Make sure the start and end of the PT page ranges are aligned to the
size of what's addressable by a single PT page, thus ensuring we count
them up correctly.
In pmap_bootstrap1(), check to see if FIXEDVA entries in machine_bootmap[]
are covered by any existing page table range, and if not, allocate additional
page table ranges to cover them.
This does not impact the one current user of FIXEDVA -- hp300 -- which
uses it to map the last page of RAM VA==PA. In the hp300 case, this
was already covered by the PTs that map the alternate SYSMAP_VA that
the hp300 uses (precisely because it needs the last VA==PA mapping).
This will eventually be used to map the I/O region VA==PA for mac68k.
Normally, we might otherwies use a TT register for that, but mac68k
runs on 68020s, so we cannot.
Two issues:
- Always include vmparams.h via <machine/vmparams.h>, because redirection
logic elsewhere relies on this.
- But akshually, isn't not even needed here because the code that would
use it is #if 0'd out anyway.
So, collect the garbage and ramble on.
Pull up following revision(s) (requested by martin in ticket #275):
share/mk/bsd.own.mk: 1.1481 via patch
PR 58762: disable MKCOMPAT for earm*.
If someone is interested in re-adding support for oabi compat library
builds, they can figure out the missing bits. But for now, stop producing
bogus compat32/debug32 sets on all evbarm builds.
Pull up following revision(s) (requested by martin in ticket #274):
usr.bin/cksum/cksum.c: 1.53
PR 60154: do not print arbitrary control characters when printing
file names while checking hashes with the -c option.
Remove the upper bound on nkmempages for virt68k, else the kmem arena
gets easily starved on VMs with large memory configs (which is sort of
the whole point of virt68k).
Partially revert previous to fix the builds.
The UVM stats functions are used (apparently) by the RUMP
"kernel", and need to be exposed as they were before, so
for now, do that.
Feel free to undo this, and fix the problem some other way.
Reduce the pmap API surface exposed to modules. Of the ones that remain,
you could argue that those shouldn't be exposed either, but our file
system <-> pager interface is not that great, among other things, so
they have to remain for now.
arcbios: enable FPU around ARCS calls
This seems required for the SGI O2 for PROM graphics console IO to work
when it scrolls. See PR port-sgimips/60204 for more details.
The issue was introduced in v1.214 of sys/arch/mips/mips/locore.S .
Notably maya@ disabled the FPU early in boot, expecting the rest of
boot to run with the FPU disabled. The FPU is being explicitly
enabled/disabled here to keep to the spirit of maya@'s above commit.
A "better" solution would be to use the pcu API and mark the thread
as needing the FPU for the duration of the arcbios call, however:
* The current MIPS pcu FPU code in sys/arch/mips/mips/mips_fpu.c
doesn't support FPU use in kernel yet; and
* this stuff is called super early during boot and once the OS starts
the ARCBIOS API is no longer supposed to be used.
call this bozohttpd 20260503, and update the CHANGES for the last 2 years
o fix the default minimum TLS version to 1.1 from 1.3. the
manual already said 1.1 was the default. fixes PR#58878.
o log the correct port with TLS connections. fixes PR#59644.
o fix use-after-free, double-free, and bounds checking problems.
from shm.
o better lint support.
o several updates for the manual. from lukem.
add D Bohdan to the contributors list.
Hide platform vmparam differences away from user-space and modules.
For modules that need to know USRSTACK (exec / compat modules), provide
a way to get that via a variable initialized in cpu_startup().
