OpenMPI compile/link problems [SOLVED]

After installing openmpi 1.6-2, I tried compiling a well-tested mpi program that
I have been running under lam mpi on Debian linux. Attempting to compile on
Arch openmpi results in missing libraries:
swallow:~/svn/conv/trunk$ make
mpicc -g -c main.c
mpicc -g -c timestep.c
mpicc -g -c thermo.c
mpicc -g -c toyrad.c
mpicc -g -c getprofile.c
mpicc -g -o conv main.o timestep.o thermo.o toyrad.o \
getprofile.o -lcdf -lm
main.o: In function `main':
/home/raymond/svn/conv/trunk/main.c:157: undefined reference to `lam_mpi_comm_world'
/home/raymond/svn/conv/trunk/main.c:158: undefined reference to `lam_mpi_comm_world'
/home/raymond/svn/conv/trunk/main.c:187: undefined reference to `lam_mpi_comm_world'
/home/raymond/svn/conv/trunk/main.c:189: undefined reference to `lam_mpi_comm_world'
/home/raymond/svn/conv/trunk/main.c:629: undefined reference to `lam_mpi_sum'
collect2: error: ld returned 1 exit status
make: *** [conv] Error 1
swallow:~/svn/conv/trunk$
I then downloaded the openmpi package from the OpenMPI website, compiled it myself,
and installed it in /usr/local. I was able to compile and run the program with no trouble.
(The --prefix /usr/local flag was needed on mpirun.)
I notice that the openmpi libraries have been moved to /usr/lib/openmpi/ in the Arch version.
Presumably Arch openmpi would work if one told the system where the libraries were (I haven't
tried it), but relocating the libraries seems like an unnecessary complication to me and a
booby trap for the unwary.
Last edited by djraymondnm (2012-09-11 12:57:18)

Compilation of C code works for me. The location of the libs should not be a problem because the linker find them with the following file :
$ cat /etc/ld.so.conf.d/openmpi.conf
/usr/lib/openmpi
I just pushed openmpi-1.6.1 in [extra]. Let me know if it works for you. If not, could you please give me a minimal example that does not compile for you ?
Last edited by stephane (2012-09-04 23:10:17)

Similar Messages

OCCI compilation/linking problems on Linux

Hi All,
I am having difficulty in compiling/linking demo programms on Linux machine. I am trying to compile a demo program with the following command:
make -f demo_rdbms.mk build EXE=occidml OBJS=occidml.o
When I try to do this I get following errors:
make -f demo_rdbms.mk build EXE=occidml OBJS=occidml.o
/usr/bin/g++ -c -I/u01/app/oracle/product/9.0.1/rdbms/demo -I/u01/app/oracle/product/9.0.1/rdbms/public -I/u01/app/oracle/product/9.0.1/plsql/public -I/u01/app/oracle/product/9.0.1/network/public occidml.cpp
cc -L/u01/app/oracle/product/9.0.1/lib/ -L/u01/app/oracle/product/9.0.1/rdbms/lib/ -o occidml occidml.o -lclntsh `cat /u01/app/oracle/product/9.0.1/lib/ldflags` `cat /u01/app/oracle/product/9.0.1/lib/sysliblist` -ldl -lm
occidml.o: In function `main':
occidml.o(.text+0x49): undefined reference to `endl(ostream &)'
occidml.o(.text+0x56): undefined reference to `cout'
occidml.o(.text+0x5b): undefined reference to `ostream::operator<<(char const *)'
occidml.o(.text+0x66): undefined reference to `ostream::operator<<(ostream &(*)(ostream &))'
occidml.o(.text+0x102): undefined reference to `endl(ostream &)'
occidml.o(.text+0x10f): undefined reference to `cout'
occidml.o(.text+0x114): undefined reference to `ostream::operator<<(char const *)'
occidml.o(.text+0x11f): undefined reference to `ostream::operator<<(ostream &(*)(ostream &))'
occidml.o(.text+0x138): undefined reference to `endl(ostream &)'
occidml.o(.text+0x145): undefined reference to `cout'
occidml.o(.text+0x14a): undefined reference to `ostream::operator<<(char const *)'
occidml.o(.text+0x155): undefined reference to `ostream::operator<<(ostream &(*)(ostream &))'
occidml.o(.text+0x16e): undefined reference to `endl(ostream &)'
occidml.o(.text+0x17b): undefined reference to `cout'
occidml.o(.text+0x180): undefined reference to `ostream::operator<<(char const *)'
occidml.o(.text+0x18b): undefined reference to `ostream::operator<<(ostream &(*)(ostream &))'
occidml.o(.text+0x1a4): undefined reference to `endl(ostream &)'
occidml.o(.text+0x1b1): undefined reference to `cout'
occidml.o(.text+0x1b6): undefined reference to `ostream::operator<<(char const *)'
occidml.o(.text+0x1c1): undefined reference to `ostream::operator<<(ostream &(*)(ostream &))'
occidml.o(.text+0x1f7): undefined reference to `endl(ostream &)'
occidml.o(.text+0x204): undefined reference to `cout'
occidml.o(.text+0x209): undefined reference to `ostream::operator<<(char const *)'
occidml.o(.text+0x214): undefined reference to `ostream::operator<<(ostream &(*)(ostream &))'
occidml.o(.text+0x22d): undefined reference to `endl(ostream &)'
I am sure that I don't have some of the environmet variables right. I will really appreciate if anyone can point me in the right direction.
Thanks for your time.
Parag

Thanks for your reply. I installed the Sun Forte Compiler 7 and tried again. The errors I got this time are:
----------------- ERRORS ---------------------
Undefined first referenced
symbol in file
__builtin_delete test.o
__length_error(const char*) test.o
__ls__7ostreamPFR7ostream_R7ostream test.o
terminate(void) test.o
createEnvironment__Q36oracle4occi11EnvironmentQ46oracle4occi11Environment4ModePvPFPvUi_PvPFPvPvUi_PvPFPvPv_v test.o
endl(ostream&) test.o
cerr test.o
__ls__7ostreamPCc test.o
__cp_pop_exception test.o
__tfQ36oracle4occi12SQLException test.o
copy__t18string_char_traits1ZcPcPCcUi test.o
__eh_rtime_match test.o
__throw test.o
length__t18string_char_traits1ZcPCc test.o
__start_cp_handler test.o
move__t18string_char_traits1ZcPcPCcUi test.o
__out_of_range(const char*) test.o
ld: fatal: Symbol referencing errors. No output written to test.exe
====================================================
Do you have any other suggestions? Though Linux sounds good, I have no choice but to develop on Solaris.
Thanks very much!

Documentation link problem [solved]

Hi, It's happening a problem with the 10gR2 documentation link.
The page for the link was not found.
See:
http://www.oracle.com/pls/db102/portal.portal_db?selected=3#index-ADV

It's OK again.

OCI compilation/linking problems

HI guys:
System: Solaris
Database: Oracle 9i
I got the following compilation errors.
Anybody who ever experienced with this situation, please share to me. Thank You.
Regards,
--Lijie
ild: (undefined symbol) CD_free -- referenced in the text segment of /data2/oracle/9.0.1/lib//libn9.a(ez_cdc.o)
ild: (undefined symbol) CD_memcmp -- referenced in the text segment of /data2/oracle/9.0.1/lib//libn9.a(pkcs12_30.o)
ild: (undefined symbol) CD_memset -- referenced in the text segment of /data2/oracle/9.0.1/lib//libn9.a(ez_cdc.o)
ild: (undefined symbol) ocieperr -- referenced in the text segment of /data2/oracle/9.0.1/lib/libclient9.a(kpxs.o)
ild: (undefined symbol) ociepmsg -- referenced in the text segment of /data2/oracle/9.0.1/lib/libclient9.a(kpxe.o)
ild: (undefined symbol) CD_memcpy -- referenced in the text segment of /data2/oracle/9.0.1/lib//libn9.a(pkcs7_30.o)

You can use the old library if you compile new code using the -compat=4 option. You will not be able to use modern C++ features in your new code, nor the C++ Standard Library. You will be resticted to 1990-style C++, as described in the ARM.
Refer to the C++ MIgration Guide that comes with the compiler for full details. (It's also available at docs.sun.com)
We very strongly recommend that you not use -compat=4 mode and old binaries. Find a way to rebuild or replace the old library, and use standard C++ instead.

Re: Fwd: Link Problems With Borland C++ 4.52

I have seen this problem before in another context, and I'll offer the
cause and solution in the hope that they will apply to the Crystal problem.
Many Windows based applications rely on PASCAL calling conventions, which
change the way parameters are handled in function/method calls. They
indicate this by placing one of the following immediately before the
function name in the prototype declarations:
- pascal_far (or something like that)
- WINAPI
- some other typedef of either of the above
For example:
int WINAPI AddTotal(int valueA, int valueB);
Unfortunately, v2.0 of Forte does not provide any mechanisms for changing
the calling conventions of the prototypes in the generated C++ wrapper
library, so when you compile that code, the linker fails. I think that the
compiler may generate different symbols depending on calling conventions,
so that's why it fails.
To fix this, don't autocompile your code, but generate the distribution, go
into the generated C++ files and look for the function prototypes (I think
you can search for FORTE_NO_PROTOTYPES), add WINAPI to the appropriate
places in the prototype definitions (see above) and use fcompile to build
the library. Instructions for fcompile are in the Interfacing With
External Systems manual.
Hope this helps,
James
At 11:05 AM 5/29/97 PDT, you wrote:
>
We are trying to wrapper Crystal Reports from Forte. I know that there
are a number of other people in this same boat, as I've seen messages
posted here at various points during the past few weeks. We are having
a particular problem with getting the compile to go through, which we
have sent in to Forte Tech Support. I'm forwarding the message I sent
Tech Support to this group in the hopes that someone here may have
already seen and resolved a similar problem.
Thanks in advance for any help you can offer!
Dale V. Georg
Indus Consultancy Services [email protected]
Mack Trucks, Inc. [email protected]
>
Date: Wed, 28 May 97 13:47:20 PDT
From: dg7077a
To: Forte Technical Support
Cc: Gardner, Steve
Subject: Link Problems With Borland C++ 4.52
Name of requestor: [Dale V. Georg / Alaiah Chandrashekar]
Company: [Indus Consultancy Services]
Phone for callback: [(610) 709-3956]
Customer Site: [Mack Trucks, Inc.]
Product: [Forte]
Version of Forte: [2.0.H.1]
Server OS: [SunOS 5.5.1]
Client OS: [Windows 3.1]
DBMS: [Oracle 7.2.3]
Reproducible?: [Yes]
Brief description: [Link Problems With Borland C++ 4.52]
Complete description of problem or question:
We are attempting to write a C-wrapper interface from Forte to Crystal
Reports' Report Engine. We are using Borland C++ version 4.52.
Unfortunately, we have been unable to get a clean compile after a day
and half of effort. We get as far as the link stage of the
compilation, and
the compiler aborts with an "Unknown symbol" error message for each
of the functions we are trying to wrapper. We have tried a number of
ideas to fix this problem, and are continuing to try to solve it on
our own,
but any help would be greatly appreciated. Please have someone call
Alaiah Chandrashekar at the number above as soon as possible.
Thanks!
Dale V. Georg
Indus Consultancy Services [email protected]
Mack Trucks, Inc. [email protected]
James Urquhart [email protected]
Product Manager phone: (510) 986-3513
Forte Software, Inc. fax: (510) 869-2092

James,
Thanks for your quick response. Yesterday we had been running down
the path of examining the calling conventions and trying to change them
to PASCAL, but without much success. After receiving your note, we
went back over it again, and this time we were able to finally to piece it
together. In addition to editing the Forte-generated .cc file to declare
the functions as PASCAL, we also had to turn off the compiler's case
sensitivity. (The Crystal .lib file had the function names in mixed case,
but the Borland compiler was generating all uppercase for the names.)
Now I had actually tried this yesterday and it didn't work (in fact it
generated a whole bunch of new errors) - because until we took a
second look at it today, I didn't realize that Borland's linker actually has
TWO flags that control case sensitivity. If you only turn one or the
other off, things can get pretty ugly looking. As soon as we turned
both of them off, the compile and link went beautifully. Again, thanks
for your help; hopefully we are over the worst of it now!
Dale
I have seen this problem before in another context, and I'll offer the
cause and solution in the hope that they will apply to the Crystalproblem.
>
Many Windows based applications rely on PASCAL callingconventions, which
change the way parameters are handled in function/method calls.They
indicate this by placing one of the following immediately before the
function name in the prototype declarations:
- pascal_far (or something like that)
- WINAPI
- some other typedef of either of the above
For example:
int WINAPI AddTotal(int valueA, int valueB);
Unfortunately, v2.0 of Forte does not provide any mechanisms forchanging
the calling conventions of the prototypes in the generated C++wrapper
library, so when you compile that code, the linker fails. I think thatthe
compiler may generate different symbols depending on callingconventions,
so that's why it fails.
To fix this, don't autocompile your code, but generate thedistribution, go
into the generated C++ files and look for the function prototypes (I think
you can search for FORTE_NO_PROTOTYPES), add WINAPI tothe appropriate
places in the prototype definitions (see above) and use fcompile tobuild
the library. Instructions for fcompile are in the Interfacing With
External Systems manual.
Hope this helps,
James
At 11:05 AM 5/29/97 PDT, you wrote:
We are trying to wrapper Crystal Reports from Forte. I know that
there
are a number of other people in this same boat, as I've seenmessages
posted here at various points during the past few weeks. We arehaving
a particular problem with getting the compile to go through, whichwe
have sent in to Forte Tech Support. I'm forwarding the message Isent
Tech Support to this group in the hopes that someone here mayhave
already seen and resolved a similar problem.
Thanks in advance for any help you can offer!
Dale V. Georg
Indus Consultancy Services [email protected]
Mack Trucks, Inc.
[email protected]
>
>>
Date: Wed, 28 May 97 13:47:20 PDT
From: dg7077a
To: Forte Technical Support
Cc: Gardner, Steve
Subject: Link Problems With Borland C++ 4.52
Name of requestor: [Dale V. Georg / AlaiahChandrashekar
Company: [Indus Consultancy Services]
Phone for callback: [(610) 709-3956]
Customer Site: [Mack Trucks, Inc.]
Product: [Forte]
Version of Forte: [2.0.H.1]
Server OS: [SunOS 5.5.1]
Client OS: [Windows 3.1]
DBMS: [Oracle 7.2.3]
Reproducible?: [Yes]
Brief description: [Link Problems With Borland C++ 4.52]
Complete description of problem or question:
We are attempting to write a C-wrapper interface from Forte to
Crystal
>>
Reports' Report Engine. We are using Borland C++ version 4.52.
Unfortunately, we have been unable to get a clean compile after aday
and half of effort. We get as far as the link stage of the
compilation, and
the compiler aborts with an "Unknown symbol" error message foreach
of the functions we are trying to wrapper. We have tried a numberof
ideas to fix this problem, and are continuing to try to solve it on
our own,
but any help would be greatly appreciated. Please have someonecall
Alaiah Chandrashekar at the number above as soon as possible.
Thanks!
Dale V. Georg
Indus Consultancy Services [email protected]
Mack Trucks, [email protected]
James Urquhart [email protected]
Product Manager phone: (510) 986-3513
Forte Software, Inc. fax: (510) 869-2092-----------------------------------------------------------------------------------
Dale V. Georg
Indus Consultancy Services [email protected]
Mack Trucks, Inc. [email protected]
[email protected]------------------

Fwd: Link Problems With Borland C++ 4.52

We are trying to wrapper Crystal Reports from Forte. I know that there
are a number of other people in this same boat, as I've seen messages
posted here at various points during the past few weeks. We are having
a particular problem with getting the compile to go through, which we
have sent in to Forte Tech Support. I'm forwarding the message I sent
Tech Support to this group in the hopes that someone here may have
already seen and resolved a similar problem.
Thanks in advance for any help you can offer!
Dale V. Georg
Indus Consultancy Services [email protected]
Mack Trucks, Inc. [email protected]
Date: Wed, 28 May 97 13:47:20 PDT
From: dg7077a
To: Forte Technical Support
Cc: Gardner, Steve
Subject: Link Problems With Borland C++ 4.52
Name of requestor: [Dale V. Georg / Alaiah Chandrashekar]
Company: [Indus Consultancy Services]
Phone for callback: [(610) 709-3956]
Customer Site: [Mack Trucks, Inc.]
Product: [Forte]
Version of Forte: [2.0.H.1]
Server OS: [SunOS 5.5.1]
Client OS: [Windows 3.1]
DBMS: [Oracle 7.2.3]
Reproducible?: [Yes]
Brief description: [Link Problems With Borland C++ 4.52]
Complete description of problem or question:
We are attempting to write a C-wrapper interface from Forte to Crystal
Reports' Report Engine. We are using Borland C++ version 4.52.
Unfortunately, we have been unable to get a clean compile after a day
and half of effort. We get as far as the link stage of the
compilation, and
the compiler aborts with an "Unknown symbol" error message for each
of the functions we are trying to wrapper. We have tried a number of
ideas to fix this problem, and are continuing to try to solve it on
our own,
but any help would be greatly appreciated. Please have someone call
Alaiah Chandrashekar at the number above as soon as possible.
Thanks!
Dale V. Georg
Indus Consultancy Services [email protected]
Mack Trucks, Inc. [email protected]
-------------

We are trying to wrapper Crystal Reports from Forte. I know that there
are a number of other people in this same boat, as I've seen messages
posted here at various points during the past few weeks. We are having
a particular problem with getting the compile to go through, which we
have sent in to Forte Tech Support. I'm forwarding the message I sent
Tech Support to this group in the hopes that someone here may have
already seen and resolved a similar problem.
Thanks in advance for any help you can offer!
Dale V. Georg
Indus Consultancy Services [email protected]
Mack Trucks, Inc. [email protected]
Date: Wed, 28 May 97 13:47:20 PDT
From: dg7077a
To: Forte Technical Support
Cc: Gardner, Steve
Subject: Link Problems With Borland C++ 4.52
Name of requestor: [Dale V. Georg / Alaiah Chandrashekar]
Company: [Indus Consultancy Services]
Phone for callback: [(610) 709-3956]
Customer Site: [Mack Trucks, Inc.]
Product: [Forte]
Version of Forte: [2.0.H.1]
Server OS: [SunOS 5.5.1]
Client OS: [Windows 3.1]
DBMS: [Oracle 7.2.3]
Reproducible?: [Yes]
Brief description: [Link Problems With Borland C++ 4.52]
Complete description of problem or question:
We are attempting to write a C-wrapper interface from Forte to Crystal
Reports' Report Engine. We are using Borland C++ version 4.52.
Unfortunately, we have been unable to get a clean compile after a day
and half of effort. We get as far as the link stage of the
compilation, and
the compiler aborts with an "Unknown symbol" error message for each
of the functions we are trying to wrapper. We have tried a number of
ideas to fix this problem, and are continuing to try to solve it on
our own,
but any help would be greatly appreciated. Please have someone call
Alaiah Chandrashekar at the number above as soon as possible.
Thanks!
Dale V. Georg
Indus Consultancy Services [email protected]
Mack Trucks, Inc. [email protected]
-------------

"Link Problem" in Incident Form does not seem to work right? How do I do it correctly?

1. The Link Problem tool in the task pane on our Incidents, do not seem to actually link the Incident to the Problem.
2. I can link the two by going to the Problem and adding it on the related items tab, but this is very time consuming in a call center environment, especially during an outage where we are relying a lot more on Problem and root cause analysis.
3. I have tried creating the incident and applying, then doing the Link Problem tool and selecting the problem, then applying again and selecting OK to close the incident. Again it doesn't work for me.
What am I doing wrong, number one?
Can we do this without manually opening the problem for every incident and attaching from there?
3. Can I build an incident template that will automate the Attachment to the problem at creation? (Tried this today to and it wouldn't work, but then again, the Link Problem might again be the culprit) This would be great for real time saving during
an outage. Once outage is found, I could create the template with all the necessary items filled out, and would like to have it automatically attach to the specific problem I choose at template creation, and that way analysts can simply apply template, and
click OK when they get a call related to the outage.

Thanks for the response Thomas!
The functionality seems to work just fine. I use the LINK PROBLEM and it pops up the problem search box, I search for and select a Problem, I then select it and click ok, and I get no error saying the relationship was not created.
The problem is, when I open the problem after that, there is never any relationship showing in related items there, as it says there should be.
Also, to point out, I am talking about newly created Incidents. I tried even using APPLY prior to doing the LINK PROBLEM but it still did not work. I suspect it may be a workflow problem, where the workflow to create the incident has not completed, there
for the relationship is not completed.... and I get no error.
In my work environment, we have analysts during outages taking a call every 30 seconds, and I need a more efficient way for this system to work. As of now, they spend 1-2 minutes doing the ticket, and that is too long.
I have decided to try a newb way, where I created an Incident (GENERIC OUTAGE) that folks can use, and every 15-30 minutes during an outage, I can open the problem and search for open incidents with that title, and attach all that I find from the problem
side. This is assuming that once attached they will not show up in the search to attach the next time lol. Also this assumes that if attached from the problem, they will auto resolve when the problem is resolved.
I am rusty on my programming, and am currently trying to get refreshed. I am sure that would come in handy for this :)
Oh and lastly, one additional tidbit that will make this clearer, with regard to work flow, we re also running our Service Manager in our help desk, virtualized, as they have not upgraded to Win 7 yet :(. So that compiles my time to add ticket issue :(.

InDesign to Adobe Acrobat to Adobe Reader Link problem

Hi - I'm responsible for a long conversation about link problems in creating file links in InDesign that did not work in Acrobat. At the end of that long discussion, I thought I solved the problem, but didn't. Since then, I've been MUCH more observant.
The problem appears to take place in Adobe Reader.
My home computer on Win7 is different than my work Computer (XP), but I have the same software on both.
At work, I discovered that when I clicked on a PDF, it worked fine, that its links opened a new PDF window, but when I clicked on links on the second document, the links failed and Microsoft returned an error message telling me that I was screwed. Because I was screwed, I couldn't see what was in the path of that error.
It turns out that for some unknown reason (which you might clarify for me), my Adobe Readers are attempting to open my documents in Internet Explorer and that just won't work.
What I observed at home is that I made 5 test documents in ID, and ported them to Acrobat. As it turns out, my default for opening PDFs is Adobe Reader, which I saw on the chrome of the first document I opened. When I clicked a link, and I got the error message, I noticed my browser was up.
When I set my default for opening in Acrobat, everything worked like a charm! How about that!
Now then, what that says to me is that Adobe Reader is to be used exclusively with a browser, so if the people who are going to read your internal PDF files with Reader without those files being on a web server of some type, you are totally screwed. The only way you can use Reader is if you click on the file and bring it up directly. Don't rely on linking to other documents.
I mean, that is what it seems to me.
Now, I am creating a big project for an environment without a webserver and obviously my sense of panic is palpable. That said, I believe that all my users can read my files using Acrobat (to be tested tomorrow), which I hope means I can breathe a bit easier.
Do any of you have any thoughts on this subject? If you are really curious, I can make a packet of test files available.

Karen_Little wrote:
The files and the PDFs are all in the same directory.
The tests are on local drives. All indesign files and their related PDFs were in the same directory.
The same for Word and its related PDFs. Note that the PDFs created by Word then read by Adobe Reader worked correctly. They all opened in Adobe Reader.
If the default program for opening PDFs is Acrobat, everything works correctly, with the added benefit of everything opening in a different window, which is what I want (yet to be tested at work). If the default program is Adobe Reader, the second generation of links do NOT work because Reader opens in Internet Explorer.
When I open a Word-created PDF with Acrobat, view the link, and click to the Actions menu, the action is "Open a file"
When I open an InDesign PDF with Acrobat and do the same, the action says "Open a web link"
Big BIG BIG bug.
Are you seeing the link commands by editing them in Acrobat? If not, try, so you can see the exact command and file or URL instruction. It still may be a bug, but you'll have better evidence when you report it here:
Adobe - Feature Request/Bug Report Form
And, you might want to try correcting a link command in Acrobat, to see if can work for both Acrobat and Reader.
HTH
Regards,
Peter
Peter Gold
KnowHow ProServices

Help in compiling/linking a DEVICE DRIVER in SOLARIS10 x86

hello!
PROBLEM:
++++++++++++++++++++++++++++++++++
i need assistance in compiling/linking a video capture
driver (bt848x) (http://bt848x.sourceforge.net/) for
solaris 10 x86.
the driver uses the /usr/lib/libc.a (specifically the _xtoll.o)
which, if im not mistaken, is not available in solaris 10
anymore. so how am i going to compile it?
BACKGROUND:
++++++++++++++++++++++++++++++++++
the original compile/link code in the makefile is:
/usr/ccs/bin/ld -o bt848x -r bt848-cards.o bt848-driver.o
bt848-kern.o i2c.o tuner.o sound-tda9875.o
sound-msp3400.o _xtoll.o
i modified it to:
/usr/ccs/bin/ld -o bt848x bt848-cards.o bt848-driver.o
bt848-kern.o i2c.o tuner.o sound-tda9875.o
sound-msp3400.o -L/usr/lib -lc
(NOTE: i removed the -r and _xtoll.o; and added
-L/usr/lib -lc)
the output was:
Undefined ____________ first referenced
symbol ______________ in file
mod_install bt848-kern.o
delay bt848-cards.o
ddi_get_instance bt848-kern.o
nodev bt848-kern.o
ddi_create_minor_node bt848-kern.o
ddi_regs_map_free bt848-kern.o
pci_config_getw bt848-driver.o
pci_config_getb bt848-driver.o
ddi_soft_state_zalloc bt848-kern.o
ddi_dma_unbind_handle bt848-kern.o
kmem_alloc i2c.o
cv_broadcast bt848-driver.o
ddi_remove_intr bt848-kern.o
ddi_get_soft_state bt848-kern.o
mod_driverops bt848-kern.o
ddi_get_lbolt bt848-cards.o
mutex_tryenter bt848-kern.o
ddi_get_iblock_cookie bt848-kern.o
nochpoll bt848-kern.o
ddi_copyin bt848-kern.o
mod_info bt848-kern.o
ddi_remove_minor_node bt848-kern.o
nulldev bt848-kern.o
ddi_dma_sync bt848-driver.o
ddi_dma_addr_bind_handle bt848-kern.o
mod_remove bt848-kern.o
ddi_prop_lookup_int_array bt848-cards.o
_pagesize bt848-driver.o
ddi_prop_op bt848-kern.o
ddi_dma_free_handle bt848-kern.o
ddi_set_driver_private bt848-kern.o
cv_wait_sig bt848-kern.o
ddi_intr_hilevel bt848-kern.o
ddi_regs_map_setup bt848-kern.o
pci_config_teardown bt848-driver.o
ddi_dma_mem_alloc bt848-kern.o
ddi_copyout bt848-kern.o
cmn_err bt848-cards.o
ddi_dma_alloc_handle bt848-kern.o
ddi_prop_exists bt848-cards.o
cv_init bt848-kern.o
pci_config_setup bt848-driver.o
cv_destroy bt848-kern.o
timeout sound-msp3400.o
untimeout sound-msp3400.o
devmap_umem_setup bt848-kern.o
ddi_prop_get_int bt848-kern.o
uiomove bt848-kern.o
kmem_free i2c.o
ddi_add_intr bt848-kern.o
ddi_report_dev bt848-kern.o
getminor bt848-kern.o
mutex_exit bt848-driver.o
ddi_prop_free bt848-cards.o
ddi_soft_state_fini bt848-kern.o
ddi_soft_state_free bt848-kern.o
ddi_soft_state_init bt848-kern.o
mutex_enter bt848-driver.o
ddi_mem_putl bt848-cards.o
ddi_mem_getl bt848-cards.o
drv_usecwait bt848-cards.o
drv_usectohz bt848-cards.o
ddi_dma_mem_free bt848-kern.o
ld: fatal: Symbol referencing errors. No output written to bt848x
QUESTIONS:
++++++++++++++++++++++++++++++++++
1.) What other libraries do i need to include so that
there would no more "symbol referencing errors" ???
2.) Is my linking code correct?
/usr/ccs/bin/ld -o bt848x bt848-cards.o bt848-driver.o
bt848-kern.o i2c.o tuner.o sound-tda9875.o
sound-msp3400.o -L/usr/lib -lc
i mean, is it ok to build a driver with dynamic libraries?
or the device driver should always be statically linked?
COMPUTER SPECS:
++++++++++++++++++++++++++++++++++
Solaris 10 (6/06 release)
SunOS 5.10 Generic_118844-26 i86pc i386 i86pc
Pentium 4 - 1.4GHz
1GB RDRAM
32 BIT
Video Capture Card: PixelView PV-BT878P+FM.RC (rev.9F)
thanks!!!
-kllc

i think i already found the answers to my questions
(i partially read SUN docs WRITING DEVICE DRIVER in
http://docs.sun.com/app/docs/doc/816-4854 specifically
the BUILDING A DEVICE DRIVER) =)
i used:
/usr/ccs/bin/ld -o bt848x -dy -r bt848-cards.o bt848-driver.o
bt848-kern.o i2c.o tuner.o sound-tda9875.o sound-msp3400.o
-L/usr/lib -lc
(NOTE: i used the flags -dy -r -L/usr/lib -lc)
so far, the TV-TUNER is working hehehehe!!
any comments/suggestions/tips from other people are still welcome!
thanks!!
-kllc

Flash Builder Link Problem

Whenever I try to access links in Flash Builder such as those on the Start Page, Device Connection Help in Run Configurations or even Find and Download other versions of Flex SDK in the Flex Configuration window it opens up a blank Notepad window with an error "The filename, directory name, or volume label syntax is incorrect."
Anyone have any idea on how to fix this? I assume it is supposed to open up a web page.

Wow, problem solved. I thought it was weird when I installed Flash Builder 4.7 and it was still happening. Apparently at some point I set Notepad.exe as the default program to open HTML files and did not realize it.
I feel kind of dumb now but as every other program was working (heck, links even worked in Flash Professional) I was so sure it was a Flash Builder problem.
Anyway, I am just glad to have the mystery solved.

Pro*C linker problems

System is Linux Mandrake 6.0 with Oracle 8i, with the 8i patch
applied.
What a nightmare that install was! Lots of permission problems,
unreadable directories, programs with no execute permission on
them and so on.
I now have a database up and running, and am trying to get Pro*C
working. The code was written on RedHat 5.2 with Oracle 8.0.5,
and worked fine there, so I know it is not a fundamental code
problem!
I have patched the pcscfg.cfg file as detailed in the patch FAQ
(with modifications for the slightly different paths under
Mandrake Linux).
When I compile the Pro*c file, I get many "undefined reference
to 'dbcon'" and "undefined reference to 'ctx'" errors from the
linker. This occurs whether I use a pre-defined list of specific
libraries in my own Makefile, or if I use the standard
demo_proc.mk makefile. I have used nm(1) on every place I can
think of, to try to find the required library/object, but to no
avail.
I appreciate that Oracle have so far only certified Oracle for
Linux against Redhat 6.0, but surely Mandrake can't be THAT
different ;-)
Does anybody have any ideas, known problems etc? I REALLY HOPE
that I don't have to switch to Redhat as the only solution - no
problem with Redhat, it would just require a tedious and
complicated system rebuild, which I don't have the time to do :-(
TIA
Mike Holden
null

Robert B gin (self) (guest) wrote:
: Robert B gin (guest) wrote:
: : Since I've last posted, I've installed RedHat 5.2 and
ditched
: : Slackware!
: : (btw, what a breeze - Hat's off to RH! very nice install!).
: : I've then installed oracle, have managed to setup
tnsnames.ora
: : so I can connect via sqlplus to my remote database (on
: VAX/VMS).
: : But now I have Pro*C linker problems with:
: : make -f demo_proc.mk build OBJS=test.o exe=test
: : (blah,blah,blah)
: : /usr/lib/crt1.o(.text+0x36): undefined reference to `main'
: : make: *** [build] Error 1
: : Does anybody out there have a similar problem?
: : Rob
: Duh! forget it... (EXE has to be caps... RTFM)
: (cryptic error message though?)
: works great!!!
Could you please tell me where I can get demo_proc.mk file
for Linux (RedHat)?
Thanks
null

OCCI Link Problem

Hi,
I am using Oracle 9.2.0.1.0, Redhat 9, gcc 2.96, glibc 2.3.2-11.9. I continually receive the following link problem:
/usr/bin/ld: warning: libclntsh.so, needed by /opt/ora9/product/9.2//lib/libvdc.so, not found (try using -rpath or -rpath-link)
/opt/ora9/product/9.2//lib/libocci9.a(occiUtil.o)(.text+0x99f): In function `oracle::occi::toUpperCase(basic_string<char, string_char_traits<char>, __default_alloc_template<true, 0> > const &)':
: undefined reference to `__ctype_toupper'
collect2: ld returned 1 exit status
Can anyone assist in resolving this issue?
Thanks
JD

Removed libocci9 and it compiled.
JD

Am I able to find my iTunes serial number? I need it for the problem solving as my problem is solely bad on my iTunes.

For the Apple problem solving section I need a serial number, my issue is that my problem regards missing contents within my iTunes. Do I have an iTunes serial number? And if so, where can I locate it? If there is not a serial number for iTunes, advice on what to do next is much appreciated.

iTunes has no serial number. If you are trying to get to Apple phone support, that is linked to the purchase of Apple hardware, so you need to enter in the serial number of an Apple device. Otherwise, for assistance with problems with iTunes itself you would need to call Apple Support and they would almost certainly charge for the call. I would instead suggest you post the issue you're trying to address here in detail. Someone here can probably offer suggestions.
If your issue is with the iTunes Store, not the iTunes application, go here:
http://www.apple.com/support/itunes/contact/
and follow the instructions to report your issue to the iTunes Store.
Regards.

OCCI Link problem (Error LNK2019: unresolved external symbol) in VS2012

Hi,
I'm having a hard time with this. I'm trying to implement OCCI within my application. Error I'm getting is:
Error     1     error LNK2019: unresolved external symbol "public: static class oracle::occi::Environment * __cdecl oracle::occi::Environment::createEnvironment(class std::basic_string<char,struct std::char_traits<char>,class std::allocator<char> > const &,class std::basic_string<char,struct std::char_traits<char>,class std::allocator<char> > const &,enum oracle::occi::Environment::Mode,void *,void * (__cdecl*)(void *,unsigned int),void * (__cdecl*)(void *,void *,unsigned int),void (__cdecl*)(void *,void *))" (?createEnvironment@Environment@occi@oracle@@SAPAV123@ABV?$basic_string@DU?$char_traits@D@std@@V?$allocator@D@2@@std@@0W4Mode@123@PAXP6APAX2I@ZP6APAX22I@ZP6AX22@Z@Z) referenced in function "private: bool __thiscall ServerDataLoader::getPreSuffixesFromDB(void)" (?getPreSuffixesFromDB@ServerDataLoader@@AAE_NXZ)     P:\code\VizTool\VizPortal\DwgDgnConverter\ServerDataLoader.obj     DwgDgnConverter
Error     2     error LNK1120: 1 unresolved externals     P:\code\VizTool\VizPortal\Debug\DwgDgnConverter.exe     1     1     DwgDgnConverter
Yes, it looks like linking problem,...but for me everything seems set-up.
- Included header files (from instantclient-sdk-windows.x64-11.2.0.3.0.zip)
- Included library path from OCCI download (11.2.0.3.0)
- Added oraocci11d.lib in linker
Please, what am I doing wrong?
Millions of thanks for any kind of feedback ;-)

Have you succeeded? I tried vs2012 and it passed compilation, but gave the run-time error of _crtisvalidheappointer. But the same code runs well with vs2010. I think Oracle needs to publish the new patch for vs2012

Element link problem in datamodel

Hi,
We created 2 datasets in Datamodel, When we tried to put element level links for both the datasets , its showing error when we generate sample xml.
The error is displayed as shown below:
XML Parsing Error: no element found
Location: http://localhost:9704/xmlpserver/servlet/xdo
Line Number 4, Column 1:
Also in bipublisher.log file, i'm getting this error:
*[2012-03-02T10:54:01.438+05:30] [bi_server1] [WARNING] [] [oracle.xdo] [tid: 38] [userId: <anonymous>] [ecid: 4b1d6eaff3f572e1:6d93021e:135c76b9078:-8000-0000000000015be5,0] [APP: bipublisher#11.1.1] java.sql.SQLException: java.io.IOException: prepare query failed[nQSError: 43113] Message returned from OBIS.[[*
*[nQSError: 27002] Near <=>: Syntax error [nQSError: 26012] .*
the column which 'm trying to link is having same datatype as VARCHAR only.
What should be done for this error? can anyone help in this issue?
Thanks & Regards,
Surya

Thanks for the response Thomas!
The functionality seems to work just fine. I use the LINK PROBLEM and it pops up the problem search box, I search for and select a Problem, I then select it and click ok, and I get no error saying the relationship was not created.
The problem is, when I open the problem after that, there is never any relationship showing in related items there, as it says there should be.
Also, to point out, I am talking about newly created Incidents. I tried even using APPLY prior to doing the LINK PROBLEM but it still did not work. I suspect it may be a workflow problem, where the workflow to create the incident has not completed, there
for the relationship is not completed.... and I get no error.
In my work environment, we have analysts during outages taking a call every 30 seconds, and I need a more efficient way for this system to work. As of now, they spend 1-2 minutes doing the ticket, and that is too long.
I have decided to try a newb way, where I created an Incident (GENERIC OUTAGE) that folks can use, and every 15-30 minutes during an outage, I can open the problem and search for open incidents with that title, and attach all that I find from the problem
side. This is assuming that once attached they will not show up in the search to attach the next time lol. Also this assumes that if attached from the problem, they will auto resolve when the problem is resolved.
I am rusty on my programming, and am currently trying to get refreshed. I am sure that would come in handy for this :)
Oh and lastly, one additional tidbit that will make this clearer, with regard to work flow, we re also running our Service Manager in our help desk, virtualized, as they have not upgraded to Win 7 yet :(. So that compiles my time to add ticket issue :(.

OpenMPI compile/link problems [SOLVED]

Similar Messages

Maybe you are looking for