Most exit handlers return <= 0 to indicate that the host needs to
handle the exit. Make kvm_handle_mmio_return() consistent with
the exit handlers in handle_exit(). This makes the code easier to
reason about, and makes it easier to add other handlers in future
patches.
No functional change intended.
Signed-off-by: Fuad Tabba <tabba@google.com>
Acked-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240423150538.2103045-15-tabba@google.com
Signed-off-by: Marc Zyngier <maz@kernel.org>
Fix the comment to clarify that __pkvm_vcpu_init_traps()
initializes traps for all VMs in protected mode, and not only
for protected VMs.
Signed-off-by: Fuad Tabba <tabba@google.com>
Acked-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240423150538.2103045-14-tabba@google.com
Signed-off-by: Marc Zyngier <maz@kernel.org>
We've added a .data section for the hypervisor, which kmemleak is
eager to parse. This clearly doesn't go well, so add the section
to kmemleak's block list.
Signed-off-by: Quentin Perret <qperret@google.com>
Signed-off-by: Fuad Tabba <tabba@google.com>
Acked-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240423150538.2103045-13-tabba@google.com
Signed-off-by: Marc Zyngier <maz@kernel.org>
pKVM maintains its own state at EL2 for tracking the host fpsimd
state. Therefore, no need to map and share the host's view with
it.
Signed-off-by: Fuad Tabba <tabba@google.com>
Reviewed-by: Mark Brown <broonie@kernel.org>
Acked-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240423150538.2103045-12-tabba@google.com
Signed-off-by: Marc Zyngier <maz@kernel.org>
Rename __tlb_switch_to_{guest,host}() to
{enter,exit}_vmid_context() in VHE code to maintain symmetry
between the nVHE and VHE TLB invalidations.
No functional change intended.
Suggested-by: Oliver Upton <oliver.upton@linux.dev>
Signed-off-by: Fuad Tabba <tabba@google.com>
Acked-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240423150538.2103045-11-tabba@google.com
Signed-off-by: Marc Zyngier <maz@kernel.org>
Typically, TLB invalidation of guest stage-2 mappings using nVHE is
performed by a hypercall originating from the host. For the invalidation
instruction to be effective, therefore, __tlb_switch_to_{guest,host}()
swizzle the active stage-2 context around the TLBI instruction.
With guest-to-host memory sharing and unsharing hypercalls
originating from the guest under pKVM, there is need to support
both guest and host VMID invalidations issued from guest context.
Replace the __tlb_switch_to_{guest,host}() functions with a more general
{enter,exit}_vmid_context() implementation which supports being invoked
from guest context and acts as a no-op if the target context matches the
running context.
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Fuad Tabba <tabba@google.com>
Acked-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240423150538.2103045-10-tabba@google.com
Signed-off-by: Marc Zyngier <maz@kernel.org>
Break-before-make (BBM) can be expensive, as transitioning via an
invalid mapping (i.e. the "break" step) requires the completion of TLB
invalidation and can also cause other agents to fault concurrently on
the invalid mapping.
Since BBM is not required when changing only the software bits of a PTE,
avoid the sequence in this case and just update the PTE directly.
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Fuad Tabba <tabba@google.com>
Acked-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240423150538.2103045-9-tabba@google.com
Signed-off-by: Marc Zyngier <maz@kernel.org>
Don't just assume that the PTE is valid when checking whether it
describes an executable or cacheable mapping.
This makes sure that we don't issue CMOs for invalid mappings.
Suggested-by: Will Deacon <will@kernel.org>
Signed-off-by: Fuad Tabba <tabba@google.com>
Acked-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240423150538.2103045-8-tabba@google.com
Signed-off-by: Marc Zyngier <maz@kernel.org>
Under certain circumstances __get_fault_info() may resolve the faulting
address using the AT instruction. Given that this is being done outside
of the host lock critical section, it is racy and the resolution via AT
may fail. We currently BUG() in this situation, which is obviously less
than ideal. Moving the address resolution to the critical section may
have a performance impact, so let's keep it where it is, but bail out
and return to the host to try a second time.
Signed-off-by: Quentin Perret <qperret@google.com>
Signed-off-by: Fuad Tabba <tabba@google.com>
Acked-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240423150538.2103045-7-tabba@google.com
Signed-off-by: Marc Zyngier <maz@kernel.org>
On the guest teardown path, pKVM will zero the pages used to back
the guest data structures before returning them to the host as
they may contain secrets (e.g. in the vCPU registers). However,
the zeroing is done using a cacheable alias, and CMOs are
missing, hence giving the host a potential opportunity to read
the original content of the guest structs from memory.
Fix this by issuing CMOs after zeroing the pages.
Signed-off-by: Quentin Perret <qperret@google.com>
Signed-off-by: Fuad Tabba <tabba@google.com>
Acked-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240423150538.2103045-6-tabba@google.com
Signed-off-by: Marc Zyngier <maz@kernel.org>
The lock is already initialized in core KVM code at
kvm_create_vm().
Fixes: 9d0c063a4d ("KVM: arm64: Instantiate pKVM hypervisor VM and vCPU structures from EL1")
Signed-off-by: Fuad Tabba <tabba@google.com>
Acked-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240423150538.2103045-5-tabba@google.com
Signed-off-by: Marc Zyngier <maz@kernel.org>
To avoid direct comparison against the fp_owner enum, add a new
function that performs the check, host_owns_fp_regs(), to
complement the existing guest_owns_fp_regs().
To check for fpsimd state ownership, use the helpers instead of
directly using the enums.
No functional change intended.
Suggested-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Fuad Tabba <tabba@google.com>
Reviewed-by: Mark Brown <broonie@kernel.org>
Acked-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240423150538.2103045-4-tabba@google.com
Signed-off-by: Marc Zyngier <maz@kernel.org>
guest_owns_fp_regs() will be used to check fpsimd state ownership
across kvm/arm64. Therefore, move it to kvm_host.h to widen its
scope.
Moreover, the host state is not per-vcpu anymore, the vcpu
parameter isn't used, so remove it as well.
No functional change intended.
Signed-off-by: Fuad Tabba <tabba@google.com>
Reviewed-by: Mark Brown <broonie@kernel.org>
Acked-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240423150538.2103045-3-tabba@google.com
Signed-off-by: Marc Zyngier <maz@kernel.org>
Since the host_fpsimd_state has been removed from kvm_vcpu_arch,
it isn't pointing to the hyp's version of the host fp_regs in
protected mode.
Initialize the host_data fpsimd_state point to the host_data's
context fp_regs on pKVM initialization.
Fixes: 51e09b5572 ("KVM: arm64: Exclude host_fpsimd_state pointer from kvm_vcpu_arch")
Signed-off-by: Fuad Tabba <tabba@google.com>
Acked-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240423150538.2103045-2-tabba@google.com
Signed-off-by: Marc Zyngier <maz@kernel.org>
Commit d5a32b60dc ("KVM: arm64: Allow userspace to change
ID_AA64MMFR{0-2}_EL1") made certain fields in these registers writable,
but in doing so, ID_AA64MMFR1_EL1_XNX was listed twice. Remove the
duplication.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Reviewed-by: Zenghui Yu <zenghui.yu@linux.dev>
Link: https://lore.kernel.org/r/E1s2AxF-00AWLv-03@rmk-PC.armlinux.org.uk
Signed-off-by: Marc Zyngier <maz@kernel.org>
Currently, if an automatically freed allocation is an error pointer that
will lead to a crash. An example of this is in wm831x_gpio_dbg_show().
171 char *label __free(kfree) = gpiochip_dup_line_label(chip, i);
172 if (IS_ERR(label)) {
173 dev_err(wm831x->dev, "Failed to duplicate label\n");
174 continue;
175 }
The auto clean up function should check for error pointers as well,
otherwise we're going to keep hitting issues like this.
Fixes: 54da6a0924 ("locking: Introduce __cleanup() based infrastructure")
Cc: <stable@vger.kernel.org>
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Commit 284f17ac13 ("mm/slub: handle bulk and single object freeing
separately") splits single and bulk object freeing in two functions
slab_free() and slab_free_bulk() which leads slab_free() to call
slab_free_hook() directly instead of slab_free_freelist_hook().
If `init_on_free` is set, slab_free_hook() zeroes the object.
Afterward, if `slub_debug=F` and `CONFIG_SLAB_FREELIST_HARDENED` are
set, the do_slab_free() slowpath executes freelist consistency
checks and try to decode a zeroed freepointer which leads to a
"Freepointer corrupt" detection in check_object().
During bulk free, slab_free_freelist_hook() isn't affected as it always
sets it objects freepointer using set_freepointer() to maintain its
reconstructed freelist after `init_on_free`.
For single free, object's freepointer thus needs to be avoided when
stored outside the object if `init_on_free` is set. The freepointer left
as is, check_object() may later detect an invalid pointer value due to
objects overflow.
To reproduce, set `slub_debug=FU init_on_free=1 log_level=7` on the
command line of a kernel build with `CONFIG_SLAB_FREELIST_HARDENED=y`.
dmesg sample log:
[ 10.708715] =============================================================================
[ 10.710323] BUG kmalloc-rnd-05-32 (Tainted: G B T ): Freepointer corrupt
[ 10.712695] -----------------------------------------------------------------------------
[ 10.712695]
[ 10.712695] Slab 0xffffd8bdc400d580 objects=32 used=4 fp=0xffff9d9a80356f80 flags=0x200000000000a00(workingset|slab|node=0|zone=2)
[ 10.716698] Object 0xffff9d9a80356600 @offset=1536 fp=0x7ee4f480ce0ecd7c
[ 10.716698]
[ 10.716698] Bytes b4 ffff9d9a803565f0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
[ 10.720703] Object ffff9d9a80356600: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
[ 10.720703] Object ffff9d9a80356610: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
[ 10.724696] Padding ffff9d9a8035666c: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
[ 10.724696] Padding ffff9d9a8035667c: 00 00 00 00 ....
[ 10.724696] FIX kmalloc-rnd-05-32: Object at 0xffff9d9a80356600 not freed
Fixes: 284f17ac13 ("mm/slub: handle bulk and single object freeing separately")
Cc: <stable@vger.kernel.org>
Co-developed-by: Chengming Zhou <chengming.zhou@linux.dev>
Signed-off-by: Chengming Zhou <chengming.zhou@linux.dev>
Signed-off-by: Nicolas Bouchinet <nicolas.bouchinet@ssi.gouv.fr>
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
We flush the rebind worker during the vm close phase, however in places
like preempt_fence_work_func() we seem to queue the rebind worker
without first checking if the vm has already been closed. The concern
here is the vm being closed with the worker flushed, but then being
rearmed later, which looks like potential uaf, since there is no actual
refcounting to track the queued worker. We can't take the vm->lock here
in preempt_rebind_work_func() to first check if the vm is closed since
that will deadlock, so instead flush the worker again when the vm
refcount reaches zero.
v2:
- Grabbing vm->lock in the preempt worker creates a deadlock, so
checking the closed state is tricky. Instead flush the worker when
the refcount reaches zero. It should be impossible to queue the
preempt worker without already holding vm ref.
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/1676
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/1591
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/1364
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/1304
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/1249
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: <stable@vger.kernel.org> # v6.8+
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240423074721.119633-4-matthew.auld@intel.com
(cherry picked from commit 3d44d67c44)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
These drivers don't use the driver_data member of struct i2c_device_id,
so don't explicitly initialize this member.
This prepares putting driver_data in an anonymous union which requires
either no initialization or named designators. But it's also a nice
cleanup on its own.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Link: https://lore.kernel.org/r/20240430085654.1028864-2-u.kleine-koenig@pengutronix.de
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Since the value for PWMOUT Target Duty Cycle register is a 9 bit
left-justified value that ranges from 0 to 511 and is contained in 2
bytes.
There is an issue that the PWM signal recorded by oscilloscope would
not be on consistently if we set PWM to 100% to the driver.
It is because the LSB of the 9 bit would always be zero if it just
left shift 8 bit for the value that write to PWMOUT Target Duty
Cycle register.
Therefore, revise the scale of the value that was written to pwm input
from 255 to 511 and modify the value to left-justified value.
Signed-off-by: Delphine CC Chiu <Delphine_CC_Chiu@wiwynn.com>
Link: https://lore.kernel.org/r/20240416022211.859483-1-Delphine_CC_Chiu@wiwynn.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Add tests for some of the more common STE update operations that we expect
to see, as well as some artificial STE updates to test the edges of
arm_smmu_write_entry. These also serve as a record of which common
operation is expected to be hitless, and how many syncs they require.
arm_smmu_write_entry implements a generic algorithm that updates an STE/CD
to any other abritrary STE/CD configuration. The update requires a
sequence of write+sync operations with some invariants that must be held
true after each sync. arm_smmu_write_entry lends itself well to
unit-testing since the function's interaction with the STE/CD is already
abstracted by input callbacks that we can hook to introspect into the
sequence of operations. We can use these hooks to guarantee that
invariants are held throughout the entire update operation.
Link: https://lore.kernel.org/r/20240106083617.1173871-3-mshavit@google.com
Tested-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Michael Shavit <mshavit@google.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/9-v9-5040dc602008+177d7-smmuv3_newapi_p2_jgg@nvidia.com
Signed-off-by: Will Deacon <will@kernel.org>
Half the code was living in arm_smmu_domain_finalise_s1(), just move it
here and take the values directly from the pgtbl_ops instead of storing
copies.
Tested-by: Nicolin Chen <nicolinc@nvidia.com>
Tested-by: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com>
Reviewed-by: Michael Shavit <mshavit@google.com>
Reviewed-by: Mostafa Saleh <smostafa@google.com>
Reviewed-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/8-v9-5040dc602008+177d7-smmuv3_newapi_p2_jgg@nvidia.com
Signed-off-by: Will Deacon <will@kernel.org>
Pull all the calculations for building the CD table entry for a mmu_struct
into arm_smmu_make_sva_cd().
Call it in the two places installing the SVA CD table entry.
Open code the last caller of arm_smmu_update_ctx_desc_devices() and remove
the function.
Remove arm_smmu_write_ctx_desc() since all callers are gone. Add the
locking assertions to arm_smmu_alloc_cd_ptr() since
arm_smmu_update_ctx_desc_devices() was the last problematic caller.
Remove quiet_cd since all users are gone, arm_smmu_make_sva_cd() creates
the same value.
The behavior of quiet_cd changes slightly, the old implementation edited
the CD in place to set CTXDESC_CD_0_TCR_EPD0 assuming it was a SVA CD
entry. This version generates a full CD entry with a 0 TTB0 and relies on
arm_smmu_write_cd_entry() to install it hitlessly.
Tested-by: Nicolin Chen <nicolinc@nvidia.com>
Tested-by: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com>
Reviewed-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/7-v9-5040dc602008+177d7-smmuv3_newapi_p2_jgg@nvidia.com
Signed-off-by: Will Deacon <will@kernel.org>
Avoid arm_smmu_attach_dev() having to undo the changes to the
smmu_domain->devices list, acquire the cdptr earlier so we don't need to
handle that error.
Now there is a clear break in arm_smmu_attach_dev() where all the
prep-work has been done non-disruptively and we commit to making the HW
change, which cannot fail.
This completes transforming arm_smmu_attach_dev() so that it does not
disturb the HW if it fails.
Tested-by: Nicolin Chen <nicolinc@nvidia.com>
Tested-by: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com>
Reviewed-by: Michael Shavit <mshavit@google.com>
Reviewed-by: Nicolin Chen <nicolinc@nvidia.com>
Reviewed-by: Mostafa Saleh <smostafa@google.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/6-v9-5040dc602008+177d7-smmuv3_newapi_p2_jgg@nvidia.com
Signed-off-by: Will Deacon <will@kernel.org>
Only the attach callers can perform an allocation for the CD table entry,
the other callers must not do so, they do not have the correct locking and
they cannot sleep. Split up the functions so this is clear.
arm_smmu_get_cd_ptr() will return pointer to a CD table entry without
doing any kind of allocation.
arm_smmu_alloc_cd_ptr() will allocate the table and any required
leaf.
A following patch will add lockdep assertions to arm_smmu_alloc_cd_ptr()
once the restructuring is completed and arm_smmu_alloc_cd_ptr() is never
called in the wrong context.
Tested-by: Nicolin Chen <nicolinc@nvidia.com>
Reviewed-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/5-v9-5040dc602008+177d7-smmuv3_newapi_p2_jgg@nvidia.com
Signed-off-by: Will Deacon <will@kernel.org>
A cleared entry is all 0's. Make arm_smmu_clear_cd() do this sequence.
If we are clearing an entry and for some reason it is not already
allocated in the CD table then something has gone wrong.
Remove case (5) from arm_smmu_write_ctx_desc().
Tested-by: Nicolin Chen <nicolinc@nvidia.com>
Tested-by: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com>
Reviewed-by: Michael Shavit <mshavit@google.com>
Reviewed-by: Nicolin Chen <nicolinc@nvidia.com>
Reviewed-by: Moritz Fischer <moritzf@google.com>
Reviewed-by: Mostafa Saleh <smostafa@google.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/4-v9-5040dc602008+177d7-smmuv3_newapi_p2_jgg@nvidia.com
Signed-off-by: Will Deacon <will@kernel.org>
Introduce arm_smmu_make_s1_cd() to build the CD from the paging S1 domain,
and reorganize all the places programming S1 domain CD table entries to
call it.
Split arm_smmu_update_s1_domain_cd_entry() from
arm_smmu_update_ctx_desc_devices() so that the S1 path has its own call
chain separate from the unrelated SVA path.
arm_smmu_update_s1_domain_cd_entry() only works on S1 domains attached to
RIDs and refreshes all their CDs. Remove case (3) from
arm_smmu_write_ctx_desc() as it is now handled by directly calling
arm_smmu_write_cd_entry().
Remove the forced clear of the CD during S1 domain attach,
arm_smmu_write_cd_entry() will do this automatically if necessary.
Tested-by: Nicolin Chen <nicolinc@nvidia.com>
Tested-by: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com>
Reviewed-by: Michael Shavit <mshavit@google.com>
Reviewed-by: Nicolin Chen <nicolinc@nvidia.com>
Reviewed-by: Mostafa Saleh <smostafa@google.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/3-v9-5040dc602008+177d7-smmuv3_newapi_p2_jgg@nvidia.com
[will: Drop unused arm_smmu_clean_cd_entry() function]
Signed-off-by: Will Deacon <will@kernel.org>
CD table entries and STE's have the same essential programming sequence,
just with different types. Use the new ops indirection to link CD
programming to the common writer.
In a few more patches all CD writers will call an appropriate make
function and then directly call arm_smmu_write_cd_entry().
arm_smmu_write_ctx_desc() will be removed.
Until then lightly tweak arm_smmu_write_ctx_desc() to also use the new
programmer by using the same logic as right now to build the target CD on
the stack, sanitizing it to meet the used rules, and then using the
writer.
Sanitizing is necessary because the writer expects that the currently
programmed CD follows the used rules. Next patches add new make functions
and new direct calls to arm_smmu_write_cd_entry() which will require this.
Signed-off-by: Michael Shavit <mshavit@google.com>
Tested-by: Nicolin Chen <nicolinc@nvidia.com>
Tested-by: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com>
Reviewed-by: Moritz Fischer <moritzf@google.com>
Reviewed-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/2-v9-5040dc602008+177d7-smmuv3_newapi_p2_jgg@nvidia.com
Signed-off-by: Will Deacon <will@kernel.org>
Prepare to put the CD code into the same mechanism. Add an ops indirection
around all the STE specific code and make the worker functions independent
of the entry content being processed.
get_used and sync ops are provided to hook the correct code.
Signed-off-by: Michael Shavit <mshavit@google.com>
Reviewed-by: Michael Shavit <mshavit@google.com>
Reviewed-by: Nicolin Chen <nicolinc@nvidia.com>
Tested-by: Nicolin Chen <nicolinc@nvidia.com>
Tested-by: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/1-v9-5040dc602008+177d7-smmuv3_newapi_p2_jgg@nvidia.com
Signed-off-by: Will Deacon <will@kernel.org>
The Qualcomm TBU debug support introduced by 414ecb0308
("iommu/arm-smmu-qcom-debug: Add support for TBUs") provides its own
driver initialisation function, which breaks the link when the core SMMU
driver is built as a module:
ld.lld: error: duplicate symbol: init_module
>>> defined at arm-smmu.c
>>> drivers/iommu/arm/arm-smmu/arm-smmu.o:(init_module)
>>> defined at arm-smmu-qcom-debug.c
>>> drivers/iommu/arm/arm-smmu/arm-smmu-qcom-debug.o:(.init.text+0x4)
Since we're late in the cycle, just make the debug features depend on a
non-modular SMMU driver for now while the initialisation is reworked to
hang off qcom_smmu_impl_init().
Signed-off-by: Will Deacon <will@kernel.org>
Profiling shows that calling nr_possible_cpus() in objpool_pop() takes
a noticeable amount of CPU (when profiled on 80-core machine), as we
need to recalculate number of set bits in a CPU bit mask. This number
can't change, so there is no point in paying the price for recalculating
it. As such, cache this value in struct objpool_head and use it in
objpool_pop().
On the other hand, cached pool->nr_cpus isn't necessary, as it's not
used in hot path and is also a pretty trivial value to retrieve. So drop
pool->nr_cpus in favor of using nr_cpu_ids everywhere. This way the size
of struct objpool_head remains the same, which is a nice bonus.
Same BPF selftests benchmarks were used to evaluate the effect. Using
changes in previous patch (inlining of objpool_pop/objpool_push) as
baseline, here are the differences:
BASELINE
========
kretprobe : 9.937 ± 0.174M/s
kretprobe-multi: 10.440 ± 0.108M/s
AFTER
=====
kretprobe : 10.106 ± 0.120M/s (+1.7%)
kretprobe-multi: 10.515 ± 0.180M/s (+0.7%)
Link: https://lore.kernel.org/all/20240424215214.3956041-3-andrii@kernel.org/
Cc: Matt (Qiang) Wu <wuqiang.matt@bytedance.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
objpool_push() and objpool_pop() are very performance-critical functions
and can be called very frequently in kretprobe triggering path.
As such, it makes sense to allow compiler to inline them completely to
eliminate function calls overhead. Luckily, their logic is quite well
isolated and doesn't have any sprawling dependencies.
This patch moves both objpool_push() and objpool_pop() into
include/linux/objpool.h and marks them as static inline functions,
enabling inlining. To avoid anyone using internal helpers
(objpool_try_get_slot, objpool_try_add_slot), rename them to use leading
underscores.
We used kretprobe microbenchmark from BPF selftests (bench trig-kprobe
and trig-kprobe-multi benchmarks) running no-op BPF kretprobe/kretprobe.multi
programs in a tight loop to evaluate the effect. BPF own overhead in
this case is minimal and it mostly stresses the rest of in-kernel
kretprobe infrastructure overhead. Results are in millions of calls per
second. This is not super scientific, but shows the trend nevertheless.
BEFORE
======
kretprobe : 9.794 ± 0.086M/s
kretprobe-multi: 10.219 ± 0.032M/s
AFTER
=====
kretprobe : 9.937 ± 0.174M/s (+1.5%)
kretprobe-multi: 10.440 ± 0.108M/s (+2.2%)
Link: https://lore.kernel.org/all/20240424215214.3956041-2-andrii@kernel.org/
Cc: Matt (Qiang) Wu <wuqiang.matt@bytedance.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Take into account CONFIG_FTRACE_VALIDATE_RCU_IS_WATCHING when validating
that RCU is watching when trying to setup rethooko on a function entry.
One notable exception when we force rcu_is_watching() check is
CONFIG_KPROBE_EVENTS_ON_NOTRACE=y case, in which case kretprobes will use
old-style int3-based workflow instead of relying on ftrace, making RCU
watching check important to validate.
This further (in addition to improvements in the previous patch)
improves BPF multi-kretprobe (which rely on rethook) runtime throughput
by 2.3%, according to BPF benchmarks ([0]).
[0] https://lore.kernel.org/bpf/CAEf4BzauQ2WKMjZdc9s0rBWa01BYbgwHN6aNDXQSHYia47pQ-w@mail.gmail.com/
Link: https://lore.kernel.org/all/20240418190909.704286-2-andrii@kernel.org/
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Paul E. McKenney <paulmck@kernel.org>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Introduce CONFIG_FTRACE_VALIDATE_RCU_IS_WATCHING config option to
control whether ftrace low-level code performs additional
rcu_is_watching()-based validation logic in an attempt to catch noinstr
violations.
This check is expected to never be true and is mostly useful for
low-level validation of ftrace subsystem invariants. For most users it
should probably be kept disabled to eliminate unnecessary runtime
overhead.
This improves BPF multi-kretprobe (relying on ftrace and rethook
infrastructure) runtime throughput by 2%, according to BPF benchmarks ([0]).
[0] https://lore.kernel.org/bpf/CAEf4BzauQ2WKMjZdc9s0rBWa01BYbgwHN6aNDXQSHYia47pQ-w@mail.gmail.com/
Link: https://lore.kernel.org/all/20240418190909.704286-1-andrii@kernel.org/
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Paul E. McKenney <paulmck@kernel.org>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Active uprobes are stored in an RB tree and accesses to this tree are
dominated by read operations. Currently these accesses are serialized by
a spinlock but this leads to enormous contention when large numbers of
threads are executing active probes.
This patch converts the spinlock used to serialize access to the
uprobes_tree RB tree into a reader-writer spinlock. This lock type
aligns naturally with the overwhelmingly read-only nature of the tree
usage here. Although the addition of reader-writer spinlocks are
discouraged [0], this fix is proposed as an interim solution while an
RCU based approach is implemented (that work is in a nascent form). This
fix also has the benefit of being trivial, self contained and therefore
simple to backport.
We have used a uprobe benchmark from the BPF selftests [1] to estimate
the improvements. Each block of results below show 1 line per execution
of the benchmark ("the "Summary" line) and each line is a run with one
more thread added - a thread is a "producer". The lines are edited to
remove extraneous output.
The tests were executed with this driver script:
for num_threads in {1..20}
do
sudo ./bench -a -p $num_threads trig-uprobe-nop | grep Summary
done
SPINLOCK (BEFORE)
==================
Summary: hits 1.396 ± 0.007M/s ( 1.396M/prod)
Summary: hits 1.656 ± 0.016M/s ( 0.828M/prod)
Summary: hits 2.246 ± 0.008M/s ( 0.749M/prod)
Summary: hits 2.114 ± 0.010M/s ( 0.529M/prod)
Summary: hits 2.013 ± 0.009M/s ( 0.403M/prod)
Summary: hits 1.753 ± 0.008M/s ( 0.292M/prod)
Summary: hits 1.847 ± 0.001M/s ( 0.264M/prod)
Summary: hits 1.889 ± 0.001M/s ( 0.236M/prod)
Summary: hits 1.833 ± 0.006M/s ( 0.204M/prod)
Summary: hits 1.900 ± 0.003M/s ( 0.190M/prod)
Summary: hits 1.918 ± 0.006M/s ( 0.174M/prod)
Summary: hits 1.925 ± 0.002M/s ( 0.160M/prod)
Summary: hits 1.837 ± 0.001M/s ( 0.141M/prod)
Summary: hits 1.898 ± 0.001M/s ( 0.136M/prod)
Summary: hits 1.799 ± 0.016M/s ( 0.120M/prod)
Summary: hits 1.850 ± 0.005M/s ( 0.109M/prod)
Summary: hits 1.816 ± 0.002M/s ( 0.101M/prod)
Summary: hits 1.787 ± 0.001M/s ( 0.094M/prod)
Summary: hits 1.764 ± 0.002M/s ( 0.088M/prod)
RW SPINLOCK (AFTER)
===================
Summary: hits 1.444 ± 0.020M/s ( 1.444M/prod)
Summary: hits 2.279 ± 0.011M/s ( 1.139M/prod)
Summary: hits 3.422 ± 0.014M/s ( 1.141M/prod)
Summary: hits 3.565 ± 0.017M/s ( 0.891M/prod)
Summary: hits 2.671 ± 0.013M/s ( 0.534M/prod)
Summary: hits 2.409 ± 0.005M/s ( 0.401M/prod)
Summary: hits 2.485 ± 0.008M/s ( 0.355M/prod)
Summary: hits 2.496 ± 0.003M/s ( 0.312M/prod)
Summary: hits 2.585 ± 0.002M/s ( 0.287M/prod)
Summary: hits 2.908 ± 0.011M/s ( 0.291M/prod)
Summary: hits 2.346 ± 0.016M/s ( 0.213M/prod)
Summary: hits 2.804 ± 0.004M/s ( 0.234M/prod)
Summary: hits 2.556 ± 0.001M/s ( 0.197M/prod)
Summary: hits 2.754 ± 0.004M/s ( 0.197M/prod)
Summary: hits 2.482 ± 0.002M/s ( 0.165M/prod)
Summary: hits 2.412 ± 0.005M/s ( 0.151M/prod)
Summary: hits 2.710 ± 0.003M/s ( 0.159M/prod)
Summary: hits 2.826 ± 0.005M/s ( 0.157M/prod)
Summary: hits 2.718 ± 0.001M/s ( 0.143M/prod)
Summary: hits 2.844 ± 0.006M/s ( 0.142M/prod)
The numbers in parenthesis give averaged throughput per thread which is
of greatest interest here as a measure of scalability. Improvements are
in the order of 22 - 68% with this particular benchmark (mean = 43%).
V2:
- Updated commit message to include benchmark results.
[0] https://docs.kernel.org/locking/spinlocks.html
[1] https://github.com/torvalds/linux/blob/master/tools/testing/selftests/bpf/benchs/bench_trigger.c
Link: https://lore.kernel.org/all/20240422102306.6026-1-jonathan.haslam@gmail.com/
Signed-off-by: Jonathan Haslam <jonathan.haslam@gmail.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
The function rethook_find_ret_addr() prints a warning message and returns 0
when the target task is running and is not the "current" task in order to
prevent the incorrect return address, although it still may return an
incorrect address.
However, the warning message turns into noise when BPF profiling programs
call bpf_get_task_stack() on running tasks in a firm with a large number of
hosts.
The callers should be aware and willing to take the risk of receiving an
incorrect return address from a task that is currently running other than
the "current" one. A warning is not needed here as the callers are intent
on it.
Link: https://lore.kernel.org/all/20240408175140.60223-1-thinker.li@gmail.com/
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
We are going to store callbacks in following change,
so this will ease up the code.
Link: https://lore.kernel.org/all/20240207153550.856536-2-jolsa@kernel.org/
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
This patch adds fprobe test cases for new print format type "%pd/%pD".The
test cases test the following items:
1. Test "%pd" type for dput();
2. Test "%pD" type for vfs_read();
This test case require enable CONFIG_HAVE_FUNCTION_ARG_ACCESS_API configuration.
Link: https://lore.kernel.org/all/20240322064308.284457-6-yebin10@huawei.com/
Signed-off-by: Ye Bin <yebin10@huawei.com>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
This patch adds test cases for new print format type "%pd/%pD".The test cases
test the following items:
1. Test README if add "%pd/%pD" type;
2. Test "%pd" type for dput();
3. Test "%pD" type for vfs_read();
This test case require enable CONFIG_HAVE_FUNCTION_ARG_ACCESS_API configuration.
Link: https://lore.kernel.org/all/20240322064308.284457-5-yebin10@huawei.com/
Signed-off-by: Ye Bin <yebin10@huawei.com>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Similar to printk() '%pd' is for fetch dentry's name from struct dentry's
pointer, and '%pD' is for fetch file's name from struct file's pointer.
Link: https://lore.kernel.org/all/20240322064308.284457-4-yebin10@huawei.com/
Signed-off-by: Ye Bin <yebin10@huawei.com>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
As like '%pd' type, this patch supports print type '%pD' for print file's
name. For example "name=$arg1:%pD" casts the `$arg1` as (struct file*),
dereferences the "file.f_path.dentry.d_name.name" field and stores it to
"name" argument as a kernel string.
Here is an example:
[tracing]# echo 'p:testprobe vfs_read name=$arg1:%pD' > kprobe_event
[tracing]# echo 1 > events/kprobes/testprobe/enable
[tracing]# grep -q "1" events/kprobes/testprobe/enable
[tracing]# echo 0 > events/kprobes/testprobe/enable
[tracing]# grep "vfs_read" trace | grep "enable"
grep-15108 [003] ..... 5228.328609: testprobe: (vfs_read+0x4/0xbb0) name="enable"
Note that this expects the given argument (e.g. $arg1) is an address of struct
file. User must ensure it.
Link: https://lore.kernel.org/all/20240322064308.284457-3-yebin10@huawei.com/
[Masami: replaced "previous patch" with '%pd' type]
Signed-off-by: Ye Bin <yebin10@huawei.com>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
During fault locating, the file name needs to be printed based on the
dentry address. The offset needs to be calculated each time, which
is troublesome. Similar to printk, kprobe support print type '%pd' for
print dentry's name. For example "name=$arg1:%pd" casts the `$arg1`
as (struct dentry *), dereferences the "d_name.name" field and stores
it to "name" argument as a kernel string.
Here is an example:
[tracing]# echo 'p:testprobe dput name=$arg1:%pd' > kprobe_events
[tracing]# echo 1 > events/kprobes/testprobe/enable
[tracing]# grep -q "1" events/kprobes/testprobe/enable
[tracing]# echo 0 > events/kprobes/testprobe/enable
[tracing]# cat trace | grep "enable"
bash-14844 [002] ..... 16912.889543: testprobe: (dput+0x4/0x30) name="enable"
grep-15389 [003] ..... 16922.834182: testprobe: (dput+0x4/0x30) name="enable"
grep-15389 [003] ..... 16922.836103: testprobe: (dput+0x4/0x30) name="enable"
bash-14844 [001] ..... 16931.820909: testprobe: (dput+0x4/0x30) name="enable"
Note that this expects the given argument (e.g. $arg1) is an address of struct
dentry. User must ensure it.
Link: https://lore.kernel.org/all/20240322064308.284457-2-yebin10@huawei.com/
Signed-off-by: Ye Bin <yebin10@huawei.com>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
It's very common with BPF-based uprobe/uretprobe use cases to have
a system-wide (not PID specific) probes used. In this case uprobe's
trace_uprobe_filter->nr_systemwide counter is bumped at registration
time, and actual filtering is short circuited at the time when
uprobe/uretprobe is triggered.
This is a great optimization, and the only issue with it is that to even
get to checking this counter uprobe subsystem is taking
read-side trace_uprobe_filter->rwlock. This is actually noticeable in
profiles and is just another point of contention when uprobe is
triggered on multiple CPUs simultaneously.
This patch moves this nr_systemwide check outside of filter list's
rwlock scope, as rwlock is meant to protect list modification, while
nr_systemwide-based check is speculative and racy already, despite the
lock (as discussed in [0]). trace_uprobe_filter_remove() and
trace_uprobe_filter_add() already check for filter->nr_systewide
explicitly outside of __uprobe_perf_filter, so no modifications are
required there.
Confirming with BPF selftests's based benchmarks.
BEFORE (based on changes in previous patch)
===========================================
uprobe-nop : 2.732 ± 0.022M/s
uprobe-push : 2.621 ± 0.016M/s
uprobe-ret : 1.105 ± 0.007M/s
uretprobe-nop : 1.396 ± 0.007M/s
uretprobe-push : 1.347 ± 0.008M/s
uretprobe-ret : 0.800 ± 0.006M/s
AFTER
=====
uprobe-nop : 2.878 ± 0.017M/s (+5.5%, total +8.3%)
uprobe-push : 2.753 ± 0.013M/s (+5.3%, total +10.2%)
uprobe-ret : 1.142 ± 0.010M/s (+3.8%, total +3.8%)
uretprobe-nop : 1.444 ± 0.008M/s (+3.5%, total +6.5%)
uretprobe-push : 1.410 ± 0.010M/s (+4.8%, total +7.1%)
uretprobe-ret : 0.816 ± 0.002M/s (+2.0%, total +3.9%)
In the above, first percentage value is based on top of previous patch
(lazy uprobe buffer optimization), while the "total" percentage is
based on kernel without any of the changes in this patch set.
As can be seen, we get about 4% - 10% speed up, in total, with both lazy
uprobe buffer and speculative filter check optimizations.
[0] https://lore.kernel.org/bpf/20240313131926.GA19986@redhat.com/
Reviewed-by: Jiri Olsa <jolsa@kernel.org>
Link: https://lore.kernel.org/all/20240318181728.2795838-4-andrii@kernel.org/
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
uprobe_cpu_buffer and corresponding logic to store uprobe args into it
are used for uprobes/uretprobes that are created through tracefs or
perf events.
BPF is yet another user of uprobe/uretprobe infrastructure, but doesn't
need uprobe_cpu_buffer and associated data. For BPF-only use cases this
buffer handling and preparation is a pure overhead. At the same time,
BPF-only uprobe/uretprobe usage is very common in practice. Also, for
a lot of cases applications are very senstivie to performance overheads,
as they might be tracing a very high frequency functions like
malloc()/free(), so every bit of performance improvement matters.
All that is to say that this uprobe_cpu_buffer preparation is an
unnecessary overhead that each BPF user of uprobes/uretprobe has to pay.
This patch is changing this by making uprobe_cpu_buffer preparation
optional. It will happen only if either tracefs-based or perf event-based
uprobe/uretprobe consumer is registered for given uprobe/uretprobe. For
BPF-only use cases this step will be skipped.
We used uprobe/uretprobe benchmark which is part of BPF selftests (see [0])
to estimate the improvements. We have 3 uprobe and 3 uretprobe
scenarios, which vary an instruction that is replaced by uprobe: nop
(fastest uprobe case), `push rbp` (typical case), and non-simulated
`ret` instruction (slowest case). Benchmark thread is constantly calling
user space function in a tight loop. User space function has attached
BPF uprobe or uretprobe program doing nothing but atomic counter
increments to count number of triggering calls. Benchmark emits
throughput in millions of executions per second.
BEFORE these changes
====================
uprobe-nop : 2.657 ± 0.024M/s
uprobe-push : 2.499 ± 0.018M/s
uprobe-ret : 1.100 ± 0.006M/s
uretprobe-nop : 1.356 ± 0.004M/s
uretprobe-push : 1.317 ± 0.019M/s
uretprobe-ret : 0.785 ± 0.007M/s
AFTER these changes
===================
uprobe-nop : 2.732 ± 0.022M/s (+2.8%)
uprobe-push : 2.621 ± 0.016M/s (+4.9%)
uprobe-ret : 1.105 ± 0.007M/s (+0.5%)
uretprobe-nop : 1.396 ± 0.007M/s (+2.9%)
uretprobe-push : 1.347 ± 0.008M/s (+2.3%)
uretprobe-ret : 0.800 ± 0.006M/s (+1.9)
So the improvements on this particular machine seems to be between 2% and 5%.
[0] https://github.com/torvalds/linux/blob/master/tools/testing/selftests/bpf/benchs/bench_trigger.c
Reviewed-by: Jiri Olsa <jolsa@kernel.org>
Link: https://lore.kernel.org/all/20240318181728.2795838-3-andrii@kernel.org/
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Move the logic of fetching temporary per-CPU uprobe buffer and storing
uprobes args into it to a new helper function. Store data size as part
of this buffer, simplifying interfaces a bit, as now we only pass single
uprobe_cpu_buffer reference around, instead of pointer + dsize.
This logic was duplicated across uprobe_dispatcher and uretprobe_dispatcher,
and now will be centralized. All this is also in preparation to make
this uprobe_cpu_buffer handling logic optional in the next patch.
Link: https://lore.kernel.org/all/20240318181728.2795838-2-andrii@kernel.org/
[Masami: update for v6.9-rc3 kernel]
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Reviewed-by: Jiri Olsa <jolsa@kernel.org>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
