On Sat, Oct 09, 2021 at 06:49:34PM +0800, Dongdong Liu wrote: > PCIe spec 5.0 r1.0 section 2.2.6.2 says: > > If an Endpoint supports sending Requests to other Endpoints (as > opposed to host memory), the Endpoint must not send 10-Bit Tag > Requests to another given Endpoint unless an implementation-specific > mechanism determines that the Endpoint supports 10-Bit Tag Completer > capability. > > Add a 10bit_tag sysfs file, write 0 to disable 10-Bit Tag Requester > when the driver does not bind the device. The typical use case is for > p2pdma when the peer device does not support 10-Bit Tag Completer. > Write 1 to enable 10-Bit Tag Requester when RC supports 10-Bit Tag > Completer capability. The typical use case is for host memory targeted > by DMA Requests. The 10bit_tag file content indicate current status of > 10-Bit Tag Requester Enable. Don't we have a hole here? We're adding knobs to control 10-Bit Tag usage, but don't we have basically the same issues with Extended (8-bit) Tags? I wonder if we should be adding a more general "tags" file that can manage both 8-bit and 10-bit tag usage. > +static struct device_attribute dev_attr_10bit_tag = __ATTR(10bit_tag, 0644, > + pci_10bit_tag_show, > + pci_10bit_tag_store); I think this should use DEVICE_ATTR(). Or even better, if the name doesn't start with a digit, DEVICE_ATTR_RW().
