|
From: Arand, T. <tho...@si...> - 2006-06-23 14:23:27
|
Hello
=20
we have a program that runs fairly well under valgrind control - there
are a few messages, but that's all.
=20
We compiled the same program (and all shared libs it uses) with the gcc
(V 3.4.3) option -fstack-check and let it again run on the same machine
(a MontaVista CarrierGrade Linux based on a 2.6 Linux kernel). Now we
get many thousands (more that 100000, after that valgrind stops
reporting) errors messages, obviously at any function entry:
=20
=3D=3D25102=3D=3D Invalid write of size 4
=3D=3D25102=3D=3D at 0x82E2561: ShwmSnmpTrap::ShwmSnmpTrap()
(ShwmAppl.cc:178)
=3D=3D25102=3D=3D by 0x82E0D1F:
ShwmAppl::ShwmMsgRegisterClient::registerMessageTypes()
(ShwmAppl.cc:181)
=3D=3D25102=3D=3D by 0x4108A9D:
GenInitLibClientStore<TspMsgWrapperRegisterClient>::init() (in
/usr/project/lib/libTspComWrapper.so.3.2.2.06061321)
=3D=3D25102=3D=3D by 0x40D8D54: GenInitLibServer::init()
(GenApplInitLib.cc:97)
=3D=3D25102=3D=3D by 0x40D8854: GenApplInitLib::initLibs()
(GenApplInitLib.cc:52)
=3D=3D25102=3D=3D by 0x40D8BF3: GenApplInitLib::doStart()
(GenApplInitLib.cc:87)
=3D=3D25102=3D=3D by 0x40D1DF4: GenApplication::start(unsigned, =
char**)
(GenApplication.cc:40)
=3D=3D25102=3D=3D by 0x82E0ACE: ShwmAppl::main(int, char**) =
(ShwmAppl.cc:141)
=3D=3D25102=3D=3D by 0x808822C: main (MainSHWM.cc:28)
=3D=3D25102=3D=3D Address 0xBEFFCD58 is not stack'd, malloc'd or =
(recently)
free'd
=20
Does anybody know what is happening?
=20
Thanks and Ciao,
Thomas Arand
=20
=20
----------------------------------------------------------------------=20
Thomas Arand Siemens AG=20
Tel. +49 - 89 - 636 76328=20
Fax. +49 - 89 - 636 76282=20
tho...@si...=20
=20
|
|
From: Julian S. <js...@ac...> - 2006-06-24 12:26:49
|
> Does anybody know what is happening? Dunno, maybe gcc included code which is messing with your stack somehow. What happens if you run V with --workaround-gcc296-bugs=yes ? J |