Re: [PATCH V4 06/16] xfs: Promote xfs_extnum_t and xfs_aextnum_t to 64 and 32-bits respectively

[Chandan Babu R <chandan.babu@oracle.com>]

On 05 Jan 2022 at 22:51, Darrick J. Wong wrote:
> On Wed, Jan 05, 2022 at 07:44:23PM +0530, Chandan Babu R wrote:
>> On 05 Jan 2022 at 05:24, Darrick J. Wong wrote:
>> > On Wed, Dec 15, 2021 at 02:49:48PM +0530, Chandan Babu R wrote:
>> >> On 14 Dec 2021 at 20:45, kernel test robot wrote:
>> >> > Hi Chandan,
>> >> >
>> >> > Thank you for the patch! Yet something to improve:
>> >> >
>> >> > [auto build test ERROR on xfs-linux/for-next]
>> >> > [also build test ERROR on v5.16-rc5]
>> >> > [If your patch is applied to the wrong git tree, kindly drop us a note.
>> >> > And when submitting patch, we suggest to use '--base' as documented in
>> >> > https://git-scm.com/docs/git-format-patch]
>> >> >
>> >> > url:    https://github.com/0day-ci/linux/commits/Chandan-Babu-R/xfs-Extend-per-inode-extent-counters/20211214-164920
>> >> > base:   https://git.kernel.org/pub/scm/fs/xfs/xfs-linux.git for-next
>> >> > config: microblaze-randconfig-r016-20211214 (https://download.01.org/0day-ci/archive/20211214/202112142335.O3Nu0vQI-lkp@intel.com/config)
>> >> > compiler: microblaze-linux-gcc (GCC) 11.2.0
>> >> > reproduce (this is a W=1 build):
>> >> >         wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
>> >> >         chmod +x ~/bin/make.cross
>> >> >         # https://github.com/0day-ci/linux/commit/db28da144803c4262c0d8622d736a7d20952ef6b
>> >> >         git remote add linux-review https://github.com/0day-ci/linux
>> >> >         git fetch --no-tags linux-review Chandan-Babu-R/xfs-Extend-per-inode-extent-counters/20211214-164920
>> >> >         git checkout db28da144803c4262c0d8622d736a7d20952ef6b
>> >> >         # save the config file to linux build tree
>> >> >         mkdir build_dir
>> >> >         COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-11.2.0 make.cross O=build_dir ARCH=microblaze SHELL=/bin/bash
>> >> >
>> >> > If you fix the issue, kindly add following tag as appropriate
>> >> > Reported-by: kernel test robot <lkp@intel.com>
>> >> >
>> >> > All errors (new ones prefixed by >>):
>> >> >
>> >> >    microblaze-linux-ld: fs/xfs/libxfs/xfs_bmap.o: in function `xfs_bmap_compute_maxlevels':
>> >> >>> (.text+0x10cc0): undefined reference to `__udivdi3'
>> >> >
>> >> 
>> >> The fix for the compilation error on 32-bit systems involved invoking do_div()
>> >> instead of using the regular division operator. I will include the fix in the
>> >> next version of the patchset.
>> >
>> > So, uh, how did you resolve this in the end?
>> >
>> > 	maxblocks = roundup_64(maxleafents, minleafrecs);
>> >
>> > and
>> >
>> > 	maxblocks = roundup_64(maxblocks, minnodrecs);
>> >
>> > ?
>> 
>> I had made the following changes,
>> 
>> 	maxblocks = maxleafents + minleafrecs - 1;
>> 	do_div(maxblocks, minleafrecs);
>> 
>> and
>> 	maxblocks += minnoderecs - 1;
>> 	do_div(maxblocks, minnoderecs);
>> 
>> roundup_64() would cause maxleafents to have a value >= its previous value
>> right?

Sorry, I meant to say "The result of roundup_64(maxleafents, minleafrecs) will
be >= than maxleafents".

The original statement was,
maxblocks = (maxleafents + minleafrecs - 1) / minleafrecs;
i.e. maxblocks would contain the number of leaf blocks required to hold
maxleafents number of records.

With maxleafents = 2^48, minleafrecs = minnoderecs = 125,
"maxblocks = (maxleafents + minleafrecs - 1) / minleafrecs" would result in,
maxblocks = (2^48 + 125 - 1) / 125
          = ~2^41

>
> roundup_64 doesn't alter its parameters, if I'm not mistaken:
>
> static inline uint64_t roundup_64(uint64_t x, uint32_t y)
> {
> 	x += y - 1;
> 	do_div(x, y);
> 	return x * y;
> }
>
          
A call to roundup_64(maxleafents, minleafrecs) would result in,
x = 2^48 + 125 - 1
x = do_div((2^48 + 125 - 1), 125) = ~2^41
x = 2^41 * 125 = ~2^48

i.e. maxblocks will not have the number of required leaf blocks.

-- 
chandan