|
From: John C. <jo...@jc...> - 2005-05-21 03:47:11
|
Hi people. Please forgive my reposting the issue to this list - it seems the bugtracking list at sf.net is very little used. Here's a short version of the problem: 1) valgrind 2.4.0 doesn't start at all (without any parameters). It stops with just a Segmentation Fault. 2) 2.2.0 does work 3) I tried to trace into valgrind 2.4.0 with gdb, but I'm no expert see below what little I managed to get. Both valgrinds were compiled here without errors. Linux 2.6.11, gcc 3.4.3 ------------Here the most important info: 'run'ning in gdb gives segfault and a one-line backtrace: #0 0xb805c600 in ?? () Really stepping gives: 311 init_sp = argv - 1; 315 VALGRIND_MAKE_READABLE(init_sp, sizeof(int)); 317 cp = getenv(VALGRINDLIB); 319 if (cp != NULL) 324 getrlimit(RLIMIT_AS, &rlim); 325 rlim.rlim_cur = rlim.rlim_max; 326 setrlimit(RLIMIT_AS, &rlim); 329 jmp_with_stack(main2, (Addr)stack + sizeof(stack)); Program received signal SIGSEGV, Segmentation fault. 0xb805c600 in ?? () Stepping into jmp_with_stack: .... 280 err = do_exec(buf, &info); 282 if (err != 0) { 290 padfile = as_openpadfile(); 291 as_pad(0, (void *)info.map_base, padfile); 293 esp = fix_auxv(init_sp, &info, padfile); 301 jmp_with_stack((void (*)(void))info.init_eip, (Addr)esp); Program received signal SIGSEGV, Segmentation fault. 0xb805c600 in ?? () |
|
From: Nicholas N. <nj...@cs...> - 2005-05-21 19:30:36
|
On Sat, 21 May 2005, John Coppens wrote: > Please forgive my reposting the issue to this list - it seems the > bugtracking list at sf.net is very little used. Here's a short version of > the problem: Can you track this on our bugzilla page at http://bugs.kde.org/enter_valgrind_bug.cgi ? Thanks. N |
|
From: Jeremy F. <je...@go...> - 2005-05-25 19:32:42
|
John Coppens wrote: >Hi people. > >Please forgive my reposting the issue to this list - it seems the >bugtracking list at sf.net is very little used. Here's a short version of >the problem: > >1) valgrind 2.4.0 doesn't start at all (without any parameters). It stops >with just a Segmentation Fault. > > Can you try building with "configure --disable-pie"? J |
|
From: John C. <jo...@jc...> - 2005-05-28 18:36:03
|
On Wed, 25 May 2005 11:46:15 -0700 Jeremy Fitzhardinge <je...@go...> wrote: > John Coppens wrote: > > >Hi people. > > > >Please forgive my reposting the issue to this list - it seems the > >bugtracking list at sf.net is very little used. Here's a short version > >of the problem: > > > >1) valgrind 2.4.0 doesn't start at all (without any parameters). It > >stops with just a Segmentation Fault. > > > > > Can you try building with "configure --disable-pie"? Hi Jeremy, Yes - that works. I wonder - if the executable was built with PIE, that's because support was detected, wasn't it? Why doesn't it work afterwards - is something missing? Thanks, John |
|
From: Jeremy F. <je...@go...> - 2005-05-29 05:28:42
|
John Coppens wrote:
>Yes - that works. I wonder - if the executable was built with PIE, that's
>because support was detected, wasn't it? Why doesn't it work afterwards -
>is something missing?
>
On some systems, the configure check works, but the final binary
doesn't. I don't know why yet.
J
|