On 10/05/2016 04:20 PM, Peter Maydell wrote: > On 5 October 2016 at 07:00, Cédric Le Goater <c...@kaod.org> wrote: >> On 10/05/2016 03:53 PM, Peter Maydell wrote: >>> Which tswap? Last time I worked through the stack of >>> what happens I thought that we had the right set of >>> swaps in the right places. >> >> The one I am talking about are under qtest_process_command(), >> see below. > > Those are correct and required, and they do not change > the overall behaviour of the system depending on the host > endianness. (They convert 32-bit values to "bag of > bytes in guest order" which is what the cpu_physical_memory_* > functions want.) > > https://lists.gnu.org/archive/html/qemu-arm/2016-07/msg00037.html > is the explanation from last time around, I think. > > If your test is giving different answers on different > host endiannesses, it is presumably because it is > incorrectly issuing different readl/etc commands > in the two cases.
OK. I think my brain is starting to see things from the right angle. Let's try that : my test is simulating a Little Endian CPU which is writing a Big Endian value, so it *always* needs to bswap() that value. cpu_to_be32 is incorrect as it implies the endianness of the host which can be wrong (when running BE). The qtest layer is a just fixing layer. Thanks, C.
