Virtualbox hangs on start in Ubuntu 8.10

[Steve]

I tried several different packages from the closed source release, and tried compiling from source. All of them give me the same error. When I start Virtualbox from the terminal with "./VirtualBox", 100% of my CPU is taken up and there is no output. There is no VBox.log file.

I did an strace and discovered that it was repeating the same few lines over and over:

select(8, [7], [], NULL, NULL)          = 1 (in [7])
read(7, "\241\10\324\3300\0 \0053\1\04\0\0\0\0\0\0"..., 4096) = 32
read(7, 0x99a345c, 4096)                = -1 EAGAIN (Resource temporarily unavailable)
select(8, [7], [7], NULL, NULL)         = 1 (out [7])
writev(7, [{"\4\2\300\0 \524\06\0213\0\0+\1\04\0\0\0"..., 32}], 1) = 32
select(8, [7], [], NULL, NULL)          = 1 (in [7])
read(7, "\1 \326\31\0\04\0\0\0\01\0\0@\4 \10\330\254"..., 4096) = 36
read(7, 0x99a345c, 4096)                = -1 EAGAIN (Resource temporarily unavailable)
select(8, [7], [7], NULL, NULL)         = 1 (out [7])
writev(7, [{"\27\02\0001\1\0", 8}], 1) = 8 

So I ran VirtualBox in gdb and added breakpints to all calls of "select". After running for a short time it started repeating the same breakpoint.

Breakpoint 142, 0xb6fc5ba0 in select () from /lib/tls/i686/cmov/libc.so.6
(gdb) step
Single stepping until exit from function select,
which has no line number information.
0xb6a55862 in xcb_wait_for_reply () from /usr/lib/libxcb.so.1
(gdb)
Single stepping until exit from function xcb_wait_for_reply,
which has no line number information.
0xb6e3f369 in _XReply () from /usr/lib/libX11.so.6
(gdb)
Single stepping until exit from function _XReply,
which has no line number information.
0xb6e2c217 in XQueryExtension () from /usr/lib/libX11.so.6
(gdb)
Single stepping until exit from function XQueryExtension,
which has no line number information.
0xb6e20771 in XInitExtension () from /usr/lib/libX11.so.6
(gdb)
Single stepping until exit from function XInitExtension,
which has no line number information.
0xb6e929a0 in XkbUseExtension () from /usr/lib/libX11.so.6
(gdb)
Single stepping until exit from function XkbUseExtension,
which has no line number information.

Breakpoint 142, 0xb6fc5ba0 in select () from /lib/tls/i686/cmov/libc.so.6
(gdb)
Single stepping until exit from function select,
which has no line number information.
0xb6a557b6 in xcb_wait_for_reply () from /usr/lib/libxcb.so.1
...

My core dump is 3.2 Mb. Let me know if you need it.

[Steve]

This is a repeat of bug #2586. My apologies, I did not know that 2586 went through.

[Frank Mehnert]

Yes, that core dump could help. Please could you e-mail it to frank _dot_ mehnert _at_ sun _dot_ com? Note that the core dump is only useful for me if you took it with an official package from virtualbox.org. In that case, please tell me which package.

[Steve]

I still don't know what the problem is, but I have confirmed that the same error happens when I try to run nessusclient. This makes me think that it is some sort of conflict on my end. I also confirmed that Virtualbox works from a livecd. I will do my best to find out what is causing this problem and post any findings.

[Nicholas Newberry]

I have similar symptoms with Windows XP running as guest on Ubuntu 8.10 64-bit. I am booting a pre-existing Windows XP installation from a writethrough vmdk.

poll([{fd=26, events=POLLIN}, {fd=31, events=POLLIN}, {fd=30, events=POLLIN}, {fd=35, events=POLLIN}, {fd=40, events=POLLIN|POLLPRI}, {fd=42, events=POLLIN|POLLPRI}, {fd=43, events=POLLIN|POLLPRI}, {fd=44, events=POLLIN|POLLPRI}], 8, 43) = 0
read(30, 0xc61974, 4096)                = -1 EAGAIN (Resource temporarily unavailable)
read(30, 0xc61974, 4096)                = -1 EAGAIN (Resource temporarily unavailable)
poll([{fd=26, events=POLLIN}, {fd=31, events=POLLIN}, {fd=30, events=POLLIN}, {fd=35, events=POLLIN}, {fd=40, events=POLLIN|POLLPRI}, {fd=42, events=POLLIN|POLLPRI}, {fd=43, events=POLLIN|POLLPRI}, {fd=44, events=POLLIN|POLLPRI}], 8, 45) = 0
read(30, 0xc61974, 4096)                = -1 EAGAIN (Resource temporarily unavailable)
read(30, 0xc61974, 4096)                = -1 EAGAIN (Resource temporarily unavailable)
poll([{fd=26, events=POLLIN}, {fd=31, events=POLLIN}, {fd=30, events=POLLIN}, {fd=35, events=POLLIN}, {fd=40, events=POLLIN|POLLPRI}, {fd=42, events=POLLIN|POLLPRI}, {fd=43, events=POLLIN|POLLPRI}, {fd=44, events=POLLIN|POLLPRI}], 8, 43) = 0
read(30, 0xc61974, 4096)                = -1 EAGAIN (Resource temporarily unavailable)
read(30, 0xc61974, 4096)                = -1 EAGAIN (Resource temporarily unavailable)

The 100% CPU usage seems to begin as soon as the bootloader on the writethrough disk is loaded. That is, it does not begin while the VirtualBox boot splash screen is displayed, but only after. No settings I have changed have had an effect. There are no processes running in the VM to account for the CPU usage; Windows XP shows the "System Idle" process accounting for most of the CPU time.

The VirtualBox process appears to be continuously reading on FD 30, which lsof shows as follows:

COMMAND    PID      USER   FD   TYPE             DEVICE    SIZE    NODE NAME
...
VirtualBo 6657 nnewberry   30u  unix 0xffff8800a18aed00           23426 socket

If there is any other data that would be helpful, let me know!

[Nicholas Newberry]

I should perhaps add that Windows does eventually boot in VirtualBox, and though the first attempt to perform any action (such as start an application, open the "Start" menu, etc.) seems very slow, it always seems to work, and repeated actions are much closer to what I would normally expect from a guest VM.

[Steve]

I have just completed a full reinstall of Ubuntu 8.10 and it would seem that everything works fine. I unfortunately never discovered what caused the bug in the first place, but everything is fine now.

[Sander van Leeuwen]

Then I guess we can close this for now. Thanks for the feedback.