CVS commit: src/sys/arch

["Jason R Thorpe" <thorpej%netbsd.org@localhost>]

Module Name:    src
Committed By:   thorpej
Date:           Tue Nov 11 15:17:05 UTC 2025

Modified Files:
        src/sys/arch/cesfic/cesfic: locore.s
        src/sys/arch/hp300/hp300: locore.s
        src/sys/arch/luna68k/luna68k: locore.s
        src/sys/arch/mvme68k/mvme68k: locore.s
        src/sys/arch/news68k/news68k: locore.s
        src/sys/arch/next68k/next68k: locore.s
        src/sys/arch/x68k/x68k: locore.s

Log Message:
Fix a ~30-year-old bug: pass the correct address to m68881_restore()
when setting the null FP context after probing for the FPU.  Thanks to
Isaki-san for the pointing it out and to Tsusui-san for the testing work
that prompted Isaki-san's analysis.

(This bug has propagated from the hp300 port to just about everywhere
else over the years; an ancient ruin in the garden. The question about
whether this is even strictly necessary will be addressed separately.)

This also fixes a more recently bug of my own making (left over from
the recent pmap_bootstrap2() changes) -- use the correct address register
when fiddling with the FP context.


To generate a diff of this commit:
cvs rdiff -u -r1.49 -r1.50 src/sys/arch/cesfic/cesfic/locore.s
cvs rdiff -u -r1.189 -r1.190 src/sys/arch/hp300/hp300/locore.s
cvs rdiff -u -r1.88 -r1.89 src/sys/arch/luna68k/luna68k/locore.s
cvs rdiff -u -r1.134 -r1.135 src/sys/arch/mvme68k/mvme68k/locore.s
cvs rdiff -u -r1.91 -r1.92 src/sys/arch/news68k/news68k/locore.s
cvs rdiff -u -r1.88 -r1.89 src/sys/arch/next68k/next68k/locore.s
cvs rdiff -u -r1.137 -r1.138 src/sys/arch/x68k/x68k/locore.s

Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.