Messages in this thread | | | Subject | Re: 2.6.18-rc5-mm1: drivers/infiniband/hw/amso1100/c2.c compile error | From | Roland Dreier <> | Date | Fri, 01 Sep 2006 13:54:04 -0700 |
| |
Russell> Sure, if you want a _non-atomic_ 64-bit write then that's Russell> possible, but many 32-bit architectures can't do a 64-bit Russell> atomic IO write and that isn't something they can "fix".
I agree completely. And going one step further: if an architecture cannot implement a 64-bit write atomically, then the precise serialization that is required is device-specific knowledge that belongs in the device driver.
(For example, in the mthca case, the only serialization required is that no writes go to the same page of MMIO space between the two 32-bit halves of the 64-bit write)
- R.
-- VGER BF report: H 0 - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |