From: Graham Perrin <grahamperrin_at_gmail.com> wrote on Date: Sun, 01 Oct 2023 19:28:36 UTC :
> User environment (1500000) inferior to kernel (1500001). > > Is this type of discrepancy ever normal/acceptable? That indicates that __FreeBSD_version was 1500000 when the uname that was run was built but the running kernel was built when __FreeBSD_version was 1500001 . (Presumes a lack of use of OSVERSION to override the getosreldate result for the kernel case.) What does "file /usr/bin/uname" report? Which does it list for: FreeBSD 15.0 (1500000) vs. FreeBSD 15.0 (1500001) ? I did not see evidence about: sys/sys/param.h:#define __FreeBSD_version 1500000 vs. sys/sys/param.h:#define __FreeBSD_version 1500001 at the various build stages in your later material. So I presume the 1500001 change was always present, even at the start of the builds. But, I saw no evidence of the commands used to do the builds. Such can control some of the rebuilt vs. not status, even with the 1500001 change being present. For example: A) From scratch builds (empty build tree at the start of the build) vs. B) META_MODE builds (that check for sys/sys/param.h having been updated) vs. C) NOCLEAN builds (without META_MODE) (A)/(B) should have lead to uname being rebuilt just based on the sys/sys/param.h change. But (C) likely has no tracking of uname being dependent on sys/sys/param.h for make to cause a rebuild. (The FreeBSD Makefiles do not list all the source code dependencies on other source code files.) There may be more alternatives that I've not covered. === Mark Millard marklmi at yahoo.com
