Sysctl OIDs were internally stored in linked lists, triggering O(n^2)
behavior when userland iterates over many of them. The slowdown is
noticeable for MIBs that have > 100 children (for example, vm.uma). But
it's unignorable for kstat.zfs when a pool has > 1000 datasets.
Convert the linked lists into RB trees. This produces a ~25x speedup
for listing kstat.zfs with 4100 datasets, and no measurable penalty for
small dataset counts.
Bump __FreeBSD_version for the KPI change.
Sponsored by: Axcient
Reviewed by: mjg
Differential Revision: https://reviews.freebsd.org/D36500
While we deal with 0 returned, some drivers directly use and check for
DMA_MAPPING_ERROR. Add the case and check for both in dma_mapping_error().
MFC after: 1 week
Reviewed by: hselasky
Differential Revision: https://reviews.freebsd.org/D36686
Add a structure definition as well as a dummy dmi_walk for now
which returns an error as not supported. Our current dmi implementation
is special but does not give access to all details but rather only
information from kenv which does not suffice all use cases.
MFC after: 1 week
Reviewed by: hselasky
Differential Revision: https://reviews.freebsd.org/D36687
This matches the return type of pmap_mapdev/bios.
Reviewed by: kib, markj
Sponsored by: DARPA
Differential Revision: https://reviews.freebsd.org/D36548
Factor out parts of pcim_iomap_regions_request_all() into
pcim_iomap_regions() now needed for a driver.
Sponsored by: The FreeBSD Foundation
MFC after: 7 days
Reviewed by: hselasky
Differential Revision: https://reviews.freebsd.org/D36654
In addition to the ones added last year add more found in modern
drivers.
Sponsored by: The FreeBSD Foundation
MFC after: 7 days
Reviewed by: hselasky
Differential Revision: https://reviews.freebsd.org/D36656
Constify "*from" arguments and add __ioread32_copy() and
__ioread64_copy() based on the already existing implementations.
Sponsored by: The FreeBSD Foundation
MFC after: 7 days
Reviewed by: hselasky
Differential Revision: https://reviews.freebsd.org/D36657
Add the devres version dmam_alloc_coherent() of dma_alloc_coherent()
along with the ancillary free function.
Sponsored by: The FreeBSD Foundation
MFC after: 7 days
Reviewed by: hselasky
Differential Revision: https://reviews.freebsd.org/D36661
Add some more defines used in drivers to make it easier to compile.
MFC after: 7 days
Reviewed by: hselasky, emaste
Differential Revision: https://reviews.freebsd.org/D36660
Add PCI vendor IDs found in ath and mt76 drivers. This should make it
easier for me and others not having to re-define them locally.
MFC after: 7 days
Reviewed by: hselasky, emaste
Differential Revision: https://reviews.freebsd.org/D36659
Add NOPs for lockdep_{,un}register_key().
Sponsored by: The FreeBSD Foundation
MFC after: 7 days
Reviewed by: hselasky, emaste
Differential Revision: https://reviews.freebsd.org/D36658
Add #defines for PCI_DEVICE_ID and repoint the PCI_VENDOR_ID one.
Add dev_is_pci().
Add pcie_capability_clear_word() according to similar implementations.
Sponsored by: The FreeBSD Foundation
MFC after: 7 days
Reviewed by: hselasky, emaste
Differential Revision: https://reviews.freebsd.org/D36653
Add devm_kmemdup() as needed by a networking driver.
Sponsored by: The FreeBSD Foundation
MFC after: 7 days
eviewed by: hselasky, emaste
Differential Revision: https://reviews.freebsd.org/D36652
seq_printf() is defined in both spl_procfs_list.c and linux_seq_file.c .
Fix this by renaming the LinuxKPI ones and use macros to invoke the
correct function.
Reported by: jfree@
Differential Revision: https://reviews.freebsd.org/D35883
MFC after: 1 week
Sponsored by: NVIDIA Networking
This diff extends LinuxKPI to support simple attribute files in debugfs.
These simple attributes are an essential component for compiling drm-kmod
with CONFIG_DEBUG_FS enabled.
This will allow for easier graphics driver debugging using
Intel's igt-gpu-tools.
Reviewed by: hselasky
Differential Revision: https://reviews.freebsd.org/D35883
Sponsored by: Google, Inc. (GSoC 2022)
The API says that CMSG_SPACE should be used for msg_controllen, but in
practice the native ABI allows you to only use CMSG_LEN for the final
(typically only) control message, and real-world software does this,
including Wayland. For freebsd32, this is in practice mostly harmless,
since control messages are generally used to carry file descriptors,
which are already 4 bytes in size and thus no padding is needed, but
they can carry other quantities that may not result in an aligned
length. This was discovered after CheriBSD's freebsd64 equivalent was
updated to match the freebsd32 implementation, as that uses 8 byte
alignment which does break the file descriptor use case, and thus
Wayland.
This used to be addressed by aligning buflen before the first iteration,
but that allowed unwanted invalid inputs and was lost in 1b1428dcc8,
with no safer equivalent put in its place.
Reviewed by: brooks, kib, markj
Obtained from: CheriBSD
Fixes: 1b1428dcc8 ("Fix a TOCTOU vulnerability in freebsd32_copyin_control().")
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D36554
Change RB_COLOR_INSERT to take a parent parameter, to avoid looking up
a value already available. Make adjustments to a linux rbtree header,
which invokes it.
Reviewed by: alc, hselasky
Differential Revision: https://reviews.freebsd.org/D36114
Fix types for various struct fields (initially added as int).
Move structs and inline functions logically together, e.g., for wowlan.
Add more skeleton functions and #defines needed for iwlwifi d3.c
in the future.
Add struct ieee80211_vht_cap (without "_ie") to net80211 and remove
duplicate definitions in LinuxKPI headers now using net80211 structs.
For now leave ieee80211_ie_vhtcap in net80211. I am not sure yet if we
actually need it as such. That'll be cleaned up with more VHT updates
in net80211 in the future.
No functional changes in currently compiled code intended.
Try to implement ieee80211_action_contains_tpc() as I ran into it with
an older iwlwifi chipset. This depends on c994352a88.
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
The non-atomic versions are required by drm-510-kmod to build on 32-bit
architectures.
Approved by: hselasky, manu, tcberner (mentor)
Differential Revision: https://reviews.freebsd.org/D36507
Change RB_INSERT_COLOR and RB_REMOVE_COLOR so that the blocks of code
that are identical except for left and right being exchanged are made
only one block with a variable to indicate left- or right-handedness.
Rename RB macros so that those not intended for external use begin
with an underscore.
Add comments to the balancing code so that another might understand it.
Reviewed by: alc, kib
MFC after: 3 weeks
Differential Revision: https://reviews.freebsd.org/D36393
Update the mac80211 ops KPI for (*conf_tx), (*assign_vif_chanctx),
(*unassign_vif_chanctx), (*start_ap), and (*stop_ap), as well as
ieee80211_beacon_get_tim() and ieee80211_beacon_get_template().
Update in-tree drivers iwlwifi and rtw88 accordingly based on upstream
changes (as well as out-of-tree ones). This was triggered by trying to
synchronize more drivers to a common state.
MFC after: 1 week
There are drivers directly accessing napi->state testing for bits
(NAPI_STATE_SCHED encountered so far). Rename the internal _flags
struct field to state and expose our internal state flag bits along
with the one official aliased.
As I left in a comment, I wished Linux would hide these accesses
behind inline functions or by other means and not public expose
the implementation details.
MFC after: 1 week
A (so far out-of-tree) driver update needs
request_partial_firmware_into_buf(). Given we load the full .ko file
using firmware(9) just do that and copy the requeste data into the
buffer (rather than poissibly only reading portions of the firmware
file).
MFC after: 1 week
For one initialise vif->hw_queues later set in lkpi_80211_txq_tx_one()
for drivers using them. This may still need slightly more cleanup once
we get QUEUE_CONTROL-supporting drivers into the tree.
For the other rework and improve how we deal with tid/ac/txq in
lkpi_80211_txq_tx_one() and cleanup old comments and unused code.
This seems to reduce (remove) "Invalid TXQ id" reports from iwlwifi.
(The assumption is that the frame(s) triggering this WARN_ONCE or the
added FreeBSD specific logging were sent with lkpi_80211_mo_tx()).
Adjust the one logging from e674ddec0b
to IWL_DEBUG_TX so that now this is fixed it is also not always logged
anymore but leave it in case we need to further debug queues in the
future.
Tested by: pstef, Kevin Oberman (rkoberman gmail.com)
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
Continue what was started in 26a3694833
in iwlwifi and extend out internal implementation of
linuxkpi_ieee80211_get_tid() by an argument as to whether "no-QoS"
answers are acceptable. For the LinuxKPI ieee80211_get_tid() set
this to false as the Linux derived drivers seem to do extra checks
for the QoS-Data frame before acquiring the tid.
Add KASSERTs to enforce the extra argument.
This allows us to use the net80211 variant in LinuxKPI for other
means explicitly documenting that we do accept a IEEE80211_NONQOS_TID.
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
For consistency rename LKPI_SCAN_RUNNING adding a LHW_ prefix.
Add a local flag LKPI_LHW_SCAN_HW mirroring the net80211
IEEE80211_FEXT_SCAN_OFFLOAD flag.
Slightly simplify the code.
Overload (*ic_scan_curchan) and (*ic_scan_mindwell) so that we can
call the net80211 implementation in case of software scan but skip it
in case of full-offload scans.
Also add a bandaid to our (*ic_set_channel) implementation to not siwtch
channels if we have an active hw_scan running.
Obtained from: bz/wireless-dev
Sponsored by: The FreeBSD Foundation (partially)
MFC after: 4 days
Implement the mac80211 (*get_antenna) call and after checking any
antenna information present query the current configuration on startup
(both informations should be identical at this point in theory).
Both the wiphy variables and function call report a bitmask not a count.
Count the bits for net80211 for as long as we get away with just a
number in ic_[rt]xstream.
Sponsored by: The FreeBSD Foundation
MFC after: 4 days
Rename TRY_HW_CRYPTO to LKPI_80211_HW_CRYPTO for consitency and make
it compileable again in case someone wants to sit down and make it
work. It's probably not too much to do. Otherwise I might eventually
get around to it.
Obtained from: bz/wireless-dev
Sponsored by: The FreeBSD Foundation
MFC after: 4 days
Going through the Frame (Sub)types the "QOS Data" being called "QOS"
scheme leads to a naming conflict for QOS_CFPOLL and QOS_CFACKPOLL
(if added). Rename QOS* to QOS_DATA* to avoid the conflict and
to also better match the standards name.
No functional changes intended.
Sponsored by: The FreeBSD Foundation
MFC after: 5 days
Reviewed by: hselasky
Differential Revision: https://reviews.freebsd.org/D36409
Enabling other driver code found that the bssid in
struct ieee80211_bss_conf is not an array but expected to be
a const pointer (const, != NULL checks).
Adjust accordingly in the header and in the LinuxKPI compat code.
There initialization now needs to be a static array always present
as we need a value before we will have a BSS (node in scan_to_auth)
as the mac80211 driver (*handlers) are expecting the pointer to be
not NULL (copying without checks).
This is a pre-req to enable d3 (CONFIG_PM[_SLEEP]) in the future.
Tested by: Tomoaki AOKI (junchoon dec.sakura.ne.jp)
Tested by: Berislav Purgar (bpurgar gmail.com)
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
When a freebsd32 caller uses all or most allowed space for control
messages (MCLBYTES == 2K) then the message may no longer fit when
the messages are padded for 64-bit alignment. Historically we've just
shrugged and said there is no ABI guarantee. We ran into this on
CheriBSD where a capsicumized 64-bit nm would fail when called with more
than 64 files.
Fix this by not gratutiously capping size of mbuf data we'll allocate
to MCLBYTES and let m_get2 allocate up to MJUMPAGESIZE (4K or larger).
Instead of hard-coding a length check, let m_get2 do it and check for a
NULL return.
Reviewed by: markj, jhb, emaste
Sponsored by: DARPA, AFRL
Differential Revision: https://reviews.freebsd.org/D36322
This is a joint work with manu.
- fixed conditions in do_i2c_transfer and i2c_transfer as linux_i2cbb
does not set adapter->algo->master_xfer but does set
adapter->algo_data;
- fixed parent bus specification for linux_i2cbb driver module;
- actually implemented iicbb_transfer method;
- added iicbb_pre_xfer and iicbb_post_xfer methods;
- removed unnecessary and harmful delays (and other extra logic) from
iicbb methods as iicbb driver already has them;
- added setting of iicbb speed based on algo_data->udelay, so that iicbb
uses correct delays;
PR: 265920
Fixes: 1961a14a47 linuxkpi: Add i2c support
MFC after: 2 weeks
Sponsored by: Beckhoff Automation GmbH & Co. KG (manu's work)
RB_ROTATE_LEFT (and it symmetric twin) modify the rb-tree, adjusting
pointers so that what started as a proper tree ends up a proper
tree. When two consecutive rotations move the same node up the tree,
some of the pointers changed in the first rotation are immediately
changed again in the second - namely, the pointer from the rising node
to its new parent, and the pointer from that parent back to the rising
node. This change removes from RB_ROTATE macros the responsibility for
managing those two pointers, and leaves it to the code that calls for
rotations to fix up those pointers afterward. That drops a comparison
and a pair of assignments from every INSERT_COLOR or REMOVE_COLOR call
that ends in a double rotation.
A side-effect of this change is that the SWAP_CHILD macro must take as
a parameter a pointer to the node that is changing children, where it
is now computed from the old child. Since this macro is called in a
couple of places besides the RB_ROTATE macros, those calls are also
affected.
Reviewed by: alc
MFC after: 3 weeks
Differential Revision: https://reviews.freebsd.org/D36266
Linux is using 7 bit addressing while FreeBSD uses 8 bit addresses
internally, but i2c(8) uses 7 bit address.
This confused me when originally doing the code and I thought that
0x50 was the 8bit EDID address while it's the 7bit address and since
I did all my testing using this I didn't noticed the problem.
Reported by: avg
PR: 265920 (somewhat)
Rather than defining the same values in two places and having to do
conflict resulution on the name in LKPI, change the defines to an
enum in net80211. In addition to de-duplication this also gives us
value checks in certain cases.
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Reviewed by: adrian
Differential Revision: https://reviews.freebsd.org/D36250
Both are needed by drm-kmod
Obtained from: drm-kmod
Sponsored by: Beckhoff Automation GmbH & Co. KG
Differential Revision: //reviews.freebsd.org/D36212
