FreeBSD git weekly: 2026-10-19 to 2026-10-25

Map the packet QoS value to a TID before looking it up in the
per node + TID AMPDU state array.

The TX A-MPDU state used to be per-AC, but I changed it to be
per-TID because it /is/ negotiated per-TID.  However this is
not aligned correctly and A-MPDU on eg the voice queue would
cause a mismatch and thus packets to get dropped.

(TIDs 8..15 are special and session based, not traffic QoS based.)

I noticed this because "ssh host" failed after a while, but
"telnet host 22" succeeded.  It turned out that frames weren't
making it to the driver transmit path and I traced it
back to being dropped here.  Traffic worked for a while until
A-MPDU TX was negotiated, and then all traffic on a TID
would fail until the 802.11 association was re-established.

Locally tested:

* iwn 6205, STA mode - forced traffic into VO AC/TID

MFC after: 1 week

Differential Revision:  https://reviews.freebsd.org/D60848

nvme: add a timeout for Format NVM commands

Format commands need a longer timeout.

Reviewed by:            seuros, chuck
Differential Revision:  https://reviews.freebsd.org/D59827

nvme: support per-request timeouts

Allow different timeouts per command.

Reviewed by:            seuros, chuck
Differential Revision:  https://reviews.freebsd.org/D59828

nvme: honor Linux passthrough command timeouts

Now that we have per-command timeouts, allow linux-compat ioctls to have
a passed-in timeout.

Reviewed by:            seuros, chuck
Differential Revision:  https://reviews.freebsd.org/D59829

nvme: Refactor namespace notifications for additions

When we add a namespace, we have the code inline to do the
notifications. Refactor ahead or reuse.

Reviewed by:            seuros
Differential Revision:  https://reviews.freebsd.org/D59831

nvme: Notify namespaces on reset sometimes

When we're resetting after a controller failure, and we fix the failure,
and we've not resetting as part of initialization, notify the all the
children devices of new namespaces. When we failed the controller
before, we removed all the namespaces.

Reviewed by:            seuros
Differential Revision:  https://reviews.freebsd.org/D59832

Triggering is only needed if device is in mmap mode, and in mmap example
the device is triggered as part of syncgroup. For non-mmap mode,
triggering is not needed at all. Triggering mmap'ed device too early
would start the IPTR/OPTR counters and that would cause those pointers
to be out of sync, so let syncgroup do the triggering once the device is
fully configured and ready for capture and playback.

MFC after:      1 week
Reviewed by:    christos
Differential Revision:  https://reviews.freebsd.org/D60839

Recognize Intel 8086:7e25 using 32-bit MMIO with a four-byte register
stride. Deassert the LPSS functional resets and program the 100 MHz
reference clock divider for a 44.2368 MHz UART clock before the UART
core claims the register window, and reprogram both on resume.

Reviewed by:    imp, adrian
Differential Revision:  https://reviews.freebsd.org/D60858

The Intel Low Power Subsystem wraps its I2C, SPI and UART functions in a
block of private registers that control the functional clock and the
function and DMA resets.

Refactored to use a single header

Reviewed by:    imp, adrian
Differential Revision:  https://reviews.freebsd.org/D60859

vmm doesn't implement it, and bhyve simply aborts.  I fixed this in
commit c83e075d72f6, but it'll still be impossible to boot newer FreeBSD
guests on older hosts without this workaround.

Reviewed by:    adrian
Fixes:          https://cgit.freebsd.org/src/commit/?id=2272fac218a8 ("hwpmc: Add support for AMD-only IRPERF in hwpmc PERF class")
Differential Revision:  https://reviews.freebsd.org/D60730

When the driver's isc_rxd_pkt_get() fails part way through a receive
pass, iflib_rxeof() jumps straight to the reset request.  The frames it
already took off the ring and chained for if_input() are neither passed
up nor freed, so up to budget - 1 frames are lost and their mbufs and
clusters leak on every such error, and the frames queued to LRO in the
pass wait for the next one.  Leave the loop instead, pass the chain up
and flush LRO as on the normal exit, then request the reset as before,
without refilling.

The path is reachable from several drivers: igc, em, igb and ixgbe
return EBADMSG for frames received with errors, which the hardware
passes up only with store-bad-packets enabled; ixl, iavf and ice for
every descriptor flagged RXE; aq, enetc, mgb and enic for descriptors
they cannot use.

Reviewed by:            gallatin
Sponsored by:           Rubicon Communications, LLC ("Netgate")
Differential Revision:  https://reviews.freebsd.org/D60805

nvmecontrol: select format by sector size

Add automatic selection of lbaf value based on a single sector size. The
lbas (the data sector size) is selected based on the power of 2 in the
sector size, and the metadata size is selected by the remainder. If an
lbaf matches these values is used, otherwise we exit with an error.
When there's a tie, we prefer the higher performance.

Manual selection of these values remains as well for times when we
select incorrectly.

Reviewed by:            seuros, chuck
Differential Revision:  https://reviews.freebsd.org/D59833

nvmecontrol: support controller-wide sector sizes

Extend sector-size selection to controller-wide formats when every
namespace has one compatible format index and matching effective
metadata and protection settings. Reject ambiguous or inconsistent
layouts so heterogeneous fleets can request a sector size without
risking an unintended controller-wide format.

Reviewed by:            seuros, chuck
Differential Revision:  https://reviews.freebsd.org/D59834

nvmecontrol: add format dry-run support

Add -n/--dry-run to resolve and print the Format NVM command without
submitting it. This lets users verify namespace selection and format
parameters before making destructive media changes.

Reviewed by:            seuros, chuck
Differential Revision:  https://reviews.freebsd.org/D59835

nvmecontrol: Use admin device for admin-passthru

When we're doing admin-passhtru, we need to use the admin device, not
the namespace device. So when we're passed in ndaX or nvdX, we need to
translate that to nvmeY for this command for the admin-passthru command,
but not the io-passthru command. Note namespace nuance in man page.

Differential Revision:  https://reviews.freebsd.org/D59878

nvmecontrol: Fix mixup between pi and pil

pi and pil were extracted from the incorrect place. Swap how we get/set
there to be correct. However, since these are always 0, this is a nop.

This will make it easier to port bhyveload's functionality into the main
bhyve executable.

No functional change intended.

Reviewed by:    novel
MFC after:      2 weeks
Differential Revision:  https://reviews.freebsd.org/D60481

Use the new array size when calling realloc().  Note that this queue
stores events read from a host evdev device, i.e., it is not
guest-controlled.

While here, use the cdefs.h __unused macro and use them correctly, i.e.,
don't mark "arg" as unused when it isn't.

Reported by:    Brian Delgado from Intel using Anthropic Opus 4.8
Fixes:          https://cgit.freebsd.org/src/commit/?id=054accac71e0 ("Add a virtio-input device emulation.")
MFC after:      2 weeks
Sponsored by:   The FreeBSD Foundation
Differential Revision:  https://reviews.freebsd.org/D60679

thr_try_gc unlock and lock again to make sure the lock order. However,
there is no reason for such roundtrip if the caller holds the correct
oder. As a result, we create a new helper function to add the thread
to gc list without unlock + lock and wire the join_common to this new helper.

Reviewed by:    kib
MFC after:      2 weeks
Differential Revision: https://reviews.freebsd.org/D60717

For better readability.

No functional change (intended).

MFC after:      3 days
Sponsored by:   The FreeBSD Foundation

No functional change (intended).

MFC after:      3 days
Sponsored by:   The FreeBSD Foundation

No functional change (intended).

MFC after:      3 days
Sponsored by:   The FreeBSD Foundation

sched_preempt() does not support operating on a remote thread.  If you
are looking to influence CPU allocation on another thread, see, e.g.,
cpuset_setthread(), which in the end calls sched_affinity().

MFC after:      3 days
Sponsored by:   The FreeBSD Foundation

MFC after:    1 week

When an application uses inotify to watch a directory,
vn_inotify_add_watch() reads the directory entries and does a name
lookup so as to induce the filesystem to create namecache entries.
Some filesystems expect the directory to be opened before VOP_READDIR
can be called, so do that.

PR:             https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=299239
Reviewed by:    kib
MFC after:      1 week
Differential Revision:  https://reviews.freebsd.org/D60782

MFC after:      3 days
Sponsored by:   The FreeBSD Foundation