[DTrace-devel] [PATCH v3 1/2] Implement the sched provider (first part)
Kris Van Hees
kris.van.hees at oracle.com
Wed May 24 06:07:58 UTC 2023
Partial implementation of the sched SDT provider. The following probes
are implemented:
sched:::dequeue (arg1 is NULL)
sched:::enqueue (arg1 is NULL)
sched:::off-cpu
sched:::on-cpu (limited trigger locations)
sched:::surrender
sched:::tick
sched:::wakeup
Signed-off-by: Kris Van Hees <kris.van.hees at oracle.com>
Reviewed-by: Nick Alcock <nick.alcock at oracle.com>
---
libdtrace/Build | 2 +
libdtrace/dt_open.c | 1 +
libdtrace/dt_prov_sched.c | 151 +++++++++++++++++++++++++
libdtrace/dt_provider.h | 1 +
test/demo/sched/dtrace.d | 1 -
test/demo/sched/howlong.d | 1 -
test/demo/sched/qlen.d | 1 -
test/demo/sched/qtime.d | 1 -
test/demo/sched/tick.d | 1 -
test/demo/sched/ticktime.d | 1 -
test/demo/sched/where.d | 1 -
test/demo/sched/whoqueue.d | 1 -
test/demo/sched/whosteal.d | 1 -
test/unittest/buffering/tst.cputime.sh | 1 -
14 files changed, 155 insertions(+), 10 deletions(-)
create mode 100644 libdtrace/dt_prov_sched.c
diff --git a/libdtrace/Build b/libdtrace/Build
index cf18a043..89972c66 100644
--- a/libdtrace/Build
+++ b/libdtrace/Build
@@ -52,6 +52,7 @@ libdtrace-build_SOURCES = dt_aggregate.c \
dt_prov_proc.c \
dt_prov_profile.c \
dt_prov_rawtp.c \
+ dt_prov_sched.c \
dt_prov_sdt.c \
dt_prov_syscall.c \
dt_prov_uprobe.c \
@@ -96,6 +97,7 @@ dt_prov_fbt.c_CFLAGS := -Wno-pedantic
dt_prov_proc.c_CFLAGS := -Wno-pedantic
dt_prov_profile.c_CFLAGS := -Wno-pedantic
dt_prov_rawtp.c_CFLAGS := -Wno-pedantic
+dt_prov_sched.c_CFLAGS := -Wno-pedantic
dt_prov_sdt.c_CFLAGS := -Wno-pedantic
dt_prov_syscall.c_CFLAGS := -Wno-pedantic
dt_prov_uprobe.c_CFLAGS := -Wno-pedantic
diff --git a/libdtrace/dt_open.c b/libdtrace/dt_open.c
index d13b859a..110d45bc 100644
--- a/libdtrace/dt_open.c
+++ b/libdtrace/dt_open.c
@@ -71,6 +71,7 @@ static const dt_provimpl_t *dt_providers[] = {
&dt_proc,
&dt_profile,
&dt_rawtp,
+ &dt_sched,
&dt_sdt,
&dt_syscall,
&dt_uprobe,
diff --git a/libdtrace/dt_prov_sched.c b/libdtrace/dt_prov_sched.c
new file mode 100644
index 00000000..8703c936
--- /dev/null
+++ b/libdtrace/dt_prov_sched.c
@@ -0,0 +1,151 @@
+/*
+ * Oracle Linux DTrace.
+ * Copyright (c) 2023, Oracle and/or its affiliates. All rights reserved.
+ * Licensed under the Universal Permissive License v 1.0 as shown at
+ * http://oss.oracle.com/licenses/upl.
+ *
+ * The 'sched' SDT provider for DTrace-specific probes.
+ */
+#include <assert.h>
+#include <errno.h>
+
+#include "dt_dctx.h"
+#include "dt_cg.h"
+#include "dt_provider_sdt.h"
+#include "dt_probe.h"
+
+static const char prvname[] = "sched";
+static const char modname[] = "vmlinux";
+
+static probe_dep_t probes[] = {
+ { "dequeue",
+ DTRACE_PROBESPEC_NAME, "fbt::dequeue_task_*:entry" },
+ { "enqueue",
+ DTRACE_PROBESPEC_NAME, "fbt::enqueue_task_*:entry" },
+ { "off-cpu",
+ DTRACE_PROBESPEC_NAME, "rawtp:sched::sched_switch" },
+ { "on-cpu",
+ DTRACE_PROBESPEC_NAME, "fbt::schedule_tail:entry" },
+ { "surrender",
+ DTRACE_PROBESPEC_NAME, "fbt::do_sched_yield:entry" },
+ { "tick",
+ DTRACE_PROBESPEC_NAME, "fbt::scheduler_tick:entry" },
+ { "wakeup",
+ DTRACE_PROBESPEC_NAME, "rawtp:sched::sched_wakeup" },
+ { NULL, }
+};
+
+static probe_arg_t probe_args[] = {
+#if 0
+ { "change-pri", 0, { 0, 0, "struct task_struct *", "lwpsinfo_t *" } },
+ { "change-pri", 1, { 0, 0, "struct task_struct *", "psinfo_t *" } },
+ { "change-pri", 2, { 1, 0, "int", } },
+#endif
+ { "dequeue", 0, { 0, 0, "struct task_struct *", "lwpsinfo_t *" } },
+ { "dequeue", 1, { 0, 0, "struct task_struct *", "psinfo_t *" } },
+ { "dequeue", 2, { 1, 0, "cpuinfo_t *", } },
+ { "dequeue", 3, { 2, 0, "int", } },
+ { "enqueue", 0, { 0, 0, "struct task_struct *", "lwpsinfo_t *" } },
+ { "enqueue", 1, { 0, 0, "struct task_struct *", "psinfo_t *" } },
+ { "enqueue", 2, { 1, 0, "cpuinfo_t *", } },
+ { "off-cpu", 0, { 0, 0, "struct task_struct *", "lwpsinfo_t *" } },
+ { "off-cpu", 1, { 0, 0, "struct task_struct *", "psinfo_t *" } },
+ { "on-cpu", },
+#if 0
+ { "preempt", },
+ { "remain-cpu", },
+ { "sleep", },
+#endif
+ { "surrender", 0, { 0, 0, "struct task_struct *", "lwpsinfo_t *" } },
+ { "surrender", 1, { 0, 0, "struct task_struct *", "psinfo_t *" } },
+ { "tick", 0, { 0, 0, "struct task_struct *", "lwpsinfo_t *" } },
+ { "tick", 1, { 0, 0, "struct task_struct *", "psinfo_t *" } },
+ { "wakeup", 0, { 0, 0, "struct task_struct *", "lwpsinfo_t *" } },
+ { "wakeup", 1, { 0, 0, "struct task_struct *", "psinfo_t *" } },
+ { NULL, }
+};
+
+static const dtrace_pattr_t pattr = {
+{ DTRACE_STABILITY_EVOLVING, DTRACE_STABILITY_EVOLVING, DTRACE_CLASS_ISA },
+{ DTRACE_STABILITY_PRIVATE, DTRACE_STABILITY_PRIVATE, DTRACE_CLASS_UNKNOWN },
+{ DTRACE_STABILITY_PRIVATE, DTRACE_STABILITY_PRIVATE, DTRACE_CLASS_UNKNOWN },
+{ DTRACE_STABILITY_EVOLVING, DTRACE_STABILITY_EVOLVING, DTRACE_CLASS_ISA },
+{ DTRACE_STABILITY_EVOLVING, DTRACE_STABILITY_EVOLVING, DTRACE_CLASS_ISA },
+};
+
+/*
+ * Provide all the "sched" SDT probes.
+ */
+static int populate(dtrace_hdl_t *dtp)
+{
+ return dt_sdt_populate(dtp, prvname, modname, &dt_sched, &pattr,
+ probe_args, probes);
+}
+
+/*
+ * Generate a BPF trampoline for a SDT probe.
+ *
+ * The trampoline function is called when a SDT probe triggers, and it must
+ * satisfy the following prototype:
+ *
+ * int dt_sched(void *data)
+ *
+ * The trampoline will populate a dt_dctx_t struct and then call the function
+ * that implements the compiled D clause. It returns the value that it gets
+ * back from that function.
+ */
+static int trampoline(dt_pcb_t *pcb, uint_t exitlbl)
+{
+ dt_irlist_t *dlp = &pcb->pcb_ir;
+ dt_probe_t *prp = pcb->pcb_probe;
+
+ if (strcmp(prp->desc->prb, "dequeue") == 0) {
+ emit(dlp, BPF_LOAD(BPF_DW, BPF_REG_0, BPF_REG_7, DMST_ARG(1)));
+ emit(dlp, BPF_STORE(BPF_DW, BPF_REG_7, DMST_ARG(0), BPF_REG_0));
+ /*
+ * FIXME: arg1 should be a pointer to cpuinfo_t for the CPU
+ * associated with the runqueue.
+ */
+ emit(dlp, BPF_STORE_IMM(BPF_DW, BPF_REG_7, DMST_ARG(1), 0));
+ } else if (strcmp(prp->desc->prb, "enqueue") == 0) {
+/*
+ * This is ugly but necessary... enqueue_task() takes a flags argument and the
+ * ENQUEUE_HEAD flag is used to indicate that the task is to be placed at the
+ * head of the queue. We need to be able to pass this special case as arg2
+ * in the enqueue probe.
+ *
+ * The flag values are found in kernel/sched/sched.h which is not exposed
+ * outside the kernel source tree.
+ */
+#define ENQUEUE_HEAD 0x10
+
+ emit(dlp, BPF_LOAD(BPF_DW, BPF_REG_0, BPF_REG_7, DMST_ARG(1)));
+ emit(dlp, BPF_STORE(BPF_DW, BPF_REG_7, DMST_ARG(0), BPF_REG_0));
+ /*
+ * FIXME: arg1 should be a pointer to cpuinfo_t for the CPU
+ * associated with the runqueue.
+ */
+ emit(dlp, BPF_STORE_IMM(BPF_DW, BPF_REG_7, DMST_ARG(1), 0));
+ emit(dlp, BPF_LOAD(BPF_DW, BPF_REG_0, BPF_REG_7, DMST_ARG(2)));
+ emit(dlp, BPF_ALU64_IMM(BPF_AND, BPF_REG_0, ENQUEUE_HEAD));
+ emit(dlp, BPF_STORE(BPF_DW, BPF_REG_7, DMST_ARG(2), BPF_REG_0));
+ } else if (strcmp(prp->desc->prb, "off-cpu") == 0) {
+ emit(dlp, BPF_LOAD(BPF_DW, BPF_REG_0, BPF_REG_7, DMST_ARG(2)));
+ emit(dlp, BPF_STORE(BPF_DW, BPF_REG_7, DMST_ARG(0), BPF_REG_0));
+ } else if (strcmp(prp->desc->prb, "surrender") == 0 ||
+ strcmp(prp->desc->prb, "tick") == 0) {
+ emit(dlp, BPF_CALL_HELPER(BPF_FUNC_get_current_task));
+ emit(dlp, BPF_STORE(BPF_DW, BPF_REG_7, DMST_ARG(0), BPF_REG_0));
+ }
+
+ return 0;
+}
+
+dt_provimpl_t dt_sched = {
+ .name = prvname,
+ .prog_type = BPF_PROG_TYPE_UNSPEC,
+ .populate = &populate,
+ .enable = &dt_sdt_enable,
+ .trampoline = &trampoline,
+ .probe_info = &dt_sdt_probe_info,
+};
diff --git a/libdtrace/dt_provider.h b/libdtrace/dt_provider.h
index 068c65a2..252a73d4 100644
--- a/libdtrace/dt_provider.h
+++ b/libdtrace/dt_provider.h
@@ -73,6 +73,7 @@ extern dt_provimpl_t dt_fbt;
extern dt_provimpl_t dt_proc;
extern dt_provimpl_t dt_profile;
extern dt_provimpl_t dt_rawtp;
+extern dt_provimpl_t dt_sched;
extern dt_provimpl_t dt_sdt;
extern dt_provimpl_t dt_syscall;
extern dt_provimpl_t dt_uprobe;
diff --git a/test/demo/sched/dtrace.d b/test/demo/sched/dtrace.d
index cedffb53..a1e9b4a8 100644
--- a/test/demo/sched/dtrace.d
+++ b/test/demo/sched/dtrace.d
@@ -4,7 +4,6 @@
* Licensed under the Universal Permissive License v 1.0 as shown at
* http://oss.oracle.com/licenses/upl.
*/
-/* @@xfail: dtv2 */
#pragma D option quiet
diff --git a/test/demo/sched/howlong.d b/test/demo/sched/howlong.d
index f68239e9..ace1b8fc 100644
--- a/test/demo/sched/howlong.d
+++ b/test/demo/sched/howlong.d
@@ -4,7 +4,6 @@
* Licensed under the Universal Permissive License v 1.0 as shown at
* http://oss.oracle.com/licenses/upl.
*/
-/* @@xfail: dtv2 */
sched:::off-cpu
/curlwpsinfo->pr_state == SSLEEP/
diff --git a/test/demo/sched/qlen.d b/test/demo/sched/qlen.d
index 029cbbf6..95025bd6 100644
--- a/test/demo/sched/qlen.d
+++ b/test/demo/sched/qlen.d
@@ -4,7 +4,6 @@
* Licensed under the Universal Permissive License v 1.0 as shown at
* http://oss.oracle.com/licenses/upl.
*/
-/* @@xfail: dtv2 */
sched:::enqueue
{
diff --git a/test/demo/sched/qtime.d b/test/demo/sched/qtime.d
index ba010f3e..344a88ea 100644
--- a/test/demo/sched/qtime.d
+++ b/test/demo/sched/qtime.d
@@ -4,7 +4,6 @@
* Licensed under the Universal Permissive License v 1.0 as shown at
* http://oss.oracle.com/licenses/upl.
*/
-/* @@xfail: dtv2 */
sched:::enqueue
{
diff --git a/test/demo/sched/tick.d b/test/demo/sched/tick.d
index 54d18d70..57376f1f 100644
--- a/test/demo/sched/tick.d
+++ b/test/demo/sched/tick.d
@@ -4,7 +4,6 @@
* Licensed under the Universal Permissive License v 1.0 as shown at
* http://oss.oracle.com/licenses/upl.
*/
-/* @@xfail: dtv2 */
sched:::tick,
sched:::enqueue
diff --git a/test/demo/sched/ticktime.d b/test/demo/sched/ticktime.d
index 3fd0bfbc..dca76dba 100644
--- a/test/demo/sched/ticktime.d
+++ b/test/demo/sched/ticktime.d
@@ -4,7 +4,6 @@
* Licensed under the Universal Permissive License v 1.0 as shown at
* http://oss.oracle.com/licenses/upl.
*/
-/* @@xfail: dtv2 */
uint64_t last[int];
diff --git a/test/demo/sched/where.d b/test/demo/sched/where.d
index 8845610d..906920d0 100644
--- a/test/demo/sched/where.d
+++ b/test/demo/sched/where.d
@@ -4,7 +4,6 @@
* Licensed under the Universal Permissive License v 1.0 as shown at
* http://oss.oracle.com/licenses/upl.
*/
-/* @@xfail: dtv2 */
sched:::on-cpu
{
diff --git a/test/demo/sched/whoqueue.d b/test/demo/sched/whoqueue.d
index 3057600c..1eb3d1b4 100644
--- a/test/demo/sched/whoqueue.d
+++ b/test/demo/sched/whoqueue.d
@@ -4,7 +4,6 @@
* Licensed under the Universal Permissive License v 1.0 as shown at
* http://oss.oracle.com/licenses/upl.
*/
-/* @@xfail: dtv2 */
#pragma D option quiet
#pragma D option nspec=4
diff --git a/test/demo/sched/whosteal.d b/test/demo/sched/whosteal.d
index 5ac091b3..5287dd99 100644
--- a/test/demo/sched/whosteal.d
+++ b/test/demo/sched/whosteal.d
@@ -4,7 +4,6 @@
* Licensed under the Universal Permissive License v 1.0 as shown at
* http://oss.oracle.com/licenses/upl.
*/
-/* @@xfail: dtv2 */
#pragma D option quiet
diff --git a/test/unittest/buffering/tst.cputime.sh b/test/unittest/buffering/tst.cputime.sh
index cde1fba0..6a420e8d 100755
--- a/test/unittest/buffering/tst.cputime.sh
+++ b/test/unittest/buffering/tst.cputime.sh
@@ -6,7 +6,6 @@
# http://oss.oracle.com/licenses/upl.
#
# @@timeout: 12
-# @@xfail: dtv2
script()
{
--
2.40.1
More information about the DTrace-devel
mailing list