Update to match recent cryptodev update
<opencrypto/cryptodev.h> recently changed a bunch of fields from
signed to unsigned. To avoid gcc stupidly complaining about
comparisons of different signedness ints (I mean, how can
for (i = 0; i < unsigned_thing; i++)
ever see i become negative?) change i from int to unsigned int.
Should fix the i386 build.
Fix:
An integer wraparound bug that could cause a NULL pointer exception.
Data sharing issues that could cause data corruption.
Locking issues that could cause double freeing or using free memory.
Issues discovered by nasm and coordinated by VulnCheck
Thank you!
libssh/Makefile: Fix comment about compat libssh.
We _do_ create a private compat libssh, not for the benefit of
applications linked against it (there aren't any, or if there were it
would be a bug) but for the benefit of applications loading
pam_ssh.so which is legitimately linked against libssh.
When I made the change a few months ago to restore building compat
libssh (by not defining NOCOMPAT) I explained this in the commit
message but neglected to update the comment.
PR lib/58648: private shared libraries should go in /usr/lib/private,
not /usr/lib
Update an old temporary usage comment for tzdata
There was a short period (years ago now) when the tzdata2netbsd
script for updating NetBSD HEAD with the current version of tzdata
was not working as desired, and a comment added here indicated that.
That is long since fixed, the script has been used again now for
ages, so remove the comment indicating things were being done by
hand, and instead include usage advice for tzdata2netbsd.
Undo earlier meaningless change
Revert my botched change (rev 1.7, 2026-04-18 19:42:21 +0000) which
had the parens in the wrong place, which made it useless (pointed out
offlist by rillig@ - thanks). The change was made unnecessary by
christos later change (rev 1.8, 2026-04-18 21:37:04 +0000), so there
is no point fixing it, just make it go away.
VM_KERNEL_PT_PAGES is an old-pmap parameter, so segregate it to
the old-pmap, and default it there to a common value that individual
platforms can override if desired.
Fix riscv sqlite3 builds
It appears that with our gcc, there's no int128_t defined for
the riscv architecture, so comment out the "|| defined(__riscv)"
(2 places) which enables it.
This should fix the riscv builds of HEAD after the recent sqlite3
update.
__HAVE_PMAP_PHYSSEG an old-68k-pmap construct, so put it and associated
declarations into <m68k/pmap_motorola.h>, and remove it from all of the
m68k vmparam.h's.
Centralize the definitions of MAXTSIZ, DFLDSIZ, MAXDSIZ, DFLSSIZ,
an MAXSSIZ across all m68k platforms. Notable callouts:
- default values for 68010 come directly from sun2, but will be suitable
for any additional 68010 systems that may appear in the future.
- Sun3 MMU dictates more conservative limits than the rest of the 68020+
crowd.
- Amiga is odd-one-out and keeps the previous values (it has an odd-ball
USRSTACK, too).