port-mips/59285: _longjmp(..., 0) makes setjmp return 0, not 1

[campbell+netbsd%mumble.net@localhost]

>Number:         59285
>Category:       port-mips
>Synopsis:       _longjmp(..., 0) makes setjmp return 0, not 1
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    port-mips-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sat Apr 12 18:35:01 +0000 2025
>Originator:     Taylor R Campbell
>Release:        current, 10, 9, ...
>Organization:
The MipsBSD _Longjmpzero
>Environment:
>Description:
Test case: lib/libc/setjmp/t_setjmp/_longjmp_zero

Duration: 1.332031 seconds
Termination reason

FAILED: setjmp returned zero after longjmp

https://releng.netbsd.org/b5reports/pmax/2025/2025.04.11.04.54.02/test.html#lib_libc_setjmp_t_setjmp__longjmp_zero

Only happens with _longjmp, not the other versions.

Cursory examination of _setjmp.S shows that nothing is done to map 0 to 1 before passing it as the return value -- though curiously the argument a1 is both _stored_ in the jmp_buf _and moved_ to the return value register v0, when I would expect only the latter is necessary:

    119 LEAF(_longjmp)
...
    130 	REG_S		a1, _SC_REGS_V0(a0)	# save return value
...
    178 	j	ra
    179 	 move	v0, a1			# get return value in 1st arg

https://nxr.netbsd.org/xref/src/lib/libc/arch/mips/gen/_setjmp.S?r=1.26#119

There are some other instructions in the middle that use v0 as a temporary (for restoring the floating-point control and status register), but that doesn't seem to touch a1.
>How-To-Repeat:
cd /usr/tests/lib/libc/setjmp
atf-run t_setjmp | atf-report
>Fix:
Yes, please!