[PATCH v4 00/15] Introduce RTE_DIV_ROUND_UP

[Thomas Monjalon]

06/10/2026 18:38, Stephen Hemminger:
> On Mon,  5 Oct 2026 19:30:36 -0700
> Joshua Washington <joshwash at google.com> wrote:
> 
> > Numerous components independently define the DIV_ROUND_UP macro, using
> > the same definition. Attempt to commonize by introducing
> > RTE_DIV_ROUND_UP to rte_common.h and replacing usages of DIV_ROUND_UP
> > with RTE_DIV_ROUND_UP. Similarly, replace DIV_CEIL, which is
> > functionally identical, to RTE_DIV_ROUND_UP.
> > 
> > Changes in v4:
> >   - eal: Fix merge conflict in release notes after rebase.
> >   - various commits: add rte_common.h includes where missing to fix
> >     compilation issues.
> >   - opae: drop RTE_DIV_ROUND_UP change due to shered osdep file for RTE
> >     and raw drivers.
> 
> Looks good, some minor AI notes on commit messages and doc

I consider the minor notes as candidates for follow-up or not relevant.

Applied (with a bit of squashing), thanks.


> Info: remaining private copies not covered by this series, candidates
> for a follow-up:
>   drivers/raw/ifpga/base/opae_osdep.h   DIV_ROUND_UP (map like cxgbe)
>   drivers/crypto/ccp/ccp_dev.h          CCP_DIV_ROUND_UP
>   drivers/common/qat/dev/qat_dev_gen_lce.c  LCE_DIV_ROUND_UP
>   drivers/net/sfc/sfc_dp.h              SFC_DIV_ROUND_UP
>   drivers/common/sxe2/sxe2_osal.h       SXE2_DIV_ROUND_UP
>   drivers/common/cnxk/roc_platform.h    PLT_DIV_CEIL

Would be nice to address in a new series, yes.


> Info: the Doxygen repeats itself. "Evaluated twice" is in @param d and
> again in the first @note; "non-negative" is in @param n and again in
> the second @note. Keep the @param text and drop the duplicate notes.

I think this repeat is fine.


> Info: nothing in DPDK defines BITS_TO_LONGS, so the new #ifndef guard
> does nothing; drop it.

It is in rnp_osdep.h, probably to allow including in a different environment.