2024-01-25 07:38:53

by yebin (H)

[permalink] [raw]
Subject: [PATCH v5 0/8] support '%pd' and '%pD' for print file name

During fault locating, the file name needs to be printed based on the
dentry/file address. The offset needs to be calculated each time, which
is troublesome. Similar to printk, kprobe supports printing file names
for dentry/file addresses.

Diff v5 vs v4:
1. Use glob_match() instead of str_has_suffix(), so remove the first patch;
2. Allocate buffer from heap for expand dentry;
3. Support "%pd/%pD" print type for fprobe;
4. Use $arg1 instead of origin register in test case;
5. Add test case for fprobe;

Diff v4 vs v3:
1. Use "argv[i][idx + 3] == 'd'" instead of "argv[i][strlen(argv[i]) - 1] == 'd'"
to judge print format in PATCH[4/7];

Diff v3 vs v2:
1. Return the index of where the suffix was found in str_has_suffix();

Diff v2 vs v1:
1. Use "%pd/%pD" print format instead of "pd/pD" print format;
2. Add "%pd/%pD" in README;
3. Expand "%pd/%pD" argument before parameter parsing;
4. Add more detail information in ftrace documentation;
5. Add test cases for new print format in selftests/ftrace;

Ye Bin (8):
tracing/probes: add traceprobe_expand_dentry_args() helper
tracing/probes: support '%pd' type for print struct dentry's name
tracing/probes: support '%pD' type for print struct file's name
tracing/probes: support '%pd/%pD' type for fprobe
tracing: add new type "%pd/%pD" in readme_msg[]
Documentation: tracing: add new type '%pd' and '%pD' for kprobe
selftests/ftrace: add kprobe test cases for VFS type "%pd" and "%pD"
selftests/ftrace: add fprobe test cases for VFS type "%pd" and "%pD"

Documentation/trace/kprobetrace.rst | 8 ++-
kernel/trace/trace.c | 2 +-
kernel/trace/trace_fprobe.c | 6 ++
kernel/trace/trace_kprobe.c | 6 ++
kernel/trace/trace_probe.c | 63 +++++++++++++++++++
kernel/trace/trace_probe.h | 2 +
.../ftrace/test.d/dynevent/fprobe_args_vfs.tc | 40 ++++++++++++
.../ftrace/test.d/kprobe/kprobe_args_vfs.tc | 43 +++++++++++++
8 files changed, 167 insertions(+), 3 deletions(-)
create mode 100644 tools/testing/selftests/ftrace/test.d/dynevent/fprobe_args_vfs.tc
create mode 100644 tools/testing/selftests/ftrace/test.d/kprobe/kprobe_args_vfs.tc

--
2.31.1



2024-01-25 07:38:59

by yebin (H)

[permalink] [raw]
Subject: [PATCH v5 1/8] tracing/probes: add traceprobe_expand_dentry_args() helper

Add traceprobe_expand_dentry_args() to expand dentry args. this API is
prepare to support "%pd" print format for kprobe.

Signed-off-by: Ye Bin <[email protected]>
---
kernel/trace/trace_probe.c | 50 ++++++++++++++++++++++++++++++++++++++
kernel/trace/trace_probe.h | 2 ++
2 files changed, 52 insertions(+)

diff --git a/kernel/trace/trace_probe.c b/kernel/trace/trace_probe.c
index 4dc74d73fc1d..0c1caf0f474a 100644
--- a/kernel/trace/trace_probe.c
+++ b/kernel/trace/trace_probe.c
@@ -1565,6 +1565,56 @@ const char **traceprobe_expand_meta_args(int argc, const char *argv[],
return ERR_PTR(ret);
}

+/* @buf: *buf must be equal to NULL. Caller must to free *buf */
+int traceprobe_expand_dentry_args(int argc, const char *argv[], char **buf)
+{
+ int i, used, ret;
+ int bufsize = MAX_DENTRY_ARGS_LEN;
+ char *tmpbuf = NULL;
+
+ if (*buf)
+ return -EINVAL;
+
+ used = 0;
+ for (i = 0; i < argc; i++) {
+ if (glob_match("*:%pd", argv[i])) {
+ char *tmp;
+ char *equal;
+
+ if (!tmpbuf) {
+ tmpbuf = kmalloc(bufsize, GFP_KERNEL);
+ if (!tmpbuf)
+ return -ENOMEM;
+ }
+
+ tmp = kstrdup(argv[i], GFP_KERNEL);
+ if (!tmp)
+ goto nomem;
+
+ equal = strchr(tmp, '=');
+ if (equal)
+ *equal = '\0';
+ tmp[strlen(argv[i]) - 4] = '\0';
+ ret = snprintf(tmpbuf + used, bufsize - used,
+ "%s%s+0x0(+0x%zx(%s)):string",
+ equal ? tmp : "", equal ? "=" : "",
+ offsetof(struct dentry, d_name.name),
+ equal ? equal + 1 : tmp);
+ kfree(tmp);
+ if (ret >= bufsize - used)
+ goto nomem;
+ argv[i] = tmpbuf + used;
+ used += ret + 1;
+ }
+ }
+
+ *buf = tmpbuf;
+ return 0;
+nomem:
+ kfree(tmpbuf);
+ return -ENOMEM;
+}
+
void traceprobe_finish_parse(struct traceprobe_parse_context *ctx)
{
clear_btf_context(ctx);
diff --git a/kernel/trace/trace_probe.h b/kernel/trace/trace_probe.h
index 850d9ecb6765..5800513439f3 100644
--- a/kernel/trace/trace_probe.h
+++ b/kernel/trace/trace_probe.h
@@ -34,6 +34,7 @@
#define MAX_ARRAY_LEN 64
#define MAX_ARG_NAME_LEN 32
#define MAX_BTF_ARGS_LEN 128
+#define MAX_DENTRY_ARGS_LEN 256
#define MAX_STRING_SIZE PATH_MAX
#define MAX_ARG_BUF_LEN (MAX_TRACE_ARGS * MAX_ARG_NAME_LEN)

@@ -402,6 +403,7 @@ extern int traceprobe_parse_probe_arg(struct trace_probe *tp, int i,
const char **traceprobe_expand_meta_args(int argc, const char *argv[],
int *new_argc, char *buf, int bufsize,
struct traceprobe_parse_context *ctx);
+extern int traceprobe_expand_dentry_args(int argc, const char *argv[], char **buf);

extern int traceprobe_update_arg(struct probe_arg *arg);
extern void traceprobe_free_probe_arg(struct probe_arg *arg);
--
2.31.1


2024-01-25 07:39:25

by yebin (H)

[permalink] [raw]
Subject: [PATCH v5 3/8] tracing/probes: support '%pD' type for print struct file's name

Similar to '%pD' for printk, use '%pD' for print struct file's name.

Signed-off-by: Ye Bin <[email protected]>
---
kernel/trace/trace_probe.c | 57 +++++++++++++++++++++++---------------
1 file changed, 35 insertions(+), 22 deletions(-)

diff --git a/kernel/trace/trace_probe.c b/kernel/trace/trace_probe.c
index 0c1caf0f474a..5ac7b0193dd8 100644
--- a/kernel/trace/trace_probe.c
+++ b/kernel/trace/trace_probe.c
@@ -12,6 +12,7 @@
#define pr_fmt(fmt) "trace_probe: " fmt

#include <linux/bpf.h>
+#include <linux/fs.h>
#include "trace_btf.h"

#include "trace_probe.h"
@@ -1577,35 +1578,47 @@ int traceprobe_expand_dentry_args(int argc, const char *argv[], char **buf)

used = 0;
for (i = 0; i < argc; i++) {
- if (glob_match("*:%pd", argv[i])) {
- char *tmp;
- char *equal;
-
- if (!tmpbuf) {
- tmpbuf = kmalloc(bufsize, GFP_KERNEL);
- if (!tmpbuf)
- return -ENOMEM;
- }
+ char *tmp;
+ char *equal;
+ size_t arg_len;

- tmp = kstrdup(argv[i], GFP_KERNEL);
- if (!tmp)
- goto nomem;
+ if (!glob_match("*:%p[dD]", argv[i]))
+ continue;

- equal = strchr(tmp, '=');
- if (equal)
- *equal = '\0';
- tmp[strlen(argv[i]) - 4] = '\0';
+ if (!tmpbuf) {
+ tmpbuf = kmalloc(bufsize, GFP_KERNEL);
+ if (!tmpbuf)
+ return -ENOMEM;
+ }
+
+ tmp = kstrdup(argv[i], GFP_KERNEL);
+ if (!tmp)
+ goto nomem;
+
+ equal = strchr(tmp, '=');
+ if (equal)
+ *equal = '\0';
+ arg_len = strlen(argv[i]);
+ tmp[arg_len - 4] = '\0';
+ if (argv[i][arg_len - 1] == 'd')
ret = snprintf(tmpbuf + used, bufsize - used,
"%s%s+0x0(+0x%zx(%s)):string",
equal ? tmp : "", equal ? "=" : "",
offsetof(struct dentry, d_name.name),
equal ? equal + 1 : tmp);
- kfree(tmp);
- if (ret >= bufsize - used)
- goto nomem;
- argv[i] = tmpbuf + used;
- used += ret + 1;
- }
+ else
+ ret = snprintf(tmpbuf + used, bufsize - used,
+ "%s%s+0x0(+0x%zx(+0x%zx(%s))):string",
+ equal ? tmp : "", equal ? "=" : "",
+ offsetof(struct dentry, d_name.name),
+ offsetof(struct file, f_path.dentry),
+ equal ? equal + 1 : tmp);
+
+ kfree(tmp);
+ if (ret >= bufsize - used)
+ goto nomem;
+ argv[i] = tmpbuf + used;
+ used += ret + 1;
}

*buf = tmpbuf;
--
2.31.1


2024-01-25 07:39:59

by yebin (H)

[permalink] [raw]
Subject: [PATCH v5 2/8] tracing/probes: support '%pd' type for print struct dentry's name

Similar to '%pd' for printk, use '%pd' for print struct dentry's name.

Signed-off-by: Ye Bin <[email protected]>
---
kernel/trace/trace_kprobe.c | 6 ++++++
1 file changed, 6 insertions(+)

diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c
index c4c6e0e0068b..7cbb43740b4f 100644
--- a/kernel/trace/trace_kprobe.c
+++ b/kernel/trace/trace_kprobe.c
@@ -779,6 +779,7 @@ static int __trace_kprobe_create(int argc, const char *argv[])
char buf[MAX_EVENT_NAME_LEN];
char gbuf[MAX_EVENT_NAME_LEN];
char abuf[MAX_BTF_ARGS_LEN];
+ char *dbuf = NULL;
struct traceprobe_parse_context ctx = { .flags = TPARG_FL_KERNEL };

switch (argv[0][0]) {
@@ -930,6 +931,10 @@ static int __trace_kprobe_create(int argc, const char *argv[])
argv = new_argv;
}

+ ret = traceprobe_expand_dentry_args(argc, argv, &dbuf);
+ if (ret)
+ goto out;
+
/* setup a probe */
tk = alloc_trace_kprobe(group, event, addr, symbol, offset, maxactive,
argc, is_return);
@@ -971,6 +976,7 @@ static int __trace_kprobe_create(int argc, const char *argv[])
trace_probe_log_clear();
kfree(new_argv);
kfree(symbol);
+ kfree(dbuf);
return ret;

parse_error:
--
2.31.1


2024-01-25 07:40:16

by yebin (H)

[permalink] [raw]
Subject: [PATCH v5 4/8] tracing/probes: support '%pd/%pD' type for fprobe

Support print type '%pd/%pD' for print dentry's or file's name.

Signed-off-by: Ye Bin <[email protected]>
---
kernel/trace/trace_fprobe.c | 6 ++++++
1 file changed, 6 insertions(+)

diff --git a/kernel/trace/trace_fprobe.c b/kernel/trace/trace_fprobe.c
index 7d2ddbcfa377..988d68e906ad 100644
--- a/kernel/trace/trace_fprobe.c
+++ b/kernel/trace/trace_fprobe.c
@@ -976,6 +976,7 @@ static int __trace_fprobe_create(int argc, const char *argv[])
char gbuf[MAX_EVENT_NAME_LEN];
char sbuf[KSYM_NAME_LEN];
char abuf[MAX_BTF_ARGS_LEN];
+ char *dbuf = NULL;
bool is_tracepoint = false;
struct tracepoint *tpoint = NULL;
struct traceprobe_parse_context ctx = {
@@ -1086,6 +1087,10 @@ static int __trace_fprobe_create(int argc, const char *argv[])
argv = new_argv;
}

+ ret = traceprobe_expand_dentry_args(argc, argv, &dbuf);
+ if (ret)
+ goto out;
+
/* setup a probe */
tf = alloc_trace_fprobe(group, event, symbol, tpoint, maxactive,
argc, is_return);
@@ -1131,6 +1136,7 @@ static int __trace_fprobe_create(int argc, const char *argv[])
trace_probe_log_clear();
kfree(new_argv);
kfree(symbol);
+ kfree(dbuf);
return ret;

parse_error:
--
2.31.1


2024-01-25 07:40:54

by yebin (H)

[permalink] [raw]
Subject: [PATCH v5 7/8] selftests/ftrace: add kprobe test cases for VFS type "%pd" and "%pD"

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.

Signed-off-by: Ye Bin <[email protected]>
---
.../ftrace/test.d/kprobe/kprobe_args_vfs.tc | 43 +++++++++++++++++++
1 file changed, 43 insertions(+)
create mode 100644 tools/testing/selftests/ftrace/test.d/kprobe/kprobe_args_vfs.tc

diff --git a/tools/testing/selftests/ftrace/test.d/kprobe/kprobe_args_vfs.tc b/tools/testing/selftests/ftrace/test.d/kprobe/kprobe_args_vfs.tc
new file mode 100644
index 000000000000..cf0599b90f1a
--- /dev/null
+++ b/tools/testing/selftests/ftrace/test.d/kprobe/kprobe_args_vfs.tc
@@ -0,0 +1,43 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-2.0
+# description: Kprobe event VFS type argument
+# requires: kprobe_events
+
+: "Test argument %pd/%pD in README"
+grep -q "%pd/%pD" README
+
+: "Test argument %pd with name"
+echo 'p:testprobe dput name=$arg1:%pd' > kprobe_events
+echo 1 > events/kprobes/testprobe/enable
+grep -q "1" events/kprobes/testprobe/enable
+echo 0 > events/kprobes/testprobe/enable
+grep "dput" trace | grep -q "enable"
+echo "" > kprobe_events
+echo "" > trace
+
+: "Test argument %pd without name"
+echo 'p:testprobe dput $arg1:%pd' > kprobe_events
+echo 1 > events/kprobes/testprobe/enable
+grep -q "1" events/kprobes/testprobe/enable
+echo 0 > events/kprobes/testprobe/enable
+grep "dput" trace | grep -q "enable"
+echo "" > kprobe_events
+echo "" > trace
+
+: "Test argument %pD with name"
+echo 'p:testprobe vfs_read name=$arg1:%pD' > kprobe_events
+echo 1 > events/kprobes/testprobe/enable
+grep -q "1" events/kprobes/testprobe/enable
+echo 0 > events/kprobes/testprobe/enable
+grep "vfs_read" trace | grep -q "enable"
+echo "" > kprobe_events
+echo "" > trace
+
+: "Test argument %pD without name"
+echo 'p:testprobe vfs_read $arg1:%pD' > kprobe_events
+echo 1 > events/kprobes/testprobe/enable
+grep -q "1" events/kprobes/testprobe/enable
+echo 0 > events/kprobes/testprobe/enable
+grep "vfs_read" trace | grep -q "enable"
+echo "" > kprobe_events
+echo "" > trace
--
2.31.1


2024-01-25 07:44:43

by yebin (H)

[permalink] [raw]
Subject: [PATCH v5 6/8] Documentation: tracing: add new type '%pd' and '%pD' for kprobe

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.

Signed-off-by: Ye Bin <[email protected]>
---
Documentation/trace/kprobetrace.rst | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/Documentation/trace/kprobetrace.rst b/Documentation/trace/kprobetrace.rst
index bf9cecb69fc9..f13f0fc11251 100644
--- a/Documentation/trace/kprobetrace.rst
+++ b/Documentation/trace/kprobetrace.rst
@@ -58,8 +58,9 @@ Synopsis of kprobe_events
NAME=FETCHARG : Set NAME as the argument name of FETCHARG.
FETCHARG:TYPE : Set TYPE as the type of FETCHARG. Currently, basic types
(u8/u16/u32/u64/s8/s16/s32/s64), hexadecimal types
- (x8/x16/x32/x64), "char", "string", "ustring", "symbol", "symstr"
- and bitfield are supported.
+ (x8/x16/x32/x64), VFS layer common type(%pd/%pD), "char",
+ "string", "ustring", "symbol", "symstr" and bitfield are
+ supported.

(\*1) only for the probe on function entry (offs == 0). Note, this argument access
is best effort, because depending on the argument type, it may be passed on
@@ -113,6 +114,9 @@ With 'symstr' type, you can filter the event with wildcard pattern of the
symbols, and you don't need to solve symbol name by yourself.
For $comm, the default type is "string"; any other type is invalid.

+VFS layer common type(%pd/%pD) is a special type, which fetches dentry's or
+file's name from struct dentry's address or struct file's address.
+
.. _user_mem_access:

User Memory Access
--
2.31.1


2024-01-25 07:45:40

by yebin (H)

[permalink] [raw]
Subject: [PATCH v5 8/8] selftests/ftrace: add fprobe test cases for VFS type "%pd" and "%pD"

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.

Signed-off-by: Ye Bin <[email protected]>
---
.../ftrace/test.d/dynevent/fprobe_args_vfs.tc | 40 +++++++++++++++++++
1 file changed, 40 insertions(+)
create mode 100644 tools/testing/selftests/ftrace/test.d/dynevent/fprobe_args_vfs.tc

diff --git a/tools/testing/selftests/ftrace/test.d/dynevent/fprobe_args_vfs.tc b/tools/testing/selftests/ftrace/test.d/dynevent/fprobe_args_vfs.tc
new file mode 100644
index 000000000000..0a5baaffc137
--- /dev/null
+++ b/tools/testing/selftests/ftrace/test.d/dynevent/fprobe_args_vfs.tc
@@ -0,0 +1,40 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-2.0
+# description: Fprobe event VFS type argument
+# requires: kprobe_events
+
+: "Test argument %pd with name for fprobe"
+echo 'f:testprobe dput name=$arg1:%pd' > dynamic_events
+echo 1 > events/fprobes/testprobe/enable
+grep -q "1" events/fprobes/testprobe/enable
+echo 0 > events/fprobes/testprobe/enable
+grep "dput" trace | grep -q "enable"
+echo "" > dynamic_events
+echo "" > trace
+
+: "Test argument %pd without name for fprobe"
+echo 'f:testprobe dput $arg1:%pd' > dynamic_events
+echo 1 > events/fprobes/testprobe/enable
+grep -q "1" events/fprobes/testprobe/enable
+echo 0 > events/fprobes/testprobe/enable
+grep "dput" trace | grep -q "enable"
+echo "" > dynamic_events
+echo "" > trace
+
+: "Test argument %pD with name for fprobe"
+echo 'f:testprobe vfs_read name=$arg1:%pD' > dynamic_events
+echo 1 > events/fprobes/testprobe/enable
+grep -q "1" events/fprobes/testprobe/enable
+echo 0 > events/fprobes/testprobe/enable
+grep "vfs_read" trace | grep -q "enable"
+echo "" > dynamic_events
+echo "" > trace
+
+: "Test argument %pD without name for fprobe"
+echo 'f:testprobe vfs_read $arg1:%pD' > dynamic_events
+echo 1 > events/fprobes/testprobe/enable
+grep -q "1" events/fprobes/testprobe/enable
+echo 0 > events/fprobes/testprobe/enable
+grep "vfs_read" trace | grep -q "enable"
+echo "" > dynamic_events
+echo "" > trace
--
2.31.1


2024-02-07 14:32:57

by Masami Hiramatsu

[permalink] [raw]
Subject: Re: [PATCH v5 4/8] tracing/probes: support '%pd/%pD' type for fprobe

On Thu, 25 Jan 2024 15:39:19 +0800
Ye Bin <[email protected]> wrote:

> Support print type '%pd/%pD' for print dentry's or file's name.
>

nit: Looks good to me. but the patch ordering seems a bit strange.
This should be next to [2/8].

Acked-by: Masami Hiramatsu (Google) <[email protected]>

Thank you,

> Signed-off-by: Ye Bin <[email protected]>
> ---
> kernel/trace/trace_fprobe.c | 6 ++++++
> 1 file changed, 6 insertions(+)
>
> diff --git a/kernel/trace/trace_fprobe.c b/kernel/trace/trace_fprobe.c
> index 7d2ddbcfa377..988d68e906ad 100644
> --- a/kernel/trace/trace_fprobe.c
> +++ b/kernel/trace/trace_fprobe.c
> @@ -976,6 +976,7 @@ static int __trace_fprobe_create(int argc, const char *argv[])
> char gbuf[MAX_EVENT_NAME_LEN];
> char sbuf[KSYM_NAME_LEN];
> char abuf[MAX_BTF_ARGS_LEN];
> + char *dbuf = NULL;
> bool is_tracepoint = false;
> struct tracepoint *tpoint = NULL;
> struct traceprobe_parse_context ctx = {
> @@ -1086,6 +1087,10 @@ static int __trace_fprobe_create(int argc, const char *argv[])
> argv = new_argv;
> }
>
> + ret = traceprobe_expand_dentry_args(argc, argv, &dbuf);
> + if (ret)
> + goto out;
> +
> /* setup a probe */
> tf = alloc_trace_fprobe(group, event, symbol, tpoint, maxactive,
> argc, is_return);
> @@ -1131,6 +1136,7 @@ static int __trace_fprobe_create(int argc, const char *argv[])
> trace_probe_log_clear();
> kfree(new_argv);
> kfree(symbol);
> + kfree(dbuf);
> return ret;
>
> parse_error:
> --
> 2.31.1
>


--
Masami Hiramatsu (Google) <[email protected]>

2024-02-07 14:40:09

by Masami Hiramatsu

[permalink] [raw]
Subject: Re: [PATCH v5 1/8] tracing/probes: add traceprobe_expand_dentry_args() helper

Hi Ye,

On Thu, 25 Jan 2024 15:39:16 +0800
Ye Bin <[email protected]> wrote:

> Add traceprobe_expand_dentry_args() to expand dentry args. this API is
> prepare to support "%pd" print format for kprobe.
>
> Signed-off-by: Ye Bin <[email protected]>
> ---
> kernel/trace/trace_probe.c | 50 ++++++++++++++++++++++++++++++++++++++
> kernel/trace/trace_probe.h | 2 ++
> 2 files changed, 52 insertions(+)
>
> diff --git a/kernel/trace/trace_probe.c b/kernel/trace/trace_probe.c
> index 4dc74d73fc1d..0c1caf0f474a 100644
> --- a/kernel/trace/trace_probe.c
> +++ b/kernel/trace/trace_probe.c
> @@ -1565,6 +1565,56 @@ const char **traceprobe_expand_meta_args(int argc, const char *argv[],
> return ERR_PTR(ret);
> }
>
> +/* @buf: *buf must be equal to NULL. Caller must to free *buf */
> +int traceprobe_expand_dentry_args(int argc, const char *argv[], char **buf)
> +{
> + int i, used, ret;
> + int bufsize = MAX_DENTRY_ARGS_LEN;

nit: this bufsize is consistent, thus you should use the MAX_DENTRY_ARGS_LEN
directly. (or add 'const' or maybe you can update bufsize as remaining size)

Thanks,

> + char *tmpbuf = NULL;
> +
> + if (*buf)
> + return -EINVAL;
> +
> + used = 0;
> + for (i = 0; i < argc; i++) {
> + if (glob_match("*:%pd", argv[i])) {
> + char *tmp;
> + char *equal;
> +
> + if (!tmpbuf) {
> + tmpbuf = kmalloc(bufsize, GFP_KERNEL);
> + if (!tmpbuf)
> + return -ENOMEM;
> + }
> +
> + tmp = kstrdup(argv[i], GFP_KERNEL);
> + if (!tmp)
> + goto nomem;
> +
> + equal = strchr(tmp, '=');
> + if (equal)
> + *equal = '\0';
> + tmp[strlen(argv[i]) - 4] = '\0';
> + ret = snprintf(tmpbuf + used, bufsize - used,
> + "%s%s+0x0(+0x%zx(%s)):string",
> + equal ? tmp : "", equal ? "=" : "",
> + offsetof(struct dentry, d_name.name),
> + equal ? equal + 1 : tmp);
> + kfree(tmp);
> + if (ret >= bufsize - used)
> + goto nomem;
> + argv[i] = tmpbuf + used;
> + used += ret + 1;
> + }
> + }
> +
> + *buf = tmpbuf;
> + return 0;
> +nomem:
> + kfree(tmpbuf);
> + return -ENOMEM;
> +}
> +
> void traceprobe_finish_parse(struct traceprobe_parse_context *ctx)
> {
> clear_btf_context(ctx);
> diff --git a/kernel/trace/trace_probe.h b/kernel/trace/trace_probe.h
> index 850d9ecb6765..5800513439f3 100644
> --- a/kernel/trace/trace_probe.h
> +++ b/kernel/trace/trace_probe.h
> @@ -34,6 +34,7 @@
> #define MAX_ARRAY_LEN 64
> #define MAX_ARG_NAME_LEN 32
> #define MAX_BTF_ARGS_LEN 128
> +#define MAX_DENTRY_ARGS_LEN 256
> #define MAX_STRING_SIZE PATH_MAX
> #define MAX_ARG_BUF_LEN (MAX_TRACE_ARGS * MAX_ARG_NAME_LEN)
>
> @@ -402,6 +403,7 @@ extern int traceprobe_parse_probe_arg(struct trace_probe *tp, int i,
> const char **traceprobe_expand_meta_args(int argc, const char *argv[],
> int *new_argc, char *buf, int bufsize,
> struct traceprobe_parse_context *ctx);
> +extern int traceprobe_expand_dentry_args(int argc, const char *argv[], char **buf);
>
> extern int traceprobe_update_arg(struct probe_arg *arg);
> extern void traceprobe_free_probe_arg(struct probe_arg *arg);
> --
> 2.31.1
>


--
Masami Hiramatsu (Google) <[email protected]>

2024-02-07 14:41:04

by Masami Hiramatsu

[permalink] [raw]
Subject: Re: [PATCH v5 7/8] selftests/ftrace: add kprobe test cases for VFS type "%pd" and "%pD"

On Thu, 25 Jan 2024 15:39:22 +0800
Ye Bin <[email protected]> wrote:

> 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.
>
> Signed-off-by: Ye Bin <[email protected]>
> ---
> .../ftrace/test.d/kprobe/kprobe_args_vfs.tc | 43 +++++++++++++++++++
> 1 file changed, 43 insertions(+)
> create mode 100644 tools/testing/selftests/ftrace/test.d/kprobe/kprobe_args_vfs.tc
>
> diff --git a/tools/testing/selftests/ftrace/test.d/kprobe/kprobe_args_vfs.tc b/tools/testing/selftests/ftrace/test.d/kprobe/kprobe_args_vfs.tc
> new file mode 100644
> index 000000000000..cf0599b90f1a
> --- /dev/null
> +++ b/tools/testing/selftests/ftrace/test.d/kprobe/kprobe_args_vfs.tc
> @@ -0,0 +1,43 @@
> +#!/bin/sh
> +# SPDX-License-Identifier: GPL-2.0
> +# description: Kprobe event VFS type argument
> +# requires: kprobe_events
> +
> +: "Test argument %pd/%pD in README"
> +grep -q "%pd/%pD" README

No. This means, if the kernel doesn't support %pd/%pD, this test fails.
But the latest version of the ftracetest will be used for testing stable
kernels (which will not support this feature).
So you should make test returning 'unsupported' for that case.

You can add a requirement as

# requires: kprobe_events "%pd/%pD":README

(e.g. tools/testing/selftests/ftrace/test.d/dynevent/eprobes_syntax_errors.tc)

Thank you,

> +
> +: "Test argument %pd with name"
> +echo 'p:testprobe dput name=$arg1:%pd' > kprobe_events
> +echo 1 > events/kprobes/testprobe/enable
> +grep -q "1" events/kprobes/testprobe/enable
> +echo 0 > events/kprobes/testprobe/enable
> +grep "dput" trace | grep -q "enable"
> +echo "" > kprobe_events
> +echo "" > trace
> +
> +: "Test argument %pd without name"
> +echo 'p:testprobe dput $arg1:%pd' > kprobe_events
> +echo 1 > events/kprobes/testprobe/enable
> +grep -q "1" events/kprobes/testprobe/enable
> +echo 0 > events/kprobes/testprobe/enable
> +grep "dput" trace | grep -q "enable"
> +echo "" > kprobe_events
> +echo "" > trace
> +
> +: "Test argument %pD with name"
> +echo 'p:testprobe vfs_read name=$arg1:%pD' > kprobe_events
> +echo 1 > events/kprobes/testprobe/enable
> +grep -q "1" events/kprobes/testprobe/enable
> +echo 0 > events/kprobes/testprobe/enable
> +grep "vfs_read" trace | grep -q "enable"
> +echo "" > kprobe_events
> +echo "" > trace
> +
> +: "Test argument %pD without name"
> +echo 'p:testprobe vfs_read $arg1:%pD' > kprobe_events
> +echo 1 > events/kprobes/testprobe/enable
> +grep -q "1" events/kprobes/testprobe/enable
> +echo 0 > events/kprobes/testprobe/enable
> +grep "vfs_read" trace | grep -q "enable"
> +echo "" > kprobe_events
> +echo "" > trace
> --
> 2.31.1
>
>


--
Masami Hiramatsu (Google) <[email protected]>