If we can't access PCI extended config space, e.g., if CONFIG_PCI_MMCONFIG=n, the IOAT driver complains: [ 0.696357] ioatdma: Intel(R) QuickData Technology Driver 4.00 [ 0.696487] ioatdma 0000:00:04.0: channel error register unreachable [ 0.696546] ioatdma 0000:00:04.0: channel enumeration error [ 0.696604] ioatdma 0000:00:04.0: Intel(R) I/OAT DMA Engine init failed But this doesn't give the user any clue about how to fix the problem. It should emit a more useful diagnostic. Dan Williams has suggested that the IOAT driver doesn't even really *need* extended config space. If that's true, it would be even better to just remove the dependency on it.
Two years later, kernel 4.0.4 and the problem is still here. Is this a correct place for such reports?
Many maintainers don't pay attention to the bugzilla. I would send email to the folks involved with that driver (from scripts/get_mainainer.pl): Dan Williams <dan.j.williams@intel.com> (reviewer,commit_signer:2/7=29%,authored:1/7=14%,removed_lines:1/14=7%) Vinod Koul <vinod.koul@intel.com> (maintainer:DMA GENERIC OFFLO...,commit_signer:6/7=86%) Dave Jiang <dave.jiang@intel.com> (commit_signer:4/7=57%,authored:3/7=43%,added_lines:20/55=36%) dmaengine@vger.kernel.org (open list:DMA GENERIC OFFLO...) linux-kernel@vger.kernel.org (open list)