| Message ID | 20210513223203.5542-7-logang@deltatee.com (mailing list archive) |
|---|---|
| State | Changes Requested |
| Delegated to: | Bjorn Helgaas |
| Headers | show |
| Series | Add new DMA mapping operation for P2PDMA | expand |
On Thu, May 13, 2021 at 04:31:47PM -0600, Logan Gunthorpe wrote: > Attempt to find the mapping type for P2PDMA pages on the first > DMA map attempt if it has not been done ahead of time. > > Previously, the mapping type was expected to be calculated ahead of > time, but if pages are to come from userspace then there's no > way to ensure the path was checked ahead of time. > > With this change it's no longer invalid to call pci_p2pdma_map_sg() > before the mapping type is calculated so drop the WARN_ON when that > is the case. Why?
On 2021-05-14 7:49 a.m., Christoph Hellwig wrote: > On Thu, May 13, 2021 at 04:31:47PM -0600, Logan Gunthorpe wrote: >> Attempt to find the mapping type for P2PDMA pages on the first >> DMA map attempt if it has not been done ahead of time. >> >> Previously, the mapping type was expected to be calculated ahead of >> time, but if pages are to come from userspace then there's no >> way to ensure the path was checked ahead of time. >> >> With this change it's no longer invalid to call pci_p2pdma_map_sg() >> before the mapping type is calculated so drop the WARN_ON when that >> is the case. > > Why? Before this change, the if the mapping type wasn't already calculated pci_p2pdma_map_sg() would just fail. This was fine for NVMe-of as it always called pci_p2pdma_distance() ahead of time which calculated the mapping type, stored it in the xarray and did not proceed if the two devices could not talk to each other. This patch makes it so if the mapping type is not already calculated at dma map time, it does the calculation. This means the dma map operation can fail if the two devices aren't able to talk to each other. Logan
diff --git a/drivers/pci/p2pdma.c b/drivers/pci/p2pdma.c index 0e0b2218eacd..4034ffa0eb06 100644 --- a/drivers/pci/p2pdma.c +++ b/drivers/pci/p2pdma.c @@ -847,11 +847,17 @@ EXPORT_SYMBOL_GPL(pci_p2pmem_publish); static enum pci_p2pdma_map_type pci_p2pdma_map_type(struct pci_dev *provider, struct pci_dev *client) { + enum pci_p2pdma_map_type ret; + if (!provider->p2pdma) return PCI_P2PDMA_MAP_NOT_SUPPORTED; - return xa_to_value(xa_load(&provider->p2pdma->map_types, - map_types_idx(client))); + ret = xa_to_value(xa_load(&provider->p2pdma->map_types, + map_types_idx(client))); + if (ret == PCI_P2PDMA_MAP_UNKNOWN) + return calc_map_type_and_dist_warn(provider, client, NULL); + + return ret; } static int __pci_p2pdma_map_sg(struct pci_p2pdma_pagemap *p2p_pgmap, @@ -899,7 +905,6 @@ int pci_p2pdma_map_sg_attrs(struct device *dev, struct scatterlist *sg, case PCI_P2PDMA_MAP_BUS_ADDR: return __pci_p2pdma_map_sg(p2p_pgmap, dev, sg, nents); default: - WARN_ON_ONCE(1); return 0; } }
Attempt to find the mapping type for P2PDMA pages on the first DMA map attempt if it has not been done ahead of time. Previously, the mapping type was expected to be calculated ahead of time, but if pages are to come from userspace then there's no way to ensure the path was checked ahead of time. With this change it's no longer invalid to call pci_p2pdma_map_sg() before the mapping type is calculated so drop the WARN_ON when that is the case. Signed-off-by: Logan Gunthorpe <logang@deltatee.com> --- drivers/pci/p2pdma.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-)