snd_ctl_new1() can return NULL when memory allocation fails. The
ice1712 driver calls snd_ctl_new1() without checking the return value
before dereferencing the pointer in multiple places (ice1712.c,
ice1724.c, aureon.c), which can lead to NULL pointer dereferences.
Add NULL checks after snd_ctl_new1() calls and return -ENOMEM if any
fails.
Assisted-by: Opencode:DeepSeek-V4-Flash
Cc: stable@vger.kernel.org
Fixes: b9a4efd61b ("ALSA: ice1712,ice1724: fix the kcontrol->id initialization")
Signed-off-by: Zhao Dongdong <zhaodongdong@kylinos.cn>
Link: https://patch.msgid.link/tencent_42E5E2AB1B6A5101F7EE8C2117F1F687BB07@qq.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
snd_ctl_new1() can return NULL when memory allocation fails.
snd_gf1_pcm_volume_control() does not check the return value before
dereferencing kctl->id.index, which can lead to a NULL pointer
dereference.
Add a NULL check after snd_ctl_new1() and return -ENOMEM if it fails.
Assisted-by: Opencode:DeepSeek-V4-Flash
Cc: stable@vger.kernel.org
Fixes: c5ae57b1bb ("ALSA: gus: Fix kctl->id initialization")
Signed-off-by: Zhao Dongdong <zhaodongdong@kylinos.cn>
Link: https://patch.msgid.link/tencent_F644A3DCAD32945D62DB2FEEBE8A996F6809@qq.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
When a device's default substream attaches to an identity domain, the SMMU
driver currently sets the device's STE between two modes:
Mode 1: Cfg=Translate, S1DSS=Bypass, EATS=1
Mode 2: Cfg=bypass (EATS is ignored by HW)
When there is an active PASID (non-default substream), mode 1 is used. And
when there is no PASID support or no active PASID, mode 2 is used.
The driver will also downgrade an STE from mode 1 to mode 2, when the last
active substream becomes inactive.
However, there are PCIe devices that demand ATS to be always on. For these
devices, their STEs have to use the mode 1 as HW ignores EATS with mode 2.
Change the driver accordingly:
- always use the mode 1
- never downgrade to mode 2
- allocate and retain a CD table (see note below)
Note that these devices might not support PASID, i.e. doing non-PASID ATS.
In such a case, the ssid_bits is set to 0. However, s1cdmax must be set to
a !0 value in order to keep the S1DSS field effective. Thus, when a master
requires ats_always_on, set its s1cdmax to at least 1, meaning that the CD
table will have a dummy entry (SSID=1) that will never be used.
Now for these devices, arm_smmu_cdtab_allocated() will always return true,
v.s. false prior to this change. When its default substream is attached to
an IDENTITY domain, its first CD is NULL in the table, which is a totally
valid case. Thus, add "!master->ats_always_on" to the condition.
Reviewed-by: Jonathan Cameron <jonathan.cameron@huawei.com>
Tested-by: Nirmoy Das <nirmoyd@nvidia.com>
Acked-by: Nirmoy Das <nirmoyd@nvidia.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Some NVIDIA GPU/NIC devices, though they don't implement CXL config space,
have many CXL-like properties. Call this kind "pre-CXL".
Similar to CXL.cache capability, these pre-CXL devices also require the ATS
function even when their RIDs are IOMMU bypassed, i.e. keep ATS "always on"
v.s. "on demand" when a non-zero PASID line gets enabled in SVA use cases.
Introduce pci_dev_specific_ats_required() quirk function to scan a list of
IDs for these devices. Then, include it in pci_ats_required().
Suggested-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Nirmoy Das <nirmoyd@nvidia.com>
Tested-by: Nirmoy Das <nirmoyd@nvidia.com>
Reviewed-by: Jonathan Cameron <jonathan.cameron@huawei.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Reviewed-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
snd_ctl_new1() can return NULL when memory allocation fails.
snd_es1938_mixer() does not check the return value before dereferencing
the pointer, which can lead to a NULL pointer dereference.
Add a NULL check after snd_ctl_new1() and return -ENOMEM if it fails.
Assisted-by: Opencode:DeepSeek-V4-Flash
Cc: stable@vger.kernel.org
Fixes: 1da177e4c3 ("Linux-2.6.12-rc2")
Signed-off-by: Zhao Dongdong <zhaodongdong@kylinos.cn>
Link: https://patch.msgid.link/tencent_E0DC65165FDF2C8982BAFB6794B854B53B0A@qq.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Controlled by IOMMU drivers, ATS can be enabled "on demand", when a given
PASID on a device is attached to an I/O page table. This is working, even
when a device has no translation on its RID (i.e., RID is IOMMU bypassed).
However, certain PCIe devices require non-PASID ATS on their RID even when
the RID is IOMMU bypassed. Call this "ATS always on" in IOMMU term.
For example, CXL spec r4.0 notes in sec 3.2.5.13 Memory Type on CXL.cache:
"To source requests on CXL.cache, devices need to get the Host Physical
Address (HPA) from the Host by means of an ATS request on CXL.io."
In other words, the CXL.cache capability requires ATS; otherwise, it can't
access host physical memory.
Introduce a new pci_ats_required() helper for the IOMMU driver to scan a
PCI device and shift ATS policies between "on demand" and "always on".
Add the support for CXL.cache devices first. Pre-CXL devices will be added
in quirks.c file.
Note that pci_ats_required() validates against pci_ats_supported(), so we
ensure that untrusted devices (e.g. external ports) will not be always on.
This maintains the existing ATS security policy regarding potential side-
channel attacks via ATS.
Cc: linux-cxl@vger.kernel.org
Suggested-by: Vikram Sethi <vsethi@nvidia.com>
Suggested-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Jonathan Cameron <jonathan.cameron@huawei.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Tested-by: Nirmoy Das <nirmoyd@nvidia.com>
Acked-by: Nirmoy Das <nirmoyd@nvidia.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Reviewed-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Smatch complains about the NULL check on "iommu" because list_entry()
can't be NULL. Clean up this code by using list_for_each_entry().
Signed-off-by: Dan Carpenter <error27@gmail.com>
Reviewed-by: Benjamin Gaignard <benjamin.gaignard@collabora.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
When -Wformat-security is enabled, it catches a call to
iommu_device_sysfs_add() that passes a string variable in
place of a format:
drivers/iommu/vsi-iommu.c: In function 'vsi_iommu_probe':
drivers/iommu/vsi-iommu.c:717:9: error: format not a string literal and no format arguments [-Werror=format-security]
717 | err = iommu_device_sysfs_add(&iommu->iommu, dev, NULL, dev_name(dev));
| ^~~
Pass this indirectly using "%s" as the format instead.
Fixes: 917ace84b7 ("iommu: Add verisilicon IOMMU driver")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Benjamin Gaignard <benjamin.gaignard@collabora.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
In init_iommu_one(), when processing IOMMU EFR attributes, the code checks
whether GASUP is enabled. If GASUP is not enabled, the code falls back to
legacy guest IR mode and then breaks out of the switch statement.
This break incorrectly skips the subsequent initialization steps that
follow the GASUP check. These initializations are independent of GASUP
support and must always be performed.
Fix this by replacing the early break with a conditional else block,
ensuring that the XTSUP check is only skipped when GASUP is not available.
Fixes: a44092e326 ("iommu/amd: Use IVHD EFR for early initialization of IOMMU features")
Reported-by: Sudheer Dantuluri <dantuluris@google.com>
Tested-by: Dheeraj Kumar Srivastava <dheerajkumar.srivastava@amd.com>
Signed-off-by: Vasant Hegde <vasant.hegde@amd.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
- Fix the Vec::reserve() doc test to properly account for the existing
vector length in the capacity assertion
- Fix an incorrect operator in the Vec::extend_with() SAFETY comment;
add a doc test demonstrating basic usage and the zero-length case
- Cleanup all imports in the alloc module and its doctests to use the
"kernel vertical" import style
-----BEGIN PGP SIGNATURE-----
iHUEABYKAB0WIQS2q/xV6QjXAdC7k+1FlHeO1qrKLgUCahOgcAAKCRBFlHeO1qrK
LnUJAP4qD/Mdo35beMgEMzql2vK1Gxcc/tKg4dc8gccHYBtXJgD/cwHcwfjuoSIH
R4a4B43Jz7r+XA1ej9hNr5Ci1f47xgQ=
=ZrNv
-----END PGP SIGNATURE-----
Merge tag 'alloc-7.2-rc1' of https://github.com/Rust-for-Linux/linux into rust-next
Pull alloc updates from Danilo Krummrich:
- Fix the 'Vec::reserve()' doctest to properly account for the existing
vector length in the capacity assertion.
- Fix an incorrect operator in the 'Vec::extend_with()' SAFETY comment;
add a doc test demonstrating basic usage and the zero-length case.
- Cleanup all imports in the alloc module and its doctests to use the
"kernel vertical" import style.
* tag 'alloc-7.2-rc1' of https://github.com/Rust-for-Linux/linux:
rust: alloc: cleanup doctest imports to "kernel vertical" style
rust: alloc: cleanup imports and use "kernel vertical" style
rust: alloc: fix `Vec::extend_with` SAFETY comment
rust: alloc: add doc test for `Vec::extend_with`
rust: alloc: fix assert in `Vec::reserve` doc test
The exynos_gem->base.size argument is a size_t rather than an
unsigned long, so adapt the printk() format string accordingly:
In file included from drivers/gpu/drm/exynos/exynos_drm_gem.c:16:
drivers/gpu/drm/exynos/exynos_drm_gem.c: In function 'exynos_drm_alloc_buf':
drivers/gpu/drm/exynos/exynos_drm_gem.c:69:49: error: format '%lx' expects argument of type 'long unsigned int', but argument 6 has type 'size_t' {aka 'unsigned int'} [-Werror=format=]
69 | DRM_DEV_DEBUG_KMS(drm_dev_dma_dev(dev), "dma_addr(0x%lx), size(0x%lx)\n",
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
70 | (unsigned long)exynos_gem->dma_addr, exynos_gem->base.size);
| ~~~~~~~~~~~~~~~~~~~~~
| |
| size_t {aka unsigned int}
The dma_addr in the same line is already printed using a cast
to unsigned long, so change that similarly to use the correct
%pad format.
Fixes: 11e898373f ("drm/exynos: Drop exynos_drm_gem.size field")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Simona Vetter <simona.vetter@ffwll.ch>
Link: https://patch.msgid.link/20260527194525.45762-1-arnd@kernel.org
gcc-16 has gained some more advanced inter-procedual optimization
techniques that enable it to inline the dummy_tlb_add_page() and
dummy_tlb_flush() function pointers into a specialized version of
__arm_v7s_unmap:
WARNING: modpost: vmlinux: section mismatch in reference: __arm_v7s_unmap+0x2cc (section: .text) -> dummy_tlb_add_page (section: .init.text)
ERROR: modpost: Section mismatches detected.
>From what I can tell, the transformation is correct, as this is only
called when __arm_v7s_unmap() is called from arm_v7s_do_selftests(),
which is also __init. Since __arm_v7s_unmap() however is not __init,
gcc cannot inline the inner function calls directly.
In debug_objects_selftest(), the same thing happens. Both the
caller and the leaf function are __init, but the IPA pulls
it into a non-init one:
WARNING: modpost: vmlinux: section mismatch in reference: lookup_object_or_alloc+0x7c (section: .text.lookup_object_or_alloc) -> is_static_object (section: .init.text)
Marking the affected functions as not "__init" would reliably avoid this
issue but is not a good solution because it removes an otherwise correct
annotation. I tried marking the functions as 'noinline', but that ended
up not covering all the affected configurations.
With some more experimenting, I found that marking these functions as
__attribute__((noipa)) is both logical and reliable.
In order to keep the syntax readable, add a custom macro for this in
include/linux/compiler_attributes.h next to other related macros and
use it to annotate both files.
Link: https://lore.kernel.org/all/abRB6g-48ZX6Yl2r@willie-the-truck/
Cc: Will Deacon <will@kernel.org>
Cc: Thomas Gleixner <tglx@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Miguel Ojeda <ojeda@kernel.org>
Cc: linux-kbuild@vger.kernel.org
Cc: stable@vger.kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Will Deacon <will@kernel.org>
Acked-by: Thomas Gleixner <tglx@kernel.org>
Acked-by: Miguel Ojeda <ojeda@kernel.org>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
All published NPU firmware versions support D0i3 delayed entry
flow, making this workaround obsolete. It was originally added as
a safety measure for potential firmware bugs.
Recent firmware dropped legacy D0i3 entry support, so the workaround
can't be used anyway. Hardcode d0i3_delayed_entry boot param to 1 to
ensure older firmware works in the correct mode.
No functional changes, just dead code cleanup.
Signed-off-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
Reviewed-by: Karol Wachowski <karol.wachowski@linux.intel.com>
Signed-off-by: Karol Wachowski <karol.wachowski@linux.intel.com>
Link: https://patch.msgid.link/20260526125521.594479-1-andrzej.kacprowski@linux.intel.com
Introduce a rwsem for protecting `ec_dev` to prevent Use-After-Free on
the `ec_dev`.
- Writers: In driver's probe() and remove().
- Readers: In file operations.
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/20260525052654.4076429-5-tzungbi@kernel.org
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
For the purposes of memory bandwidth calculations tile4
should not be considered the same as tileY. Make it so.
This should not actually change anything as the affected
code only applies to pre-MTL integrated GPUs, which don't
have tile4.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260522200346.17377-11-ville.syrjala@linux.intel.com
Reviewed-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Remove the deinterleave fallback calculation from the TGL+ codepath.
The fallback is using the ICL deinterleave calculation which was never
in the TGL+ algorithm. All supported memory types have the correct
deinterleave already specified for TGL+ anyway, so this is dead code.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260522200346.17377-10-ville.syrjala@linux.intel.com
Reviewed-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Bspec asks us to round down instead of closest doing the /100 for
the PM demand bandwidth. Make it so.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260522200346.17377-9-ville.syrjala@linux.intel.com
Reviewed-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
We have several copies of the same memory peak bandwidth calculations,
and the rounding directions are all over the place in some of them.
Unify it all into one small function (with rounding matching what Bspec
says).
Note that 'channel_width' is always a multiple of 8 anyway, so for
'channel_width / 8' the rounding direction doesn't actually matter.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260522200346.17377-8-ville.syrjala@linux.intel.com
Reviewed-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
DEPROGBWPCLIMIT is specified in %, so divide by 100 instead of 10.
Fortunately the deprobbwlimit is much lower than the peak memory
bandwidth on BMG, so whether we take 60% or 600% of the peak
bandwidth doesn't matter as the min() will pick the lower
deprobbwlimit anyway.
Eg. on the BMG here I get (with or without the fix):
QGV 0: deratedbw=33600 peakbw=48000
QGV 1: deratedbw=53000 peakbw=456000
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260522200346.17377-7-ville.syrjala@linux.intel.com
Reviewed-by: Michał Grzelak <michal.grzelak@intel.com>
The '8/num_channels' in the clperchgroup is supposed to be rounded
down according to the spec. Make it so.
Not sure we can ever actually have a non-power of two number of
channels, so this might not matter.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260522200346.17377-6-ville.syrjala@linux.intel.com
Reviewed-by: Michał Grzelak <michal.grzelak@intel.com>
For some reason we're rounding up when calculating the deinterleave
value. But the spec says we should round down. Fix it.
But I suppose this doesn't actually matter since the deinterleave
values should always be power of two. The only exception is therefore
the deinterleave==1 case, which gets handled by the max(..., 1).
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260522200346.17377-5-ville.syrjala@linux.intel.com
Reviewed-by: Michał Grzelak <michal.grzelak@intel.com>
The DRAM bandwidth value should be rounded down, not up.
Bspec: 64631
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260522200346.17377-4-ville.syrjala@linux.intel.com
Reviewed-by: Michał Grzelak <michal.grzelak@intel.com>
Reviewed-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Fix up the total mess when calculating the DCLK
frequency. Some codepaths are trying to do both DIV_ROUND_UP()
and an open coded "round to nearest" at the same time. The
MTL+ codepath was the only one that was correct (using
DIV_ROUND_CLOSEST()).
Let's unify all of them, and borrow the actual '100/6'
approach from adl_calc_psf_bw() so that we get even less
rounding errors.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260522200346.17377-3-ville.syrjala@linux.intel.com
Reviewed-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
The TGL+ bw code has an off by one error on the num_planes
calculation, and tgl_max_bw_index() incorrectly bumps
the num_planes to 1 from 0.
That approach made sense on ICL where num_planes is more or
less a minimum number of planes to consider for the group,
but on TGL+ num_planes really is a maximum number of planes,
so these adjustments no longer make any sense there.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260522200346.17377-2-ville.syrjala@linux.intel.com
Reviewed-by: Michał Grzelak <michal.grzelak@intel.com>
Introduce an event relayer mechanism. Instead of each open file
registering directly with `ec_dev->event_notifier`, the platform device
registers a single relayer notifier. Individual files then register
with a local subscribers list in `chardev_pdata`.
This allows the driver to safely disconnect from the event chain
`ec_dev->event_notifier` during cros_ec_chardev_remove(), preventing
events from being delivered to open files after the device is removed,
while still allowing those files to be closed safely later.
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/20260525052654.4076429-4-tzungbi@kernel.org
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Move `ec_dev` and `cmd_offset` from `chardev_priv` to `chardev_pdata` as
they are per-device properties but not per-open-file properties.
Hold a reference to `chardev_pdata` for each open file to ensure the
data remains valid even if the underlying platform device is removed.
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/20260525052654.4076429-3-tzungbi@kernel.org
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Introduce struct chardev_pdata to hold platform driver data.
The platform driver data is allocated by kzalloc() instead of devm
variant, allowing for managed cleanup that can eventually extend beyond
device removal if files are still open.
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/20260525052654.4076429-2-tzungbi@kernel.org
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
ipv6_rpl_srh_decompress() computes:
outhdr->hdrlen = (((n + 1) * sizeof(struct in6_addr)) >> 3);
hdrlen is __u8. For n >= 127 the result exceeds 255 and silently
truncates. With n=127 (cmpri=15, cmpre=15, pad=0, hdrlen=16):
(128 * 16) >> 3 = 256, truncated to 0 as __u8
The caller in ipv6_rpl_srh_rcv() then places the compressed header
at buf + ((ohdr->hdrlen + 1) << 3). With hdrlen=0 this is buf + 8,
but the decompressed region occupies buf[0..2055] (8-byte header
plus 128 full addresses). The compressed header overlaps the
decompressed data, and ipv6_rpl_srh_compress() writes into this
overlap, corrupting the routing header of the forwarded packet.
The existing guard at exthdrs.c:546 checks (n + 1) > 255, which
prevents n+1 from overflowing unsigned char (the segments_left
field), but does not prevent the computed hdrlen from overflowing
__u8. n=127 passes because 128 <= 255, yet hdrlen=256 does not
fit.
Tighten the bound to (n + 1) > 127. This caps n at 126, giving
hdrlen = (127 * 16) >> 3 = 254, which fits in __u8. The compressed
header then lands at buf + ((254 + 1) << 3) = buf + 2040, exactly
past the decompressed region (buf[0..2039]). No overlap. 127
segments is well beyond any realistic RPL deployment.
Fixes: 8610c7c6e3 ("net: ipv6: add support for rpl sr exthdr")
Signed-off-by: Rahul Chandelkar <rc@rexion.ai>
Link: https://patch.msgid.link/20260525154031.2290876-1-rc@rexion.ai
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Now that KVM correctly handles Host-Only and Guest-Only bits in the
event selector MSRs, allow the guest to set them if the vCPU advertises
SVM and uses the mediated PMU.
Signed-off-by: Jim Mattson <jmattson@google.com>
Signed-off-by: Yosry Ahmed <yosry@kernel.org>
Link: https://patch.msgid.link/20260527234711.4175166-14-yosry@kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
Reprogram PMU counters on nested transitions for the mediated PMU, to
re-evaluate Host-Only and Guest-Only bits and enable/disable the PMU
counters accordingly. For example, if Host-Only is set and Guest-Only is
cleared, a counter should be disabled when entering guest mode and
enabled when exiting guest mode.
According to the APM, when EFER.SVME is cleared, setting Host-Only or
Guest-Only disables the counter, so also trigger counter reprogramming
when EFER.SVME is toggled.
Counters setting any of Host-Only and Guest-Only bits are already being
tracked in pmc_has_mode_specific_enables, use the bitmap to reprogram
these counters.
Reprogram the counters synchronously on nested VMRUN/#VMEXIT and
EFER.SVME toggling. This is necessary as these instructions are counted
based on the new CPU state (after the instruction is retired in
hardware). Hence, the PMU needs to be updated before instruction
emulation is completed and kvm_pmu_instruction_retired() is called.
Defer reprogramming the counters when force leaving guest mode through
svm_leave_nested() to avoid potentially reading stale state (e.g.
incorrect EFER). All flows force leaving nested are non-architectural,
so accuracy is irrelevant.
Refactor a helper out of kvm_pmu_request_reprogram_counters() that
accepts a boolean allowing synchronous vs deferred reprogramming, and
use that from SVM code to support both scenarios.
Signed-off-by: Yosry Ahmed <yosry@kernel.org>
Link: https://patch.msgid.link/20260527234711.4175166-13-yosry@kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
Instead of always checking of a counter needs to be disabled for
mode-specific reasons (e.g. Host-Only/Guest-Only bits in SVM), add a
bitmap to track such counters. Set the bit for counters using either
Host-Only or Guest-Only bits in EVENTSEL on SVM.
This bitmap will also be reused in following changes to selectively
apply changes to such counters.
Signed-off-by: Yosry Ahmed <yosry@kernel.org>
Link: https://patch.msgid.link/20260527234711.4175166-12-yosry@kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
Introduce an optional per-vendor PMU callback for checking if a counter
is disabled in the current mode, and register a callback on AMD to
disable a counter based on the vCPU's setting of Host-Only or Guest-Only
EVENT_SELECT bits with the mediated PMU.
If EFER.SVME is set, all events are counted if both bits are set or
cleared. If only one bit is set, the counter is disabled if the vCPU
context does not match the set bit.
If EFER.SVME is cleared, the counter is disabled if any of the bits is
set, otherwise all events are counted. Note that a Linux guest correctly
handles this and clears Host-Only when EFER.SVME is cleared, see commit
1018faa6cf ("perf/x86/kvm: Fix Host-Only/Guest-Only counting with SVM
disabled").
The callback is made from pmc_is_locally_enabled(), which is used for
the mediated PMU when updating eventsel_hw in
kvm_mediated_pmu_refresh_eventsel_hw(), as well as when checking what
PMCs count instructions/branches for emulation in
kvm_pmu_recalc_pmc_emulation().
Host-Only and Guest-Only bits are currently reserved, so this change is
a noop, but the bits will be allowed with mediated PMU in a following
change when fully supported.
Originally-by: Jim Mattson <jmattson@google.com>
Signed-off-by: Yosry Ahmed <yosry@kernel.org>
Link: https://patch.msgid.link/20260527234711.4175166-11-yosry@kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
Add definitions for KVM_X86_PMU_OP_OPTIONAL_RET0() to resolve to
__static_call_return0, similar to KVM_X86_OP_OPTIONAL_RET0(). Move the
definition of kvm_pmu_call() to pmu.h, and add declarations for the
static PMU calls in the header to allow making callbacks from the header
in following changes.
Suggested-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Yosry Ahmed <yosry@kernel.org>
Link: https://patch.msgid.link/20260527234711.4175166-10-yosry@kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
If the guest disables the counter (by clearing
ARCH_PERFMON_EVENTSEL_ENABLE), KVM still performs the PMU filter lookup,
even though it doesn't end up changing eventsel_hw. Check if the
counter is enabled by the guest before doing the potentially expensive
PMU filter lookup.
Suggested-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Yosry Ahmed <yosry@kernel.org>
Link: https://patch.msgid.link/20260527234711.4175166-9-yosry@kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
Do a single atomic OR using the atomic overlay of reprogram_pmi bitmask,
instead of one atomic set_bit() call per counter.
Signed-off-by: Yosry Ahmed <yosry@kernel.org>
Link: https://patch.msgid.link/20260527234711.4175166-8-yosry@kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
Rename reprogram_counters() to kvm_pmu_request_counters_reprogram()
clarifying that it is more similar to
kvm_pmu_request_counter_reprogram(), and less similar to
reprogram_counter(). The kvm_pmu_* prefix is also appropriate as the
function is exposed in the header.
Opportunistically rename the argument from 'diff' to 'counters'.
No functional change intended.
Signed-off-by: Yosry Ahmed <yosry@kernel.org>
Link: https://patch.msgid.link/20260527234711.4175166-7-yosry@kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
The declaration and definition of enable_pmu/enable_mediated_pmu
semantically belongs in pmu.h and pmu.c, and more importantly, pmu.h
uses enable_mediated_pmu and relies on the caller including x86.h.
There is already precedence for other module params defined outside of
x86.c, so move enable_pmu/enable_mediated_pmu to pmu.c.
No functional change intended.
Signed-off-by: Yosry Ahmed <yosry@kernel.org>
Link: https://patch.msgid.link/20260527234711.4175166-6-yosry@kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
A successful VMRUN retires in guest mode and should be counted by the
PMU as a guest instruction. Move the call to
kvm_pmu_instruction_retired() after potentially entering guest mode,
such that VMRUN is counted correctly.
The PMU event will be matched against L2's CPL, but otherwise this does
not change the behavior in terms of guest vs. host, because KVM does
not virtualize Host-Only/Guest-Only PMC controls yet, so all
instructions are counted regardless of the vCPU's host/guest state. But
this change is needed for the incoming support for Host-Only/Guest-Only
controls to count VMRUN correctly.
Signed-off-by: Yosry Ahmed <yosry@kernel.org>
Link: https://patch.msgid.link/20260527234711.4175166-5-yosry@kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
The code paths for advancing RIP and retiring the instruction for RIP
are very similar whether or not caching vmcb12 succeeds. The only
difference is handling mapping failures (i.e. EFAULT).
Pull the mapping failure handling out and unify the calls to
svm_skip_emulated_instruction() and kvm_pmu_instruction_retired(), but
return immediately after if copying and caching vmcb12 failed. A nice
side effect of this is that the FIXME comment is now above the only code
path calling svm_skip_emulated_instruction().
Signed-off-by: Yosry Ahmed <yosry@kernel.org>
Link: https://patch.msgid.link/20260527234711.4175166-4-yosry@kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
If svm_skip_emulation_instruction() fails, then RIP could not be
advanced correctly (e.g. decode failure when NextRIP is not available).
KVM will exit to userspace to handle the emulation failure, but only
after stuffing the wrong RIP into vmcb01 and entering guest mode.
Bail early and exit to userspace before committing any side-effects of
emulating the VMRUN (e.g. entering guest mode).
Fixes: c8e16b78c6 ("x86: KVM: svm: eliminate hardcoded RIP advancement from vmrun_interception()")
Signed-off-by: Yosry Ahmed <yosry@kernel.org>
Link: https://patch.msgid.link/20260527234711.4175166-3-yosry@kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
According to the APM, TF on VMRUN causes a #DB after VMRUN completes on
the _host_ side. However, KVM injects a #DB in L2 context instead (or
exits to userspace if KVM_GUESTDBG_SINGLESTEP is set) in
kvm_skip_emulated_instruction().
Avoid single-step handling on VMRUN by open-coding the rest of
kvm_skip_emulated_instruction() in nested_svm_vmrun(). This doesn't look
pretty, but following changes will need to open-code
kvm_pmu_instruction_retired() anyway, and will cleanup the code. This
ignores TF on VMRUN instead of injecting a spurious exception into
L2. Document this virtualization hole with a FIXME.
Note that a failed VMRUN would have been correctly single-stepped, but
now TF is always ignored for consistency and simplicity purposes. VMX
does not support TF on a successful VMLAUNCH/VMRESUME, so it's unlikely
that single-stepping VMRUN properly is important, especially if it's
only for failed VMRUNs.
Fixes: c8e16b78c6 ("x86: KVM: svm: eliminate hardcoded RIP advancement from vmrun_interception()")
Signed-off-by: Yosry Ahmed <yosry@kernel.org>
Link: https://patch.msgid.link/20260527234711.4175166-2-yosry@kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
When inserting an EDT packet with time before flow->time_next_packet,
update the flow and possibly queue next delivery time.
Reinsert the flow into the q->delayed rb-tree to position correctly
and to have fq_check_throttled set wake-up at the right next time.
Factor RB tree insertion out fq_flow_set_throttled to avoid open
coding twice.
EDT packets do not take precedence over queue rate limit. Skip this
new step if a queue limit is set. EDT packets do take precedence over
per-socket rate limits, as can be seen from fq_dequeue reading
sk_pacing_rate if !skb->tstamp.
With this change the so_txtime selftest sends packets in the expected
order.
Signed-off-by: Willem de Bruijn <willemb@google.com>
Link: https://patch.msgid.link/20260526134109.2624493-1-willemdebruijn.kernel@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Louis-Alexis Eyraud says:
====================
Introduce Airoha AN8801R series Gigabit Ethernet PHY driver
This series introduces the Airoha AN8801R Gigabit Ethernet PHY initial
support.
The Airoha AN8801R is a low power single-port Ethernet PHY Transceiver
with Single-port serdes interface for 1000Base-X/RGMII.
This chip is compliant with 10Base-T, 100Base-TX and 1000Base-T IEEE
802.3(u,ab) and supports:
- Energy Efficient Ethernet (802.3az)
- Full Duplex Control Flow (802.3x)
- auto-negotiation
- crossover detect and autocorrection,
- Wake-on-LAN with Magic Packet
- Jumbo Frame up to 9 Kilobytes.
This PHY also supports up to three user-configurable LEDs, which are
usually used for LAN Activity, 100M, 1000M indication.
The series provides the devicetree binding and the driver that have been
written by AngeloGioacchino Del Regno, based on downstream
implementation ([1]). The driver allows setting up PHY LEDs, 10/100M,
1000M speeds, and Wake on LAN and PHY interrupts.
Since v2, the series also adds the air_phy_lib library, which goal is to
share common code between air_en8811h and air_an8801 drivers, and its use
in them. The first shared functions are the existing BuckPbus register
accessors and air_phy_read/write_page functions coming from air_en8811h
driver.
The series is based on net-next kernel tree (sha1: 90d03ee2c5) and
I have tested it on Mediatek Genio 720-EVK board (that integrates an
Airoha AN8801RIN/A Ethernet PHY) with early board hardware enablement
patches.
[1]: https://gitlab.com/mediatek/aiot/bsp/linux/-/blob/mtk-v6.6/drivers/net/phy/an8801.c
====================
Link: https://patch.msgid.link/20260526-add-airoha-an8801-support-v5-0-01aea8dee69b@collabora.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
To ensure that the Airoha AN8801R PHY uses the maximum available link
speed, an additional register write is needed to configure the function
mode for either 1G or 100M/10M operation after link detection.
So, in air_an8801 driver, implement a custom read_status callback, that
after genphy_read_status determines the link speed, sets the bit 0 of
the link mode register (REG_LINK_MODE) if the detected speed is 1Gbps,
or unsets it otherwise.
Signed-off-by: Louis-Alexis Eyraud <louisalexis.eyraud@collabora.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20260526-add-airoha-an8801-support-v5-6-01aea8dee69b@collabora.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Introduce a driver for the Airoha AN8801R Series Gigabit Ethernet
PHY; this currently supports setting up PHY LEDs, 10/100M, 1000M
speeds, and Wake on LAN and PHY interrupts.
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Louis-Alexis Eyraud <louisalexis.eyraud@collabora.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20260526-add-airoha-an8801-support-v5-5-01aea8dee69b@collabora.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Rename the BuckPBus register accessors functions present in air_phy_lib
and their calls in air_en8811h driver, so all exported functions start
with the same prefix.
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Louis-Alexis Eyraud <louisalexis.eyraud@collabora.com>
Link: https://patch.msgid.link/20260526-add-airoha-an8801-support-v5-4-01aea8dee69b@collabora.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
In preparation of Airoha AN8801R PHY support, move the BuckPBus
register accessors and definitions, present in air_en8811h driver,
into the Airoha PHY shared code (air_phy_lib), so they will be usable
by the new driver without duplicating them.
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Louis-Alexis Eyraud <louisalexis.eyraud@collabora.com>
Link: https://patch.msgid.link/20260526-add-airoha-an8801-support-v5-3-01aea8dee69b@collabora.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
In preparation of Airoha AN8801R PHY support, split out the interface
functions that will be common between the already present air_en8811h
driver and the new one, and put them into a new library named
air_phy_lib.
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Louis-Alexis Eyraud <louisalexis.eyraud@collabora.com>
Link: https://patch.msgid.link/20260526-add-airoha-an8801-support-v5-2-01aea8dee69b@collabora.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add a new binding to support the Airoha AN8801R Series Gigabit
Ethernet PHY.
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Reviewed-by: Rob Herring (Arm) <robh@kernel.org>
Signed-off-by: Louis-Alexis Eyraud <louisalexis.eyraud@collabora.com>
Link: https://patch.msgid.link/20260526-add-airoha-an8801-support-v5-1-01aea8dee69b@collabora.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Quan Sun reported [1] a stack overflow in cls_bpf_offload_cmd().
Reproducer on netdevsim: add a skip_sw cls_bpf filter, set the
bpf_tc_accept debugfs knob to 0, then `tc filter replace`. The replace
calls tc_setup_cb_replace() which fails. cls_bpf_offload_cmd() then
swaps prog/oldprog and recursively calls itself to roll back. But
bpf_tc_accept=0 makes the rollback fail too, which triggers yet another
rollback frame with the same arguments, and so on until the stack is
exhausted.
bpf_tc_accept is just a convenient knob for the reproducer. Any driver
whose tc_setup_cb_replace() fails twice in a row can hit the same loop,
so this is not a netdevsim-only issue.
Two ways to fix it:
1) Have the rollback call tc_setup_cb_add() on oldprog instead of
re-entering cls_bpf_offload_cmd().
2) Mark the rollback frame with a flag and skip a second-level
rollback from inside it.
Go with (2). It is the smaller change and keeps the original behaviour:
the rollback still goes through tc_setup_cb_replace(), so the driver
gets one real chance to restore its state. If that attempt also fails,
we just return the original error instead of recursing.
[1]: https://lore.kernel.org/bpf/ce5a6005-3c5e-4696-9e05-eba9461dc860@std.uestc.edu.cn/T/#u
Fixes: 102740bd94 ("cls_bpf: fix offload assumptions after callback conversion")
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Jiayuan Chen <jiayuan.chen@linux.dev>
Link: https://patch.msgid.link/20260526025529.24382-1-jiayuan.chen@linux.dev
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
