t_arc4random: Fix test to guarantee global state initialization.
Previously, if pthread_atfork failed, arc4random_global_buf would
fail to initialize the global state as intended. (There are separate
tests for intentional failure of global state initialization.) It is
unlikely for pthread_atfork to fail, but with RLIMIT_AS cranked down
to zero, it may. To ensure that the global state is successfully
initialized for the tests that are trying to exercise it, call
arc4random from another thread first before forcing failure of the
local state initialization.
PR lib/59117: arc4random has some failure modes it shouldn't
fix the m68ksf build.
avoid using bare ${MACHINE_ARCH} when referring to subdirs that
mknative-gcc creates, and avoid overwriting ${GCC_MACHINE_ARCH}
if it is already set.
fix a test in libgomp, and explain why it exists.
sun4v: Try to use the hypervisor interrupt api version 1 if version 2 fails.
Version 3 systems are working ok - trying to fix the setup when a version < 2
system is used.
Not tested on yet.
exec_elf: Nix diagnostic messages for all (unknown) fd.o notes.
Same as for unknown GNU notes. Shaves off a tiny bit of code in
non-DIAGNOSTIC builds, saves us some trouble when fd.o invariably
adds more metadata not relevant to execution semantics.
PR kern/60901: Unknown elf note type 1081871370
hppa: Enable fortify/ssp in userland for progs/libs that opt into it.
The `ssp' part doesn't actually do much (and really I'm not sure why
these options are mixed together) because hppa stack grows up, so
writing past the end of an array won't overwrite the return address,
so stack smashing in this way is extremely unlikely. But the
`fortify' part, which instruments functions like memcpy/read/getcwd
with bounds checks where statically known, is still useful to catch
buffer overruns.
ok `I guess' nick@
PR lib/60858: fortuitous embarrassment: fortify is all kinds of busted
Update bind.keys from src/external/mpl/dist/bind.keys.
managed-keys is ancient history and just generated warnings doing
nothing useful. The proper keyword to use is trust-anchors and
this file is now current. The latest root key is only partially
specified as it wasn't published at the time this file was updated,
but enough information is provided for named to download the full
key using RFC 5011.
Fixes PR/52465 and PR/60864
Fix compiler core dumps on MacOS on Apple Silicon (PR toolchain/60904),
by disabling the use of the broken arm64 asm that comes with the present
version of GMP (we can re-enable it when we upgrade).
beaglebone: fix sysctl hw.clk crashing the kernel
The clock gate driver on the beaglebone black did not provide a clock rate function, but a clock rate function is mandatory. This caused a null pointer dereference when running "sysctl hw.clk".
sun4v: Introduce a SAVE_WINREGS macro for future debugging.
To be used for tracing stability issues in the kernel.
The SAVE_WINREGS macro can be called from relevant places
in the locore.s file to dump the contents of all the register windows.
Function for printing out the contents of the debug area will be
added later.
Import rsu(4), a driver for Realtek's 802.11n FullMAC USB Wi-Fi adapters.
This is a port of the OpenBSD/FreeBSD driver by Damien Bergamini.
The hardware is less common as a discrete adapter than urtwn(4),
perhaps due to being more expensive, or being targeted at embedded
systems with slow CPUs. Adapters pop up on eBay now and then from
manufacturers like Sitecom. I have two such devices from scouring
eBay.
It is a work-in-progress - the device attaches and detaches but is
not yet usable, so not enabled yet. Committing mostly for easier
review and collaboration.
