Am 10.06.2007 um 18:35 schrieb Rolf Fokkens:
Sorry - immidiately after being absolutely sure about this, the
same problem also occured with rtl8139 emulation. Well, it's an
issue anyhow.
Rolf
Rolf Fokkens wrote:
On both qemu (no-kqemu) and qemu-kvm (no-kvm or with kvm) release
0.9.0 the following happens when using ne2000 (default) emulation:
UG: waring at kernel/softirq.c:138/local_bh_enable() (Not tainted)
[<c042b0cf>] local_bh_enable+ /0x92
[..........] cond_resched_sof +0x2c/0x42
[..........] release_sock+0x4f/0 9d
[..........] inet_stream_connect+0x116/0x1ff
[..........] autoremove_wake_function+0x0/0x35
[..........] sys_connect+0x82/0xad
[..........] d_instantiate+0x5c/ x60
[..........] sock_attach_fd+0x70/0xcf
[..........] selinux_file_alloc_security+0x1f/0x40
[..........] get_empty_filp+0x99/0x151
[..........] __sock_create+0x16e/0x1a8
[..........] fd_install+0x1e/0x47
[..........] sock_map_fd+0x41/0x47
[..........] sys_socketcall+0xac/0x261
[..........] syscall_call+0x7/0xb
The virtual host runs fedora 7, kernel-2.6.21-1.3194.fc7. When
using windows-XP as a virtual host a network hang also occurs, but
I have no details about it.
The mentioned problems (for both Linux and Windows XP) do not
occur when using rtl8139 emulation.
The problem might rather be Fedora 7 itself - someone recently
reported they changed some of their ATA libraries and with the recent
Fedora and QEMU CVS updates it no longer boots at all for me...
NE2000 appears to work fine on my other guests.
Andreas
