tests/make: remove filemon path from expected debug output
That variable is not essential for the test, and it differs between the
standard configuration and the c90-stdbool configuration from
test-variants.mk.
MaybeSubMake: look for expansion of ${.MAKE}
Only looking for the equivalent of ${.MAKE:T}
and calling it a match if preceded by '/' is too aggressive.
Look for the expansion of ${.MAKE} and if that is a full path also
look for the basename, but in each case, only call it a match if at the
start of the command line or preceded by space.
Note: over enthusiastic setting of OP_SUBMAKE is harmless, though
it can impact meta mode, requiring an explicit .META on some targets.
gcc/vax: (gcc12) adjust the constraints on the output operands of the
'*extzv_aligned' instruction patterns, removing the 'earlyclobber'
modifier, to be consistent with the constraints on the output operands
of the related 'extv_aligned' instruction patterns.
Remove CRYPTO options from man page
Since we do not build the code with CRYPTO defined, documenting
the CRYPTO options (pretending that they might work) in the man
page is misleading at best.
Rather than removing the text for the options completely, include
it conditioned on the C *roff number register being > 0 (and define
it to be 0 so the formatters don't format the CRYPTO sections of
the man page).
Add a comment to the Makefile indicating that if the commented
out section of it which would enable building with CRYPTO is
ever re-enabled, to also change the init of the C number register
in nc.1 from 0 to 1, so the options will return to the manual.
Clean up code, NFCI.
Add the /* $NetBSD$ */ header line, which wasn't added
when this was originally imported (NetBSD __RCSID was).
Move more of the CRYPTO related code into #ifdef CRYPTO
than was done before (global var decls, etc).
The -e option is CRYPTO related, does nothing without
the CRYPTO code, so move it out of the regular options,
retaining it only with CRYPTO.
The V option however (which was included with the CRYPTO
options in the getopts() arg string) is not CRYPTO related,
it is OpenBSD, and (when building for OpenBSD) would apply
to either CRYPTO or not. Fix that (though it makes no
difference when we exclude both CRYPTO & OpenBSD options).
Don't do run time tests that can never be relevant (like
[12 lines not shown]
sh code enhancements (NFCI).
More error checking, printf instead of echo, quoting only when needed,
sh ! rather than test ! (that kind of thing).
Tested via a full release + images build of HEAD/vax - the only current
user of this script.
Instead of hard-coding the ramdisksize both in the distrib makefiles
and the kernel, provide a script to grab the size from the kernel so
we can define it in one place.
zfs: reject read() on directory
right now, netbsd in general allows read() on directory for
the compatibility with historical applications. (i have not
seen such an application by myself though. is anyone around
here still keeping such ancient binaries? i'm curious if such
a binary still works on today's UFS.)
this commit makes zfs reject such an attempt because zfs is
not prepared to produce the historical UFS dirent structure.
