Skip to contentSkip to conversation

Fix double-free when linker processes empty sections

PATCH In review Eduard Zingerman 4 participants 6 messages Mar 28, 2023

Subsystems (4)
lore
Share & tools
lore

Eduard Zingerman

author maintainer
#0 lore raw

cover Fix double-free when linker processes empty sections

Fixes double-free error in linker.c:bpf_linker__free() caused by realloc(..., 0) call in linker.c:extend_sec() (such a call "frees" memory every second time :). The error is triggered when object files with empty sections of the same name are processed by linker.

- The first patch extends progs/linked_funcs[12].c to trigger the error upon tests compilation;
- The second patch contains detailed description of the error, fix and appropriate attributions.

Eduard Zingerman (2):
selftests/bpf: Test if bpftool linker handles empty sections
libbpf: Fix double-free when linker processes empty sections

 tools/lib/bpf/linker.c                            | 14 +++++++++++++-
 tools/testing/selftests/bpf/progs/linked_funcs1.c |  3 +++
 tools/testing/selftests/bpf/progs/linked_funcs2.c |  3 +++
 3 files changed, 19 insertions(+), 1 deletion(-)
Signature · 1 line
2.40.0
author maintainer
#1 lore raw

PATCH 1/2 selftests/bpf: Test if bpftool linker handles empty sections

Adds two empty functions to linked_funcs[12].c. The functions are annotated as "naked" and go to a separate section. This section ends up having size 0. bpftool linker merges content for sections with identical names. This tests if it can handle empty sections.

---
 tools/testing/selftests/bpf/progs/linked_funcs1.c | 3 +++
 tools/testing/selftests/bpf/progs/linked_funcs2.c | 3 +++
 2 files changed, 6 insertions(+)
tools/testing/selftests/bpf/progs/linked_funcs1.c +3 -0
index c4b49ceea967..029bb5022ba2 100644
@@ -86,4 +86,7 @@ int BPF_PROG(handler1, struct pt_regs *regs, long id)
 	return 0;
 }
 
+SEC(".empty_section")
+__naked void empty_function1(void) {}
+
 char LICENSE[] SEC("license") = "GPL";
tools/testing/selftests/bpf/progs/linked_funcs2.c +3 -0
index 013ff0645f0c..4547c8dfc689 100644
@@ -86,4 +86,7 @@ int BPF_PROG(handler2, struct pt_regs *regs, long id)
 	return 0;
 }
 
+SEC(".empty_section")
+__naked void empty_function2(void) {}
+
 char LICENSE[] SEC("license") = "GPL";
Signature · 1 line
2.40.0
#2 lore raw

Re: [PATCH bpf-next 1/2] selftests/bpf: Test if bpftool linker handles empty sections

Quoted 29 lines · Eduard Zingerman

Adds two empty functions to linked_funcs[12].c. The functions are annotated as "naked" and go to a separate section. This section ends up having size 0. bpftool linker merges content for sections with identical names. This tests if it can handle empty sections.

---
 tools/testing/selftests/bpf/progs/linked_funcs1.c | 3 +++
 tools/testing/selftests/bpf/progs/linked_funcs2.c | 3 +++
 2 files changed, 6 insertions(+)
tools/testing/selftests/bpf/progs/linked_funcs1.c +3 -0
index c4b49ceea967..029bb5022ba2 100644
@@ -86,4 +86,7 @@ int BPF_PROG(handler1, struct pt_regs *regs, long id)
        return 0;
 }
 
+SEC(".empty_section")
+__naked void empty_function1(void) {}
+
 char LICENSE[] SEC("license") = "GPL";
tools/testing/selftests/bpf/progs/linked_funcs2.c +0 -0
index 013ff0645f0c..4547c8dfc689 100644

tools/testing/selftests/bpf/progs/linked_funcs2.c int BPF_PROG(handler2, struct pt_regs *regs, long id) · line 90

Quoted 5 lines
tools/testing/selftests/bpf/progs/linked_funcs2.c +2 -0
@@ -86,4 +86,7 @@ int BPF_PROG(handler2, struct pt_regs *regs, long id)
        return 0;
 }
 
+SEC(".empty_section")
+__naked void empty_function2(void) {}

These empty section functions make this whole BPF object file invalid from libbpf's standpoint. It didn't feel worth it to add this confusion just to test this edge case in realloc() handling. So I dropped this patch and only applied libbpf fix. Pushed to bpf-next, thanks!

Quoted 5 lines
tools/testing/selftests/bpf/progs/linked_funcs2.c +1 -1
+
 char LICENSE[] SEC("license") = "GPL";
--

2.40.0

author maintainer
#3 lore raw

PATCH 2/2 libbpf: Fix double-free when linker processes empty sections

In reply to Eduard Zingerman

Double-free error in bpf_linker__free() was reported by James Hilliard. The error is caused by miss-use of realloc() in extend_sec(). The error occurs when two files with empty sections of the same name are linked:
- when first file is processed:
- extend_sec() calls realloc(dst->raw_data, dst_align_sz)

    with dst->raw_data == NULL and dst_align_sz == 0;
  - dst->raw_data is set to a special pointer to a memory block of
    size zero;
- when second file is processed:
  - extend_sec() calls realloc(dst->raw_data, dst_align_sz)
    with dst->raw_data == <special pointer> and dst_align_sz == 0;
  - realloc() "frees" dst->raw_data special pointer and returns NULL;
  - extend_sec() exits with -ENOMEM, and the old dst->raw_data value
    is preserved (it is now invalid);

- eventually, bpf_linker__free() attempts to free dst->raw_data again.

This patch fixes the bug by avoiding -ENOMEM exit for dst_align_sz == 0. The fix was suggested by Andrii Nakryiko <[email protected]>.

---
 tools/lib/bpf/linker.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)
tools/lib/bpf/linker.c +13 -1
index d7069780984a..5ced96d99f8c 100644
@@ -1115,7 +1115,19 @@ static int extend_sec(struct bpf_linker *linker, struct dst_sec *dst, struct src
 
 	if (src->shdr->sh_type != SHT_NOBITS) {
 		tmp = realloc(dst->raw_data, dst_final_sz);
-		if (!tmp)
+		/* If dst_align_sz == 0, realloc() behaves in a special way:
+		 * 1. When dst->raw_data is NULL it returns:
+		 *    "either NULL or a pointer suitable to be passed to free()" [1].
+		 * 2. When dst->raw_data is not-NULL it frees dst->raw_data and returns NULL,
+		 *    thus invalidating any "pointer suitable to be passed to free()" obtained
+		 *    at step (1).
+		 *
+		 * The dst_align_sz > 0 check avoids error exit after (2), otherwise
+		 * dst->raw_data would be freed again in bpf_linker__free().
+		 *
+		 * [1] man 3 realloc
+		 */
+		if (!tmp && dst_align_sz > 0)
 			return -ENOMEM;
 		dst->raw_data = tmp;
 
Signature · 1 line
2.40.0
Tested-by on v1 2/2
#4 lore raw

Re: [PATCH bpf-next 2/2] libbpf: Fix double-free when linker processes empty sections

Quoted 16 lines · Eduard Zingerman

Double-free error in bpf_linker__free() was reported by James Hilliard. The error is caused by miss-use of realloc() in extend_sec(). The error occurs when two files with empty sections of the same name are linked:
- when first file is processed:
- extend_sec() calls realloc(dst->raw_data, dst_align_sz)

    with dst->raw_data == NULL and dst_align_sz == 0;
  - dst->raw_data is set to a special pointer to a memory block of
    size zero;
- when second file is processed:
  - extend_sec() calls realloc(dst->raw_data, dst_align_sz)
    with dst->raw_data == <special pointer> and dst_align_sz == 0;
  - realloc() "frees" dst->raw_data special pointer and returns NULL;
  - extend_sec() exits with -ENOMEM, and the old dst->raw_data value
    is preserved (it is now invalid);
Quoted 7 lines

- eventually, bpf_linker__free() attempts to free dst->raw_data again.

This patch fixes the bug by avoiding -ENOMEM exit for dst_align_sz == 0. The fix was suggested by Andrii Nakryiko <[email protected]>.

Quoted 33 lines
---
 tools/lib/bpf/linker.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)
tools/lib/bpf/linker.c +13 -2
index d7069780984a..5ced96d99f8c 100644
@@ -1115,7 +1115,19 @@ static int extend_sec(struct bpf_linker *linker, struct dst_sec *dst, struct src
 
        if (src->shdr->sh_type != SHT_NOBITS) {
                tmp = realloc(dst->raw_data, dst_final_sz);
-               if (!tmp)
+               /* If dst_align_sz == 0, realloc() behaves in a special way:
+                * 1. When dst->raw_data is NULL it returns:
+                *    "either NULL or a pointer suitable to be passed to free()" [1].
+                * 2. When dst->raw_data is not-NULL it frees dst->raw_data and returns NULL,
+                *    thus invalidating any "pointer suitable to be passed to free()" obtained
+                *    at step (1).
+                *
+                * The dst_align_sz > 0 check avoids error exit after (2), otherwise
+                * dst->raw_data would be freed again in bpf_linker__free().
+                *
+                * [1] man 3 realloc
+                */
+               if (!tmp && dst_align_sz > 0)
                        return -ENOMEM;
                dst->raw_data = tmp;
 
--

2.40.0

#5 lore raw
In reply to Eduard Zingerman
This series was applied to bpf/bpf-next.git (master) by Andrii Nakryiko <[email protected]>:

Hello:

This series was applied to bpf/bpf-next.git (master) by Andrii Nakryiko <[email protected]>:

Quoted 11 lines · On Tue, 28 Mar 2023 03:47:36 +0300 you wrote:

Fixes double-free error in linker.c:bpf_linker__free() caused by realloc(..., 0) call in linker.c:extend_sec() (such a call "frees" memory every second time :). The error is triggered when object files with empty sections of the same name are processed by linker.

- The first patch extends progs/linked_funcs[12].c to trigger the error upon tests compilation;
- The second patch contains detailed description of the error, fix and appropriate attributions.

[...]

Here is the summary with links:
- [bpf-next,1/2] selftests/bpf: Test if bpftool linker handles empty sections (no matching commit)
- [bpf-next,2/2] libbpf: Fix double-free when linker processes empty sections

    https://git.kernel.org/bpf/bpf-next/c/d08ab82f59d5

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html