https://bugzilla.kernel.org/show_bug.cgi?id=215740 --- Comment #31 from Alan Stern (stern@xxxxxxxxxxxxxxxxxxx) --- I don't know why my local test didn't reproduce the problem -- and I haven't tried to find out what's happening. My best guess is that the allocations of hub->buffer and hub->status end up (by random chance) being non-adjacent, so they aren't in the same cache line. But I haven't even run the test patches attached to this bug report. Regardless, changing the allocation size to 64 merely hides the problem. The real problem is that the DMA debugging code is wrong and needs to be fixed. It should not complain about two different mappings of the same cache line on x86, unless the mappings actually overlap. -- You may reply to this email to add a comment. You are receiving this mail because: You are watching the assignee of the bug.
