CVE-2021-47139

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net: hns3: put off calling register_netdev() until client initialize complete<br /> <br /> Currently, the netdevice is registered before client initializing<br /> complete. So there is a timewindow between netdevice available<br /> and usable. In this case, if user try to change the channel number<br /> or ring param, it may cause the hns3_set_rx_cpu_rmap() being called<br /> twice, and report bug.<br /> <br /> [47199.416502] hns3 0000:35:00.0 eth1: set channels: tqp_num=1, rxfh=0<br /> [47199.430340] hns3 0000:35:00.0 eth1: already uninitialized<br /> [47199.438554] hns3 0000:35:00.0: rss changes from 4 to 1<br /> [47199.511854] hns3 0000:35:00.0: Channels changed, rss_size from 4 to 1, tqps from 4 to 1<br /> [47200.163524] ------------[ cut here ]------------<br /> [47200.171674] kernel BUG at lib/cpu_rmap.c:142!<br /> [47200.177847] Internal error: Oops - BUG: 0 [#1] PREEMPT SMP<br /> [47200.185259] Modules linked in: hclge(+) hns3(-) hns3_cae(O) hns_roce_hw_v2 hnae3 vfio_iommu_type1 vfio_pci vfio_virqfd vfio pv680_mii(O) [last unloaded: hclge]<br /> [47200.205912] CPU: 1 PID: 8260 Comm: ethtool Tainted: G O 5.11.0-rc3+ #1<br /> [47200.215601] Hardware name: , xxxxxx 02/04/2021<br /> [47200.223052] pstate: 60400009 (nZCv daif +PAN -UAO -TCO BTYPE=--)<br /> [47200.230188] pc : cpu_rmap_add+0x38/0x40<br /> [47200.237472] lr : irq_cpu_rmap_add+0x84/0x140<br /> [47200.243291] sp : ffff800010e93a30<br /> [47200.247295] x29: ffff800010e93a30 x28: ffff082100584880<br /> [47200.254155] x27: 0000000000000000 x26: 0000000000000000<br /> [47200.260712] x25: 0000000000000000 x24: 0000000000000004<br /> [47200.267241] x23: ffff08209ba03000 x22: ffff08209ba038c0<br /> [47200.273789] x21: 000000000000003f x20: ffff0820e2bc1680<br /> [47200.280400] x19: ffff0820c970ec80 x18: 00000000000000c0<br /> [47200.286944] x17: 0000000000000000 x16: ffffb43debe4a0d0<br /> [47200.293456] x15: fffffc2082990600 x14: dead000000000122<br /> [47200.300059] x13: ffffffffffffffff x12: 000000000000003e<br /> [47200.306606] x11: ffff0820815b8080 x10: ffff53e411988000<br /> [47200.313171] x9 : 0000000000000000 x8 : ffff0820e2bc1700<br /> [47200.319682] x7 : 0000000000000000 x6 : 000000000000003f<br /> [47200.326170] x5 : 0000000000000040 x4 : ffff800010e93a20<br /> [47200.332656] x3 : 0000000000000004 x2 : ffff0820c970ec80<br /> [47200.339168] x1 : ffff0820e2bc1680 x0 : 0000000000000004<br /> [47200.346058] Call trace:<br /> [47200.349324] cpu_rmap_add+0x38/0x40<br /> [47200.354300] hns3_set_rx_cpu_rmap+0x6c/0xe0 [hns3]<br /> [47200.362294] hns3_reset_notify_init_enet+0x1cc/0x340 [hns3]<br /> [47200.370049] hns3_change_channels+0x40/0xb0 [hns3]<br /> [47200.376770] hns3_set_channels+0x12c/0x2a0 [hns3]<br /> [47200.383353] ethtool_set_channels+0x140/0x250<br /> [47200.389772] dev_ethtool+0x714/0x23d0<br /> [47200.394440] dev_ioctl+0x4cc/0x640<br /> [47200.399277] sock_do_ioctl+0x100/0x2a0<br /> [47200.404574] sock_ioctl+0x28c/0x470<br /> [47200.409079] __arm64_sys_ioctl+0xb4/0x100<br /> [47200.415217] el0_svc_common.constprop.0+0x84/0x210<br /> [47200.422088] do_el0_svc+0x28/0x34<br /> [47200.426387] el0_svc+0x28/0x70<br /> [47200.431308] el0_sync_handler+0x1a4/0x1b0<br /> [47200.436477] el0_sync+0x174/0x180<br /> [47200.441562] Code: 11000405 79000c45 f8247861 d65f03c0 (d4210000)<br /> [47200.448869] ---[ end trace a01efe4ce42e5f34 ]---<br /> <br /> The process is like below:<br /> excuting hns3_client_init<br /> |<br /> register_netdev()<br /> | hns3_set_channels()<br /> | |<br /> hns3_set_rx_cpu_rmap() hns3_reset_notify_uninit_enet()<br /> | |<br /> | quit without calling function<br /> | hns3_free_rx_cpu_rmap for flag<br /> | HNS3_NIC_STATE_INITED is unset.<br /> | |<br /> | hns3_reset_notify_init_enet()<br /> | |<br /> set HNS3_NIC_STATE_INITED call hns3_set_rx_cpu_rmap()-- crash<br /> <br /> Fix it by calling register_netdev() at the end of function<br /> hns3_client_init().