lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [day] [month] [year] [list]
Message-ID: <20250829102050.232291-1-seokwoo.chung130@gmail.com>
Date: Fri, 29 Aug 2025 19:20:50 +0900
From: Ryan Chung <seokwoo.chung130@...il.com>
To: Steven Rostedt <rostedt@...dmis.org>,
	Masami Hiramatsu <mhiramat@...nel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@...icios.com>,
	linux-trace-kernel@...r.kernel.org,
	linux-kernel@...r.kernel.org,
	Ryan Chung <seokwoo.chung130@...il.com>
Subject: [PATCH v2] trace_fprobe.c: TODO: handle filter, nofilter or symbol list

This v2 addresses the TODO in trace_fprobe to handle comma-separated
symbol lists and the '!' prefix. Tokens starting with '!' are collected
as "nofilter", and the others as "filter", then passed to
register_fprobe() accordingly. Empty tokens are rejected and errors are
reported with trace_probe_log_err().

Questions for maintainers (to confirm my understanding):
  * Parsing location: Masami suggested doing the parsing in the parse
    stage (e.g., via parse_symbol_and_return()). v2 keeps the logic in
    __register_trace_fprobe(), but I can move the call into the parsing
    path in v3 if that is the preferred place. Is that correct?
  * Documentation: I plan to update the user-facing docs for fprobe
    syntax. Is Documentation/trace/ the right place (e.g., 
    Documentation/trace/fprobetrace.rst)?

Link: https://lore.kernel.org/linux-trace-kernel/20250812162101.5981-1-seokwoo.chung130@gmail.com/
Signed-off-by: Ryan Chung <seokwoo.chung130@...il.com>
---

Changes in v2:
  * Classify '!' tokens as nofilter, others as filter; pass both to
    register_fprobe().
  * Reject empty tokens; log errors with trace_probe_log_*().
  * Use __free(kfree) for temporary buffers.
  * Keep subject and style per "Submitting patches" (tabs, wrapping).
  * No manual dedup (leave to ftrace).

 kernel/trace/trace_fprobe.c | 48 +++++++++++++++++++++++++++++++++++--
 1 file changed, 46 insertions(+), 2 deletions(-)

diff --git a/kernel/trace/trace_fprobe.c b/kernel/trace/trace_fprobe.c
index b36ade43d4b3..d731d9754a39 100644
--- a/kernel/trace/trace_fprobe.c
+++ b/kernel/trace/trace_fprobe.c
@@ -815,6 +815,11 @@ static int trace_fprobe_verify_target(struct trace_fprobe *tf)
 static int __register_trace_fprobe(struct trace_fprobe *tf)
 {
 	int i, ret;
+	const char *p, *q;
+	size_t spec_len, flen = 0, nflen = 0, tlen;
+	bool have_f = false, have_nf = false;
+	char *filter __free(kfree) = NULL;
+	char *nofilter __free(kfree) = NULL;
 
 	/* Should we need new LOCKDOWN flag for fprobe? */
 	ret = security_locked_down(LOCKDOWN_KPROBES);
@@ -835,8 +840,47 @@ static int __register_trace_fprobe(struct trace_fprobe *tf)
 	if (trace_fprobe_is_tracepoint(tf))
 		return __regsiter_tracepoint_fprobe(tf);
 
-	/* TODO: handle filter, nofilter or symbol list */
-	return register_fprobe(&tf->fp, tf->symbol, NULL);
+	spec_len = strlen(tf->symbol);
+	filter = kzalloc(spec_len + 1, GFP_KERNEL);
+	nofilter = kzalloc(spec_len + 1, GFP_KERNEL);
+	if (!filter || !nofilter)
+		return -ENOMEM;
+
+	p = tf->symbol;
+	for (p = tf->symbol; p; p = q ? q + 1 : NULL) {
+		q = strchr(p, ',');
+		tlen = q ? (size_t)(q-p) : strlen(p);
+
+		/* reject empty token */
+		if (!tlen) {
+			trace_probe_log_set_index(1);
+			trace_probe_log_err(0, BAD_TP_NAME);
+			return -EINVAL;
+		}
+
+		if (*p == '!') {
+			if (tlen == 1) {
+				trace_probe_log_set_index(1);
+				trace_probe_log_err(0, BAD_TP_NAME);
+				return -EINVAL;
+			}
+			if (have_nf)
+				nofilter[nflen++] = ',';
+			memcpy(nofilter + nflen, p + 1, tlen - 1);
+			nflen += tlen - 1;
+			have_nf = true;
+		} else {
+			if (have_f)
+				filter[flen++] = ',';
+			memcpy(filter + flen, p, tlen);
+			flen += tlen;
+			have_f = true;
+		}
+	}
+
+	return register_fprobe(&tf->fp,
+			have_f ? filter : NULL,
+			have_nf ? nofilter : NULL);
 }
 
 /* Internal unregister function - just handle fprobe and flags */
-- 
2.43.0


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ