pci_resource: use existing BARs even if mem/io space is left disabled
It's possible that firmware assigned resources to a device and left
memory / io space accesses disabled. Let's assume if we see a non-zero
address in a BAR that it was assigned by firmware.
Should help PR# kern/60427 and kern/60463
gicv3: don't wait for RWP to clear when unblocking irqs
The RWP bit does not track writes to ISENABLER<n>, only ICENABLER<n>. So
don't bother waiting for it to clear when unblocking IRQs.
This is ok; pic_block_irqs is expected to be synchronous (no IRQs allowed
after the call) but not pic_unblock_irqs (some time can pass before an IRQ
is raised).
Pull up the following, requested by kre in ticket #1339:
external/public-domain/tz/dist/CONTRIBUTING up to 1.1.1.11
external/public-domain/tz/dist/Makefile up to 1.7
external/public-domain/tz/dist/NEWS up to 1.9
external/public-domain/tz/dist/README up to 1.4
external/public-domain/tz/dist/TZDATA_VERSION up to 1.47
external/public-domain/tz/dist/asia up to 1.17
external/public-domain/tz/dist/backward up to 1.13
external/public-domain/tz/dist/backzone up to 1.5
external/public-domain/tz/dist/northamerica up to 1.8
external/public-domain/tz/dist/southamerica up to 1.6
external/public-domain/tz/dist/theory.html up to 1.7
external/public-domain/tz/dist/version up to 1.20
external/public-domain/tz/dist/zone.tab up to 1.7
external/public-domain/tz/dist/zone1970.tab up to 1.8
external/public-domain/tz/dist/zonenow.tab up to 1.9
doc/3RDPARTY (apply patch)
Updated tzdata to 2026d (via local 2026dgtz).
Pull up following revision(s) (requested by bouyer in ticket #1338):
sys/arch/xen/xen/xbdback_xenbus.c: revision 1.109
Fix a brain-o in a KASSERT that could cause a guest to panic a DIAGNOSTIC
kernel: to assert that an I/O is already in flight we want
xbdi_pendingreqs > 0, not xbdi_pendingreqs > 1
While there, fix an error path that could also cause a KASSERT to fire
or a state corruption: reset xio_need_bounce to 0 before calling
xbdback_io_error() when the bounce buffer is not fully set up.
Pull up following revision(s) (requested by bouyer in ticket #464):
sys/arch/xen/xen/xbdback_xenbus.c: revision 1.109
Fix a brain-o in a KASSERT that could cause a guest to panic a DIAGNOSTIC
kernel: to assert that an I/O is already in flight we want
xbdi_pendingreqs > 0, not xbdi_pendingreqs > 1
While there, fix an error path that could also cause a KASSERT to fire
or a state corruption: reset xio_need_bounce to 0 before calling
xbdback_io_error() when the bounce buffer is not fully set up.
Pull up following revision(s) (requested by tls in ticket #461):
sys/net/npf/files.npf: revision 1.26
sys/net/bpf_filter.c: revision 1.74
sys/net/bpf.c: revision 1.259
sys/net/npf/npf_os.c: revision 1.24
sys/net/npf/npf_os.c: revision 1.25
Fix npf crash in kernels without "pseudo-device bpfilter".
Actually fix panic on non-modular kernels with npf and no bpf.
Fix commit of badly-applied patch that brought us npf_os.c 1.24.
arm64: fpu: avoid unnecessarily reads of MVFR1_EL1
The contents of MVFR1_EL1 do not change at runtime and it is already
saved in struct cpu_info. Use the cached value in fpu_state_load() as it
avoids a guest exit when running under a hypervisor with HCR_EL2.TID3 = '1'.
Pull up following revision(s) (requested by jdc in ticket #460):
sys/dev/i2c/lm75.c: revision 1.51
Fix setting limits:
check for NULL limits pointer (restore)
handle degc (user) or raw (restore initial) values
Pull up following revision(s) (requested by jdc in ticket #459):
sys/dev/i2c/adm1026.c: revision 1.14
sys/dev/i2c/adm1026.c: revision 1.15
sys/dev/i2c/adm1026.c: revision 1.16
sys/dev/i2c/adm1026.c: revision 1.17
sys/dev/i2c/adm1026reg.h: revision 1.6
share/man/man4/adm1026hm.4: revision 1.2
share/man/man4/adm1026hm.4: revision 1.3
share/man/man4/adm1026hm.4: revision 1.4
Remove the "multi_read" logic now that the problem in the pcf8584 driver
has been fixed (r1.23 - use repeat start).
Handle cases where we have a gap in the sensors configured/present with
a map of envsys sensor to chip sensor.
Use a variable to hold the register number instead of passing a constant
to the iic routines.
[26 lines not shown]
Pull up following revision(s) (requested by jdc in ticket #458):
sys/arch/sparc64/sparc64/autoconf.c: revision 1.247 (patch)
sys/arch/sparc64/sparc64/ofw_patch.c: revision 1.10 (patch)
sys/arch/sparc64/sparc64/ofw_patch.h: revision 1.5
Reduce the amount of patching conditions in autoconf.c by moving the
machine model checks to ofw_patch.c. We now have one patch routine for
i2c buses and another for i2c devices.
Consolidate the patching checks for parent bus type into a single check.
Remove the "multi_read" property for V210, V240 and V440 - no longer needed
because the pcfiic driver was fixed.
While here, add a "fan_mask" property for the V240, which has only 3 fans
in the first group of 4.
Pull up following revision(s) (requested by jdc in ticket #457):
sys/dev/ic/bq4802reg.h: revision 1.1
sys/dev/ic/bq4802reg.h: revision 1.2
sys/arch/sparc64/conf/GENERIC: revision 1.245 (via patch)
sys/arch/sparc64/conf/files.sparc64: revision 1.169 (via patch)
sys/arch/sparc64/dev/bq4802_ebus.c: revision 1.1
sys/arch/sparc64/dev/bq4802_ebus.c: revision 1.2
Add a driver for the Texas Instruments bq4802Y/bq4802LY RTC chip,
used in the Sun Ultra 45.
Add bq4802rtc for the Ultra 45 RTC.
Add bq4802rtc (RTC for U45).
Use the correct setting for the stop bit, so that the clock runs when the
power is off (instead of vice versa). Make this clearer in the header.
Pull up following revision(s) (requested by jdc in ticket #456):
sys/dev/ic/pcf8584var.h: revision 1.7 (patch)
sys/arch/sparc64/dev/pcfiic_ebus.c: revision 1.10 (patch)
sys/dev/ic/pcf8584.c: revision 1.23 (patch)
When reading from an i2c device register, use repeat start on the bus.
This avoids the situation where another master gets the bus between our
register select (i.e. write) and the following read.
Add a delay after accessing our registers for the PIC implementation
of the PCF8584 (e.g. on the Sun Ultra 45).
Change swapregs to flags to handle multiple quirks and set the flags
appropriately in the Ebus frontend.
Pull up following revision(s) (requested by tls in ticket #455):
sys/uvm/uvm_pdpolicy_clockpro.c: revision 1.28
sys/uvm/uvm_pdpolicy_clockpro.c: revision 1.29
sys/uvm/uvm_pdpolicy_clockpro.c: revision 1.30
sys/uvm/uvm_pdpolicy_clock.c: revision 1.43
Fix a very brief race which could cause page states to be corrupted.
Observed under heavy physio load from an 8-way parallel dump|restore
workload used in filesystem testing - manifested as intents being
scribbled over while the interlock was momentarily not held
in uvmpdpol_selectvictim. I've attempted to apply the same fix to
clockpro, which looks like it needs it too, but clockpro was on fire
when I got there and I didn't put it out (still does not compile).
Fix an obvious typo which applied the wrong intent. Still does not compile.
fix build with clockpro
Pull up following revision(s) (requested by sborrill in ticket #454):
sys/compat/linux/common/linux_file.c: revision 1.135
Fix edge-case panic in copy_file_range when off_in or off_out are not NULL.
Found by Henrique Brito as part of GSoC 2024.
Patch to 2026d (previous was 2026c)
Changes to code
zic now rejects Link targets that would have invalid names, and
more efficiently processes Expires, Leap and Rule lines with years
far in the past or future. (Thanks to Darren Carreras.)
zic now ports to systems that report lack of link support via
EINVAL, ENOSYS or EPERM errno values. (Thanks to Tom Lane.)
When tzset and related functions encounter a TZif file that is too
large for them, they now consistently fail instead of sometimes
silently ignoring excess parts of the file.
localtime-related functions no longer mishandle extreme timestamps
when given TZif files holding some unlikely timezone histories.
(Problem reported by David Sarkisyan.)
[12 lines not shown]
PR/60655: vmt(4) probes the VMware backdoor unconditionally on aarch64
On ACPI systems only probe for vmt(4) when Hypervisor_Id in FADT reports
VMWare.
This helps avoid buggy hypervisors that don't implement a RAZ mdccsr_el0
Patch from Showta Ishizaki and slightly updated by me.
