcom'.
Program terminated with signal 10, Bus error.
BUS_ADRALN - Invalid address alignment. Please refer to the following link that
helps in handling unaligned data: http://docs.hp.com/en/7730/newhelp0610/pragmas.htm#pragma-pack-ex3
#0 0x60000000c1e77db0:0 in uwx_self_init_context+0x170 ()
from /usr/lib/hpux32/libunwind.so.1

It seems the patch you referred earlier is related with the one i am getting.

Will applying the patch "PHSS_40541" instead of moving to latest compiler resolves the isssue?

One more thing, gdb says,
"BUS_ADRALN - Invalid address alignment."

So it is better to check for the code that has address alignment issues.. right?

Apart from this, i also want to try this
"Otherwise you could create a .i file with "-E -.i" and chop up that file until the problem goes away. "

I have used -E -.i options and created .i file. Could you please let me know how to proceed further.

Let me know your sugesstions.

Thanks

regards,
Srikrishna Erra.

[Dennis Handly]

>Here is the gdb info

Where is the complete stack trace?

>Will applying the patch PHSS_40541 instead of moving to latest compiler resolves the issue?

Of course not. It just enables a stack trace, which may tell us what's wrong.

>So it is better to check for the code that has address alignment issues.. right?

No, since "the code" is the compiler and problem isn't alignment issues but bad/corrupted pointers.

>Could you please let me know how to proceed further.

(I still haven't seen those points yet.)

First you need to compile that .i file to see if you can duplicate the problem.
Then you need to do a binary search in the source code phase space to isolate the problem.

[Srikrishna Erra]

Hello Dennis,

I am very sorry.
Here is the full stack trace...

$ gdb /vob/hpux-intel/opt/aCC/lbin/ctcom -c core
HP gdb 5.9 for HP Itanium (32 or 64 bit) and target HP-UX 11.2x.
Copyright 1986 - 2001 Free Software Foundation, Inc.
Hewlett-Packard Wildebeest 5.9 (based on GDB) is covered by the
GNU General Public License. Type "show copying" to see the conditions to
change it and/or distribute copies. Type "show warranty" for warranty/support.
..
Core was generated by `ctcom'.
Program terminated with signal 10, Bus error.
BUS_ADRALN - Invalid address alignment. Please refer to the following link that
helps in handling unaligned data: http://docs.hp.com/en/7730/newhelp0610/pragmas
.htm#pragma-pack-ex3
#0 0x60000000c1e77db0:0 in uwx_self_init_context+0x170 ()
from /usr/lib/hpux32/libunwind.so.1
(gdb) bt
#0 0x60000000c1e77db0:0 in uwx_self_init_context+0x170 ()
from /usr/lib/hpux32/libunwind.so.1
#1 0x60000000c1e873a0:0 in U_STACK_TRACE+0x200 ()
from /usr/lib/hpux32/libunwind.so.1
#2 0x5745220:0 in sighandler () at Compiler.C:1710
#3
#4 0x60000000c1e77db0:0 in uwx_self_init_context+0x170 ()
from /usr/lib/hpux32/libunwind.so.1
#5 0x60000000c1e873a0:0 in U_STACK_TRACE+0x200 ()
from /usr/lib/hpux32/libunwind.so.1
#6 0x5762b60:0 in ErrorHandler::toolError () at ErrorHandler.C:466
#7 0x49bb750:0 in fe_message () at fe_error.C:280
#8 0x5a20080:0 in UT_AssertFailure () at com_util.C:284
#9 0x5588fd0:0 in $cold__Z30TCG_Add_Array_Element_Constantyyjj+0x160 ()
#10 0x55711e0:0 in $cold__ZN17CxxStringConstant29createVariablePointerConstantEv
+0x190 ()
#11 0x4a0fc50:0 in CxxStringConstant::emitCode () at UC.ExprCodegen.C:6626
#12 0x511b280:0 in processArgument () at UC.ExprCodegen.C:4034
#13 0x511b660:0 in processArgumentList () at UC.ExprCodegen.C:4074
#14 0x511b600:0 in processArgumentList () at UC.ExprCodegen.C:4074
#15 0x4c207c0:0 in Call::emitCode ()
at /CLO/Components/HPCXX/Src/CPlusPlus/Expression.h:3835
#16 0x4c4a3a0:0 in Convert::emitCode () at UC.ExprCodegen.C:4786
---Type to continue, or q to quit---c
#17 0x4c34660:0 in ExprStmt::emitCodeForStatement ()
at UC.StatementCodegen.C:73
#18 0x4c815c0:0 in Statement::emitCodeDriver () at UC.StatementCodegen.C:1126
#19 0x4b79920:0 in Function::doCode () at UC.DeclCodegen.C:1813
#20 0x49f2ac0:0 in Function::code () at DeclCodegen.C:1066
#21 0x4bf0010:0 in Compiler::ProcessExternalDeclaration () at Compiler.C:3002
#22 0x4cb7a00:0 in Compiler::ProcessExternalDeclarationList ()
at Compiler.C:2652
#23 0x4d58fd0:0 in doReduction1 () at Grammar.y:4430
#24 0x4d6a840:0 in yyparse () at Parser.C:2806
#25 0x48e5fe0:0 in Compiler::DoCompile () at Compiler.C:1028
#26 0x48834d0:0 in Compiler::DoCompile () at Compiler.C:1350
#27 0x489c6b0:0 in Compiler::DoCompileFile () at Compiler.C:1390
#28 0x48bcdb0:0 in main () at Compiler.C:2012
(gdb)


Thanks in advance.

regards,
Srikrishna Erra.

[Srikrishna Erra]

Hi,
Did you get a chance to look into the trace.

Any sugesstions?

regards,
Srikrishna Erra.

[Srikrishna Erra]

Hello all,
I have generated .i file of the source program which i am getting Coredump in the Compiler code.

Then i have tried compiling the .i file by choping up the .i file until i found the code source that resulting in Core Dump.

My source file has below code which is resulting in Coredump

"ICULocaleService *tService = new ICUBreakIteratorService();"

Here ICULocaleService is a base class and
ICUBreakIteratorService is derived one.

Derived class "ICUBreakIteratorService"
has below constructor definition.

class ICUBreakIteratorService : public ICULocaleService {
public:
ICUBreakIteratorService()
: ICULocaleService(UNICODE_STRING("Break Iterator", 14))
{
UErrorCode status = U_ZERO_ERROR;
registerFactory(new ICUBreakIteratorFactory(), status);
}
....
....


When i remove instantiation and leave the code which declares only the object then there is no coredump.
i.e ICULocaleService *tService;

Coredump comes when object is instantiated.

Thinking that there may be Casting problem, i have also tried using "dynamic_cast" and tried with below modified code but the same result, it coredumps again.

ICULocaleService myobj, *tService;
tService = dynamic_cast(&myobj);


I don't think the code has to be suspected because it is valid one.

Any help in this regards is really appreciated.

Thanks in advance.

regards,
Srikrishna Erra.

[Dennis Handly]

>Any help in this regards is really appreciated.

Refer to my Jun 1 06:01:34 comments.

[Srikrishna Erra]

I am done with assigning points for all the replies.

[Dennis Handly]

>I am done with assigning points for all the replies.

You still have 10 replies in 4 threads to do:
http://forums.itrc.hp.com/service/forums/pageList.do?userId=WW2137&listType=unassigned

Have you tried installing PHSS_40541 or its replacement PHSS_41182?
Or even tried PHSS_29655, A.05.52?

#9 0x5588fd0:0 $cold__Z30TCG_Add_Array_Element_Constantyyjj

This may be the string in this constructor:
: ICULocaleService(UNICODE_STRING("Break Iterator", 14))

Perhaps create a global with this value and use that instead:
UNICODE_STRING temp_work("Break Iterator", 14);
: ICULocaleService(temp_work)

>declares only the object then there is no coredump.
>i.e. ICULocaleService *tService;

This is a POD, so it can't fail.

>I have also tried using "dynamic_cast" and tried with below modified code

The original had an implicit static_cast. And &myobj doesn't need a cast, it's of the correct type.

Does it fail with just this: ICULocaleService myobj;

[Srikrishna Erra]

Hi,
>>You still have 10 replies in 4 threads to do:
http://forums.itrc.hp.com/service/forums/pageList.do?userId=WW2137&listType=unassigned

I am done with this.

>>Have you tried installing PHSS_40541 or its replacement PHSS_41182?
Or even tried PHSS_29655, A.05.52?
#9 0x5588fd0:0 $cold__Z30TCG_Add_Array_Element_Constantyyjj

I haven't tried this. I have to check with my Admin person on this(doesn't have permisson to run even swlist). Will revert back to you once i got this stuff.


>>This may be the string in this constructor:
: ICULocaleService(UNICODE_STRING("Break Iterator", 14))
Perhaps create a global with this value and use that instead:
UNICODE_STRING temp_work("Break Iterator", 14);
: ICULocaleService(temp_work)


Actually UNICODE_STRING is a macro which inturns call UnicodeString Class constructor. It is defined as

#define UNICODE_STRING ( cs,
_length ) U_NAMESPACE_QUALIFIER UnicodeString(TRUE, (const UChar *)L ## cs, _length)


>>Does it fail with just this: ICULocaleService myobj;

I had done/tested below changes

ICULocaleService *tService = new ICUBreakIteratorService();

Changed to

ICULocaleService myobj, *tService;
tService = dynamic_cast(&myobj);

Could you please clarify what actually you are suggesting me to do?


BTW, we are using IBM ICU(International Components for Unicode). Earlier we use to have ICU 3.6 version and we were able to compile it with the same version of compiler. Now we had moved to ICU 4.4.1 and got these issues.

The Block of code which we are getting Coredump has not been changed from ICU 3.6 to ICU 4.4.1


Thanks for the quick reply.


regards,
Srikrishna Erra.

[Dennis Handly]

>I am done with this.

(Thanks)

>UNICODE_STRING is a macro which in turns call UnicodeString Class constructor.

Ok, make a global of that type with that "Break Iterator" value.

>>Does it fail with just this: ICULocaleService myobj;

>Changed to:
ICULocaleService myobj, *tService;
tService = dynamic_cast(&myobj);

>Could you please clarify what actually you are suggesting me to do?

Basically the hard part is constructing the object. Doing pointer assignments (tService) is easier.

So does just "ICULocaleService myobj;" work?
Or using the derived class:
ICUBreakIteratorService foo;

>Now we had moved to ICU 4.4.1 and got these issues.

Probably unrelated other than you have new/different code.

>The Block of code which we are getting Coredump has not been changed from ICU 3.6 to ICU 4.4.1

The abort may have to do with internal heap corruption in the compiler so any differences may trigger it.

>A.05.50 [Jan 06 2003]

Where did you get this compiler?

The official A.05.50 compiler had this date:
A.05.50 [May 15 2003]

[Srikrishna Erra]

Hi,
Actually, i had checked preprocessed code (in .i file) generated for .cpp code
": ICULocaleService(UNICODE_STRING("Break Iterator", 14))".

Then i replaced existing .cpp code with equivalent preprocessed code.

: ICULocaleService(UNICODE_STRING("Break Iterator", 14))
replaced with
: ICULocaleService(UnicodeString("Break Iterator", 14, UnicodeString::kI
nvariant))

If i compile .cpp file with above changes now then no issues.

.cpp file is getting compiled succesfully.

This is somewhat strange behaviour.


>>So does just "ICULocaleService myobj;" work?
Or using the derived class:
ICUBreakIteratorService foo;

I had tried just "ICULocaleService *tService " and it gives no error.

Problem comes when it gets instantiated i.e.
ICULocaleService *tService = new ICUBreakIteratorService();

Later in our code, tserive is again assigned to another variable and used further i.e.

gService = tService;
tService = NULL;

gService is defined as
static U_NAMESPACE_QUALIFIER ICULocaleService* gService = NULL;

So creating an instance is required.


>>Where did you get this compiler?
The official A.05.50 compiler had this date:
A.05.50 [May 15 2003]

Sorry, i have no IDEA and have to ask My ADMIN team.


Thanks

regards,
Srikrishna Erra.

[Dennis Handly]

>If I compile .cpp file with above changes now then no issues.

Great.

>I had tried just "ICULocaleService *tService" and it gives no error.

Because declaring pointers is trivial.

>Problem comes when it gets instantiated i.e.
ICULocaleService *tService = new ICUBreakIteratorService();

Right, it has to construct the object.

>I have no IDEA and have to ask My ADMIN team.

Ok.

[Srikrishna Erra]

Hi.
I am afraid about the Compiler behaviour as it behaves strange.

What could be the difference
-- in compiling original cpp code directly
AND
-- compiling cpp code by having its equivalent preprocessed code included in it.


Can i trust the Compiler in this case and proceed futher?


Thanks

regards,
Srikrishna Erra.

[Dennis Handly]

>What could be the difference
>in compiling original cpp code directly
>compiling cpp code by having its equivalent preprocessed code included in it.

Possibly different heap layout in the compiler.

>Can I trust the compiler in this case and proceed further?

You don't really have much of a choice unless you update the compiler.

I assume you have way to test your changes?

[Srikrishna Erra]

Hi,
I had found the root cause of this issue.

My configure script assuming UTF16 string literal support is provided by the compier used i.e. passing UTF16 strings like u"Stringliteral" but it seems the compiler(aCC 05.50) i am using is not supporting above kind of string literals.

Issue is with my Configure script. The script is not checking whether it is supported or not. It simply assuming the support is existed and passing string literals in the above form.

I just need confirmation for
Does Compiler (aCC 05.05) really not supports UTF16 String literals or should "install any patch/do any changes" to get the support?

Thanks

regadrs,
Srikrishna Erra.

[Dennis Handly]

>i.e. passing UTF16 strings like u"Stringliteral" but it seems the compiler (A.05.50) is not supporting above kind of string literals.

This should be supported as long as the chars are ascii.

>Does compiler (A.05.50) really not supports UTF16 String literals

It should provide limited support for them.

>or should "install any patch/do any changes" to get the support?

A.05.36 thru A.05.41 abort. Perhaps your "prerelease" A.05.50 doesn't have the fix?

Well, you could always go to the latest, A.06.25.01.