"value":"In the Linux kernel, the following vulnerability has been resolved:\n\nVMCI: Fix memcpy() run-time warning in dg_dispatch_as_host()\n\nSyzkaller hit 'WARNING in dg_dispatch_as_host' bug.\n\nmemcpy: detected field-spanning write (size 56) of single field \"&dg_info->msg\"\nat drivers/misc/vmw_vmci/vmci_datagram.c:237 (size 24)\n\nWARNING: CPU: 0 PID: 1555 at drivers/misc/vmw_vmci/vmci_datagram.c:237\ndg_dispatch_as_host+0x88e/0xa60 drivers/misc/vmw_vmci/vmci_datagram.c:237\n\nSome code commentry, based on my understanding:\n\n544 #define VMCI_DG_SIZE(_dg) (VMCI_DG_HEADERSIZE + (size_t)(_dg)->payload_size)\n/// This is 24 + payload_size\n\nmemcpy(&dg_info->msg, dg, dg_size);\n\tDestination = dg_info->msg ---> this is a 24 byte\n\t\t\t\t\tstructure(struct vmci_datagram)\n\tSource = dg --> this is a 24 byte structure (struct vmci_datagram)\n\tSize = dg_size = 24 + payload_size\n\n{payload_size = 56-24 =32} -- Syzkaller managed to set payload_size to 32.\n\n 35 struct delayed_datagram_info {\n 36 struct datagram_entry *entry;\n 37 struct work_struct work;\n 38 bool in_dg_host_queue;\n 39 /* msg and msg_payload must be together. */\n 40 struct vmci_datagram msg;\n 41 u8 msg_payload[];\n 42 };\n\nSo those extra bytes of payload are copied into msg_payload[], a run time\nwarning is seen while fuzzing with Syzkaller.\n\nOne possible way to fix the warning is to split the memcpy() into\ntwo parts -- one -- direct assignment of msg and second taking care of payload.\n\nGustavo quoted:\n\"Under FORTIFY_SOURCE we should not copy data across multiple members\nin a structure.\""
