Re: [PULL 06/35] spapr: Fail CAS if option vector table cannot be parsed

[Peter Maydell]

On Mon, 3 Feb 2020 at 06:11, David Gibson <address@hidden> wrote:
>
> From: Greg Kurz <address@hidden>
>
> Most of the option vector helpers have assertions to check their
> arguments aren't null. The guest can provide an arbitrary address
> for the CAS structure that would result in such null arguments.
> Fail CAS with H_PARAMETER and print a warning instead of aborting
> QEMU.
>
> Signed-off-by: Greg Kurz <address@hidden>
> Reviewed-by: Philippe Mathieu-Daudé <address@hidden>
> Message-Id: <address@hidden>
> Signed-off-by: David Gibson <address@hidden>
> ---
>  hw/ppc/spapr_hcall.c | 8 ++++++++
>  1 file changed, 8 insertions(+)

Hi; Coverity points out that this change introduces a
memory leak (CID 1421924):

>
> diff --git a/hw/ppc/spapr_hcall.c b/hw/ppc/spapr_hcall.c
> index f1799b1b70..ffb14641f9 100644
> --- a/hw/ppc/spapr_hcall.c
> +++ b/hw/ppc/spapr_hcall.c
> @@ -1703,7 +1703,15 @@ static target_ulong 
> h_client_architecture_support(PowerPCCPU *cpu,
>      ov_table = addr;
>
>      ov1_guest = spapr_ovec_parse_vector(ov_table, 1);

spapr_ovec_parse_vector() allocates memory...

> +    if (!ov1_guest) {
> +        warn_report("guest didn't provide option vector 1");
> +        return H_PARAMETER;
> +    }
>      ov5_guest = spapr_ovec_parse_vector(ov_table, 5);
> +    if (!ov5_guest) {
> +        warn_report("guest didn't provide option vector 5");
> +        return H_PARAMETER;

...but if we take this early exit code path it is never freed
(via spapr_ovec_cleanup()).

> +    }
>      if (spapr_ovec_test(ov5_guest, OV5_MMU_BOTH)) {
>          error_report("guest requested hash and radix MMU, which is 
> invalid.");
>          exit(EXIT_FAILURE);

All the other error paths in the function either precede
allocation of the vectors or just call exit() rather than
returning, so this is the only leak.

thanks
-- PMM