Using Fedora Core 4, the net ads add command fails with the following error (followed by the usual trace): *** glibc detected *** net: free(): invalid pointer: 0x004d27f0 ======= Backtrace: ========= /lib/libc.so.6[0x2b7424] /lib/libc.so.6(__libc_free+0x77)[0x2b795f] /lib/libcom_err.so.2(remove_error_table+0x4b)[0x128abb] /usr/lib/libkrb5.so.3[0x470823] /usr/lib/libkrb5.so.3[0x4705c7] /usr/lib/libkrb5.so.3[0x4c13ba] /lib/ld-linux.so.2[0xaf2058] /lib/libc.so.6(exit+0xc5)[0x27ec69] /lib/libc.so.6(__libc_start_main+0xce)[0x268dee] net[0x8e70f1] ---------------------------------------------------------------------------- It looks like the repsonse to invalid 'free()' calls has changed. You can get around this by manually setting MALLOC_CHECK to either 0 (ignore) or 1 (report and continue) but that's not at all obvious at the outset. :-) Thanks, Garth D'Attilo NCAR/ACD Systems Administrator
Hmmm... Looks like that should be "net ads join"... -Garth
any chance of getting a backtrace with debug symbols (compile with --enable-debug)?
Version 3.0.20pre2 compiled with --enable-debug. Debug level set to 10. Everything seems to progress normally until this point. Please let me know if you want the entire log file or if I have misunderstood what you are asking for. Thanks in advance, Garth ----------------------------------------------------------------------- [2005/07/21 17:54:28, 2] utils/net.c:main(897) return code = 0 *** glibc detected *** /usr/bin/net: free(): invalid pointer: 0x001c8db0 *** ======= Backtrace: ========= /lib/libc.so.6[0x2c7424] /lib/libc.so.6(__libc_free+0x77)[0x2c795f] /lib/libcom_err.so.2(remove_error_table+0x4b)[0x1c9abb] /usr/lib/libkrb5.so.3[0x1668c4] /usr/lib/libkrb5.so.3[0x1665c7] /usr/lib/libkrb5.so.3[0x1b79da] /lib/ld-linux.so.2[0x7f7058] /lib/libc.so.6(exit+0xc5)[0x28ec69] /lib/libc.so.6(__libc_start_main+0xce)[0x278dee] /usr/bin/net[0xa270f1]
Garth, I reproduce the same issue locally. This looks like a bug in the kerberos libs which I think redhat will have to handle with the MIT folks. I've copied Jay Fenlason at redaht on this bug.
This is a bug in the e2fsprogs on FC4. Confirmed by RH developers. Not our bug.