|
[Valgrind-users] Re: [Bug 71046] New: Unable to run CVS HEAD "Executable is mapped
outside of range"
From: <ar...@de...> - 2003-12-29 14:44:32
|
I've tested this and I think it's related to a prelinked stage2 binary. When I prelink (http://www.gentoo.org/doc/en/prelink-howto.xml) the system, particulary /usr/lib/valgrind/stage2 (from valgrind-snapshot debian package), valgrind stops working. If I undo the prelinking (prelink -u) on this file, valgrind will work again. There may be some hardcoded address within the code of stage2 because what prelink does is (roughly) to relocate symbols of an ELF file and hence, it does change the binary addresses somehow. I don't know if this can be fixed or in the future, prelink and valgrind cannot coexist together. Thanks. Crispin Flowerday <gn...@fl...> writes: > ------- You are receiving this mail because: ------- > You are the assignee for the bug, or are watching the assignee. > > http://bugs.kde.org/show_bug.cgi?id=71046 > Summary: Unable to run CVS HEAD "Executable is mapped outside of > range" > Product: valgrind > Version: unspecified > Platform: unspecified > OS/Version: Linux > Status: UNCONFIRMED > Severity: normal > Priority: NOR > Component: general > AssignedTo: js...@ac... > ReportedBy: gn...@fl... > > > While trying to run valgrind from the current CVS HEAD, I found it wouldn't run > on my system: > > $ /opt/valgrind/bin/valgrind /bin/ls > Executable is mapped outside of range 0x80d3000-0xbffff000, was 0x8048000-0x8131920 > failed to load /opt/valgrind/lib/valgrind/stage2: Cannot allocate memory > > (I have add the "was %p-%p" to show where the mmap actually occured.) > > System (debian sid): > > Linux 2.6.0 (hand compiled) > libc6 2.3.2.ds1-10 (debian package) > > I you need any more information, e.g strace, I'll be happy to help. -- Andrés Roldán <ar...@de...> GPG Key-ID: 0xB29396EB http://people.fluidsignal.com/~aroldan |
|
From: Jeremy F. <je...@go...> - 2003-12-29 17:34:27
|
On Mon, 2003-12-29 at 06:45, Andrés Roldán wrote: > I've tested this and I think it's related to a prelinked stage2 binary. > > When I prelink (http://www.gentoo.org/doc/en/prelink-howto.xml) the system, > particulary /usr/lib/valgrind/stage2 (from valgrind-snapshot debian package), > valgrind stops working. > > If I undo the prelinking (prelink -u) on this file, valgrind will work again. > > There may be some hardcoded address within the code of stage2 because what > prelink does is (roughly) to relocate symbols of an ELF file and hence, it > does change the binary addresses somehow. > > I don't know if this can be fixed or in the future, prelink and valgrind > cannot coexist together. Well, I just ran prelink on stage2 in Fedora and it seemed to work fine. I'm not quite sure what prelink means on an executable (as opposed to a .so file). Could you send me copies of before and after stage2? It might me that the answer is "don't do that then". J |