On Fri, 2018-03-16 at 12:47 +0800, Coly Li wrote: > The code still does not work with your patch on big endian machine (e.g. > s390x). When I fix the bcache s390x bug, I did something like this patch > in my first try. It turns out more things need to be fixed, > 1, some cache_sb members only swapped to cpu endianness when read in, > not swapped to little endian when write out. For example data_offset, > nbuckets, nr_in_set, nr_this_dev ... > 2, checksum is calculated in cpu endiannness, not explicitly in little > endian. > > Also user space tools need to update to support the endianness issue. > > How about leave the endianness problem to me? I will pick code piece > from your patch and set 'From: Bart Van Assche <bart.vanassche@xxxxxxx>' > on the patch(s), and continue my fix for bcache on s390x. Hello Coly, That sounds fine to me. Please let me know if you would like me to repost this series and/or address the feedback that I received for this patch series. Thanks, Bart.
