On 27/10/17 17:42, Philippe Mathieu-Daudé wrote: > Hi Mark, > > On 10/25/2017 12:59 PM, Mark Cave-Ayland wrote: >> This patchset aims to tidy-up the sparc32_dma code by improving the >> modelling of the espdma/ledma devices using both QOM and the memory >> API which didn't exist when the code was first written. >> >> The result is that it is now possible to remove both the iommu_opaque >> and is_ledma workarounds from the code, and the code for wiring up >> the espdma/ledma and respective devices is also a lot more readable. >> >> Signed-off-by: Mark Cave-Ayland <mark.cave-ayl...@ilande.co.uk> > > The whole series: > > Tested-by: Philippe Mathieu-Daudé <f4...@amsat.org> > > If you don't accept my comments (or don't have time) about keeping > "hw/sparc/sparc32_dma.h" generic and moving network/scsi parts in > "hw/sparc/sun4m.h" you can still add to your series: > > Acked-by: Philippe Mathieu-Daudé <f4...@amsat.org>
Thanks for the review, I've added your R-B tags to the individual patches. Note that while potentially I could move the network/scsi parts to hw/sparc/sun4m.h I feel that it's a slightly better match for the SPARC32 DMA container device to remain in sparc32_dma.c. So for these patches I've just added your A-B tag. > Also while testing your series on a Debian image, I noted your series > results faster, I timed: > > master: 104s > your series: 85s (>20% faster!) Really? Is that for just this patchset or also with the v2 IOMMU patchset applied on top? I can't immediately see how moving the logic into sparc32_dma.c could make a difference here... ATB, Mark.