On Thu, Oct 20, 2016 at 10:57 AM, Chris Wilson <chris at chris-wilson.co.uk> wrote: > On Thu, Oct 20, 2016 at 10:39:47AM -0400, Sean Paul wrote: >> On Thu, Oct 20, 2016 at 2:47 AM, Chris Wilson <chris at chris-wilson.co.uk> >> wrote: >> > I plan to usurp the short name of struct fence for a core kernel struct, >> > and so I need to rename the specialised fence/timeline for DMA >> > operations to make room. >> > >> > A consensus was reached in >> > https://lists.freedesktop.org/archives/dri-devel/2016-July/113083.html >> > that making clear this fence applies to DMA operations was a good thing. >> > Since then the patch has grown a bit as usage increases (and we wanted >> > to wait until after the sw_sync destage), so hopefully it remains a good >> > thing! >> > >> > I've pushed the patch to >> > >> > https://cgit.freedesktop.org/~ickle/linux-2.6/commit/?h=dma-fence&id=a4a27d87e4fccefb9f9d55f9c73ccab8f2ade3cc >> > >> > as I suspect it might be too large for the list. >> > >> > Highlights of the change: >> > >> > struct fence -> struct dma_fence (and all fence*() to dma_fence*()) >> > struct fence_array -> struct dma_fence_array >> > >> > 107 files changed, 1174 insertions(+), 1136 deletions(-) >> > >> > Mostly generated with the provided cocci spatch, but even with >> > --include-headers it seemed to miss some headers, and I never quite >> > worked out the substition commands for macros, so there is some grep and >> > replace. >> >> For the curious (like me), here's the diff between ickle's HEAD and >> what the cocci script outputs: >> >> http://hastebin.com/udizixoyed.go > > Hmm, looks like I had a few changes for dma-fence-array that didn't make > it back into the changelog. And still missed most of the trace_*(), as > only one was used outside. > > Most recently used spatch: > https://people.freedesktop.org/~ickle/fence.spatch
Thanks. I also notice you get more coverage if you break the struct renames out into their own rules (instead of the | operation). For instance, look at drivers/dma-buf/dma-buf.c dma_buf_poll_cb takes both fence and fence_cb. With your current spatch, it requires 2 passes to change both arguments. However if you split them into their own rules, it takes one pass: @@ @@ - struct fence + struct dma_fence @@ @@ - struct fence_ops + struct dma_fence_ops @@ @@ - struct fence_cb + struct dma_fence_cb @@ @@ - struct fence_array + struct dma_fence_array @@ @@ - enum fence_flag_bits + enum dma_fence_flag_bits @@ @@ Disclaimer: I am *not* a cocci expert Sean > -Chris > > -- > Chris Wilson, Intel Open Source Technology Centre
