Fix: handle default switch case with -EINVAL in __lttng_counter_add

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Introduce lttng smp_store_release smp_load_acquire wrappers

Kernels prior to 3.14 do not implement smp_store_release nor
smp_load_acquire. Implement our own wrappers with smp_mb instead for
those older kernels.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

fix: include byteorder.h for 'cpu_to_*'

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ie9da5b1d17b7311ac0d6898c3d23a6c4d592927a

Fix: notifier: use store-release/load-acquire for error counter

The "record_error" operation is executed concurrently with setting the
error counter in the notifier group without locking, so we need to
explicitly provide existance guarantees.

The only visible transition is from NULL -> !NULL, because the only
situation reverting the error counter back to NULL is on destruction of
the notification group, after an RCU synchronisation guarantees that no
record_error can observe this pointer anymore.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Fix: event-notifier: Groups may not have an error counter

Issue
=====
If no error counter is attached to an event notifier group, calling
the `record_error()` function leads to a NULL pointer dereference.

Solution
========
Check if the error_counter field is allocated before dereferencing it.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I7ba68f9ae5b7b66cea538cf030bc5e1287d6eace

Bump LTTNG_KERNEL_EVENT_NOTIFIER_PADDING to 32 bytes

Allows using 3 more 64-bit integers, and one extension pointer.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Fix: remove dead code in msgpack.c

Identified by Coverity. Comparing unsigned types < 0 is never true.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Fix: counter: cast UINT*_MAX to 64-bit signed type before negative

(long long)-UINT32_MAX leads to value 1 which is not what we expect.
This is due to implicit type promotion from unsigned to signed 32-bit
integer.

Apply this to 8-bit and 16-bit types as well even though they are
not affected by this issue to keep things regular.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Fix: include limits.h wrapper from libcounter

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Fix: include kernel.h in wrapper/limits.h

linux/kernel.h is present in older kernels, but linux/limits.h is not.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Introduce limits wrapper

Older kernels (v3.13.11 and older on the CI) miss the U32_MAX
macro. Implement them in a wrapper.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Introduce lazy system call event creation

Introduce lazy system call event creation. If only specific system call
events are enabled, only create the metadata for those.

Align implementation of system call event probes for recording events
with their notifier counterpart.

This is part of the background work required for a future feature (trace
hit counters and map support) which brings benefits on its own.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Implement event notifier error counter

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I566d34ab92aaa642f50407d913b16c4a6b8479a2

Implement lib counter

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I47424817ca874357d5ae4349a048cce889988a76

capture: Replace FIXME by explanation of the limit

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Implement capturing payload on event notifier

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ibdaa07ecfe36ecd1c340942821503ea4cb494e2c

Fix: comment related to filter bytecode list

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Fix: use vmalloc for filter bytecode allocation

Given that the filter bytecode can reach 64kB, use vmalloc rather than
kmalloc to minimize the chances of failure on systems with fragmented
memory.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Implement enum and sequence capture serialization functions

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I2a342e5e1d74e7c2f676b21822d8b7f171d68ebd

bytecode: Add `OBJECT_TYPE_{UN,}SIGNED_ENUM` type

Enumerations are currently converted to their integer counterparts as
soon as they are encountered.

In order to use them in captures, we need to differentiate the
enumerations from integers for the entirety of the interpretation.

This commit adds the `OBJECT_TYPE_SIGNED_ENUM` and
`OBJECT_TYPE_UNSIGNED_ENUM` to keep track of these objects

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ic0ab518588d7250190e42576d9baba2e8d8ce94a

Generalize `lttng_enabler_link_bytecode()` bytecode list

So it can be used for both filter and capture bytecode runtimes

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I14dd5eb1eb018d358446d68ba14aa376d1647545

Cleanup: rename `_lttng_filter_link_bytecode()` -> `link_bytecode()`

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I8a9b5d97eae502181ae8673caf39e4f54fdbdeb6

Add `interpreter_funcs` to `lttng_bytecode_runtime`

Background
==========
The current filter interpreter function signature looks like this:

uint64_t lttng_bytecode_filter_interpret(void *filter_data,
              const char *filter_stack_data);

The upcoming capture interpreter function will need an output parameter
to extract the top of stack register. It will look like this:

uint64_t lttng_bytecode_capture_interpret(void *filter_data,
              const char *filter_stack_data, struct output_register *output);

Problems
========
We can't reuse the same function pointer field in `struct
lttng_bytecode_runtime` as both interpreter functions will have
different signatures.

We can't change the signature of this existing filter function because
it's used in the tracepoint probes.

Solution
========
Add a union of callbacks to hold both interpreter functions. This also
doesn't change the layout of the `struct lttng_bytecode_runtime`
objects.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ic4423100e1f97654fe43c0927d3b81de2d1d724f

bytecode: rename `lttng_filter_sync_state()` -> `lttng_bytecode_filter_sync_state()`

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I2d63001ed633236a98eba6398ad06fccf83fde2c

Rename filter bytecode types and files

File renames:
- filter-bytecode.h -> bytecode.h
- lttng-filter-interpreter.c -> lttng-bytecode-interpreter.c
- lttng-filter-specialize.c -> lttng-bytecode-specialize.c
- lttng-filter-validator.c -> lttng-bytecode-validator.c
- lttng-filter.c -> lttng-bytecode.c
- lttng-filter.h -> lttng-bytecode.h

Function renames:
- `lttng_filter_interpret_bytecode_false()` -> `lttng_bytecode_filter_interpret_false()`
- `lttng_filter_interpret_bytecode()` -> `lttng_bytecode_filter_interpret()`
- `lttng_filter_specialize_bytecode()` -> `lttng_bytecode_specialize()`
- `lttng_filter_validate_bytecode()` -> `lttng_bytecode_validate()`

Type renames
- `filter_opcode_t` to `bytecode_opcode_t`

Enum renames:
- `enum filter_op` to `enum bytecode_op`
- `FILTER_OP_*` to `BYTECODE_OP_*`
- `enum lttng_filter_ret` -> `enum lttng_bytecode_interpreter_ret`
- `LTTNG_FILTER_DISCARD` -> `LTTNG_INTERPRETER_DISCARD`
- `LTTNG_FILTER_RECORD_FLAG` -> `LTTNG_INTERPRETER_RECORD_FLAG`

Define renames:
- `FILTER_STACK_EMPTY` -> `INTERPRETER_STACK_EMPTY`
- `FILTER_STACK_LEN`-> `INTERPRETER_STACK_LEN`
- `FILTER_MAX_DATA_LEN` -> `BYTECODE_MAX_DATA_LEN`

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I8cd3a15c8ba19aa1d66e611913a1b08e51d94a43

bytecode: generalize `struct lttng_filter_bytecode_node`

Rename `struct lttng_filter_bytecode_node` to `struct
lttng_bytecode_node` so it can be used by capture bytecode as well.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I27162522bb20d0fdce6af551fbd982a791d1067c

Add msgpack implementation for serializing captures

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I7d9cfd4906c5c047cfb4fc9467b293c4e895523d

bytecode: allow interpreter to return any type

The bytecode interpreter when used by capture bytecode needs to return
types other than an integer or dynamic type.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I099749183fbd0622f258f9c38e37fdb167493a0b

bytecode: propagate `rev_bo` of element

When specializing and executing bytecode.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I86aea5b5733f92c56564c6352bd6dcb85f6a2d30

bytecode: set register type to `REG_PTR` even if not used

There was no need to set the field when using filter as the next
instruction would assume that the top of stack is a `REG_PTR`.

With the upcoming capture feature, we need to ensure this field is
consistent for extraction.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I9c60416dd452949e584fadd70b15cdc3d402aa46

Add `lttng_bytecode_interpret_format_output()` for top of stack extraction

This new static function will be used to extract the register on the top of
stack after the execution of the bytecode. This is currently not used by the
filter bytecode but will be used by capture bytecode.

The returned value is saved in a tagged union struct named `struct
lttng_interpreter_output` and can be used by the caller of the interpreter
function.

Typically, this struct will be allocated on the stack to avoid dynamic
allocation inside the tracepoint probes.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I1cfd3ab6e84b7e308c48ed7a8a9555a3e338eea7

bytecode: add `REG_U64` interpreter register type

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I53c12a54cbd416617834982bbd2b7cf528d41a76

Fix: filter validator: refuse string and star glob input to bitwise operation

The validator refuses input ax=string,bx=unknown, but accepts input
ax=unknown,bx=string. Both inputs should be refused.

The same goes for the error glob input.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Fix: bytecode: Validate register type for instructions expecting unknown type

The bytecode validator allows unknown type as input for some
instructions which are not specialized. The interpreter therefore needs
to check the register type for their input.

Thie requires that every instruction in the interpreter sets the
register type of the output it populates (unless it is unchanged).

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I3339c36340645937b801f6bf6dbf517d06416a14

Cleanup: Rename filter functions/fields to mention "filter"

This will be cleaner when we introduce the capture bytecode functions
and fields.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I04aca5bfd31f2526b24fe3a4b2e8f2b1c1b482f9

Implement event notifiers for syscalls

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I31e60d0d758b93ada11d208f583d71f05168c014

Fix: syscalls: address of statically allocated element never null

This check is intended to confirm that the table element for that syscall
is indeed populated but checked that the element is NULL. This was never
the case because the address of an element of a statically allocated
array cannot be NULL.

Fix this by check if the function pointer is NULL instead. This means
that the element is not populated.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I1d769d6609fa4517199f022e1a262c4494c8f63a

Rename LTTNG_SYSCALL_MATCH_ -> LTTNG_KERNEL_SYSCALL_MATCH_

This is done to keep the same name scheme used to all ABI enums.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I8e1010ab21a47b7f1e519df498acd230315cdc26

Allow LTTNG_KERNEL_SYSCALL_{ENTRY, EXIT}

Signed-off-by: Francis Deslauriers <fdeslaur@gmail.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I1ea097797da5db474f2f33d779f66254b7979c46

syscalls: extract `lttng_syscall_filter_enable()` for reuse

The syscall event notifiers will reuse the concept of syscall filtering
to avoid needlessly preparing arguments for disabled syscalls.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I64395a031e526e8485e10b4b72f653058c8d0a38

Cleanup: syscall: remove unused `syscall_name` field

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I5d37b253348ac4812602d89b6be39a7abd1be4ff

fix: adjust version range for trace_find_free_extent()

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Iaa6088092cf58b4d29d55f3ff9586c57ae272302

Improve the release script

  * Use git-archive, this removes all custom code to cleanup the repo, it
    can now be used in an unclean repo as the code will be exported from
    a specific tag.
  * Add parameters, this will allow using the script on any machine
    while keeping the default behavior for the maintainer.

Change-Id: I9f29d0e1afdbf475d0bbaeb9946ca3216f725e86
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Add release maintainer script

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

fix: include order for older kernels

Fixes a build failure on v3.0 and v3.1.

Change-Id: Ic48512d2aa5ee46678e67d147b92dba6d0959615
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

fix: tracepoint: Optimize using static_call() (v5.10)

See upstream commit :

  commit d25e37d89dd2f41d7acae0429039d2f0ae8b4a07
  Author: Steven Rostedt (VMware) <rostedt@goodmis.org>
  Date:   Tue Aug 18 15:57:52 2020 +0200

    tracepoint: Optimize using static_call()

    Currently the tracepoint site will iterate a vector and issue indirect
    calls to however many handlers are registered (ie. the vector is
    long).

    Using static_call() it is possible to optimize this for the common
    case of only having a single handler registered. In this case the
    static_call() can directly call this handler. Otherwise, if the vector
    is longer than 1, call a function that iterates the whole vector like
    the current code.

Change-Id: I739dd84d62cc1a821b8bd8acff74fa29aa25d22f
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

fix: KVM: x86/mmu: Return unique RET_PF_* values if the fault was fixed (v5.10)

See upstream commit :

  commit c4371c2a682e0da1ed2cd7e3c5496f055d873554
  Author: Sean Christopherson <sean.j.christopherson@intel.com>
  Date:   Wed Sep 23 15:04:24 2020 -0700

    KVM: x86/mmu: Return unique RET_PF_* values if the fault was fixed

    Introduce RET_PF_FIXED and RET_PF_SPURIOUS to provide unique return
    values instead of overloading RET_PF_RETRY.  In the short term, the
    unique values add clarity to the code and RET_PF_SPURIOUS will be used
    by set_spte() to avoid unnecessary work for spurious faults.

    In the long term, TDX will use RET_PF_FIXED to deterministically map
    memory during pre-boot.  The page fault flow may bail early for benign
    reasons, e.g. if the mmu_notifier fires for an unrelated address.  With
    only RET_PF_RETRY, it's impossible for the caller to distinguish between
    "cool, page is mapped" and "darn, need to try again", and thus cannot
    handle benign cases like the mmu_notifier retry.

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ie0855c78852b45f588e131fe2463e15aae1bc023

fix: kvm: x86/mmu: Add TDP MMU PF handler (v5.10)

See upstream commit :

  commit bb18842e21111a979e2e0e1c5d85c09646f18d51
  Author: Ben Gardon <bgardon@google.com>
  Date:   Wed Oct 14 11:26:50 2020 -0700

    kvm: x86/mmu: Add TDP MMU PF handler

    Add functions to handle page faults in the TDP MMU. These page faults
    are currently handled in much the same way as the x86 shadow paging
    based MMU, however the ordering of some operations is slightly
    different. Future patches will add eager NX splitting, a fast page fault
    handler, and parallel page faults.

    Tested by running kvm-unit-tests and KVM selftests on an Intel Haswell
    machine. This series introduced no new failures.

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ie56959cb6c77913d2f1188b0ca15da9114623a4e

fix: KVM: x86: Add intr/vectoring info and error code to kvm_exit tracepoint (v5.10)

See upstream commit :

  commit 235ba74f008d2e0936b29f77f68d4e2f73ffd24a
  Author: Sean Christopherson <sean.j.christopherson@intel.com>
  Date:   Wed Sep 23 13:13:46 2020 -0700

    KVM: x86: Add intr/vectoring info and error code to kvm_exit tracepoint

    Extend the kvm_exit tracepoint to align it with kvm_nested_vmexit in
    terms of what information is captured.  On SVM, add interrupt info and
    error code, while on VMX it add IDT vectoring and error code.  This
    sets the stage for macrofying the kvm_exit tracepoint definition so that
    it can be reused for kvm_nested_vmexit without loss of information.

    Opportunistically stuff a zero for VM_EXIT_INTR_INFO if the VM-Enter
    failed, as the field is guaranteed to be invalid.  Note, it'd be
    possible to further filter the interrupt/exception fields based on the
    VM-Exit reason, but the helper is intended only for tracepoints, i.e.
    an extra VMREAD or two is a non-issue, the failed VM-Enter case is just
    low hanging fruit.

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I638fa29ef7d8bb432de42a33f9ae4db43259b915

fix: ext4: fast commit recovery path (v5.10)

See upstream commit :

  commit 8016e29f4362e285f0f7e38fadc61a5b7bdfdfa2
  Author: Harshad Shirwadkar <harshadshirwadkar@gmail.com>
  Date:   Thu Oct 15 13:37:59 2020 -0700

    ext4: fast commit recovery path

    This patch adds fast commit recovery path support for Ext4 file
    system. We add several helper functions that are similar in spirit to
    e2fsprogs journal recovery path handlers. Example of such functions
    include - a simple block allocator, idempotent block bitmap update
    function etc. Using these routines and the fast commit log in the fast
    commit area, the recovery path (ext4_fc_replay()) performs fast commit
    log recovery.

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ia65cf44e108f2df0b458f0d335f33a8f18f50baa

fix: btrfs: make ordered extent tracepoint take btrfs_inode (v5.10)

See upstream commit :

  commit acbf1dd0fcbd10c67826a19958f55a053b32f532
  Author: Nikolay Borisov <nborisov@suse.com>
  Date:   Mon Aug 31 14:42:40 2020 +0300

    btrfs: make ordered extent tracepoint take btrfs_inode

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I096d0801ffe0ad826cfe414cdd1c0857cbd2b624

fix: btrfs: tracepoints: output proper root owner for trace_find_free_extent() (v5.10)

See upstream commit :

  commit 437490fed3b0c9ae21af8f70e0f338d34560842b
  Author: Qu Wenruo <wqu@suse.com>
  Date:   Tue Jul 28 09:42:49 2020 +0800

    btrfs: tracepoints: output proper root owner for trace_find_free_extent()

    The current trace event always output result like this:

     find_free_extent: root=2(EXTENT_TREE) len=16384 empty_size=0 flags=4(METADATA)
     find_free_extent: root=2(EXTENT_TREE) len=16384 empty_size=0 flags=4(METADATA)
     find_free_extent: root=2(EXTENT_TREE) len=8192 empty_size=0 flags=1(DATA)
     find_free_extent: root=2(EXTENT_TREE) len=8192 empty_size=0 flags=1(DATA)
     find_free_extent: root=2(EXTENT_TREE) len=4096 empty_size=0 flags=1(DATA)
     find_free_extent: root=2(EXTENT_TREE) len=4096 empty_size=0 flags=1(DATA)

    T's saying we're allocating data extent for EXTENT tree, which is not
    even possible.

    It's because we always use EXTENT tree as the owner for
    trace_find_free_extent() without using the @root from
    btrfs_reserve_extent().

    This patch will change the parameter to use proper @root for
    trace_find_free_extent():

    Now it looks much better:

     find_free_extent: root=5(FS_TREE) len=16384 empty_size=0 flags=36(METADATA|DUP)
     find_free_extent: root=5(FS_TREE) len=8192 empty_size=0 flags=1(DATA)
     find_free_extent: root=5(FS_TREE) len=16384 empty_size=0 flags=1(DATA)
     find_free_extent: root=5(FS_TREE) len=4096 empty_size=0 flags=1(DATA)
     find_free_extent: root=5(FS_TREE) len=8192 empty_size=0 flags=1(DATA)
     find_free_extent: root=5(FS_TREE) len=16384 empty_size=0 flags=36(METADATA|DUP)
     find_free_extent: root=7(CSUM_TREE) len=16384 empty_size=0 flags=36(METADATA|DUP)
     find_free_extent: root=2(EXTENT_TREE) len=16384 empty_size=0 flags=36(METADATA|DUP)
     find_free_extent: root=1(ROOT_TREE) len=16384 empty_size=0 flags=36(METADATA|DUP)

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I1d674064d29b31417e2acffdeb735f5052a87032

fix: objtool: Rename frame.h -> objtool.h (v5.10)

See upstream commit :

  commit 00089c048eb4a8250325efb32a2724fd0da68cce
  Author: Julien Thierry <jthierry@redhat.com>
  Date:   Fri Sep 4 16:30:25 2020 +0100

    objtool: Rename frame.h -> objtool.h

    Header frame.h is getting more code annotations to help objtool analyze
    object files.

    Rename the file to objtool.h.

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ic2283161bebcbf1e33b72805eb4d2628f4ae3e89

Revert "Implement event notifiers for syscalls"

This reverts commit 8ced8896fe832af52b749d429b8eceb872a83d1b.

This commit was not ready and was committed by error.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Fix: ressource leak in id tracker

Memory leak found by Coverity:

CID 1412251 (#2 of 2): Resource leak (RESOURCE_LEAK)
21. leaked_storage: Variable head going out of scope leaks the storage it points to.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Implement event notifiers for syscalls

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Change-Id: Ic8f17feb45aef6e933252908c761d3241123cfe4

lttng-syscalls.c: extract function calling actual probe

This function will be reused by the event notifier infrastructure.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Iad25a44202d74eac8f75af108eb8297d82303d63

Namespace syscall code relating to events

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ia1c3cf01d82681dfc77c2786ab58259085d349c8

Implement event notifiers for uprobes

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I28ec581f412c1633d0cdb675020ad2c642c6c768

Namespace uprobe functions relating to events

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I4130b5e9fe88ee5121646a3303b262266db658be

doc: event notifier on kretprobe is not supported

The kretprobe behavior is to fire twice (entry and exit of target
function), placing an event notifier on such function does not make
sense at first glance. If we come up with a use case it will be quite
easy to enable.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I6f214501706d1ef170c81b80a1f82c039d687502

Implement event notifiers for kprobes

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I69075c51f9df4ae89457967d96863dcf370d4570

Namespace kprobe functions relating to events

The event notifier support for kprobe will soon be introduce and kprobe
code will be reused.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ia44aea762c158e922c1fafed381fca6919bea188

Implement event notifiers for tracepoints

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I256cc2c54179255402c5b7bc7d439508f0a6adbf

Implement event notifier probes

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I807b344edb65ad8d1343187b987f693113029794

Fix: event notifier: adapt read iterator state to poll expectations

When completing to read a subbuffer, ensure that the state of the
iterator is moved forward so the "put_subbuf" is performed before
returning to the user, so poll() will not return POLLIN when there
is actually no data available to read.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ia94b5dcb0c72d8548325b1004f214044f50fd191

Fix: event-notifier: do not flush packet if it only contains subbuf header

This poll behavior returns POLLIN in situations where there is
actually no event to read, which causes read to block when it
should not.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I59e56cd4da9907b6f9ccdc14c6037f0f72e4505e

Implement lttng_event_notifier_group_notif_fops read, poll, open, release ABI

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ia95c67306226202cfd10f3745ddeecb76b1ef1a7

Implement event notifier send notification

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Change-Id: Ibdab5a25439da7d4c26e480c41cdd655a7d58d82

Add event notifier and event notifier enabler

Idea
====
The purpose of the event notifiers is to allow the session daemon to
react to events in the tracer. For example, the user will be able to
start or stop tracing on a session when a specific tracepoint is fired.

An event notifier is really similar to a regular event. The main
difference is that when the tracepoint is fired the action of the event
notifier is to notify the session daemon about it. This mechanism will
use a special purpose ring buffer to expose these notifications to
userspace.

Unlike regular events, there are no claim on the timeliness of such
notifications.

Implementation
==============
This commit adds structures and functions related to event notifiers
mimicking what we currently do with regular events.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I962e6c7051693d6e4a79f89758f8bf1ebda6c148

Implement event notifier group create

Event notifier groups will contain the event notifiers.
All event notifiers of a group will share the same ring buffer to save
the tracer notifications.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I7c38fbfd26517d00c8de38ca1981da623d570529

Add token to `struct lttng_kernel_event`

This token is provided by the user when registering an event rule to the
kernel. It is going to used to identify messages from event notifiers, and
counter bucket owners in the upcoming event notifier and map counter
features.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I58cd6b6e33b97cc21e50cc0b36bef4b9e4224423

lttng-events: move lttng_transport_find earlier in source file

Will be used by upcoming event notifier code which is earlier in source
file.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Idb7be8c682258195166f3b513fee2aa98656de35

lib ring buffer: move subbuffer_consume_record into LTTNG_RING_BUFFER_COUNT_EVENTS ifdef

When event accounting is disabled, counting of event records consumed by
the iterator should be disabled as well, otherwise it triggers
CHAN_WARN_ON() because the accounting of events produced is not
performed.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Id8b9e657ee420886b409be1f05ef08a0807fefdc

lib ring buffer iterator: introduce lib_ring_buffer_put_current_record

Ensure that the current subbuffer is put after client code has read the
payload of the current record.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Id2173ea67213f7ef8e7395b49c5aa8fff0aefffc

Introduce event notifier lib ring buffer client

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I89da147ee956f5759c49bd992bf33fe760d79591

lttng_abi_create_stream_fd: expect fd name as parameter

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ic9711863b58307d3ed6cc782efc78e4f59345950

LTTng ring buffer clients: expect void pointer as private data to create channel

Triggers will create a channel without using the lttng_channel objects,
so allow any type of private data.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I0725616c84e401c9fcbf00a405a2e2d0f1078979

lib ring buffer: use irq_work for wakeup by writer

Using irq_work (like perf does) allows using an interrupt handler
firing soon after the instrumentation execution to issue the wakeups.

This allows the RING_BUFFER_WAKEUP_BY_WRITER ring buffer configuration
to be entirely lock-free, which allows using it in NMI context for
general tracing purposes.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I842ff15736f53d1283cf953804d803f70779652b

Rename `lttng_event_{get,put}()` to `lttng_event_desc_{get,put}()`

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I99a8b4cdf191555c28da5a38a1e65661421fd7fc

Cleanup: extract function to borrow hashlist bucket

This is going to reused by the trigger system.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ie6d032374c3991d0a75ad4737e7f082fbc1a74b1

Decouple `struct lttng_event` from filter code

The filter infrastructure will be used by event notifiers and decoupling
this will allow for massive code reuse.

Of all `struct lttng_event`'s fields, filter code needs:
1. The `const struct lttng_event_desc *desc` field,
2. The `struct cds_list_head bytecode_runtime_head` list.

These fields are used to do the tracepoint field relocation
(`apply_field_reloc()` and `specialize_event_payload_lookup()`).

Considering that only these two field are needed, we can pass them
directly to these functions.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: If569b7d315700660aa84241d112668f2451b715a

Rename `lttng_create_*_if_missing()` in anticipation of event notifiers

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I58a799f992a53215ff04896b783e7ebe31965b7c

Extract event enabler fields to specialized struct

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I356d9b91c6e20c288ca931a4d449a54b67f3937c

Docs: explain why unused `lttng_enabler::ctx` is kept around

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: If2c6b9203ea324bb1ff4051b0a705e7303dbf3a6

Rename `enum lttng_enabler_type` to `_format_type`

This will avoid confusion between the different types of enablers
(event notifier enablers and event enablers).

- Enabler format types describe the way the event name matching is done.

- Enabler types will describe the type of enablers (event
  notifier vs event)

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ic71d05159c5f244d0b1ad74f9c0ee6247fcdfbbb

Test: add signed value and enum for testings of event notifier capture

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I4be725e3ed1e2f94420f4cdcf5ab6ac7962e2464

Cleanup: remove usage of enum in ABI structures

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I3730f7c0341028b25231c368166ee6e5fd74fa5d

Fix: type mismatch in clone instrumentation

The data and metadata types should all agree to use "unsigned long",
else it triggers babeltrace trace parsing errors.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

syscalls: Make clone()'s `flags` field a 2 enum struct.

The clone system call has a flags field, whose values are defined in
uapi/linux/sched.h file. This field is now a struct made of 2
enumerations to make the values more readable/meaningful.

The `flags` field has two parts:
  1. exit signal: the least significant byte of the `unsigned long` is
      the signal the kernel need to send to the parent process on child
      exit,
  2. clone options: the remaining bytes of the `unsigned long` is used a
      bitwise flag for the clone options.

Those 2-in-1 fields should be printed using two different CTF fields.

Here's an example babeltrace output of the clone system call:
  syscall_entry_clone: { cpu_id = 2 }, { flags = { exit_signal = ( "SIGCHLD" : container = 0x11 ), options = ( "CLONE_CHILD_CLEARTID" | "CLONE_CHILD_SETTID" : container = 0x12000 ) }

Change-Id: Ic375b59fb3b6564f036e1af24d66c0c7069b47d6
Signed-off-by: Geneviève Bastien <gbastien@versatic.net>
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

fix: strncpy equals destination size warning

Some versions of GCC when called with -Wstringop-truncation will warn
when doing a copy of the same size as the destination buffer with
strncpy :

  ‘strncpy’ specified bound 256 equals destination size [-Werror=stringop-truncation]

Since we unconditionally write '\0' in the last byte, reduce the copy
size by one.

Change-Id: Idb907c9550817a06fc0dffc489740f63d440e7d4
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>

Set version to 2.13-pre

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I4124a7d9d9c2f7a36816b7e498ffd37ae27da604

Cleanup: lttng-syscalls: silence warning about uninitialized bitmap variable

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Add 'kernel_read' wrapper for kernels < v4.14

See upstream commit:

  commit bdd1d2d3d251c65b74ac4493e08db18971c09240
  Author: Christoph Hellwig <hch@lst.de>
  Date:   Fri Sep 1 17:39:13 2017 +0200

    fs: fix kernel_read prototype

    Use proper ssize_t and size_t types for the return value and count
    argument, move the offset last and make it an in/out argument like
    all other read/write helpers, and make the buf argument a void pointer
    to get rid of lots of casts in the callers.

Change-Id: I825c3fcbcc17e9b46e2a661fadc66b52a94eb2da
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

fix: Use 'kernel_read' to read from procfs

Use the 'kernel_read' helper to read files in procfs, it's present in
the kernel since the 2.6 series and does the right thing on kernels that
require the set_fs dance and newer one which don't.

Change-Id: I1a53fda379e0bb9acc79331626925bbdba63d727
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

fix: don't allow userspace copy to read kernel memory

This patch fixes a security issue which allows the root user to read
arbitrary kernel memory. Considering the security model used in LTTng
userspace tooling for kernel tracing, this bug also allows members of
the 'tracing' group to read arbitrary kernel memory.

Calls to __copy_from_user_inatomic() where wrongly enclosed in
set_fs(KERNEL_DS) defeating the access_ok() calls and allowing to read
from kernel memory if a kernel address is provided.

Remove all set_fs() calls around __copy_from_user_inatomic().

As a side effect this will allow us to support v5.10 which should remove
set_fs().

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I35e4562c835217352c012ed96a7b8f93e941381e

fix: Add a 1MB limit to lttng_strlen_user_inatomic

The previous implementation was unbounded which could result in long
loops with preemption turned off.

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I85afcd879258735bb2e7502f6016fcb2d3974cf7

fix: Adjust ranges for Ubuntu 4.15.0-119 kernel

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ie32f70f810c8fc756fbd31ab129aeb35500790f7

fix: Adjust ranges for Ubuntu HWE 5.0 kernels

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I36f2c3485dcc6ccb74ea86a7ce66fcb1662d060b

Fix: system call filter table

The system call filter table has effectively been unused for a long
time due to system call name prefix mismatch. This means the overhead of
selective system call tracing was larger than it should have been because
the event payload preparation would be done for all system calls as soon
as a single system call is traced.

However, fixing this underlying issue unearths several issues that crept
unnoticed when the "enabler" concept was introduced (after the original
implementation of the system call filter table).

Here is a list of the issues which are resolved here:

- Split lttng_syscalls_unregister into an unregister and destroy
  function, thus awaiting for a grace period (and therefore quiescence
  of the users) after unregistering the system call tracepoints before
  freeing the system call filter data structures. This effectively fixes
  a use-after-free.

- The state for enabling "all" system calls vs enabling specific system
  calls (and sequences of enable-disable) was incorrect with respect to
  the "enablers" semantic. This is solved by always tracking the
  bitmap of enabled system calls, and keeping this bitmap even when
  enabling all system calls. The sc_filter is now always allocated
  before system call tracing is registered to tracepoints, which means
  it does not need to be RCU dereferenced anymore.

Padding fields in the ABI are reserved to select whether to:

- Trace either native or compat system call (or both, which is the
  behavior currently implemented),
- Trace either system call entry or exit (or both, which is the
  behavior currently implemented),
- Select the system call to trace by name (behavior currently
  implemented) or by system call number,

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>