On 03/11/2021 18:07, Daniel P. Berrangé wrote:
> On Mon, Nov 01, 2021 at 10:21:35AM +0000, Dov Murik wrote:
>> Commit cff03145ed3c ("sev/i386: Introduce sev_add_kernel_loader_hashes
>> for measured linux boot", 2021-09-30) introduced measured direct boot
>> with -kernel, using an OVMF-designated hashes table which QEMU fills.
>>
>> However, no checks are performed on the validity of the hashes area
>> designated by OVMF. Specifically, if OVMF publishes the
>> SEV_HASH_TABLE_RV_GUID entry but it is filled with zeroes, this will
>> cause QEMU to write the hashes entries over the first page of the
>> guest's memory (GPA 0).
>>
>> Add validity checks to the published area. If the hashes table area's
>> base address is zero, or its size is too small to fit the aligned hashes
>> table, warn and skip the hashes entries addition. In such case, the
>> following warning will be displayed:
>>
>> qemu-system-x86_64: warning: SEV: OVMF's hashes table area is invalid
>> (base=0x0 size=0x0)
>>
>> Signed-off-by: Dov Murik <dovmu...@linux.ibm.com>
>> Reported-by: Brijesh Singh <brijesh.si...@amd.com>
>> ---
>> target/i386/sev.c | 8 ++++++--
>> 1 file changed, 6 insertions(+), 2 deletions(-)
>>
>> diff --git a/target/i386/sev.c b/target/i386/sev.c
>> index 682b8ccf6c..a20ddb545e 100644
>> --- a/target/i386/sev.c
>> +++ b/target/i386/sev.c
>> @@ -1201,13 +1201,18 @@ bool
>> sev_add_kernel_loader_hashes(SevKernelLoaderContext *ctx, Error **errp)
>> uint8_t kernel_hash[HASH_SIZE];
>> uint8_t *hashp;
>> size_t hash_len = HASH_SIZE;
>> - int aligned_len;
>> + int aligned_len = ROUND_UP(sizeof(SevHashTable), 16);
>>
>> if (!pc_system_ovmf_table_find(SEV_HASH_TABLE_RV_GUID, &data, NULL)) {
>> warn_report("SEV: kernel specified but OVMF has no hash table
>> guid");
>> return false;
>> }
>> area = (SevHashTableDescriptor *)data;
>> + if (!area->base || area->size < aligned_len) {
>> + warn_report("SEV: OVMF's hashes table area is invalid (base=0x%x
>> size=0x%x)",
>> + area->base, area->size);
>> + return false;
>> + }
>
> I think warn_report is likely a bad idea.
>
> If someone's use case is relying on the hashs being populated, then
> we need to be able to error_report and exit, not carry on with a
> known broken setup.
>
Thanks you. As I wrote elsewhere, I will put the whole thing under a
flag of sev-guest. If the flag is ON then I will error_report and exit
if the GUID is missing or if the address/size is wrong, as you suggest.
-Dov
