In the Linux kernel, the following vulnerability has been resolved:
KVM: arm64: Unregister redistributor for failed vCPU creation
Alex reports that syzkaller has managed to trigger a use-after-free when tearing down a VM:
BUG: KASAN: slab-use-after-free in kvmputkvm+0x300/0xe68 virt/kvm/kvm_main.c:5769 Read of size 8 at addr ffffff801c6890d0 by task syz.3.2219/10758
CPU: 3 UID: 0 PID: 10758 Comm: syz.3.2219 Not tainted 6.11.0-rc6-dirty #64 Hardware name: linux,dummy-virt (DT) Call trace: dumpbacktrace+0x17c/0x1a8 arch/arm64/kernel/stacktrace.c:317 showstack+0x2c/0x3c arch/arm64/kernel/stacktrace.c:324 dumpstack lib/dumpstack.c:93 [inline] dumpstacklvl+0x94/0xc0 lib/dumpstack.c:119 printreport+0x144/0x7a4 mm/kasan/report.c:377 kasanreport+0xcc/0x128 mm/kasan/report.c:601 _asanreportload8noabort+0x20/0x2c mm/kasan/reportgeneric.c:381 kvmputkvm+0x300/0xe68 virt/kvm/kvmmain.c:5769 kvmvmrelease+0x4c/0x60 virt/kvm/kvmmain.c:1409 _fput+0x198/0x71c fs/filetable.c:422 fput+0x20/0x30 fs/filetable.c:450 taskworkrun+0x1cc/0x23c kernel/taskwork.c:228 donotifyresume+0x144/0x1a0 include/linux/resumeusermode.h:50 el0svc+0x64/0x68 arch/arm64/kernel/entry-common.c:169 el0t64synchandler+0x90/0xfc arch/arm64/kernel/entry-common.c:730 el0t64sync+0x190/0x194 arch/arm64/kernel/entry.S:598
Upon closer inspection, it appears that we do not properly tear down the MMIO registration for a vCPU that fails creation late in the game, e.g. a vCPU w/ the same ID already exists in the VM.
It is important to consider the context of commit that introduced this bug by moving the unregistration out of _kvmvgicvcpudestroy(). That change correctly sought to avoid an srcu v. configlock inversion by breaking up the vCPU teardown into two parts, one guarded by the configlock.
Fix the use-after-free while avoiding lock inversion by adding a special-cased unregistration to _kvmvgicvcpudestroy(). This is safe because failed vCPUs are torn down outside of the config_lock.