Re: [Qemu-devel] [TestDays] info mtree abort

[Stefan Weil]

Am 12.11.2011 03:05, schrieb Andreas Färber:
> Hello,
>
> On current qemu.git master for qemu-system-x86_64 I observe crashes
> similar to this one when running info mtree on the SDL monitor console:
>
> *** glibc detected ***
> /home/andreas/QEMU/qemu-rl78/rl78-softmmu/qemu-system-rl78: free():
> invalid pointer: 0x4545454545454545 ***
>
> #0 0x00007ffff6308d95 in raise () from /lib64/libc.so.6
> #1 0x00007ffff630a2ab in abort () from /lib64/libc.so.6
> #2 0x00007ffff6342ece in __libc_message () from /lib64/libc.so.6
> #3 0x00007ffff6348c06 in malloc_printerr () from /lib64/libc.so.6
> #4 0x00000000004e9024 in mtree_print_mr (mon_printf=0x4ef720
> <monitor_printf>, f=0xf89b10, mr=<optimized out>, level=0, base=0,
> alias_print_queue=0x7fffffffd610)
> at /home/andreas/QEMU/qemu-rl78/memory.c:1405
> #5 0x00000000004ec8b6 in mtree_info (mon_printf=0x4ef720
> <monitor_printf>, f=0xf89b10) at 
> /home/andreas/QEMU/qemu-rl78/memory.c:1417
>
> Regards,
> Andreas


qemu-system-i386 also has the same problem. It does not crash
(so most users won't notice anything going wrong), but Valgrind
shows the bug, too:

==29459== Invalid read of size 4
==29459==    at 0x81ED59A: mtree_print_mr (memory.c:1404)
==29459==    by 0x81ED61E: mtree_info (memory.c:1417)
==29459==    by 0x81F1781: do_info_mtree (monitor.c:2247)
==29459==    by 0x81EE916: do_info (monitor.c:698)
==29459==    by 0x81F4234: handle_user_command (monitor.c:4130)
==29459==    by 0x81F57A6: monitor_command_cb (monitor.c:4746)
==29459==    by 0x8147992: readline_handle_byte (readline.c:370)
==29459==    by 0x81F572A: monitor_read (monitor.c:4732)
==29459==    by 0x81385AF: qemu_chr_be_write (qemu-char.c:163)
==29459==    by 0x809D1B1: kbd_send_chars (console.c:1124)
==29459==    by 0x809D440: kbd_put_keysym (console.c:1183)
==29459==    by 0x8163FBA: handle_keydown (sdl.c:724)
==29459==  Address 0x12f882c8 is 8 bytes inside a block of size 16 free'd
==29459==    at 0x4024B3A: free (vg_replace_malloc.c:366)
==29459==    by 0x8116580: free_and_trace (vl.c:2149)
==29459==    by 0x4098FC5: g_free (in /lib/libglib-2.0.so.0.2400.1)
==29459==    by 0x81ED5AA: mtree_print_mr (memory.c:1405)
==29459==    by 0x81ED61E: mtree_info (memory.c:1417)
==29459==    by 0x81F1781: do_info_mtree (monitor.c:2247)
==29459==    by 0x81EE916: do_info (monitor.c:698)
==29459==    by 0x81F4234: handle_user_command (monitor.c:4130)
==29459==    by 0x81F57A6: monitor_command_cb (monitor.c:4746)
==29459==    by 0x8147992: readline_handle_byte (readline.c:370)
==29459==    by 0x81F572A: monitor_read (monitor.c:4732)
==29459==    by 0x81385AF: qemu_chr_be_write (qemu-char.c:163)
==29459==
==29459== Invalid read of size 4
==29459==    at 0x81ED59A: mtree_print_mr (memory.c:1404)
==29459==    by 0x81ED57B: mtree_print_mr (memory.c:1400)
==29459==    by 0x81ED68D: mtree_info (memory.c:1423)
==29459==    by 0x81F1781: do_info_mtree (monitor.c:2247)
==29459==    by 0x81EE916: do_info (monitor.c:698)
==29459==    by 0x81F4234: handle_user_command (monitor.c:4130)
==29459==    by 0x81F57A6: monitor_command_cb (monitor.c:4746)
==29459==    by 0x8147992: readline_handle_byte (readline.c:370)
==29459==    by 0x81F572A: monitor_read (monitor.c:4732)
==29459==    by 0x81385AF: qemu_chr_be_write (qemu-char.c:163)
==29459==    by 0x809D1B1: kbd_send_chars (console.c:1124)
==29459==    by 0x809D440: kbd_put_keysym (console.c:1183)
==29459==  Address 0x12f88508 is 8 bytes inside a block of size 16 free'd
==29459==    at 0x4024B3A: free (vg_replace_malloc.c:366)
==29459==    by 0x8116580: free_and_trace (vl.c:2149)
==29459==    by 0x4098FC5: g_free (in /lib/libglib-2.0.so.0.2400.1)
==29459==    by 0x81ED5AA: mtree_print_mr (memory.c:1405)
==29459==    by 0x81ED57B: mtree_print_mr (memory.c:1400)
==29459==    by 0x81ED68D: mtree_info (memory.c:1423)
==29459==    by 0x81F1781: do_info_mtree (monitor.c:2247)
==29459==    by 0x81EE916: do_info (monitor.c:698)
==29459==    by 0x81F4234: handle_user_command (monitor.c:4130)
==29459==    by 0x81F57A6: monitor_command_cb (monitor.c:4746)
==29459==    by 0x8147992: readline_handle_byte (readline.c:370)
==29459==    by 0x81F572A: monitor_read (monitor.c:4732)
==29459==

(more bug messages for the same problem follow, I removed them here)

Valgrind was started like this from QEMU's source directory:

valgrind -v --error-limit=no --track-origins=yes --smc-check=all \
    --leak-check=full bin/debug/386/i386-softmmu/qemu-system-i386 -L 
pc-bios

Regards,
Stefan Weil