On Thu, Sep 13 2018 at 8:45pm -0400, Randy Dunlap <rdunlap@xxxxxxxxxxxxx> wrote: > On 9/13/18 11:51 AM, Mike Snitzer wrote: > > On Thu, Sep 13 2018 at 1:28pm -0400, > > Randy Dunlap <rdunlap@xxxxxxxxxxxxx> wrote: > > > >> On 9/12/18 10:27 PM, Stephen Rothwell wrote: > >>> Hi all, > >>> > >>> News: there will be no linux-next releases on Friday or Monday. > >>> > >>> Changes since 20180912: > >>> > >> > >> on i386: > >> > >> ERROR: "__udivdi3" [drivers/md/dm-thin-pool.ko] undefined! > > > > Well, as I pointed out in reply that nobody will see to the buildbots: > > > > There is something off in this report... I cannot reproduce. It is > > almost like the warning was generated when building an older version of > > this change, but then reported against the latest commit. > > > > I switched to sector_div() specifically because of the undefined > > __udivdi3 error. > > > > So I'm ignoring this given I cannot reproduce when using 'make ARCH=i386' > > > gcc --version > gcc (SUSE Linux) 4.8.5 > > # CONFIG_LBDAF is not set > > Perhaps you could try to reproduce it with the attached randconfig file. Ah, yeap.. sector_div() is only viable for use with sector_t. dm_block_t is typedef'd to uint64_t -- so that explains it. Need to use div_u64() instead. Thanks, Mike
