From mboxrd@z Thu Jan 1 00:00:00 1970 From: hch@lst.de (Christoph Hellwig) Date: Sun, 22 Nov 2015 09:49:07 +0100 Subject: NVMe mega patchbomb for Linux 4.5-rc (resend) In-Reply-To: References: <1448090427-18749-1-git-send-email-hch@lst.de> Message-ID: <20151122084907.GA11401@lst.de> On Sun, Nov 22, 2015@08:14:38AM +0530, Madhu (Macaque Labs) wrote: > With respect to the split of the driver into PCIe and PCIe agnostic parts, > would you deem the split complete ? Or is there work still to be done. I'd say about 95% done relative to my initial target. There's some trivial code moves that I have queued up after this series, and then we need to look into a proper controller state machine. But as always in Linux we have moving targets, so things could change in the future. > A good exercise would be to graft the upcoming SRIO driver and see > how that coupling works. SRIO as in Serial Rapid I/O? How does that driver differ from the PCIe one? I'd expect it to still use MMIO access, the DMA API and a similar queueing model as PCIe, in which case duplicating the whole PCIe driver would seem wasteful. Maybe you can post what you have currenly? > As part of our proposed lightstor standard (NVMe + extensions), And the specs for your SRIO transport as well as your extension. Did you bring either of them up in the NVMe technical working group?