Amdgpu driver does a lot of memory allocations in FPU-protected sections
of code for certain display cores, e.g. for DCN30. This does not work
currently on FreeBSD as its malloc function can not be run within a
critical section. Allocate memory for FPU context to overcome such
restriction.
Sponsored by: Serenity Cyber Security, LLC
Reviewed by: manu (previous version), markj
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D42822
It does a Read-Modify-Write operation using clear and set bitmasks on
PCI Express Capability Register at pos. As certain PCI Express
Capability Registers are accessed concurrently in RMW fashion, hence
require locking which is handled transparently to the caller.
Sponsored by: Serenity CyberSecurity, LLC
Reviewed by: manu, bz
MFC after: 1 week
MFC TODO: Move pcie_cap_lock to bottom to preserve KBI compatibility
Differential Revision: https://reviews.freebsd.org/D42821
and bitmap_shift_right() functions to linux/bitmap.h
They perform calculation of two bitmaps intersection,
copying the contents of u32 array of bits to bitmap and
logical right shifting of the bits in a bitmap.
Sponsored by: Serenity Cyber Security, LLC
Reviewed by: manu
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D42812
Cancel a work not waiting for it to finish.
Sponsored by: Serenity Cyber Security, LLC
Reviewed by: manu, kib
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D42811
Linux uses it only if SHRINKER_DEBUG config option is enabled. Ignore it.
Sponsored by: Serenity Cyber Security, LLC
Reviewers: manu, bz
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D42809
ida_alloc_min() allocates an unused ID. between min and INT_MAX.
While here allow end parameter of ida_simple_get() be larger than
INT_MAX. Linux caps the value to INT_MAX.
Sponsored by: Serenity Cyber Security, LLC
Reviewers: manu, bz
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D42806
It is implemented like ktime_get_raw_ns() function but uses less
precise getnanouptime() call.
Sponsored by: Serenity Cyber Security, LLC
Reviewed by: manu, bz
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D42804
And move it along BUILD_BUG family in to
linux/build_bug.h to match Linux.
Sponsored by: Serenity Cyber Security, LLC
Reviewed by: manu, bz
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D42800
sys/rman.h defines `resource` structure which conflicts with the Linux
structure of the same name. To fix that remove reference to sys/rman.h
from linux/pci.h and move resource management code to linux_pci.c.
Update consumers which were depending on linux/pci.h pollution.
No functional changes intended.
Sponsored by: Serenity Cyber Security, LLC
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D42792
[Why]
The amdgpu DRM driver started to use it in Linux 5.18.
Reviewed by: manu
Approved by: manu
Differential Revision: https://reviews.freebsd.org/D43020
[Why]
The i915 DRM driver in Linux 5.18 relies on this indirect include.
Reviewed by: manu
Approved by: manu
Differential Revision: https://reviews.freebsd.org/D43030
[Why]
This header and str_yes_no() are now used by the i915 DRM driver in
Linux 5.18.
Reviewed by: manu
Approved by: manu
Differential Revision: https://reviews.freebsd.org/D43014
[Why]
This callback is being used by the amdgpu DRM driver in Linux 5.18.
[How]
The callback is called after `suspend_late()`.
Reviewed by: emaste, manu
Approved by: emaste, manu
Differential Revision: https://reviews.freebsd.org/D43029
[Why]
The passed structure may not have a `.store` field. This is the case in
the amdgpu DRM driver starting with Linux 5.18.
Reviewed by: manu
Approved by: manu
Differential Revision: https://reviews.freebsd.org/D43023
[Why]
The amdgpu DRM driver started to use it in Linux 5.18.
[How]
The function is a no-op as I'm not sure how to implement this with
sysctls yet.
Reviewed by: manu
Approved by: manu
Differential Revision: https://reviews.freebsd.org/D43022
[Why]
The amdgpu DRM driver started to use it in Linux 5.18.
Reviewed by: manu
Approved by: manu
Differential Revision: https://reviews.freebsd.org/D43021
[Why]
`linux_compat.c` is already too long. I will need to add `struct kset`
in a follow-up commit, so let's move the existing `struct kobject` code
to its own file.
Reviewed by: manu
Approved by: manu
Differential Revision: https://reviews.freebsd.org/D43019
[Why]
This is consistent with Linux.
[How]
The definition is moved from <linux/shmem_fs.h> to <linux/pagemap.h> and
the latter is included from the former. This is how it is done on Linux.
Prototypes are also expanded with argument names. I got a build failure
in the DRM 5.18 drivers because the compiler considered that the
`pgoff_t` argument was there twice.
Reviewed by: manu
Approved by: manu
Differential Revision: https://reviews.freebsd.org/D43018
This change seems to break some drivers such as the mlx5*(4) drivers.
As kib@ says:
> According to the 'official' Linux kernel documentation, the GFP_KERNEL
> flag implies sleepable context.
It was introduced while working on the new vt(4)/DRM integration [1].
During this work, doing sleepable allocations broke vt(4) and the DRM
drivers. However, I made further improvements and some locking-related
fixed to the new integration without revisiting the need for it.
After more testing, the improvements to the integration mentionned above
seems to make the change to `GFP_KERNEL` unneeded now. I can thus
revert it to restore expectations of other drivers.
This reverts commit 14dcd40983.
[1] https://github.com/freebsd/drm-kmod/pull/243
Reviewed by: kib
Approved by: kib
Differential Revision: https://reviews.freebsd.org/D42962
Fix defines and structures to use proper types.
Bring in basic ni->sta synchronization, some channel width handling,
and overload the net80211 functions so that we can talk to
driver/firmware to setup parameters. We will likely not need one
or two of those but it is good for tracing currently.
Cover HT and bits of VHT code in LinuxKPI behind apropriate #ifdef
which are currently not enabled (like LKPI_80211_HW_CRYPTO) until
confirmed to work.
Last, IEEE80211_AMPDU_RX_START made some firmware unhappy.
This will allow others to work on it and test as well.
Sponsored by: The FreeBSD Foundation
MFC after: 10 days
Just like it was done for accept(2) in cfb1e92912, use same approach
for two simplier syscalls that return socket addresses. Although,
these two syscalls aren't performance critical, this change generalizes
some code between 3 syscalls trimming code size.
Following example of accept(2), provide VNET-aware and INVARIANT-checking
wrappers sopeeraddr() and sosockaddr() around protosw methods.
Reviewed by: tuexen
Differential Revision: https://reviews.freebsd.org/D42694
[Why]
Some files in DRM rely on this indirect include to use `struct rb_*`.
Reviewed by: manu
Approved by: manu
Differential Revision: https://reviews.freebsd.org/D42835
Apply the following automated changes to try to eliminate
no-longer-needed sys/cdefs.h includes as well as now-empty
blank lines in a row.
Remove /^#if.*\n#endif.*\n#include\s+<sys/cdefs.h>.*\n/
Remove /\n+#include\s+<sys/cdefs.h>.*\n+#if.*\n#endif.*\n+/
Remove /\n+#if.*\n#endif.*\n+/
Remove /^#if.*\n#endif.*\n/
Remove /\n+#include\s+<sys/cdefs.h>\n#include\s+<sys/types.h>/
Remove /\n+#include\s+<sys/cdefs.h>\n#include\s+<sys/param.h>/
Remove /\n+#include\s+<sys/cdefs.h>\n#include\s+<sys/capsicum.h>/
Sponsored by: Netflix
... instead of `M_WAITOK`.
[Why]
The reason is that in some places in the DRM drivers (in particular, the
framebuffer management code), kmalloc() is called from a non-sleepable
context, such as after a call to mtx_lock(8) with an MTX_DEF mutex.
If `GFP_KERNEL` is defined as `M_WAITOK`, we hit an assertion from
witness(4).
[How]
The definition of `GFP_KERNEL` is changed to `M_NOWAIT`. This means that
callers should verify the return value of kmalloc(). Fortunately, this
is always the case in Linux.
Reviewed by: bz, emaste, manu
Approved by: manu
Differential Revision: https://reviews.freebsd.org/D42054
Substituting 'uma_align_cache' by the appropriately named accessor
uma_get_cache_align_mask() made apparent that dma_get_cache_alignment()
was off by one, since it was defined to be the mask derived from the
alignment value.
Reviewed by: markj, bz
MFC after: 2 weeks
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D42264
Create the uma_get_cache_align_mask() accessor and put it in a separate
private header so as to minimize namespace pollution in header/source
files that need only this function and not the whole 'uma.h' header.
Make sure the accessors have '_mask' as a suffix, so that callers are
aware that the real alignment is the power of two that is the mask plus
one. Rename the stem to something more explicit. Rename
uma_set_cache_align_mask()'s single parameter to 'mask'.
Hide 'uma_align_cache' to ensure that it cannot be set in any other way
then by a call to uma_set_cache_align_mask(), which will perform sanity
checks in a further commit. While here, rename it to
'uma_cache_align_mask'.
This is also in preparation for some further changes, such as improving
the sanity checks, eliminating internal resolving of UMA_ALIGN_CACHE and
changing the type of the 'uma_cache_align_mask' variable.
Reviewed by: markj, kib
MFC after: 2 weeks
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D42258
[Why]
This function takes an offset and a length as argument, not a physical
address and a number of pages.
This misuse caused the `set_memory_*()` and
`arch_io_reserve_memtype_wc()` functions to return EINVAL.
Another problem was the fact that they returned errors as a positive
integer, whereas Linux uses negative integers.
[How]
Physical addresses and number of pages are converted to offset+length in
the `set_memory_*()` functions.
`arch_io_reserve_memtype_wc()` now calls `pmap_change_attr()` directly
instead of using `set_memory_wc()`.
Reviewed by: manu
Approved by: manu
Differential Revision: https://reviews.freebsd.org/D42053
It seems the mac_header can be set to offset 0 for frames received.
Remove the warning splattering messages to the console for each packet.
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
So we can use it in non-linuxkpi sources.
Reviewed by: emaste, mmel
Differential Revision: https://reviews.freebsd.org/D41767
Sponsored by: Beckhoff Automation GmbH & Co. KG
Commit 3f686532c9 tried to fix an issue with not properly starting
at the first page in the sg list to prevent a panic. This worked but
with the side effect of incrementing "s" during the final iteration
causing it to be NULL since the list had ended.
In cases non-DEBUG kernels this causes a panic with drm-5.15, since
"s" is NULL when we later pass it to sg_mark_end().
This change decouples the iteration sg from the return value so that
it is never incremented past the final page in the chain.
MFC after: 3 days
Reviewed by: manu
Differential Revision: https://reviews.freebsd.org/D41574
This is a set of updates of the mt76 driver based on wireless-testing
(wt-2023-05-11) 711dca0ca3d77414f8f346e564e9c8640147f40d (after v6.4-rc1),
This adds support for mt7996 as well.
(wt-2023-06-09) 7bd20e011626ccc3ad53e57873452b1716fcfaaa (after v6.4-rc5),
(wt-2023-07-24) 62e409149b62a285e89018e49b2e115757fb9022 (after v6.5-rc3),
(wt-2023-08-06) 2a220a15be657a24868368892e3e2caba2115283 (after v6.5-rc4).
The current version of LinuxKPI lacks support for "page pool" which
needs enhancing and updating a decade or so old shortcut mapping
struct page directly to struct vm_page.
MFC after: 20 days
