Re: [PATCH 0/2] hw/nvme: Fix CVE-2021-3929 (DMA re-entrancy exploitation)

[Klaus Jensen <its@irrelevant.dk>]

[-- Attachment #1: Type: text/plain, Size: 2640 bytes --]

On Dec 16 20:55, Klaus Jensen wrote:
> On Dec 16 20:13, Klaus Jensen wrote:
> > On Dec 16 18:55, Philippe Mathieu-Daudé wrote:
> > > Now that the DMA API allow passing MemTxAttrs argument and
> > > returning MemTxResult (with MEMTX_BUS_ERROR in particular),
> > > we can restrict the NVMe controller to memories (prohibitting
> > > accesses by the DMA engine to devices) and block yet another
> > > DMA re-entrancy attack.
> > > 
> > > I'll will try to get a reproducer (and authorization to commit
> > > it as qtest) from the reporter.
> > > 
> > > Based-on: <20211216123558.799425-1-philmd@redhat.com>
> > > "hw: Have DMA API take MemTxAttrs arg & propagate MemTxResult (part 2)"
> > > https://lore.kernel.org/qemu-devel/20211216123558.799425-1-philmd@redhat.com/
> > > 
> > > Philippe Mathieu-Daudé (2):
> > >   hw/nvme/ctrl: Do not ignore DMA access errors
> > >   hw/nvme/ctrl: Prohibit DMA accesses to devices (CVE-2021-3929)
> > > 
> > >  hw/nvme/ctrl.c | 9 +++++----
> > >  1 file changed, 5 insertions(+), 4 deletions(-)
> > > 
> > 
> > LGTM.
> > 
> > Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
> 
> Ugh. Jumped the gun here.
> 
> This all looked fine, but since this prohibits DMA to other devices it
> breaks DMA'ing to a controller memory buffer on another device, which is
> a used feature of some setups.
> 
> I think we need to fix this like e1000 did?

Something like this maybe?

This fixes CVE-2021-3929 "locally" by denying DMA to the iomem of the
device itself. This still allows DMA to MMIO regions of other devices
(e.g. doing P2P DMA to the controller memory buffer of another NVMe
device).


diff --git a/hw/nvme/ctrl.c b/hw/nvme/ctrl.c
index 5f573c417b3d..9a79f6728867 100644
--- a/hw/nvme/ctrl.c
+++ b/hw/nvme/ctrl.c
@@ -357,6 +357,16 @@ static inline void *nvme_addr_to_pmr(NvmeCtrl *n, hwaddr addr)
     return memory_region_get_ram_ptr(&n->pmr.dev->mr) + (addr - n->pmr.cba);
 }

+static inline bool nvme_addr_is_iomem(NvmeCtrl *n, hwaddr addr)
+{
+    hwaddr hi, lo;
+
+    lo = n->bar0.addr;
+    hi = lo + int128_get64(n->bar0.size);
+
+    return addr >= lo && addr < hi;
+}
+
 static int nvme_addr_read(NvmeCtrl *n, hwaddr addr, void *buf, int size)
 {
     hwaddr hi = addr + size - 1;
@@ -614,6 +624,10 @@ static uint16_t nvme_map_addr(NvmeCtrl *n, NvmeSg *sg, hwaddr addr, size_t len)

     trace_pci_nvme_map_addr(addr, len);

+    if (nvme_addr_is_iomem(n, addr)) {
+        return NVME_DATA_TRAS_ERROR;
+    }
+
     if (nvme_addr_is_cmb(n, addr)) {
         cmb = true;
     } else if (nvme_addr_is_pmr(n, addr)) {

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]