Re: warning: RTTI symbol not found for class

[ali_anwar <ali_anwar at codesourcery dot com>]

On 11/21/2012 11:32 PM, Jan Kratochvil wrote:
> Hello Ali,
>
> On Wed, 21 Nov 2012 19:06:22 +0100, ali_anwar wrote:
> > While debugging a cpp demo (containing the std::cout call) when I
> > issue "p std::cout", I get following warning: "warning: RTTI symbol
> > not found for class 'std::ostream'".
>
> I do not have it reproducible.  When you have not provided a GDB testcase
> could you at least provide OS and a reproducer there?  This way the patch does
> not fix anything to me.
>
> Fedora 18 x86_64
> GNU gdb (GDB) 7.5.50.20121120-cvs
> cat>cout.C<<HERE
> #include<iostream>
> int main () { std::cout<<  "foo"<<  std::endl; }
> HERE
> g++ -o cout cout.C -Wall -g
> gdb -q ./cout -ex start -ex 'p std::cout' -ex c -ex q
> $1 = {<std::basic_ios<char, std::char_traits<char>  >>  =<invalid address>, _vptr.basic_ostream = 0x0}
>
> (<invalid address>  does not look great but this patch does not fix it)

I tried it for powerpc-eabi. But following should reproduce the issue in 
your scenario as well:

gdb -q ./cout -ex start -ex s -ex fin -ex c -ex q
[snip]
2	int main () { std::cout << "foo" << std::endl; }
Value returned is $1 = warning: RTTI symbol not found for class 
'std::ostream'
[snip]


I am on Ubuntu 10.04 32 bit.

> > Do we have any fix for this issue if not then is it ok to have this
> > work around until we get a proper fix?
>
> It has a regression:
>
> -PASS: gdb.cp/bs15503.exp: print s.length()
> -PASS: gdb.cp/bs15503.exp: print s[0]
> -PASS: gdb.cp/bs15503.exp: print s[s.length()-1]
> +FAIL: gdb.cp/bs15503.exp: print s.length()
> +FAIL: gdb.cp/bs15503.exp: print s[0]
> +FAIL: gdb.cp/bs15503.exp: print s[s.length()-1]
> -PASS: gdb.cp/bs15503.exp: print (const char *) s.substr(0,4)
> -PASS: gdb.cp/bs15503.exp: print (const char *) (s=s.substr(0,4))
> +FAIL: gdb.cp/bs15503.exp: print (const char *) s.substr(0,4)
> +FAIL: gdb.cp/bs15503.exp: print (const char *) (s=s.substr(0,4))
I am facing following 6 failures without the proposed change on Ubuntu 
10.04 32 bit.

FAIL: gdb.cp/bs15503.exp: print s.length()
FAIL: gdb.cp/bs15503.exp: print s[0]
FAIL: gdb.cp/bs15503.exp: print s[s.length()-1]
FAIL: gdb.cp/bs15503.exp: print (const char *) s
FAIL: gdb.cp/bs15503.exp: print (const char *) s.substr(0,4)
FAIL: gdb.cp/bs15503.exp: print (const char *) (s=s.substr(0,4))

After applying patch I faced only following 5 issue:

FAIL: gdb.cp/bs15503.exp: print s.length()
FAIL: gdb.cp/bs15503.exp: print s[0]
FAIL: gdb.cp/bs15503.exp: print s[s.length()-1]
FAIL: gdb.cp/bs15503.exp: print (const char *) s.substr(0,4)
FAIL: gdb.cp/bs15503.exp: print (const char *) (s=s.substr(0,4))


> It has also a second regression but I understand that can be ignored as it
> just tests what you have changed:
>
> -PASS: gdb.cp/no-dmgl-verbose.exp: DMGL_VERBOSE-demangled f(std::string) is not defined
> +FAIL: gdb.cp/no-dmgl-verbose.exp: DMGL_VERBOSE-demangled f(std::string) is not defined
>
>
> I do not find the patch finished for a review when it still regresses.
>
> (It regresses on Fedora 18 x86_64 in the case it does not regress for you.)
Let me try on some other machine and verify it.

Thanks,
-Ali