RISC-V: KVM: Don't zero-out PMU snapshot area before freeing data

With the latest Linux-6.11-rc3, the below NULL pointer crash is observed
when SBI PMU snapshot is enabled for the guest and the guest is forcefully
powered-off.

  Unable to handle kernel NULL pointer dereference at virtual address 0000000000000508
  Oops [#1]
  Modules linked in: kvm
  CPU: 0 UID: 0 PID: 61 Comm: term-poll Not tainted 6.11.0-rc3-00018-g44d7178dd77a #3
  Hardware name: riscv-virtio,qemu (DT)
  epc : __kvm_write_guest_page+0x94/0xa6 [kvm]
   ra : __kvm_write_guest_page+0x54/0xa6 [kvm]
  epc : ffffffff01590e98 ra : ffffffff01590e58 sp : ffff8f80001f39b0
   gp : ffffffff81512a60 tp : ffffaf80024872c0 t0 : ffffaf800247e000
   t1 : 00000000000007e0 t2 : 0000000000000000 s0 : ffff8f80001f39f0
   s1 : 00007fff89ac4000 a0 : ffffffff015dd7e8 a1 : 0000000000000086
   a2 : 0000000000000000 a3 : ffffaf8000000000 a4 : ffffaf80024882c0
   a5 : 0000000000000000 a6 : ffffaf800328d780 a7 : 00000000000001cc
   s2 : ffffaf800197bd00 s3 : 00000000000828c4 s4 : ffffaf800248c000
   s5 : ffffaf800247d000 s6 : 0000000000001000 s7 : 0000000000001000
   s8 : 0000000000000000 s9 : 00007fff861fd500 s10: 0000000000000001
   s11: 0000000000800000 t3 : 00000000000004d3 t4 : 00000000000004d3
   t5 : ffffffff814126e0 t6 : ffffffff81412700
  status: 0000000200000120 badaddr: 0000000000000508 cause: 000000000000000d
  [<ffffffff01590e98>] __kvm_write_guest_page+0x94/0xa6 [kvm]
  [<ffffffff015943a6>] kvm_vcpu_write_guest+0x56/0x90 [kvm]
  [<ffffffff015a175c>] kvm_pmu_clear_snapshot_area+0x42/0x7e [kvm]
  [<ffffffff015a1972>] kvm_riscv_vcpu_pmu_deinit.part.0+0xe0/0x14e [kvm]
  [<ffffffff015a2ad0>] kvm_riscv_vcpu_pmu_deinit+0x1a/0x24 [kvm]
  [<ffffffff0159b344>] kvm_arch_vcpu_destroy+0x28/0x4c [kvm]
  [<ffffffff0158e420>] kvm_destroy_vcpus+0x5a/0xda [kvm]
  [<ffffffff0159930c>] kvm_arch_destroy_vm+0x14/0x28 [kvm]
  [<ffffffff01593260>] kvm_destroy_vm+0x168/0x2a0 [kvm]
  [<ffffffff015933d4>] kvm_put_kvm+0x3c/0x58 [kvm]
  [<ffffffff01593412>] kvm_vm_release+0x22/0x2e [kvm]

Clearly, the kvm_vcpu_write_guest() function is crashing because it is
being called from kvm_pmu_clear_snapshot_area() upon guest tear down.

To address the above issue, simplify the kvm_pmu_clear_snapshot_area() to
not zero-out PMU snapshot area from kvm_pmu_clear_snapshot_area() because
the guest is anyway being tore down.

The kvm_pmu_clear_snapshot_area() is also called when guest changes
PMU snapshot area of a VCPU but even in this case the previous PMU
snaphsot area must not be zeroed-out because the guest might have
reclaimed the pervious PMU snapshot area for some other purpose.

Fixes: c2f41ddbcdd7 ("RISC-V: KVM: Implement SBI PMU Snapshot feature")
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
Link: https://lore.kernel.org/r/20240815170907.2792229-1-apatel@ventanamicro.com
Signed-off-by: Anup Patel <anup@brainfault.org>

diff --git a/arch/riscv/kvm/vcpu_pmu.c b/arch/riscv/kvm/vcpu_pmu.c
index bcf41d6e0df0e36f68022a884a0204bb466e4199..2707a51b082ca7773b1797328bccd11795b65ec4 100644 (file)
--- a/arch/riscv/kvm/vcpu_pmu.c
+++ b/arch/riscv/kvm/vcpu_pmu.c
@@ -391,19 +391,9 @@ int kvm_riscv_vcpu_pmu_read_hpm(struct kvm_vcpu *vcpu, unsigned int csr_num,
 static void kvm_pmu_clear_snapshot_area(struct kvm_vcpu *vcpu)
 {
        struct kvm_pmu *kvpmu = vcpu_to_pmu(vcpu);
-       int snapshot_area_size = sizeof(struct riscv_pmu_snapshot_data);
 
-       if (kvpmu->sdata) {
-               if (kvpmu->snapshot_addr != INVALID_GPA) {
-                       memset(kvpmu->sdata, 0, snapshot_area_size);
-                       kvm_vcpu_write_guest(vcpu, kvpmu->snapshot_addr,
-                                            kvpmu->sdata, snapshot_area_size);
-               } else {
-                       pr_warn("snapshot address invalid\n");
-               }
-               kfree(kvpmu->sdata);
-               kvpmu->sdata = NULL;
-       }
+       kfree(kvpmu->sdata);
+       kvpmu->sdata = NULL;
        kvpmu->snapshot_addr = INVALID_GPA;
 }