Skip to contentSkip to conversation

perf annotate: Fix NULL pointer dereference in loongarch_call__parse

PATCH Superseded Jianping Liu 2 participants 2 messages Apr 23 perf events

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 perf annotate: Fix NULL pointer dereference in loongarch_call__parse

loongarch_call__parse() dereferences the return value of strchr() without NULL check:

  name = strchr(endptr, '<');
  name++;

When objdump output for a 'bl' (branch-and-link) instruction does not contain a symbol name delimited by '<' and '>' (e.g., when the call target has no associated symbol), strchr() returns NULL, and the subsequent name++ produces a wild pointer (0x1). This leads to a segmentation fault when strchr() is later called with this invalid pointer:

  #0  __strchr_lasx () from /lib64/libc.so.6
  #1  loongarch_call__parse () at arch/loongarch/annotate/instructions.c:29
  #2  disasm_line__init_ins () at util/annotate.c:1205

...

The objdump line triggering the crash looks like:

bl 98824 # 0x9000000000641ea0

Note the absence of "<function_name>" after the address.

Add a NULL check for the return value of strchr() before dereferencing it, consistent with s390_call__parse() which handles the same case correctly.

Fixes 4ca0d340ce20 ("perf annotate: Fix instruction association and parsing for LoongArch")
---
 tools/perf/util/annotate-arch/annotate-loongarch.c | 3 +++
 1 file changed, 3 insertions(+)
tools/perf/util/annotate-arch/annotate-loongarch.c +3 -0
index c2addca77320..0179062a6cf9 100644
@@ -29,6 +29,9 @@ static int loongarch_call__parse(const struct arch *arch, struct ins_operands *o
 	ops->target.addr = strtoull(c, &endptr, 16);
 
 	name = strchr(endptr, '<');
+	if (name == NULL)
+		return -1;
+
 	name++;
 
 	if (arch->objdump.skip_functions_char &&
Signature · 1 line
2.43.7
#1 lore raw
Sashiko AI review found 1 potential issue(s): - [Medium] The fix correctly prevents a NULL pointer dereference, but improperly aborts instruction parsing by returning `-1`. This strips the `is_call`…

Sashiko AI review found 1 potential issue(s):
- [Medium] The fix correctly prevents a NULL pointer dereference, but improperly aborts instruction parsing by returning -1. This strips the is_call attribute and degrades perf annotate functionality, whereas it should bypass name extraction and resolve the target using the already extracted address. --

commit 807b803f1f78132524dc6117283161821fab508a
Author: Jianping Liu <[email protected]>

perf annotate: Fix NULL pointer dereference in loongarch_call__parse

This commit prevents a segmentation fault in loongarch_call__parse() by adding a NULL check for the return value of strchr() when parsing objdump output. It handles cases where a branch-and-link instruction lacks a symbol name.

Quoted 3 lines

Add a NULL check for the return value of strchr() before dereferencing it, consistent with s390_call__parse() which handles the same case correctly.

Does s390_call__parse() actually handle this correctly without regressions?

Returning -1 causes disasm_line__init_ins() to receive a negative return value and set dl->ins.ops = NULL. This makes ins__is_call() evaluate to false, meaning the instruction completely loses its call semantics in the TUI.

tools/perf/util/annotate-arch/annotate-loongarch.c static int loongarch_call__parse(const struct arch *arch, struct ins_operands *o · line 35

Quoted 11 lines
tools/perf/util/annotate-arch/annotate-loongarch.c +3 -0
@@ -29,6 +29,9 @@ static int loongarch_call__parse(const struct arch *arch, struct ins_operands *o
 	ops->target.addr = strtoull(c, &endptr, 16);
 
 	name = strchr(endptr, '<');
+	if (name == NULL)
+		return -1;
+
 	name++;

Could this code bypass the name extraction and proceed directly to symbol resolution using the already extracted target address?

Since ops->target.addr is successfully parsed right above this check, skipping the name parsing and returning 0 instead of aborting with -1 would preserve the is_call attribute. This would allow the TUI to still draw branch arrows and navigate the call.

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1