Re: [PATCH] tpm_crb: mark command buffer as dirty on request completion

[Stefan Berger]

On 4/11/22 10:47, Anthony PERARD via wrote:
> From: Anthony PERARD <anthony.perard@citrix.com>
>
> At the moment, there doesn't seems to be any way to know that QEMU
> made modification to the command buffer. This is potentially an issue
> on Xen while migrating a guest, as modification to the buffer after
> the migration as started could be ignored and not transfered to the
> destination.
>
> Mark the memory region of the command buffer as dirty once a request
> is completed.
>
> Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
Reviewed-by: Stefan Berger <stefanb@linux.ibm.com>



> ---
>
> I have only read code to find out whether the tpm-crb device was fine
> with regards to migration, and I don't think there's anything that
> could mark the memory region as dirty once a request is completed.
>
> There is one call to memory_region_get_ram_ptr(), but nothing seems to
> be done with the pointer is regards to ram migration. Am I wrong?
>
> Thanks.
> ---
>   hw/tpm/tpm_crb.c | 1 +
>   1 file changed, 1 insertion(+)
>
> diff --git a/hw/tpm/tpm_crb.c b/hw/tpm/tpm_crb.c
> index aa9c00aad3..67db594c48 100644
> --- a/hw/tpm/tpm_crb.c
> +++ b/hw/tpm/tpm_crb.c
> @@ -197,6 +197,7 @@ static void tpm_crb_request_completed(TPMIf *ti, int ret)
>           ARRAY_FIELD_DP32(s->regs, CRB_CTRL_STS,
>                            tpmSts, 1); /* fatal error */
>       }
> +    memory_region_set_dirty(&s->cmdmem, 0, CRB_CTRL_CMD_SIZE);
>   }
>
>   static enum TPMVersion tpm_crb_get_version(TPMIf *ti)