On 10/12/2017 02:51 PM, Sitsofe Wheeler wrote: > On 12 October 2017 at 15:36, Jens Axboe <axboe@xxxxxxxxx> wrote: >> >> Just go through them one-by-one. I think some of these were likely >> introduced with the nsec conversion, which I guess isn't too >> surprising... > > With all the latest patches applied these are the only warnings which > look faintly legitimate: > > io_u.c:1272:28: runtime error: unsigned integer overflow: 0 - 1 cannot > be represented in type 'unsigned int' > > Might be harmless because we goto out just after td->file_service_left > has wrapped around. Might be worth it to clean up, I'll do that. > ioengines.c:270:22: runtime error: unsigned integer overflow: 59 - 63 > cannot be represented in type 'unsigned int' > > It does look like td->io_u_in_flight is over decremented but it's not > clear how I/Os that weren't already in flight somehow completed... Hmm that is odd. I'll look into that. > stat.c:1529:20: runtime error: unsigned integer overflow: 3370715872 + > 1125639672 cannot be represented in type 'unsigned int' > stat.c:1531:23: runtime error: unsigned integer overflow: 3370715859 + > 1125639668 cannot be represented in type 'unsigned int' > stat.c:1533:25: runtime error: unsigned integer overflow: 3370715862 + > 1125639669 cannot be represented in type 'unsigned int' > > The last three are all down to io_u_* structure members being 32 bit. > Would we want to make them 64 bit and bump the protocol version? Yes, I think we should make them 64-bit and just bump the protocol version. -- Jens Axboe -- To unsubscribe from this list: send the line "unsubscribe fio" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
