On Thu, Jan 6, 2022 at 4:49 PM Christophe JAILLET <christophe.jaillet@xxxxxxxxxx> wrote: > > In [1], Christoph Hellwig has proposed to remove the wrappers in > include/linux/pci-dma-compat.h. > > Some reasons why this API should be removed have been given by Julia > Lawall in [2]. > > A coccinelle script has been used to perform the needed transformation. > It can be found in [3]. > > > It has been hand modified to use 'dma_set_mask_and_coherent()' instead of > 'pci_set_dma_mask()/pci_set_consistent_dma_mask()' when applicable. > This is less verbose. > > The explicit 'ret = -EIO;' has been removed because > 'dma_set_mask_and_coherent()' returns 0 or -EIO, so its return code can be > used directly. > > > [1]: https://lore.kernel.org/kernel-janitors/20200421081257.GA131897@xxxxxxxxxxxxx/ > [2]: https://lore.kernel.org/kernel-janitors/alpine.DEB.2.22.394.2007120902170.2424@hadrien/ > [3]: https://lore.kernel.org/kernel-janitors/20200716192821.321233-1-christophe.jaillet@xxxxxxxxxx/ > > Signed-off-by: Christophe JAILLET <christophe.jaillet@xxxxxxxxxx> > Reviewed-by: Xu Yilun <yilun.xu@xxxxxxxxx> This is a correct conversion of the driver, but I'd prefer to keep this separate from the pci-dma-compat series. > + ret = dma_set_mask_and_coherent(&pcidev->dev, DMA_BIT_MASK(64)); > + if (ret) > + ret = dma_set_mask_and_coherent(&pcidev->dev, DMA_BIT_MASK(32)); > + if (ret) { > dev_err(&pcidev->dev, "No suitable DMA support available.\n"); > goto disable_error_report_exit; The code looks a bit suspicous, both the old and the new version. If the device ends up on a bus that can only do 32-bit DMA, shouldn't it remember this and make sure it only does GFP_DMA32 allocations for buffers that are passed to the device? Arnd
