[libvirt] [PATCH 04/12] python: avoid unlikely sign extension bug

[Matthias Bolte]

2011/6/6 Eric Blake <eblake at redhat.com>:
> Detected by Coverity.  cpumap was allocated with a value of
> (unsigned short)*(int), which is an int computation, and then
> promotes to size_t.  On a 64-bit platform, this fails if bit
> 32 of the product is set (because of sign extension giving
> a HUGE value to malloc), even though a naive programmer would
> assume that since the first value is unsigned, the product
> is also unsigned and at most 4GB would be allocated.
>
> Won't bite in practice (the product should never be that large),
> but worth using the right types to begin with, so that we are
> now computing (unsigned short)*(size_t).
>
> * python/libvirt-override.c (libvirt_virDomainGetVcpus): Use
> correct type.
> ---
>  python/libvirt-override.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/python/libvirt-override.c b/python/libvirt-override.c
> index 763df00..974decb 100644
> --- a/python/libvirt-override.c
> +++ b/python/libvirt-override.c
> @@ -414,7 +414,7 @@ libvirt_virDomainGetVcpus(PyObject *self ATTRIBUTE_UNUSED,
>     virDomainInfo dominfo;
>     virVcpuInfoPtr cpuinfo = NULL;
>     unsigned char *cpumap = NULL;
> -    int cpumaplen, i;
> +    size_t cpumaplen, i;
>     int i_retval;
>
>     if (!PyArg_ParseTuple(args, (char *)"O:virDomainGetVcpus",

ACK.

Matthias