The ctx_hw_stats_ext DMA address was not correctly passed to the
firmware during the HWRM_STAT_CTX_ALLOC allocation, causing stats to not
populate for Thor2. Passing the correct DMA length resolved the issue
MFC-After: 3 days
Differential-Revision: https://reviews.freebsd.org/D49732
(cherry picked from commit fec0e2064818f991867c9851a837012ea31774da)
The completion event type 76 is not supported by the driver. Instead of
flooding the dmesg with "Unknown event type" messages when this event
occurs, move the print under debug level.
MFC-After: 3 days
Differential-Revision: https://reviews.freebsd.org/D49731
(cherry picked from commit 6450d937955fcd1ab9034c49d53306e882c4a281)
Doorbell offset :
For Thor controllers doorbell offset was always hardcoded to 0x10000 for
PF devices where as for Thor2 controllers doorbell offset will be
legacy_l2_db_size_kb value provided by firmware through hwrm_func_qcfg
command.
CQ Toggle & Epoch bits support :
In order to handle out of order doorbell handling as part of Dropped
Doorbell Recovery, HW expects two changes in the driver in data path.
- First change is the epoch bit changes while updating the producer
indexes of Tx. This epoch bit is toggled by the driver, each time the
queue is wrapped for that specific doorbell.
- The second change is to add a toggle bit pair to each ARM type
doorbell. This includes the CQ_ARMALL, CQ_ARMSE, CQ_ARMENA
doorbells. The toggle bit pair in context is incremented by the chip
each time a new NQE completion is generated by the chip. To keep the
driver in-sync, the toggle bit pair will be passed in the NQE to the
host completion. This will be the toggle bit pair value that the host
must use to setup the next NQE operation. The driver will pass that
latest toggle bit pair value into the ARM type doorbells it generates to
the chip. The doorbell clients will compare the toggle bit pair in each
doorbell with the value in context. If the values match, the doorbell
will be honored. If the values do not match, the doorbell will be
discarded.
MFC-After: 3 days
Differential-Revision: https://reviews.freebsd.org/D49730
(cherry picked from commit 39c0b8b7994b0d339bffb0b17291c4a2b14cae3a)
Added support for RX V3 completion record types-
CMPL_BASE_TYPE_RX_TPA_START_V3 and CMPL_BASE_TYPE_RX_L2_V3.
MFC-After: 3 days
Differential-Revision: https://reviews.freebsd.org/D49728
(cherry picked from commit 526d74f6ca55c7a4b9c0c745d13e94c7a7bb6e0b)
Add backing store V2 support.
Thor2 controllers supports only the V2 support.
MFC-After: 3 days
Reviewed-by: Warner Losh <imp@FreeBSD.org>
Differential-Revision: https://reviews.freebsd.org/D49727
(cherry picked from commit d5ce906da7c55085f93fce096ebb8bc44a3cffe5)
Update HSI header to support Thor2 controllers.
MFC-After: 3 days
Differential-Revision: https://reviews.freebsd.org/D49726
(cherry picked from commit bb90baed6c275495b03adc5569346a59fce2a3c8)
Using one taskqueue group with single thread to execute all admin
tasks may lead to unexpected timeouts when long running task (e.g.
handling a reset after FW update) for one interface prevents
tasks from other interfaces being executed. Taskqueue group API
doesn't let to dynamically add threads, and pre-allocating thread
for each CPU as it's done for traffic queues would be a waste
of resources on systems with small number of interfaces. Replace
global taskqueue group for admin tasks with taskqueue allocated
for each interface to allow independent execution.
Signed-off-by: Krzysztof Galazka <krzysztof.galazka@intel.com>
Reviewed by: imp, jhb
Pull Request: https://github.com/freebsd/freebsd-src/pull/1336
(cherry picked from commit 3ad01642fe9e241124553f2f18fd365ffea5d20b)
INTRng can leak resources when INTx interrupts are re-routed, which is
typically harmless but can be fatal when devices are (repeatedly) hot
plugged into PCI buses on INTRng systems. Re-routing INTx interrupts
is nonetheless still necessary on some systems, and identifying whether
the re-routing should be enabled or disabled seems to be nontrivial.
Add a hw.pci.intx_reroute sysctl/tunable so systems which don't want
legacy PCI interrupt re-routing can turn it off. This is probably not
the best fix but it's something which can be safely included in FreeBSD
14.3.
Co-Authored-by: jhb
Reviewed by: bz, jhb
MFC after: 3 days
Sponsored by: Amazon
Differential Revision: https://reviews.freebsd.org/D49849
(cherry picked from commit 2187ec93ada1c7399f2f3537920f6277bec4a0ef)
Move the necessary extra logics (i.e., noise_remote_enable() and
TAILQ_INSERT_TAIL()) from wg_ioctl_set() to wg_peer_alloc(), and thus
make it easier to be called. Actually, the updated version is more
asymmetric to wg_peer_destroy() and thus less likely to be misused.
Meanwhile, rename it to wg_peer_create() to look more consistent with
wg_peer_destroy().
Reviewed by: aly_aaronly.me (diff), markj
Obtained from: DragonflyBSD 902964ab24ba (with some changes)
(cherry picked from commit 7121e9414f294d116caeadd07ebd969136d3a631)
Future firmwares will report these types to the driver. These
transceivers work already but are misidentified as a different type.
Sponsored by: Chelsio Communications
(cherry picked from commit c22b297062e1440676973a8aa89cbad1571e22f9)
The driver uses bus_reset_child on its parent to reset itself but that
performs an FLR whereas the hardware needs a Conventional Reset[1] for
full re-initialization. Add routines that perform conventional hot
reset and use them instead. The available reset mechanisms are:
* PCIe secondary bus reset (default)
* PCIe link bounce
hw.cxgbe.reset_method can be used to override the default. The internal
PL_RST is also available but is for testing only.
[1] 6.6.1 in PCI Express® Base Specification 5.0 version 1.0
Sponsored by: Chelsio Communications
(cherry picked from commit 011e3d0b8b90a4330f14b2cb7da45ed7b805ed10)
Add a new hw_all_ok() routine and use it to avoid hardware access in the
public control interfaces (ifnet ioctls, ifmedia calls, etc.). Continue
to use hw_off_limits() in the private ioctls/sysctls and other debug
code. Retire adapter_stopped() as it's of no use by itself.
This fixes problems where ifnet slow-path operations would enter a
synch_op just before set_adapter_hwstatus(false) and touch the hardware
when it's not safe to do so.
Sponsored by: Chelsio Communications
(cherry picked from commit e19d84979a183deb37ce6d7e385c3ccf02a3c8c7)
It was always set to PCATCH because the driver tested (INTR_OK) instead
of (flags & INTR_OK). Fit a WITNESS_WARN in a single line while here.
Sponsored by: Chelsio Communications
(cherry picked from commit 04bf43505bae1bb20d315a44e977d97aed3e5733)
There is no need to include private PCI headers in the driver.
Sponsored by: Chelsio Communications
(cherry picked from commit 762d32354a18517c28933ddc29f9d3d855e450b1)
An L2 table entry isn't associated with a particular SMT (Source MAC
Table) entry.
Sponsored by: Chelsio Communications
(cherry picked from commit f79fba05a016d53e054d6f587213889c3e31b4db)
The driver does minimal initialization in this mode and suspend/resume
should ignore resources that aren't setup. This is for debug only.
kenv hw.cxgbe.sos="1"
kldload if_cxgbe
devctl suspend t6nex0
devctl resume t6nex0
Sponsored by: Chelsio Communications
(cherry picked from commit f4ab14044c1de35b1aefad5449bddc5a1272f8d9)
This change introduces new sysctl handlers that allow the user to change
RX/TX ring sizes. As before, the default ring sizes will come from the
device (usually 1024). We also get the max/min limits from the device.
In the case min values are not provided we have statically defined
constants for the min values. Additionally, if the modify ring option is
not enabled on the device, changing ring sizes via sysctl will not be
possible. When changing ring sizes, the interface turns down
momentarily while allocating/freeing resources as necessary.
Signed-off-by: Vee Agarwal <veethebee@google.com>
Reviewed by: markj
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D49428
(cherry picked from commit 22fe926a62b7bca771d46502dd6a8c202f25b5be)
This change introduces new sysctl handlers that allow the user to change
RX/TX queue counts. As before, the default queue counts will be the max
value the device can support. When chaning queue counts, the interface turns
down momentarily while allocating/freeing resources as necessary.
Signed-off-by: Vee Agarwal <veethebee@google.com>
Reviewed by: markj
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D49427
(cherry picked from commit e0464f74d5579e1538ce741b0a15e6604dbc53c4)
Every tx and rx ring has its own queue-page-list (QPL) that serves as
the bounce buffer. Previously we were allocating QPLs for all queues
before the queues themselves were allocated and later associating a QPL
with a queue. This is avoidable complexity: it is much more natural for
each queue to allocate and free its own QPL.
Signed-off-by: Vee Agarwal <veethebee@google.com>
Reviewed by: markj
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D49426
(cherry picked from commit f8ed8382daf4b9a97056b1dba4fe4e5cb4f7485c)
PCIe mandates a 5 second delay between when the "Attention Button" is
pressed and when the associated device is detached; this is to allow
for the button to be pressed a second time to cancel the ejection. On
some systems this 5 second delay may not be desireable; so introduce a
hw.pci.pcie_hp_detach_timeout sysctl (which can also be set as a loader
tunable) which specifies the timeout in milliseconds (default 5000).
If set to zero, the device is detached immediately.
Reviewed by: jhb
MFC after: 2 weeks
Sponsored by: Amazon
Differential Revision: https://reviews.freebsd.org/D49585
(cherry picked from commit 9be42ee6c9c741052bdc49f13bc953bf88a24423)
whu@ cannot reproduce the assertion failure which led to these ifdefs
being added in the first place, and since they appear wrong, i.e., the
assertions ought to apply to all platforms, let's remove them.
This reverts commits 0af5a0cd27 and
6f7b1310b6.
PR: 285681
Tested by: whu
MFC after: 2 weeks
(cherry picked from commit 54a3920dc9b3b5a47cdaaa3132b4fcf1c448a737)
- The BUS_DMA_KEEP_PG_OFFSET flag is needed, since
storvsc_xferbuf_prepare() assumes that only the first segment may have
a non-zero offset, and that all following segments are page-sized and
-aligned.
- storvsc_xferbuf_prepare() handles 64-bit bus addresses, so avoid
unneeded bouncing on i386.
PR: 285681
Reported by: dim
Tested by: dim, whu
MFC after: 2 weeks
(cherry picked from commit a319ba694538a38429115aaaf1d4b3946ea3a8b5)
On some EC2 instances, there is a race between removing a device from
the system and making the PCI bus stop reporting the presence of the
device. As a result, a PCI BUS_RESCAN performed immediately after
the _EJ0 method returns "sees" the device which is being ejected, which
then causes problems later (e.g. we won't recognize a new device being
plugged into that slot because we never knew it was vacant).
On other operating systems the bus is synchronously marked as needing
to be rescanned but the rescan does not occur until O(1) seconds later.
Create a new ACPI_Q_DELAY_BEFORE_EJECT_RESCAN quirk and set it in EC2
AMIs, and add a 10 ms DELAY between _EJ0 and BUS_RESCAN when tht quirk
is set.
Reviewed by: jhb
MFC after: 1 month
Sponsored by: Amazon
Differential Revision: https://reviews.freebsd.org/D49252
(cherry picked from commit 55c3348ed78fb1d0891e8bb51a8948f95da3560b)
Fix a copy-and-paste error in acpi_pcib_request_feature where the
child device was passed into acpi_pcib_osc rather than the pcib
device.
Reviewed by: garga, jhb
Fixes: ba1904937d9a ("acpica: Extract _OSC parsing to a common file")
Sponsored by: Arm Ltd
Differential Revision: https://reviews.freebsd.org/D48285
(cherry picked from commit b714eacb790551f35de03831b88ad7b7e1502102)
bcd2bin() must not be called with a value greater or equal to
LIBKERN_LEN_BCD2BIN.
MFC after: 2 weeks
Sponsored by: The FreeBSD Foundation
(cherry picked from commit 516e24e57987d184cce70e7f31443653aa1a5e63)
Using printf() with '%s' can lead to arbitrary long printing (although,
usually, a NUL byte should appear quite quickly) and trying to print
unprintable characters.
Instead, print in hexadecimal the exact bytes that are compared to the
expected signature.
MFC after: 2 weeks
Sponsored by: The FreeBSD Foundation
(cherry picked from commit bb04712e37723d112b2fad28af4b035ef35a25be)
This helps figuring out quickly why no SMBIOS device appears in this case.
MFC after: 2 weeks
Sponsored by: The FreeBSD Foundation
(cherry picked from commit fdf08ac1e9f9baac4fcf4af8f3bf7a34d3ea0009)
Only allow the length tolerance (0x1e instead of 0x1f) for a 32-bit
entry point, as there was no 64-bit entry point in the erroneous SMBIOS
v2.1 standard and assigning the length with 0x1f does not make sense in
this case.
While here, fix accessing the major/minor versions via 'eps' even in the
64-bit entry point case (not causing any practical problem thus far as
the entry point length is greater than any SMBIOS revisions in
existence, so the comparison guarding the fixup would not pass).
MFC after: 2 weeks
Sponsored by: The FreeBSD Foundation
(cherry picked from commit f6cbd6b6d2ccd672e4807128ce0d07db333d4335)
Consistently with what the probe and attach methods are doing.
MFC after: 2 weeks
Sponsored by: The FreeBSD Foundation
(cherry picked from commit 3907feff7ab732b918ba3f5993e7a6718fffea14)
While here, de-indent most of the code by simply bailing out if 'addr'
is still 0 after the various detection methods have been tried.
Reviewed by: emaste, imp
MFC after: 2 weeks
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D49180
(cherry picked from commit 67d510f0c07afd89e51e337e5abec47f4483ecd9)
Additionally, on verbose boot, print the entry point revision as
a diagnostic/debugging help.
PR: 284460
Reviewed by: markj, imp (both older version)
MFC after: 2 weeks
Event: February src bug-busting session
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D49179
(cherry picked from commit e421a6615dc257b71ac5255336dddf5c99cd12aa)
When booted from BIOS (i.e., not EFI), also search for a 64-bit version
of the SMBIOS Entry Point.
This allows us to detect and report the proper SMBIOS version with
BIOSes that only provide the v3 table, as happens on Hetzner virtual
machines.
For machines that provide both, leverage the v3 table in priority
consistently with the EFI case.
PR: 284460
Reviewed by: markj, imp (both older version)
MFC after: 2 weeks
Relnotes: yes
Event: February src bug-busting session
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D49179
(cherry picked from commit bc7f6508363c5cf4544044e00bbaf71de8f0168d)
Get smbios working on arm64 where it seems to be
exclusively smbios version 3.x
The "interesting" thing here is that the smbios table seems to be
RAM in the EFI runtime services table. This makes it owned by "ram0",
and not io memory. That prevents bus_alloc_resource() from being able
to claim it, since ram0 already owns it. According to jhb, this is
how things are supposed to work. Eg, bus_alloc_resource() is meant
to be used with IO memory, not physical memory. Following his
suggestion, I converted the driver to simply use pmap_mapbios().
This is a prerequisite for getting IPMI to attach via the SSIF
attachment on arm64 servers, where all IPMI that I've seen
uses SSIF.
Note that this change is based on initial work by Allan Jude in
https://reviews.freebsd.org/D28739.
Reviewed by: imp
Sponsored by: Netflix, Ampere Computing LLC (D28739)
Differential Revision: https://reviews.freebsd.org/D42592
(cherry picked from commit ba0e4d7971e05ee64281a4fc49a2fb408c8ad816)
This call was meant to be the default case in the first place, but
somehow missed this.
Reported by: glebius
Fixes: 4021fa32d92d ("sound: Simplify pcm/feeder_mixer.c")
MFC after: 1 week
Sponsored by: The FreeBSD Foundation
(cherry picked from commit b6420b5ea5bcdeb859a2b3357e5dbaafe7aaff88)
Lock around uses of devclass_*() and replace leftover
CTLFLAG_NEEDGIANTs with CTLFLAG_MPSAFE.
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Reviewed by: imp, jhb
Differential Revision: https://reviews.freebsd.org/D46700
(cherry picked from commit 35400672df83e337f8792df1972a15003b603930)
Reduce ifdefs, and add aliases for the unsigned formats.
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Reviewed by: dev_submerge.ch
Differential Revision: https://reviews.freebsd.org/D48009
(cherry picked from commit a4aff024fd53a38ba08bbf5309589e1865ffe024)
Even though the OSS manual [1] advises against using AFMT_FLOAT, there
are applications that expect the sound driver to support it, and might
not work properly without it.
This patch adds AFMT_F32_LE|BE (as well as AFMT_FLOAT for OSS
compatibility) in sys/soundcard.h and implements AFMT_F32_LE|BE <->
AFMT_S32_LE|BE conversion functions. As a result, applications can
write/read floats to/from sound(4), but internally, because sound(4)
works with integers, we convert floating point samples to integer ones,
before doing any processing.
The reason for encoding/decoding IEEE754s manually, instead of using
fpu_kern(9), is that fpu_kern(9) is not supported by all architectures,
and also introduces significant overhead.
The IEEE754 encoding/decoding implementation has been written by Ariff
Abdullah [2].
[1] http://manuals.opensound.com/developer/AFMT_FLOAT.html
[2] https://people.freebsd.org/~ariff/utils/ieee754.c
PR: 157050, 184380, 264973, 280612, 281390
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Reviewed by: emaste
Differential Revision: https://reviews.freebsd.org/D47638
(cherry picked from commit e1bbaa71d62c8681a576f9f5bedf475c7541bd35)
Make vchanrate and vchanformat reflect the primary channel's software
buffer's rate and format respectively. Fix previous inconsistencies.
Get rid of the initializations in vchan_create() and move them to
chn_init().
Without the feeder_rate_round check in sysctl_dev_pcm_vchanrate(), we
can set the software rate to anything between feeder_rate_min and
feeder_rate_max. If we keep the check, however, the rate is limited to
whatever the driver's min/max is, which can be a problem if, for
example, the driver supports only a single rate, in which case we won't
be able to set anything other than the driver rate.
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D48961
(cherry picked from commit e372211be5c56e218e974a4478be9aa80bfca064)
Check if CTS bit is set in the mailbox message before waiting for ACK.
Otherwise ACK will never be received causing the function to timeout. Add
a note for ixgbe_write_mbx that it should be called while holding a lock.
Fixes: 6d243d2 ("net/ixgbe/base: introduce new mailbox API")
Cc: stable@dpdk.org
Signed-off-by: Norbert Ciosek <norbertx.ciosek@intel.com>
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Obtained from: DPDK (1f119e4)
(cherry picked from commit 1580f8d9c1740e0c54554e6c185573d34f2dcf76)
Summary:
The device IDs for these were in the driver's list of PCI ids to attach
to, but igc_set_mac_type() had never been setup to set the correct mac
type for these devices. Fix this by adding these IDs to the switch block
in order for them to be recognized by the driver instead of returning an
error.
This fixes the igc(4) attach for the I226-K LOM on the ASRock Z790
PG-ITX/TB4 motherboard, allowing it to be recognized and used.
Signed-off-by: Eric Joyner <erj@FreeBSD.org>
Reviewed by: kbowling@
Relnotes: yes
Differential Revision: https://reviews.freebsd.org/D49147
(cherry picked from commit 7ee310c80ea7b336972f53cc48b8c3d03029941e)
