Hi Gerd, I found this problem once I sent the patches :(
I checked the uapi definitions and found it is usually called pad to do the aligning. So I changed the 'resv' to 'pad' in the patch but forgot to update it in the last patch and did not test after the "small" change. Next time I will test even when the change is very small. I will change this in the next version. Sorry for the mistake. >-----Original Message----- >From: Gerd Hoffmann [mailto:[email protected]] >Sent: Wednesday, May 31, 2017 4:59 PM >To: Chen, Xiaoguang <[email protected]>; >[email protected]; [email protected]; intel- >[email protected]; [email protected]; >[email protected]; Lv, Zhiyuan <[email protected]>; intel-gvt- >[email protected]; Wang, Zhi A <[email protected]>; Tian, Kevin ><[email protected]> >Subject: Re: [PATCH v6 0/6] drm/i915/gvt: Dma-buf support for GVT-g > >On Wed, 2017-05-31 at 02:29 +0000, Chen, Xiaoguang wrote: >> Hi Gerd, >> >> It is based on 4.12.0-rc1 > >Applies, good. >But then fails to build: > >error: ‘struct vfio_vgpu_dmabuf_info’ has no member named ‘resv’ > >gvt/kvmgt.c:611:11: note: in expansion of macro ‘offsetofend’ > minsz = offsetofend(struct vfio_vgpu_dmabuf_info, resv); > >/me wonders how this was tested ... > >cheers, > Gerd
