[ace-users] ACE TAO: Upgrade from 5.5.4 to 6.5.7
jwillemsen at remedy.nl
Tue Mar 17 03:28:17 CDT 2020
Maybe something you can report back to Intel so that they can fix it in
their next compiler update?
On 3/16/20 12:11 PM, Yorick wrote:
> Hi All,
> Just to give an update so this issue can be closed.
> We figured out after investigation that there must be a bug in the
> Intel Compiler we are using.
> Reducing the optimization from -O3 to -O2 solved this issue.
> --- ACE_wrappers/include/makeinclude/platform_linux_icc.GNU Fri Nov 22
> 09:26:59 2019
> +++ ACE_wrappers/include/makeinclude/platform_linux_icc.GNU Fri Mar 13
> 13:07:57 2020
> @@ -84,7 +84,7 @@
> LIBS += -lrt
> -OCFLAGS += -O3
> +OCFLAGS += -O2
> # Disable floating point optimizer solves problem with
> # min/max float values in the TAO_IDL compiler tests
> I hope this helps anyone encountering the same issue.
> Many thanks for the quick responses.
> Yorick Bosman
> Op do 12 mrt. 2020 om 17:02 schreef Yorick <yorick.bosman at gmail.com
> <mailto:yorick.bosman at gmail.com>>:
> Hi Johnny and Adam,
> Thanks for the quick reply.
> With valgrind I could eventually found out an issue with ACE / our
> code (not sure where the responsibility lies)
> So, the issue is when we call CORBA::ORB_init with some strange
> char* argv;
> argv = "argument0";
> argv = "argument1";
> argv = "argument2";
> argv = "argument3";
> argv = "argument4";
> int argc = 3;
> // The issue occurs here, argv has more values after the count of
> 3 (specified by argc)
> CORBA::ORB_init ( argc, argv, "something" );
> When I force argv = "\0"; it seems to work, although i
> get an other issues somewhere else.
> To be complete: everything works just fine with ACE 5.5.4. Unit
> tests seem to work just fine.
> ace-users mailing list
> ace-users at list.isis.vanderbilt.edu
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the ace-users