Since commit 19df0c5abcb9d4e9 ("LinuxKPI: make __kmalloc() play by the
rules"), kmalloc() systematically allocates contiguous physical memory,
as it should. However, kvzalloc() was left defined in terms of
kmalloc(), which makes it allocate contiguous physical memory too. This
is a too stringent restriction, as kvzalloc() is supposed to be a simple
page-zeroing wrapper around kvmalloc().
According to Linux's documentation ("memory-allocation.rst"), kvmalloc()
first tries to allocate contiguous memory, falling back to
non-contiguous one if that fails. Thus, callers are already supposed to
deal with the possibility of non-contiguous memory being returned.
Reviewed by: bz
Fixes: 19df0c5abcb9 ("LinuxKPI: make __kmalloc() play by the rules")
MFC after: 10 days
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D51247
(cherry picked from commit 986edb19a49c7d7d3050c759d9b0826283492ebf)
Forgotten on commit to main/-CURRENT:
PR: 277476
Given the internal field now stores a __wsum csum (after we added
the type) also make sure csum_unfold() returns a __wsum.
Sponsored by: The FreeBSD Foundation
Fixes: 59481c7db234
(cherry picked from commit 36ca21722c2700e00e41444a29aeabf246eea90d)
There's one case which needs skb_copy_header(); add a dummy function
for now until we get to the code which uses it to test an implementation.
Sponsored by: The FreeBSD Foundation
(cherry picked from commit 84c5998ccb0b3e8a401d1f2b5a32649e88847d63)
Restructure struct_group() into a wrapper around __struct_group().
Various callers have arguments annotated as /* no foo */ so try to
add all we can find.
Sponsored by: The FreeBSD Foundation
Reviewed by: dumbbell
Differential Revision: https://reviews.freebsd.org/D50157
(cherry picked from commit 54d0e1772fb6b83d72725764e0a5d3ead8337673)
ktime_get_boottime_seconds() is needed by an updated iwlwifi driver.
Sposored by: The FreeBSD Foundation
Reviewed by: dumbbell
Differential Revision: https://reviews.freebsd.org/D50005
(cherry picked from commit 93b82146c43b8ecad4fd7694ec1daf6d79cca65c)
Needed by mediatek/mt76 wireless driver.
While here adjust the way is_multicast_ether_addr() is implemented
to not look weird. [1]
Sponsored by: The FreeBSD Foundation
Suggested by: emaste [1]
Reviewed by: emaste
Differential Revision: https://reviews.freebsd.org/D50152
(cherry picked from commit a4618caa43c16e6e672c27dac1943fd833f8bd8f)
It may be that once we implement freeing of chained tables
_devcd_free_sgtable() will become our _lkpi_dev_coredumpsg_free()
but further investigations need to happen. For now make an
updated iwlwifi driver happy which should not need more.
Sponsored by: The FreeBSD Foundation
Reviewed by: dumbbell
Differential Revision: https://reviews.freebsd.org/D50006
(cherry picked from commit 71576ed24b56143429a3b6774a7862965c2dea88)
Start to sort through the netdevice features to match them with
FreeBSD. Annotate them a bit more verbose though the names are
pretty telling already.
While here adjust style(9) as well.
Lastly add the bit definitions for use with printf(9) %b as names
read easier than bitmasks. We will use that in LinuxKPI 802.11.
No functional changes.
Sponsored by: The FreeBSD Foundation
(cherry picked from commit 4b6b9c1368b346f7093153d14e7931403ac3202d)
In order to be able to use MODULE_DEVICE_TABLE() with multiple bus
attachments, factor out the bus-specfic MODULE_PNP_INFO() and place
it next to the structure defining the table.
As it turns out bnxt(4) has been using the MODULE_DEVICE_TABLE() with
PCI attachments for the "auxillary" bus so far. That makes little sense.
Define the MODULE_PNP_INFO() to nothing for that. We may consider
pulling these LinucKPI bits in semi-native drivers into LinuxKPI
one day as that route is not really sustainabke.
Sponsored by: The FreeBSD Foundation
Reviewed by: imp, dumbbell
Differential Revision: https://reviews.freebsd.org/D51049
(cherry picked from commit 2f5666c1727c949491f73e6c3277b7b542131714)
While we currently only use MODULE_DEVICE_TABLE() for pci (or so I
thought [*]), this may soon change. Remove a hard coded "pci" for
the bus and use the bus name passed in as _bus.
[*] see follow-up change
Sponsored by: The FreeBSD Foundation
Reviewed by: imp, dumbbell
Differential Revision: https://reviews.freebsd.org/D51048
(cherry picked from commit f58e032410c93d67c83d6bce8a3e6711f377a4be)
dma_map_resource() and dma_unmap_resource() functions are used to map
and unmap of memory-mapped IO resources so they can be accessed by
the device.
Required by drm-kmod v5.5 to be built.
While for drm-kmod 6.6-lts only amdgpu uses these in one place and
the code apparently has not been exercised a lot it has been around
like this for years. With changing the underlying implementations
bring it into the tree; should we find errors with it they can also
be fixed here.
Bump __FreeBSD version to be able to detect this change.
Obtained from: D30933 (by wulf)
Sponsored by: The FreeBSD Foundation
Reviewed by: dumbbell
Differential Revision: https://reviews.freebsd.org/D49625
(cherry picked from commit a02180cf60a6a0102669b678e9c81ad9f1aa4d91)
Rename linux_dma_unmap() to lkpi_dma_unmap(), and linux_dma_map_phys()
to lkpi_dma_map_phys() so that we get the full set of function
arguments (direction and attributes were missing).
Leave the old functions as wrappers as they are called from drm-kmod
linuxkpi/bsd still, and leaving them also allows us to MFC this
change.
Add missing bus_dmamap_sync() calls. Rather than inlining them
in each inline function push them down to the actual implementation.
From there do not inline them either but call the appropriate
function dealing with the sync so that we do not have the same
code splattered in many places.
Adhere to the DMA_ATTR_SKIP_CPU_SYNC attribute and skip synching
when requested.
A previous, less complete version of this change already allowed
me to load firmware on arm64 for iwlwifi (again).
There are four more places which are not currently done as the
functions which came with an OFED sync a long time ago seem to be
unused these days. Leaving a pr_debug("TODO") call there.
Also dma_[un]map_resource() when brought in from drm-kmod/linuxkpi/bsd
(D30933) should be adjusted to pass the full arguments as the amdgpu
callers are requesting to skip synching.
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D45294
(cherry picked from commit f99d393f7f787ff080c12b92243adcc310ac7c58)
Migrate most LinuxKPI 802.11 definitions for HE IEs to net80211.
During that process also properly define them as most of them only
had dummy values. Some of the definitions are sparse; that is, only
the bits used by drivers so far were listed and annotated with the
standards section.
There seems to be little point to mangle the names and have two copies
of all these bit field definitions. We can add "_S" (shift/mask)
variants to those we need in net80211 (if we do).
Sponsored by: The FreeBSD Foundation
Reviewed by: adrian
Differential Revision: https://reviews.freebsd.org/D50677
(cherry picked from commit 2ab7cbdc346e2a174a7bee4ce44c4f78b1ca093b)
In lkpi_sta_sync_ht_from_ni() we are casting the information element (IE)
data to this struct as net80211 does not save the mcs/rx_mask.
Not marking the structs packed results in the mcs information not being
copied correctly and rx_nss calculations possibly being off. Even worse,
at a later stage iwlwifi mvm/rs.c would extract the mcs values from the
same field to pass to the firmware which will than crash in an assert
as we would want to do HT with an empty MCS set.
Sponsored by: The FreeBSD Foundation
Discovered: while testing other features on arm64.
Approved by: re (cperciva)
(cherry picked from commit 6cbf8a04b1864d933506301cabe213a9c918d968)
(cherry picked from commit 9ccf7aeaa5dc5ceb3bf896d84ee50294b841043b)
Add more structs, fields, flags, defines.
Sponsored by: The FreeBSD Foundation
Approved by: re (cperciva)
(cherry picked from commit 675e6b1c68ba71af1853952f7df4d84399340009)
(cherry picked from commit 6a0dbc0a1246c9825f0892bb6dd957f5d081057a)
Define time64_t to int64_t for all supported architectures
unconditionally.
Sponsored by: The FreeBSD Foundation
Reviewed by: imp, markj, emaste
Differential Revision: https://reviews.freebsd.org/D50004
(cherry picked from commit 8b51cd07f69e9a5052f660d96a8feb7af4b387c1)
Use a macro to automatically gather the length of the array while
we can.
Sponsored by: The FreeBSD Foundation
Reviewed by: emaste
Differential Revision: https://reviews.freebsd.org/D50007
(cherry picked from commit f94d7319540b9b2256be60c2c666efe0b6e635ef)
Add pci_info() needed for a driver update. While here prefix
__VA_ARGS__ with ## for pci_err and pci_info in case we are only
passed a string without format arguments.
Sponsored by: The FreeBSD Foundation
Reviewed by: emaste
Differential Revision: https://reviews.freebsd.org/D50009
(cherry picked from commit e29d72ac3ddd9b97d945c5c64bc817e501bb113c)
An updated rtw88 driver requires minimal changes to leds.h. On
the positive side we can now remove the __DUMMY from the enum as we
actually have a valid field name.
Sponsored by: The FreeBSD Foundation
Reviewed by: emaste
Differential Revision: https://reviews.freebsd.org/D50002
(cherry picked from commit 69880fede78fc769ad417e5a984fb9fee7815d8b)
Some code relies on header pollution (or self-sustainability).
Add cleanup.h to mutex.h as that is one case it seems to be used with.
Sponsored by: The FreeBSD Foundation
Reviewed by: emaste
Differential Revision: https://reviews.freebsd.org/D50001
(cherry picked from commit 059136a95aca2e550b62ef40ab25a184d8141124)
This is needed for rtw89 updates.
Sponsored by: The FreeBSD Foundation
Reviewed by: emaste, zarychtam_plan-b.pwste.edu.pl
Differential Revision: https://reviews.freebsd.org/D50000
(cherry picked from commit c8e2458de590d7427ef3311412b8ddbc326cd3b5)
Move some structs into the appropriate header to be visible.
Add new fields to structs and enums.
Remove arguments from two functions (one function currently unused
by drivers in the tree, for the other the argument was unused).
Adjust the iwlwifi accordingly. This is in preparation for new
driver versions to allow a smooth transition.
Sponsored by: The FreeBSD Foundation
(cherry picked from commit e6010da63903cc6d9aac0077ab861cfd6bae2aa7)
TKIP has a special key buffer. While keylen only identifies the
key length, the two MIC are appended. The LinuxKPI offsets for
these were unfortunately never set correct which lead to constant
Micheal counter measures as the MIC never was correct when calculations
were offloaded to firmware. This is the first half of the fix.
Sponsored by: The FreeBSD Foundation
(cherry picked from commit a23908728efce08cfd10b13ed34de1361d0a980f)
Fix the types of a few remaining fields.
Shuffle the fields around so the important ones align to 64byte on a
64bit platform.
Sponsored by: The FreeBSD Foundation
(cherry picked from commit 59481c7db2340c9e64f9d3d9d05dafc2d72c157e)
Initially we saved the exact allocation length for contigfree(9).
contigfree can now be replaced by free(9) so there is no need to
remember that value anymore. Removing it also simplifies the
linuxkpi_kfree_skb() code.
Sponsored by: The FreeBSD Foundation
(cherry picked from commit 73e3969fd435eea296b44f3bd40ff76c042c9a36)
Seems this is the correct type for the csum field in an sk_buff.
Add it as uint32_t __bitwise__.
Sponsored by; The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D49831
(cherry picked from commit c925e0a77606d07e4d93204a1d35530176c4b344)
Sort out some (though not yet all) changes needed for CONFIG_PM_SLEEP
and respectively WoWLAN support (so we can compile it in). This is
just one further step towards proper suspend/resume support.
Sponsored by: The FreeBSD Foundation
(cherry picked from commit b42c339c59fd00f2d97f6854b17b54b340019d0c)
Define the last missing PN length and for consistency use them
in struct ieee80211_key_seq (even though a 6 is a 6 is a 6).
Sponsored by: The FreeBSD Foundation
(cherry picked from commit c8a6676eb652b59fee545dc3e24ffec21bd0cd8d)
Move the implementation of print_hex_dump() into linux_compat.c as
lkpi_hex_dump() taking an extra function pointer and argument.
Add two internal wrappers for printf and sbuf_printf to get a common
function definition (sbuf_printf takes the extra argument).
Use these to implement print_hex_dump() and the newly added
seq_hex_dump().
This allows us to re-use the same implementation of print_hex_dump()
for both functions without duplicating the code.
Initial implementation: D49381 by dumbbell
Sponsored by: The FreeBSD Foundation
Reviewed by: dumbbell
Differential Revision: https://reviews.freebsd.org/D49637
(cherry picked from commit 4359672e656af5e6758d0aae147212e58f445f5b)
Move kfree() into slab.c as an implementation and hide the private
function linux_kfree_async() entirely.
Remove a ; at the end of a define and sort some defines into place.
Remove extern from function declarations and move the closer to where
they belong.
Sort the functions into "base allocator/free" functions--these have
an implementation in slab.c and are ensuring contiguous physical
memory allocations.
Followed by inline functions using these base allocators to implement
their functionality; vmalloc/kvalloc, and misc functions.
Sponsored by: The FreeBSD Foundation
Reviewed by: dumbbell
Differential Revision: https://reviews.freebsd.org/D49572
(cherry picked from commit a3e6f97bf57c1d323487d07369aec66542f995ce)
Implement krealloc_array() using krealloc(). Implement krealloc()
doing the various size checks ourselves and use realloc() or kmalloc()
depending on old and new allocation sizes.
This way we can ensure that allocated memory stays physically contiguous.
Sponsored by: The FreeBSD Foundation
Suggested by: jhb (see D46657)
Reviewed by: jhb, markj
Differential Revision: https://reviews.freebsd.org/D49571
(cherry picked from commit 1c95d401ebe5075ebb38b57638830713a496f107)
With mallocarray() we cannot guarantee that any size larger than
PAGE_SIZE will be contiguous. Switch kmalloc_array() and
kmalloc_array_node() to use __kmalloc()/lkpi___kmalloc_node() as their
underlying implementation which now does provide that guarantee.
Likewise adjust kcalloc_node() to use kmalloc_array_node().
This means we only have two (plain + _node) underlying allocation
routines for the entire category of functions.
Also adjust kvmalloc() and kvmalloc_array() to be a "mirrored"
implementation to their non-v counterparts. These may return
non-contiguous memory so can use malloc().
Sponsored by: The FreeBSD Foundation
Reviewed by: jhb
Extra thanks to: jhb for helping sorting this out
Differential Revision: https://reviews.freebsd.org/D46657
(cherry picked from commit 1c81ebec74d8a71c770f7835d3bc6e02c13467a0)
This is to fix slowdowns with drm-kmod that get worse over time as
physical memory become more fragmented (and probably also depending on
other factors).
Based on information posted in this bug report:
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=277476
By default, linux_alloc_pages() retries failed allocations by calling
vm_page_reclaim_contig() to attempt to free contiguous physical memory
pages. vm_page_reclaim_contig() does not always succeed and calling it
can be very slow even when it fails. When physical memory is very
fragmented, vm_page_reclaim_contig() can end up being called (and
failing) after every allocation attempt. This could cause very
noticeable graphical desktop hangs (which could last seconds).
The drm-kmod code in question attempts to allocate multiple contiguous
pages at once but does not actually require them to be contiguous. It
can fallback to doing multiple smaller allocations when larger
allocations fail. It passes alloc_pages() the __GFP_NORETRY flag in this
case.
This patch makes linux_alloc_pages() fail early (without retrying) when
this flag is passed.
[olce: The problem this patch fixes is longer and longer GUI freezes as
a machine's memory gets filled and becomes fragmented, when using amdgpu
from DRM kmod 5.15 and DRM kmod 6.1 (DRM kmod 5.10 is unaffected; newer
Linux kernel introduced an "optimization" by which a pool of pages is
filled preferentially with contiguous pages, which triggered the problem
for us). The original commit message above evokes freezes lasting
seconds, but I occasionally witnessed some lasting tens of minutes,
rendering a machine completely useless.
The patch has been reviewed for its potential impacts to other LinuxKPI
parts and our existing DRM kmods' code. In particular, there is no
other user of __GFP_NORETRY/GFP_NORETRY with Linux's alloc_pages*()
functions in our tree or DRM kmod ports.
It has also been tested extensively, by me for months against 14-STABLE
and sporadically on -CURRENT on a RX580, and by several others as
reported below and as is visible in more details in the quoted bugzilla
PR and in the initial drm-kmod issue at
https://github.com/freebsd/drm-kmod/issues/302, on a variety of other
AMD GPUs (several RX580, RX570, Radeon Pro WX5100, Green Sardine 5600G,
Ryzen 9 4900H with embedded Renoir).]
PR: 277476
Reported by: Josef 'Jeff' Sipek <jeffpc@josefsipek.net>
Reviewed by: olce
Tested by: many (olce, Pierre Pronchery, Evgenii Khramtsov, chaplina, rk)
MFC after: 2 weeks
Relnotes: yes
Sponsored by: The FreeBSD Foundation (review and part of testing)
(cherry picked from commit 718d1928f8748fe4429c011296f94f194d63c695)
This fixes a black screen issue with the i915 DRM driver from Linux v6.8
Fixes: c4e0746e7d ("LinuxKPI: Add helper macros IS_ALIGNED and DIV_ROUND_DOWN_ULL.")
Pull request: https://github.com/freebsd/freebsd-src/pull/1612
(cherry picked from commit ef037a028af7cc331e36ca6c8e2eb5612cdb1130)
Make a pass over skbuff.h:
- implement some missing bits,
- sprinkle some const,
- add locking and read/write_once calls as needed to provide
synchronization as expected by Linux,
- fix some typos,
- remove return from void functions,
- adjust tracing macros.
Sponsored by: The FreeBSD Foundation
PR: 283903 (rtw88 skb leak)
Tested by: Guillaume Outters (guillaume-freebsd outters.eu)
Tested by: oleg.nauman gmail.com
Differential Revision: https://reviews.freebsd.org/D49101
(cherry picked from commit 2ab4a41956159e7c974979693cb0b13cf552128e)
Implement cfg80211_chandef_create() to work with HT. Update enum
with HT channel types. When calling the function from LinuxKPI 802.11
code, pass in NL80211_CHAN_HT20 if HT is supported rather than NO_HT.
Sponsored by: The FreeBSD Foundation
(cherry picked from commit 11450726d3cf8d76d2113d7cabe995f8ef31ec5a)
Implement the mac80211 (*sta_statistics) to obtain station statistics.
Fix up struct station_info for that.
Add supporting funtions to decode station_info and rate_info_bw fields
to make the output more usable to human beings.
Add a per-VIF (VAP) sysctl to export that information. This is mostly
used for now to see what the driver/fw thinks about its TX rate.
We should later gather this information in different ways and pass it
to net80211 to export it to user space so we can finally have nice
station information (also for the non-AP side).
Sponsored by: The FreeBSD Foundation
(cherry picked from commit 408394185fe9be0c158189b7671bde316ac14ac9)
rtw88 seems to have an skb leak still.
Add a WARN_ONCE to __skb_queue_purge() just to make sure there is no
glitch due to missing locking.
Also add a rolling error reporting for skb allocation failures in
LinuxKPI to gather some more information if possible about queue
states.
This is a corrected version of what was initially part of D48474.
Sponsored by: The FreeBSD Foundation
(cherry picked from commit bcf1d8ee355213eef4a125c0b8518f1cb4f35df4)
(cherry picked from commit 1432e0f20ccde8eaa23d3d3e20fca2989cd344b6)
Given I was looking at the struct make it more packed at the beginning
at least. In fact it did not shrink but the tx_time_est got expanded.
Sponsored by: The FreeBSD Foundation
(cherry picked from commit 470aaf42c3c0661821a8744ae7c29b096adb4aa3)
Convert a lot more LinuxKPI rx_status fields to net80211 rx_stats
bits for as much as we can see fit. Factor the entire logic out
into its own function as it got quite long.
Now only net80211 needs to start using more of these values and
report them.
Also fix some related fields and struct definitions in LinuxKPI.
Sponsored by: The FreeBSD Foundation
(cherry picked from commit 49010ba706d39fba0d8bc128d96e6122e65eb8ad)
Add the mask and shift for the VHT Extended NSS BW Support field.
Document them in net80211 and further related bitmasks in LinuxKPI.
Sponsored by: The FreeBSD Foundation
Reviewed by: adrian
Differential Revision: https://reviews.freebsd.org/D48975
(cherry picked from commit 75fb66d8241d0487baf482c975c1de3b55d9a315)
net80211 node ni_chw currently encodes the channel width as Mhz number.
LinuxKPI 802.11 uses enum ieee80211_sta_rx_bw for the same.
Rather than keeping the "20" and "40" throughout the code (eventually
expanded to 80/160/320) switch them over to use the enum throughout.
Once we get to 320Mhz channel widths we would otherwise also need to
extend the uint8_t in struct ieee80211_node; making
enum ieee80211_sta_rx_bw __packed allows us for lots more channel
widths without breaking the KBI.
Sponsored by: The FreeBSD Foundation
Reviewed by: adrian
Differential Revision: https://reviews.freebsd.org/D47891
(cherry picked from commit ca389486a9599768e0ba69dca13c208020623083)
(cherry picked from commit 2c8b0d6205f6f98855773e3a82640b50abb2f2f6)
Linux has a header file that defines an ilog2 function and some simple
functions/macros that use it: roundup_pow_of_two, is_power_of_2,
rounddown_pow_of_two, and order_base_2. This change moves three of
those simple functions (all but is_power_of_2) from linuxkpi to
libkern. It also deletes a few implementations of these functions
that have previously been copied into code for various device drivers,
so that they can use the libkern version. The is_power_of_2 macro was
not moved because powerof2 in param.h provides almost the same service
already (except that they disagree about whether 0 is a power of two).
Since the linux definitions of these functions were copied into
FreeBSD 11 years ago, linux has improved them, and this change
provides those improvements. In particular, a giant table of log
values for evaluating ilog2 for constant values is no longer
necessary.
Reviewed by: alc, markj (previous version)
Differential Revision: https://reviews.freebsd.org/D45536
(cherry picked from commit c8b0c33b03ac072413b27bed2bdae2ae27426f3a)
The kernel source contains several definitions of an ilog2 function;
some are slower than necessary, and one of them is incorrect.
Elimininate them all and define an ilog2 macro in libkern to replace
them, in a way that is fast, correct for all argument types, and, in a
GENERIC kernel, includes a check for an invalid zero parameter.
Folks at Microsoft have verified that having a correct ilog2
definition for their MANA driver doesn't break it.
Reviewed by: alc, markj, mhorne (older version), jhibbits (older version)
Differential Revision: https://reviews.freebsd.org/D45170
Differential Revision: https://reviews.freebsd.org/D45235
(cherry picked from commit b0056b31e90029553894d17c441cbb2c06d31412)
No functional change (intended).
Reviewed by: kib
MFC after: 4 days
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D48497
(cherry picked from commit 9b7157c690723279d448fc7ea9c1f9374ad4511a)
THIS_MODULE is used to differentiate modules on Linux. We currently
completely stub out any Linux struct module usage, but THIS_MODULE
is still used to populate the "owner" fields of various drivers.
Even though we don't actually dereference these "owner" fields they
are still used by drivers to check if devices/dmabufs/etc come
from different modules. For example, during DRM GEM import some
drivers check if the dmabuf's owner matches the dev's owner. If
they match because they are both NULL drivers may incorrectly think
two resources come from the same module.
This adds a general purpose __this_linker_file which will point to
the linker file of the module that uses it. We can then use that
pointer to have a valid value for THIS_MODULE.
Reviewed by: bz, jhb
Differential Revision: https://reviews.freebsd.org/D44306
(cherry picked from commit 28a59100b54ff245a45fbd328266266f1c14eb8c)
