[PATCH 5/5] drm/i915/gvt: Assign NULL to the pointer after memory free.

[Colin Xu]

On 2019-04-08 13:55, Aleksei Gimbitskii wrote:
> The klocwork static code analyzer complains about using pointer after
> being freed, because further we pass it to the gvt_vgpu_err() function.
> Assign pointer to be NULL intentionaly, to meet requirements of the code
> analyzer.
>
> This patch fixed the issue #648 reported as error by klocwork.
>
> Signed-off-by: Aleksei Gimbitskii <aleksei.gimbitskii at intel.com>
> Cc: Zhenyu Wang <zhenyuw at linux.intel.com>
> Cc: Zhi Wang <zhi.a.wang at intel.com>
> ---
>   drivers/gpu/drm/i915/gvt/gtt.c | 1 +
>   1 file changed, 1 insertion(+)
>
> diff --git a/drivers/gpu/drm/i915/gvt/gtt.c b/drivers/gpu/drm/i915/gvt/gtt.c
> index 7dc09b389e59..3cf3984aae41 100644
> --- a/drivers/gpu/drm/i915/gvt/gtt.c
> +++ b/drivers/gpu/drm/i915/gvt/gtt.c
> @@ -1098,6 +1098,7 @@ static struct intel_vgpu_ppgtt_spt *ppgtt_populate_spt_by_guest_entry(
>   
>   err_free_spt:
>   	ppgtt_free_spt(spt);
> +	spt = NULL;

Logically it's ok to explicit set spt to NULL. However from debug purpose label "err_free_spt"
is jumped to due to failure of intel_vgpu_enable_page_track or ppgtt_populate_spt, not same as
label "err". It's better to make dbgmsg more clear about which spt fails in such case so that
the lifecycle of a spt in dbgmsg is more clear for different situations.

>   err:
>   	gvt_vgpu_err("fail: shadow page %p guest entry 0x%llx type %d\n",
>   		     spt, we->val64, we->type);

-- 
Best Regards,
Colin Xu