Skip to contentSkip to conversation

tracing: Simple fix and clean up for 5.0

GIT PULL New Steven Rostedt 1 participant 3 messages Mar 4, 2019

Subsystems (2)
lore
Share & tools
lore

Steven Rostedt

author maintainer
#0 lore raw

Linus,

This is a "pre-pull". It's only one small fix and one small clean up. I'm testing a few small patches for my real pull request which will come at a later time. The second patch depends on your tree anyway so I included it along with the urgent fix.

A small fix Pavel sent me back in august was accidentally lost due to it being placed with some other patches that failed some tests, and was rebased out of my local tree. Which was a regression that caused event filters not to handle negative numbers.

The clean up is from Masami that realized that the code in kprobes that calls probe_mem_read() wrapper, which is to be used in code used by both kprobes and uprobes, was only in code for kprobes. It should not use the wrapper there, but instead call probe_kernel_read() directly.

Please pull the latest trace-v5.0-pre tree, which can be found at:

git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace.git
trace-v5.0-pre

Tag SHA1: 6c5b335a8c8d7d98c09d10e15058f8325ff2cf9d Head SHA1: 49ef5f45701c3dedc6aa6efee5d03756fea0f99d

Masami Hiramatsu (1):
tracing/kprobes: Use probe_kernel_read instead of probe_mem_read

Pavel Tikhomirov (1):
tracing: Fix event filters and triggers to handle negative numbers

----

 kernel/trace/trace_events_filter.c | 5 ++++-
 kernel/trace/trace_kprobe.c        | 2 +-
 2 files changed, 5 insertions(+), 2 deletions(-)
author maintainer
#1 lore raw

PATCH 1/2 tracing: Fix event filters and triggers to handle negative numbers

From: Pavel Tikhomirov <[email protected]>

Then tracing syscall exit event it is extremely useful to filter exit codes equal to some negative value, to react only to required errors. But negative numbers does not work:

[root@snorch sys_exit_read]# echo "ret == -1" > filter
bash: echo: write error: Invalid argument [root@snorch sys_exit_read]# cat filter
ret == -1

        ^

parse_error: Invalid value (did you forget quotes)?

Similar thing happens when setting triggers.

These is a regression in v4.17 introduced by the commit mentioned below, testing without these commit shows no problem with negative numbers.

Fixes 80765597bc58 ("tracing: Rewrite filter logic to be simpler and faster")
---
 kernel/trace/trace_events_filter.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
kernel/trace/trace_events_filter.c +4 -1
index 27821480105e..217ef481fbbb 100644
@@ -1301,7 +1301,7 @@ static int parse_pred(const char *str, void *data,
 		/* go past the last quote */
 		i++;
 
-	} else if (isdigit(str[i])) {
+	} else if (isdigit(str[i]) || str[i] == '-') {
 
 		/* Make sure the field is not a string */
 		if (is_string_field(field)) {
@@ -1314,6 +1314,9 @@ static int parse_pred(const char *str, void *data,
 			goto err_free;
 		}
 
+		if (str[i] == '-')
+			i++;
+
 		/* We allow 0xDEADBEEF */
 		while (isalnum(str[i]))
 			i++;
Signature · 1 line
2.20.1
author maintainer
#2 lore raw

PATCH 2/2 tracing/kprobes: Use probe_kernel_read instead of probe_mem_read

In reply to Steven Rostedt

From: Masami Hiramatsu <[email protected]>

Use probe_kernel_read() instead of probe_mem_read() because probe_mem_read() is a kind of wrapper for switching memory read function between uprobes and kprobes.

---
 kernel/trace/trace_kprobe.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/trace/trace_kprobe.c +1 -1
index 9eaf07f99212..99592c27465e 100644
@@ -865,7 +865,7 @@ fetch_store_strlen(unsigned long addr)
 	u8 c;
 
 	do {
-		ret = probe_mem_read(&c, (u8 *)addr + len, 1);
+		ret = probe_kernel_read(&c, (u8 *)addr + len, 1);
 		len++;
 	} while (c && ret == 0 && len < MAX_STRING_SIZE);
 
Signature · 1 line
2.20.1