When a tracepoint goes through the 0 -> 1 transition, tracepoint_add_func()
invokes the subsystem's ext->regfunc() before attempting to install the
new probe via func_add(). If func_add() then fails (for example, when
allocate_probes() cannot allocate a new probe array under memory pressure
and returns -ENOMEM), the function returns the error without calling the
matching ext->unregfunc(), leaving the side effects of regfunc() behind
with no installed probe to justify them.
For syscall tracepoints this is particularly unpleasant: syscall_regfunc()
bumps sys_tracepoint_refcount and sets SYSCALL_TRACEPOINT on every task.
After a leaked failure, the refcount is stuck at a non-zero value with no
consumer, and every task continues paying the syscall trace entry/exit
overhead until reboot. Other subsystems providing regfunc()/unregfunc()
pairs exhibit similarly scoped persistent state.
Mirror the existing 1 -> 0 cleanup and call ext->unregfunc() in the
func_add() error path, gated on the same condition used there so the
unwind is symmetric with the registration.
Fixes: 8cf868affd ("tracing: Have the reg function allow to fail")
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Link: https://patch.msgid.link/20260413190601.21993-1-devnexen@gmail.com
Signed-off-by: David Carlier <devnexen@gmail.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
This is the exact same thing as the 'alloc_obj()' version, only much
smaller because there are a lot fewer users of the *alloc_flex()
interface.
As with alloc_obj() version, this was done entirely with mindless brute
force, using the same script, except using 'flex' in the pattern rather
than 'objs*'.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This was done entirely with mindless brute force, using
git grep -l '\<k[vmz]*alloc_objs*(.*, GFP_KERNEL)' |
xargs sed -i 's/\(alloc_objs*(.*\), GFP_KERNEL)/\1)/'
to convert the new alloc_obj() users that had a simple GFP_KERNEL
argument to just drop that argument.
Note that due to the extreme simplicity of the scripting, any slightly
more complex cases spread over multiple lines would not be triggered:
they definitely exist, but this covers the vast bulk of the cases, and
the resulting diff is also then easier to check automatically.
For the same reason the 'flex' versions will be done as a separate
conversion.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is the result of running the Coccinelle script from
scripts/coccinelle/api/kmalloc_objs.cocci. The script is designed to
avoid scalar types (which need careful case-by-case checking), and
instead replace kmalloc-family calls that allocate struct or union
object instances:
Single allocations: kmalloc(sizeof(TYPE), ...)
are replaced with: kmalloc_obj(TYPE, ...)
Array allocations: kmalloc_array(COUNT, sizeof(TYPE), ...)
are replaced with: kmalloc_objs(TYPE, COUNT, ...)
Flex array allocations: kmalloc(struct_size(PTR, FAM, COUNT), ...)
are replaced with: kmalloc_flex(*PTR, FAM, COUNT, ...)
(where TYPE may also be *VAR)
The resulting allocations no longer return "void *", instead returning
"TYPE *".
Signed-off-by: Kees Cook <kees@kernel.org>
The current use of guard(preempt_notrace)() within __DECLARE_TRACE()
to protect invocation of __DO_TRACE_CALL() means that BPF programs
attached to tracepoints are non-preemptible. This is unhelpful in
real-time systems, whose users apparently wish to use BPF while also
achieving low latencies. (Who knew?)
One option would be to use preemptible RCU, but this introduces
many opportunities for infinite recursion, which many consider to
be counterproductive, especially given the relatively small stacks
provided by the Linux kernel. These opportunities could be shut down
by sufficiently energetic duplication of code, but this sort of thing
is considered impolite in some circles.
Therefore, use the shiny new SRCU-fast API, which provides somewhat faster
readers than those of preemptible RCU, at least on Paul E. McKenney's
laptop, where task_struct access is more expensive than access to per-CPU
variables. And SRCU-fast provides way faster readers than does SRCU,
courtesy of being able to avoid the read-side use of smp_mb(). Also,
it is quite straightforward to create srcu_read_{,un}lock_fast_notrace()
functions.
Link: https://lore.kernel.org/all/20250613152218.1924093-1-bigeasy@linutronix.de/
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexei Starovoitov <ast@kernel.org>
Link: https://patch.msgid.link/20260126231256.499701982@kernel.org
Co-developed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Steps on the way to v6.15-rc1
Fix conflicts in:
include/linux/sched.h
kernel/sched/core.c
Change-Id: I6dbf9818d73f442a33ec7871ebc70fab42655559
Signed-off-by: Tudor Ambarus <tudordana@google.com>
In commit a9cfb8778c ("tracing: Introduce tracepoint extended
structure"), the tracepoint structure changed to save space when not
enabled. Update the DEFINE_HOOK_FN() to work properly with these
changes as well.
Fixes: a9cfb8778c ("tracing: Introduce tracepoint extended structure")
Change-Id: Ic1b190e502b175a528c7dda5f33f32244f9c834d
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
In commit 4a8840af5f ("tracepoints: Use new static branch API"), the
tracing infrastructure finally moved over to use the "new" static
branch apis fully, breaking the android tracepoint macros/functions.
Convert our out-of-tree tracepoint logic to use the static branch apis
as well, fixing everything up properly.
Fixes: 4a8840af5f ("tracepoints: Use new static branch API")
Change-Id: I9cd70f1c9efc255b4d67b63664971f35d5830183
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Steps on the way to 6.12-rc1
Bug: 367265496
Change-Id: I0558de38f29cc97e194b80b6c6a79f67e9c0ebaa
Signed-off-by: Matthias Maennich <maennich@google.com>
Add for_each_tracepoint_in_module() function to iterate tracepoints in
a module. This API is needed for handling tracepoints in a loading
module from tracepoint_module_notifier callback function.
This also update for_each_module_tracepoint() to pass the module to
callback function so that it can find module easily.
Link: https://lore.kernel.org/all/172397778740.286558.15781131277732977643.stgit@devnote2/
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Without it, some builds rightly complain:
kernel/tracepoint.c:840:5:
error: no previous prototype for 'android_rvh_probe_register' [-Werror=missing-prototypes]
int android_rvh_probe_register(struct tracepoint *tp, void *probe, void *data)
^~~~~~~~~~~~~~~~~~~~~~~~~~
Fixes: e0cfc5fd8c ("ANDROID: vendor_hooks: Allow multiple attachments to restricted hooks")
Signed-off-by: Lee Jones <joneslee@google.com>
Change-Id: Ib23a9d5cf50aadb59db85d3f4e9b4cca05f93643
In the case of keeping the system running, the preferred method for
tracing the kernel is dynamic tracing (kprobe), but the drawback of
this method is that events are lost, especially when tracing packages
in the network stack.
Livepatching provides a potential solution, which is to reimplement the
function you want to replace and insert a static tracepoint.
In such a way, custom stable static tracepoints can be expanded without
rebooting the system.
Link: https://lkml.kernel.org/r/20221102160236.11696-1-iecedge@gmail.com
Signed-off-by: Jianlin Lv <iecedge@gmail.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Commit 2852ca7fba ("panic: Taint kernel if tests are run")
introduced a new taint type, TAINT_TEST, to signal that an
in-kernel test module has been loaded.
TAINT_TEST taint type defaults into a 'bad_taint' list for
kernel tracing and blocks the creation of trace events. This
causes a problem for CXL testing where loading the cxl_test
module makes all CXL modules out-of-tree, blocking any trace
events.
Trace events are in development for CXL at the moment and this
issue was found in test with v6.0-rc1.
Link: https://lkml.kernel.org/r/20220829171048.263065-1-alison.schofield@intel.com
Fixes: 2852ca7fba ("panic: Taint kernel if tests are run")
Reported-by: Ira Weiny <ira.weiny@intel.com>
Suggested-by: Dan Williams <dan.j.williams@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: David Gow <davidgow@google.com>
Signed-off-by: Alison Schofield <alison.schofield@intel.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Fix function name in tracepoint.c kernel-doc comment
to remove a warning found by clang_w1.
kernel/tracepoint.c:589: warning: expecting prototype for register_tracepoint_notifier(). Prototype was for register_tracepoint_module_notifier() instead
kernel/tracepoint.c:613: warning: expecting prototype for unregister_tracepoint_notifier(). Prototype was for unregister_tracepoint_module_notifier() instead
Link: https://lkml.kernel.org/r/20210816052430.16539-1-zhaoxiao@uniontech.com
Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: zhaoxiao <zhaoxiao@uniontech.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
State transitions from 1->0->1 and N->2->1 callbacks require RCU
synchronization. Rather than performing the RCU synchronization every
time the state change occurs, which is quite slow when many tracepoints
are registered in batch, instead keep a snapshot of the RCU state on the
most recent transitions which belong to a chain, and conditionally wait
for a grace period on the last transition of the chain if one g.p. has
not elapsed since the last snapshot.
This applies to both RCU and SRCU.
This brings the performance regression caused by commit 231264d692
("Fix: tracepoint: static call function vs data state mismatch") back to
what it was originally.
Before this commit:
# trace-cmd start -e all
# time trace-cmd start -p nop
real 0m10.593s
user 0m0.017s
sys 0m0.259s
After this commit:
# trace-cmd start -e all
# time trace-cmd start -p nop
real 0m0.878s
user 0m0.000s
sys 0m0.103s
Link: https://lkml.kernel.org/r/20210805192954.30688-1-mathieu.desnoyers@efficios.com
Link: https://lore.kernel.org/io-uring/4ebea8f0-58c9-e571-fd30-0ce4f6f09c70@samba.org/
Cc: stable@vger.kernel.org
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: "Paul E. McKenney" <paulmck@kernel.org>
Cc: Stefan Metzmacher <metze@samba.org>
Fixes: 231264d692 ("Fix: tracepoint: static call function vs data state mismatch")
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
On a 1->0->1 callbacks transition, there is an issue with the new
callback using the old callback's data.
Considering __DO_TRACE_CALL:
do { \
struct tracepoint_func *it_func_ptr; \
void *__data; \
it_func_ptr = \
rcu_dereference_raw((&__tracepoint_##name)->funcs); \
if (it_func_ptr) { \
__data = (it_func_ptr)->data; \
----> [ delayed here on one CPU (e.g. vcpu preempted by the host) ]
static_call(tp_func_##name)(__data, args); \
} \
} while (0)
It has loaded the tp->funcs of the old callback, so it will try to use the old
data. This can be fixed by adding a RCU sync anywhere in the 1->0->1
transition chain.
On a N->2->1 transition, we need an rcu-sync because you may have a
sequence of 3->2->1 (or 1->2->1) where the element 0 data is unchanged
between 2->1, but was changed from 3->2 (or from 1->2), which may be
observed by the static call. This can be fixed by adding an
unconditional RCU sync in transition 2->1.
Note, this fixes a correctness issue at the cost of adding a tremendous
performance regression to the disabling of tracepoints.
Before this commit:
# trace-cmd start -e all
# time trace-cmd start -p nop
real 0m0.778s
user 0m0.000s
sys 0m0.061s
After this commit:
# trace-cmd start -e all
# time trace-cmd start -p nop
real 0m10.593s
user 0m0.017s
sys 0m0.259s
A follow up fix will introduce a more lightweight scheme based on RCU
get_state and cond_sync, that will return the performance back to what it
was. As both this change and the lightweight versions are complex on their
own, for bisecting any issues that this may cause, they are kept as two
separate changes.
Link: https://lkml.kernel.org/r/20210805132717.23813-3-mathieu.desnoyers@efficios.com
Link: https://lore.kernel.org/io-uring/4ebea8f0-58c9-e571-fd30-0ce4f6f09c70@samba.org/
Cc: stable@vger.kernel.org
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: "Paul E. McKenney" <paulmck@kernel.org>
Cc: Stefan Metzmacher <metze@samba.org>
Fixes: d25e37d89d ("tracepoint: Optimize using static_call()")
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Because of the significant overhead that retpolines pose on indirect
calls, the tracepoint code was updated to use the new "static_calls" that
can modify the running code to directly call a function instead of using
an indirect caller, and this function can be changed at runtime.
In the tracepoint code that calls all the registered callbacks that are
attached to a tracepoint, the following is done:
it_func_ptr = rcu_dereference_raw((&__tracepoint_##name)->funcs);
if (it_func_ptr) {
__data = (it_func_ptr)->data;
static_call(tp_func_##name)(__data, args);
}
If there's just a single callback, the static_call is updated to just call
that callback directly. Once another handler is added, then the static
caller is updated to call the iterator, that simply loops over all the
funcs in the array and calls each of the callbacks like the old method
using indirect calling.
The issue was discovered with a race between updating the funcs array and
updating the static_call. The funcs array was updated first and then the
static_call was updated. This is not an issue as long as the first element
in the old array is the same as the first element in the new array. But
that assumption is incorrect, because callbacks also have a priority
field, and if there's a callback added that has a higher priority than the
callback on the old array, then it will become the first callback in the
new array. This means that it is possible to call the old callback with
the new callback data element, which can cause a kernel panic.
static_call = callback1()
funcs[] = {callback1,data1};
callback2 has higher priority than callback1
CPU 1 CPU 2
----- -----
new_funcs = {callback2,data2},
{callback1,data1}
rcu_assign_pointer(tp->funcs, new_funcs);
/*
* Now tp->funcs has the new array
* but the static_call still calls callback1
*/
it_func_ptr = tp->funcs [ new_funcs ]
data = it_func_ptr->data [ data2 ]
static_call(callback1, data);
/* Now callback1 is called with
* callback2's data */
[ KERNEL PANIC ]
update_static_call(iterator);
To prevent this from happening, always switch the static_call to the
iterator before assigning the tp->funcs to the new array. The iterator will
always properly match the callback with its data.
To trigger this bug:
In one terminal:
while :; do hackbench 50; done
In another terminal
echo 1 > /sys/kernel/tracing/events/sched/sched_waking/enable
while :; do
echo 1 > /sys/kernel/tracing/set_event_pid;
sleep 0.5
echo 0 > /sys/kernel/tracing/set_event_pid;
sleep 0.5
done
And it doesn't take long to crash. This is because the set_event_pid adds
a callback to the sched_waking tracepoint with a high priority, which will
be called before the sched_waking trace event callback is called.
Note, the removal to a single callback updates the array first, before
changing the static_call to single callback, which is the proper order as
the first element in the array is the same as what the static_call is
being changed to.
Link: https://lore.kernel.org/io-uring/4ebea8f0-58c9-e571-fd30-0ce4f6f09c70@samba.org/
Cc: stable@vger.kernel.org
Fixes: d25e37d89d ("tracepoint: Optimize using static_call()")
Reported-by: Stefan Metzmacher <metze@samba.org>
tested-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
All internal use cases for tracepoint_probe_register() is set to not ever
be called with the same function and data. If it is, it is considered a
bug, as that means the accounting of handling tracepoints is corrupted.
If the function and data for a tracepoint is already registered when
tracepoint_probe_register() is called, it will call WARN_ON_ONCE() and
return with EEXISTS.
The BPF system call can end up calling tracepoint_probe_register() with
the same data, which now means that this can trigger the warning because
of a user space process. As WARN_ON_ONCE() should not be called because
user space called a system call with bad data, there needs to be a way to
register a tracepoint without triggering a warning.
Enter tracepoint_probe_register_may_exist(), which can be called, but will
not cause a WARN_ON() if the probe already exists. It will still error out
with EEXIST, which will then be sent to the user space that performed the
BPF system call.
This keeps the previous testing for issues with other users of the
tracepoint code, while letting BPF call it with duplicated data and not
warn about it.
Link: https://lore.kernel.org/lkml/20210626135845.4080-1-penguin-kernel@I-love.SAKURA.ne.jp/
Link: https://syzkaller.appspot.com/bug?id=41f4318cf01762389f4d1c1c459da4f542fe5153
Cc: stable@vger.kernel.org
Fixes: c4f6699dfc ("bpf: introduce BPF_RAW_TRACEPOINT")
Reported-by: syzbot <syzbot+721aa903751db87aa244@syzkaller.appspotmail.com>
Reported-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Tested-by: syzbot+721aa903751db87aa244@syzkaller.appspotmail.com
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Allow up to two attachments to restricted vendor hooks to enable OEM
and vendor coexistence.
Priorities are not exposed to callers on purpose. Life's too short to
re-order the callback array with concurrent readers.
Bug: 183720636
Signed-off-by: Quentin Perret <qperret@google.com>
Change-Id: I5c7aca7f69e581b4197388478d47e0da6d2893e6
On architectures using the generic syscall entry code the architecture
independent syscall work is moved to flags in thread_info::syscall_work.
This removes architecture dependencies and frees up TIF bits.
Define SYSCALL_WORK_SYSCALL_TRACEPOINT, use it in the generic entry code
and convert the code which uses the TIF specific helper functions to use
the new *_syscall_work() helpers which either resolve to the new mode for
users of the generic entry code or to the TIF based functions for the other
architectures.
Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Andy Lutomirski <luto@kernel.org>
Link: https://lore.kernel.org/r/20201116174206.2639648-6-krisman@collabora.com
Naresh reported a bug that appears to be a side effect of the static
calls. It happens when going from more than one tracepoint callback to
a single one, and removing the first callback on the list. The list of
tracepoint callbacks holds data and a function to call with the
parameters of that tracepoint and a handler to the associated data.
old_list:
0: func = foo; data = NULL;
1: func = bar; data = &bar_struct;
new_list:
0: func = bar; data = &bar_struct;
CPU 0 CPU 1
----- -----
tp_funcs = old_list;
tp_static_caller = tp_interator
__DO_TRACE()
data = tp_funcs[0].data = NULL;
tp_funcs = new_list;
tracepoint_update_call()
tp_static_caller = tp_funcs[0] = bar;
tp_static_caller(data)
bar(data)
x = data->item = NULL->item
BOOM!
To solve this, add a tracepoint_synchronize_unregister() between
changing tp_funcs and updating the static tracepoint, that does both a
synchronize_rcu() and synchronize_srcu(). This will ensure that when
the static call is updated to the single callback that it will be
receiving the data that it registered with.
Fixes: d25e37d89d ("tracepoint: Optimize using static_call()")
Reported-by: Naresh Kamboju <naresh.kamboju@linaro.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/linux-next/CA+G9fYvPXVRO0NV7yL=FxCmFEMYkCwdz7R=9W+_votpT824YJA@mail.gmail.com
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.
[peterz: updated to new interface]
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: https://lore.kernel.org/r/20200818135805.279421092@infradead.org
While auditing all module notifiers I noticed a whole bunch of fail
wrt the return value. Notifiers have a 'special' return semantics.
As is; NOTIFY_DONE vs NOTIFY_OK is a bit vague; but
notifier_from_errno(0) results in NOTIFY_OK and NOTIFY_DONE has a
comment that says "Don't care".
From this I've used NOTIFY_DONE when the function completely ignores
the callback and notifier_to_error() isn't used.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Reviewed-by: Robert Richter <rric@kernel.org>
Acked-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Link: https://lore.kernel.org/r/20200818135804.385360407@infradead.org
Pull tracing updates from Steven Rostedt:
"The main changes in this release include:
- Add user space specific memory reading for kprobes
- Allow kprobes to be executed earlier in boot
The rest are mostly just various clean ups and small fixes"
* tag 'trace-v5.3' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace: (33 commits)
tracing: Make trace_get_fields() global
tracing: Let filter_assign_type() detect FILTER_PTR_STRING
tracing: Pass type into tracing_generic_entry_update()
ftrace/selftest: Test if set_event/ftrace_pid exists before writing
ftrace/selftests: Return the skip code when tracing directory not configured in kernel
tracing/kprobe: Check registered state using kprobe
tracing/probe: Add trace_event_call accesses APIs
tracing/probe: Add probe event name and group name accesses APIs
tracing/probe: Add trace flag access APIs for trace_probe
tracing/probe: Add trace_event_file access APIs for trace_probe
tracing/probe: Add trace_event_call register API for trace_probe
tracing/probe: Add trace_probe init and free functions
tracing/uprobe: Set print format when parsing command
tracing/kprobe: Set print format right after parsed command
kprobes: Fix to init kprobes in subsys_initcall
tracepoint: Use struct_size() in kmalloc()
ring-buffer: Remove HAVE_64BIT_ALIGNED_ACCESS
ftrace: Enable trampoline when rec count returns back to one
tracing/kprobe: Do not run kprobe boot tests if kprobe_event is on cmdline
tracing: Make a separate config for trace event self tests
...
Carlos Llamas
David Carlier
Linus Torvalds
Kees Cook
Steven Rostedt
Tudor Ambarus
Huang Shijie
Greg Kroah-Hartman
Mathieu Desnoyers
Josh Poimboeuf
Matthias Maennich
Masami Hiramatsu (Google)
Lee Jones
Jianlin Lv
Zhen Lei
Alison Schofield
zhaoxiao
Lee Jones
Quentin Perret
Gabriel Krisman Bertazi
Gustavo A. R. Silva
Peter Zijlstra