Skip to contentSkip to conversation

tracing: A few fixes for 6.5

PATCH New Steven Rostedt 1 participant 4 messages Jul 5, 2023

Subsystems (3)
lore
Share & tools
lore

Steven Rostedt

author maintainer
#0 lore raw

cover tracing: A few fixes for 6.5

Tracing fixes for 6.5:

- Fix bad git merge of #endif in arm64 code
A merge of the arm64 tree caused #endif to go into the wrong place

- Fix crash on lseek of write access to tracefs/error_log Opening error_log as write only, and then doing an lseek() causes a kernel panic, because the lseek() handle expects a "seq_file" to exist (which is not done on write only opens). Use tracing_lseek() that tests for this instead of calling the default seq lseek handler.

- Check for negative instead of -E2BIG for error on strscpy() returns Instead of testing for -E2BIG from strscpy(), to be more robust, check for less than zero, which will make sure it catches any error that strscpy() may someday return.

git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace.git
trace/core

Head SHA1: fddca7db4a4c17f7333793dfb5308d80c76d2896

Arnd Bergmann (1):
arm64: ftrace: fix build error with CONFIG_FUNCTION_GRAPH_TRACER=n

Mateusz Stachyra (1):
tracing: Fix null pointer dereference in tracing_err_log_open()

Steven Rostedt (Google) (1):
tracing/boot: Test strscpy() against less than zero for error

----

 arch/arm64/kernel/asm-offsets.c | 2 +-
 kernel/trace/trace.c            | 2 +-
 kernel/trace/trace_boot.c       | 8 ++++----
 3 files changed, 6 insertions(+), 6 deletions(-)
author maintainer
#1 lore raw

PATCH 1/3 tracing: Fix null pointer dereference in tracing_err_log_open()

From: Mateusz Stachyra <[email protected]>

Fix an issue in function 'tracing_err_log_open'.
The function doesn't call 'seq_open' if the file is opened only with write permissions, which results in 'file->private_data' being left as null. If we then use 'lseek' on that opened file, 'seq_lseek' dereferences 'file->private_data' in 'mutex_lock(&m->lock)', resulting in a kernel panic. Writing to this node requires root privileges, therefore this bug has very little security impact.

Tracefs node: /sys/kernel/tracing/error_log

Example Kernel panic:

Unable to handle kernel NULL pointer dereference at virtual address 0000000000000038
Call trace:
 mutex_lock+0x30/0x110
 seq_lseek+0x34/0xb8
 __arm64_sys_lseek+0x6c/0xb8
 invoke_syscall+0x58/0x13c
 el0_svc_common+0xc4/0x10c
 do_el0_svc+0x24/0x98
 el0_svc+0x24/0x88
 el0t_64_sync_handler+0x84/0xe4
 el0t_64_sync+0x1b4/0x1b8
Code: d503201f aa0803e0 aa1f03e1 aa0103e9 (c8e97d02)
---[ end trace 561d1b49c12cf8a5 ]---
Kernel panic - not syncing: Oops: Fatal exception
---
 kernel/trace/trace.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/trace/trace.c +1 -1
index 64a4dde073ef..3d34e6fea6b2 100644
@@ -8135,7 +8135,7 @@ static const struct file_operations tracing_err_log_fops = {
 	.open           = tracing_err_log_open,
 	.write		= tracing_err_log_write,
 	.read           = seq_read,
-	.llseek         = seq_lseek,
+	.llseek         = tracing_lseek,
 	.release        = tracing_err_log_release,
 };
 
Signature · 1 line
2.39.2
author maintainer
#2 lore raw

PATCH 2/3 arm64: ftrace: fix build error with CONFIG_FUNCTION_GRAPH_TRACER=n

In reply to Steven Rostedt

From: Arnd Bergmann <[email protected]>

It appears that a merge conflict ended up hiding a newly added constant in some configurations:

arch/arm64/kernel/entry-ftrace.S: Assembler messages:
arch/arm64/kernel/entry-ftrace.S:59: Error: undefined symbol FTRACE_OPS_DIRECT_CALL used as an immediate value

FTRACE_OPS_DIRECT_CALL is still used when CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS is enabled, even if CONFIG_FUNCTION_GRAPH_TRACER is disabled, so change the ifdef accordingly.

Cc Will Deacon <[email protected]>
Cc Mark Rutland <[email protected]>
Cc Donglin Peng <[email protected]>
Fixes 3646970322464 ("arm64: ftrace: Enable HAVE_FUNCTION_GRAPH_RETVAL")
---
 arch/arm64/kernel/asm-offsets.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm64/kernel/asm-offsets.c +1 -1
index 757d01a68ffd..5ff1942b04fc 100644
@@ -213,9 +213,9 @@ int main(void)
   DEFINE(FGRET_REGS_X7,			offsetof(struct fgraph_ret_regs, regs[7]));
   DEFINE(FGRET_REGS_FP,			offsetof(struct fgraph_ret_regs, fp));
   DEFINE(FGRET_REGS_SIZE,		sizeof(struct fgraph_ret_regs));
+#endif
 #ifdef CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
   DEFINE(FTRACE_OPS_DIRECT_CALL,	offsetof(struct ftrace_ops, direct_call));
-#endif
 #endif
   return 0;
 }
Signature · 1 line
2.39.2
author maintainer
#3 lore raw

PATCH 3/3 tracing/boot: Test strscpy() against less than zero for error

In reply to Steven Rostedt

From: "Steven Rostedt (Google)" <[email protected]>

Instead of checking for -E2BIG, it is better to just check for less than zero of strscpy() for error. Testing for -E2BIG is not very robust, and the calling code does not really care about the error code, just that there was an error.

One of the updates to convert strlcpy() to strscpy() had a v2 version that changed the test from testing against -E2BIG to less than zero, but I took the v1 version that still tested for -E2BIG.

---
 kernel/trace/trace_boot.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
kernel/trace/trace_boot.c +4 -4
index 5fe525f1b8cc..7ccc7a8e155b 100644
@@ -31,7 +31,7 @@ trace_boot_set_instance_options(struct trace_array *tr, struct xbc_node *node)
 
 	/* Common ftrace options */
 	xbc_node_for_each_array_value(node, "options", anode, p) {
-		if (strscpy(buf, p, ARRAY_SIZE(buf)) == -E2BIG) {
+		if (strscpy(buf, p, ARRAY_SIZE(buf)) < 0) {
 			pr_err("String is too long: %s\n", p);
 			continue;
 		}
@@ -87,7 +87,7 @@ trace_boot_enable_events(struct trace_array *tr, struct xbc_node *node)
 	const char *p;
 
 	xbc_node_for_each_array_value(node, "events", anode, p) {
-		if (strscpy(buf, p, ARRAY_SIZE(buf)) == -E2BIG) {
+		if (strscpy(buf, p, ARRAY_SIZE(buf)) < 0) {
 			pr_err("String is too long: %s\n", p);
 			continue;
 		}
@@ -486,7 +486,7 @@ trace_boot_init_one_event(struct trace_array *tr, struct xbc_node *gnode,
 
 	p = xbc_node_find_value(enode, "filter", NULL);
 	if (p && *p != '\0') {
-		if (strscpy(buf, p, ARRAY_SIZE(buf)) == -E2BIG)
+		if (strscpy(buf, p, ARRAY_SIZE(buf)) < 0)
 			pr_err("filter string is too long: %s\n", p);
 		else if (apply_event_filter(file, buf) < 0)
 			pr_err("Failed to apply filter: %s\n", buf);
@@ -494,7 +494,7 @@ trace_boot_init_one_event(struct trace_array *tr, struct xbc_node *gnode,
 
 	if (IS_ENABLED(CONFIG_HIST_TRIGGERS)) {
 		xbc_node_for_each_array_value(enode, "actions", anode, p) {
-			if (strscpy(buf, p, ARRAY_SIZE(buf)) == -E2BIG)
+			if (strscpy(buf, p, ARRAY_SIZE(buf)) < 0)
 				pr_err("action string is too long: %s\n", p);
 			else if (trigger_process_regex(file, buf) < 0)
 				pr_err("Failed to apply an action: %s\n", p);
Signature · 1 line
2.39.2