CVE-2024-46679

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> ethtool: check device is present when getting link settings<br /> <br /> A sysfs reader can race with a device reset or removal, attempting to<br /> read device state when the device is not actually present. eg:<br /> <br /> [exception RIP: qed_get_current_link+17]<br /> #8 [ffffb9e4f2907c48] qede_get_link_ksettings at ffffffffc07a994a [qede]<br /> #9 [ffffb9e4f2907cd8] __rh_call_get_link_ksettings at ffffffff992b01a3<br /> #10 [ffffb9e4f2907d38] __ethtool_get_link_ksettings at ffffffff992b04e4<br /> #11 [ffffb9e4f2907d90] duplex_show at ffffffff99260300<br /> #12 [ffffb9e4f2907e38] dev_attr_show at ffffffff9905a01c<br /> #13 [ffffb9e4f2907e50] sysfs_kf_seq_show at ffffffff98e0145b<br /> #14 [ffffb9e4f2907e68] seq_read at ffffffff98d902e3<br /> #15 [ffffb9e4f2907ec8] vfs_read at ffffffff98d657d1<br /> #16 [ffffb9e4f2907f00] ksys_read at ffffffff98d65c3f<br /> #17 [ffffb9e4f2907f38] do_syscall_64 at ffffffff98a052fb<br /> <br /> crash&gt; struct net_device.state ffff9a9d21336000<br /> state = 5,<br /> <br /> state 5 is __LINK_STATE_START (0b1) and __LINK_STATE_NOCARRIER (0b100).<br /> The device is not present, note lack of __LINK_STATE_PRESENT (0b10).<br /> <br /> This is the same sort of panic as observed in commit 4224cfd7fb65<br /> ("net-sysfs: add check for netdevice being present to speed_show").<br /> <br /> There are many other callers of __ethtool_get_link_ksettings() which<br /> don&amp;#39;t have a device presence check.<br /> <br /> Move this check into ethtool to protect all callers.