Hi. Original post is not yet delivered... On Fri, 10 Dec 2021 23:25:11 +0000 (UTC) "Bjoern A. Zeeb" <[email protected]> wrote:
> On Sat, 11 Dec 2021, Tomoaki AOKI wrote: > > Hi, > > > As no one have done heads-up... > > > > On src, "git: df38ada2931f - main - modules: increase MAXMODNAME and > > provide backward compat" [1] mandates includes of sys/module.h after > > sys/param.h. > > That order is suggested by style.9 as well; I think we had only one > case in base not doing that. Missing param.h was more likely but > still minimal. Maybe many porters don't read it. ;-) > > All modules in base are fixed by the commit itself but modules on ports > > are not. > > I am sorry; I hadn't thought about about the possible build issue > there. The major concern on the change was the internal backward > compat issue. Would be difficult for src committers to do thorough build tests for ports. > > Maintainers having any port having kmod need checking and, if affected, > > fix it. > > > > At least, I've bitten by sysutils/acpi_call and x11/nvidia-driver. > > For x11/nvidia-driver, I've uploaded patch on already-reported Bug > > 260318 [2]. Not yet for sysutils/acpi_call. > > Thank you for that! I'd hope that's all of them. > > Is it in both cases include order or is param.h missing at all? Both cases include param.h but mis-ordered. I wonder if includes are alphabetically ordered on many ports. (Excluding all known-to-be-mandatory orders.) > > [1] > > https://lists.freebsd.org/archives/dev-commits-src-main/2021-December/001971.html > > > > [2] https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=260318 > > -- > Bjoern A. Zeeb r15:7 > -- Tomoaki AOKI <[email protected]>
