Skip to contentSkip to conversation

enforce W^X for trampoline and dispatcher

PATCH Applied Song Liu 5 participants 7 messages Sep 26, 2022 – Sep 28, 2022

Subsystems (5)
Memory management x86 BPF BPF core +1 more subsystems
lore
Share & tools
lore

Song Liu

author reviewer
#0 lore raw

v2 cover enforce W^X for trampoline and dispatcher

Changes v1 => v2:
1. Update arch_prepare_bpf_dispatcher to use a RO image and a RW buffer. (Alexei) Note: I haven't found an existing test to cover this part, so this part was tested manually (comparing the generated dispatcher is the same).

Jeff Layton reported CPA W^X warning linux-next [1]. It turns out to be W^X issue with bpf trampoline and bpf dispatcher. Fix these by:

1. Use bpf_prog_pack for bpf_dispatcher;
2. Set memory permission properly with bpf trampoline.

[1] https://lore.kernel.org/lkml/[email protected]/

Song Liu (2):
bpf: use bpf_prog_pack for bpf_dispatcher
bpf: Enforce W^X for bpf trampoline

 arch/x86/net/bpf_jit_comp.c | 16 ++++++++--------
 include/linux/bpf.h         |  4 ++--
 include/linux/filter.h      |  5 +++++
 kernel/bpf/core.c           |  9 +++++++--
 kernel/bpf/dispatcher.c     | 27 +++++++++++++++++++++------
 kernel/bpf/trampoline.c     | 22 +++++-----------------
 6 files changed, 48 insertions(+), 35 deletions(-)

--
2.30.2

author reviewer
#1 lore raw

PATCH v2 1/2 bpf: use bpf_prog_pack for bpf_dispatcher

Allocate bpf_dispatcher with bpf_prog_pack_alloc so that bpf_dispatcher can share pages with bpf programs.

arch_prepare_bpf_dispatcher() is updated to provide a RW buffer as working area for arch code to write to.

This also fixes CPA W^X warnning like:

CPA refuse W^X violation: 8000000000000163 -> 0000000000000163 range: ...

---
 arch/x86/net/bpf_jit_comp.c | 16 ++++++++--------
 include/linux/bpf.h         |  3 ++-
 include/linux/filter.h      |  5 +++++
 kernel/bpf/core.c           |  9 +++++++--
 kernel/bpf/dispatcher.c     | 27 +++++++++++++++++++++------
 5 files changed, 43 insertions(+), 17 deletions(-)
arch/x86/net/bpf_jit_comp.c +8 -8
index ae89f4143eb4..f62c05385fa7 100644
@@ -2243,7 +2243,7 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 	return ret;
 }
 
-static int emit_bpf_dispatcher(u8 **pprog, int a, int b, s64 *progs)
+static int emit_bpf_dispatcher(u8 **pprog, int a, int b, s64 *progs, u8 *image, u8 *buf)
 {
 	u8 *jg_reloc, *prog = *pprog;
 	int pivot, err, jg_bytes = 1;
@@ -2259,12 +2259,12 @@ static int emit_bpf_dispatcher(u8 **pprog, int a, int b, s64 *progs)
 		EMIT2_off32(0x81, add_1reg(0xF8, BPF_REG_3),
 			    progs[a]);
 		err = emit_cond_near_jump(&prog,	/* je func */
-					  (void *)progs[a], prog,
+					  (void *)progs[a], image + (prog - buf),
 					  X86_JE);
 		if (err)
 			return err;
 
-		emit_indirect_jump(&prog, 2 /* rdx */, prog);
+		emit_indirect_jump(&prog, 2 /* rdx */, image + (prog - buf));
 
 		*pprog = prog;
 		return 0;
@@ -2289,7 +2289,7 @@ static int emit_bpf_dispatcher(u8 **pprog, int a, int b, s64 *progs)
 	jg_reloc = prog;
 
 	err = emit_bpf_dispatcher(&prog, a, a + pivot,	/* emit lower_part */
-				  progs);
+				  progs, image, buf);
 	if (err)
 		return err;
 
@@ -2303,7 +2303,7 @@ static int emit_bpf_dispatcher(u8 **pprog, int a, int b, s64 *progs)
 	emit_code(jg_reloc - jg_bytes, jg_offset, jg_bytes);
 
 	err = emit_bpf_dispatcher(&prog, a + pivot + 1,	/* emit upper_part */
-				  b, progs);
+				  b, progs, image, buf);
 	if (err)
 		return err;
 
@@ -2323,12 +2323,12 @@ static int cmp_ips(const void *a, const void *b)
 	return 0;
 }
 
-int arch_prepare_bpf_dispatcher(void *image, s64 *funcs, int num_funcs)
+int arch_prepare_bpf_dispatcher(void *image, void *buf, s64 *funcs, int num_funcs)
 {
-	u8 *prog = image;
+	u8 *prog = buf;
 
 	sort(funcs, num_funcs, sizeof(funcs[0]), cmp_ips, NULL);
-	return emit_bpf_dispatcher(&prog, 0, num_funcs - 1, funcs);
+	return emit_bpf_dispatcher(&prog, 0, num_funcs - 1, funcs, image, buf);
 }
 
 struct x64_jit_data {
include/linux/bpf.h +2 -1
index edd43edb27d6..9ae155c75014 100644
@@ -946,6 +946,7 @@ struct bpf_dispatcher {
 	struct bpf_dispatcher_prog progs[BPF_DISPATCHER_MAX];
 	int num_progs;
 	void *image;
+	void *rw_image;
 	u32 image_off;
 	struct bpf_ksym ksym;
 };
@@ -964,7 +965,7 @@ int bpf_trampoline_unlink_prog(struct bpf_tramp_link *link, struct bpf_trampolin
 struct bpf_trampoline *bpf_trampoline_get(u64 key,
 					  struct bpf_attach_target_info *tgt_info);
 void bpf_trampoline_put(struct bpf_trampoline *tr);
-int arch_prepare_bpf_dispatcher(void *image, s64 *funcs, int num_funcs);
+int arch_prepare_bpf_dispatcher(void *image, void *buf, s64 *funcs, int num_funcs);
 #define BPF_DISPATCHER_INIT(_name) {				\
 	.mutex = __MUTEX_INITIALIZER(_name.mutex),		\
 	.func = &_name##_func,					\
include/linux/filter.h +5 -0
index 98e28126c24b..efc42a6e3aed 100644
@@ -1023,6 +1023,8 @@ extern long bpf_jit_limit_max;
 
 typedef void (*bpf_jit_fill_hole_t)(void *area, unsigned int size);
 
+void bpf_jit_fill_hole_with_zero(void *area, unsigned int size);
+
 struct bpf_binary_header *
 bpf_jit_binary_alloc(unsigned int proglen, u8 **image_ptr,
 		     unsigned int alignment,
@@ -1035,6 +1037,9 @@ void bpf_jit_free(struct bpf_prog *fp);
 struct bpf_binary_header *
 bpf_jit_binary_pack_hdr(const struct bpf_prog *fp);
 
+void *bpf_prog_pack_alloc(u32 size, bpf_jit_fill_hole_t bpf_fill_ill_insns);
+void bpf_prog_pack_free(struct bpf_binary_header *hdr);
+
 static inline bool bpf_prog_kallsyms_verify_off(const struct bpf_prog *fp)
 {
 	return list_empty(&fp->aux->ksym.lnode) ||
kernel/bpf/core.c +7 -2
index d1be78c28619..711fd293b6de 100644
@@ -825,6 +825,11 @@ struct bpf_prog_pack {
 	unsigned long bitmap[];
 };
 
+void bpf_jit_fill_hole_with_zero(void *area, unsigned int size)
+{
+	memset(area, 0, size);
+}
+
 #define BPF_PROG_SIZE_TO_NBITS(size)	(round_up(size, BPF_PROG_CHUNK_SIZE) / BPF_PROG_CHUNK_SIZE)
 
 static DEFINE_MUTEX(pack_mutex);
@@ -864,7 +869,7 @@ static struct bpf_prog_pack *alloc_new_pack(bpf_jit_fill_hole_t bpf_fill_ill_ins
 	return pack;
 }
 
-static void *bpf_prog_pack_alloc(u32 size, bpf_jit_fill_hole_t bpf_fill_ill_insns)
+void *bpf_prog_pack_alloc(u32 size, bpf_jit_fill_hole_t bpf_fill_ill_insns)
 {
 	unsigned int nbits = BPF_PROG_SIZE_TO_NBITS(size);
 	struct bpf_prog_pack *pack;
@@ -905,7 +910,7 @@ static void *bpf_prog_pack_alloc(u32 size, bpf_jit_fill_hole_t bpf_fill_ill_insn
 	return ptr;
 }
 
-static void bpf_prog_pack_free(struct bpf_binary_header *hdr)
+void bpf_prog_pack_free(struct bpf_binary_header *hdr)
 {
 	struct bpf_prog_pack *pack = NULL, *tmp;
 	unsigned int nbits;
kernel/bpf/dispatcher.c +21 -6
index 2444bd15cc2d..fa64b80b8bca 100644
@@ -85,12 +85,12 @@ static bool bpf_dispatcher_remove_prog(struct bpf_dispatcher *d,
 	return false;
 }
 
-int __weak arch_prepare_bpf_dispatcher(void *image, s64 *funcs, int num_funcs)
+int __weak arch_prepare_bpf_dispatcher(void *image, void *buf, s64 *funcs, int num_funcs)
 {
 	return -ENOTSUPP;
 }
 
-static int bpf_dispatcher_prepare(struct bpf_dispatcher *d, void *image)
+static int bpf_dispatcher_prepare(struct bpf_dispatcher *d, void *image, void *buf)
 {
 	s64 ips[BPF_DISPATCHER_MAX] = {}, *ipsp = &ips[0];
 	int i;
@@ -99,12 +99,12 @@ static int bpf_dispatcher_prepare(struct bpf_dispatcher *d, void *image)
 		if (d->progs[i].prog)
 			*ipsp++ = (s64)(uintptr_t)d->progs[i].prog->bpf_func;
 	}
-	return arch_prepare_bpf_dispatcher(image, &ips[0], d->num_progs);
+	return arch_prepare_bpf_dispatcher(image, buf, &ips[0], d->num_progs);
 }
 
 static void bpf_dispatcher_update(struct bpf_dispatcher *d, int prev_num_progs)
 {
-	void *old, *new;
+	void *old, *new, *tmp;
 	u32 noff;
 	int err;
 
@@ -117,8 +117,14 @@ static void bpf_dispatcher_update(struct bpf_dispatcher *d, int prev_num_progs)
 	}
 
 	new = d->num_progs ? d->image + noff : NULL;
+	tmp = d->num_progs ? d->rw_image + noff : NULL;
 	if (new) {
-		if (bpf_dispatcher_prepare(d, new))
+		/* Prepare the dispatcher in d->rw_image. Then use
+		 * bpf_arch_text_copy to update d->image, which is RO+X.
+		 */
+		if (bpf_dispatcher_prepare(d, new, tmp))
+			return;
+		if (IS_ERR(bpf_arch_text_copy(new, tmp, PAGE_SIZE / 2)))
 			return;
 	}
 
@@ -140,9 +146,18 @@ void bpf_dispatcher_change_prog(struct bpf_dispatcher *d, struct bpf_prog *from,
 
 	mutex_lock(&d->mutex);
 	if (!d->image) {
-		d->image = bpf_jit_alloc_exec_page();
+		d->image = bpf_prog_pack_alloc(PAGE_SIZE, bpf_jit_fill_hole_with_zero);
 		if (!d->image)
 			goto out;
+		d->rw_image = bpf_jit_alloc_exec(PAGE_SIZE);
+		if (!d->rw_image) {
+			u32 size = PAGE_SIZE;
+
+			bpf_arch_text_copy(d->image, &size, sizeof(size));
+			bpf_prog_pack_free((struct bpf_binary_header *)d->image);
+			d->image = NULL;
+			goto out;
+		}
 		bpf_image_ksym_add(d->image, &d->ksym);
 	}
 
Signature · 1 line
2.30.2
author reviewer
#2 lore raw

PATCH v2 2/2 bpf: Enforce W^X for bpf trampoline

In reply to Song Liu

Mark the trampoline as RO+X after arch_prepare_bpf_trampoline, so that the trampoine follows W^X rule strictly. This will turn off warnings like

CPA refuse W^X violation: 8000000000000163 -> 0000000000000163 range: ...

Also remove bpf_jit_alloc_exec_page(), since it is not used any more.

---
 include/linux/bpf.h     |  1 -
 kernel/bpf/trampoline.c | 22 +++++-----------------
 2 files changed, 5 insertions(+), 18 deletions(-)
include/linux/bpf.h +0 -1
index 9ae155c75014..5161fac0513f 100644
@@ -1008,7 +1008,6 @@ int arch_prepare_bpf_dispatcher(void *image, void *buf, s64 *funcs, int num_func
 void bpf_dispatcher_change_prog(struct bpf_dispatcher *d, struct bpf_prog *from,
 				struct bpf_prog *to);
 /* Called only from JIT-enabled code, so there's no need for stubs. */
-void *bpf_jit_alloc_exec_page(void);
 void bpf_image_ksym_add(void *data, struct bpf_ksym *ksym);
 void bpf_image_ksym_del(struct bpf_ksym *ksym);
 void bpf_ksym_add(struct bpf_ksym *ksym);
kernel/bpf/trampoline.c +5 -17
index 41b67eb83ab3..6f7b939321d6 100644
@@ -116,22 +116,6 @@ bool bpf_prog_has_trampoline(const struct bpf_prog *prog)
 		(ptype == BPF_PROG_TYPE_LSM && eatype == BPF_LSM_MAC);
 }
 
-void *bpf_jit_alloc_exec_page(void)
-{
-	void *image;
-
-	image = bpf_jit_alloc_exec(PAGE_SIZE);
-	if (!image)
-		return NULL;
-
-	set_vm_flush_reset_perms(image);
-	/* Keep image as writeable. The alternative is to keep flipping ro/rw
-	 * every time new program is attached or detached.
-	 */
-	set_memory_x((long)image, 1);
-	return image;
-}
-
 void bpf_image_ksym_add(void *data, struct bpf_ksym *ksym)
 {
 	ksym->start = (unsigned long) data;
@@ -404,9 +388,10 @@ static struct bpf_tramp_image *bpf_tramp_image_alloc(u64 key, u32 idx)
 		goto out_free_im;
 
 	err = -ENOMEM;
-	im->image = image = bpf_jit_alloc_exec_page();
+	im->image = image = bpf_jit_alloc_exec(PAGE_SIZE);
 	if (!image)
 		goto out_uncharge;
+	set_vm_flush_reset_perms(image);
 
 	err = percpu_ref_init(&im->pcref, __bpf_tramp_image_release, 0, GFP_KERNEL);
 	if (err)
@@ -483,6 +468,9 @@ static int bpf_trampoline_update(struct bpf_trampoline *tr, bool lock_direct_mut
 	if (err < 0)
 		goto out;
 
+	set_memory_ro((long)im->image, 1);
+	set_memory_x((long)im->image, 1);
+
 	WARN_ON(tr->cur_image && tr->selector == 0);
 	WARN_ON(!tr->cur_image && tr->selector);
 	if (tr->cur_image)
Signature · 1 line
2.30.2
#3 lore raw
In reply to Song Liu
This series was applied to bpf/bpf-next.git (master) by Alexei Starovoitov <[email protected]>:

Hello:

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

Quoted 10 lines · On Mon, 26 Sep 2022 11:47:37 -0700 you wrote:

Changes v1 => v2:
1. Update arch_prepare_bpf_dispatcher to use a RO image and a RW buffer. (Alexei) Note: I haven't found an existing test to cover this part, so this part was tested manually (comparing the generated dispatcher is the same).

Jeff Layton reported CPA W^X warning linux-next [1]. It turns out to be W^X issue with bpf trampoline and bpf dispatcher. Fix these by:

[...]

Here is the summary with links:
- [v2,bpf-next,1/2] bpf: use bpf_prog_pack for bpf_dispatcher

    https://git.kernel.org/bpf/bpf-next/c/19c02415da23
  - [v2,bpf-next,2/2] bpf: Enforce W^X for bpf trampoline
    https://git.kernel.org/bpf/bpf-next/c/5b0d1c7bd572

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
Tested-by on the v2 series
#4 lore raw
In reply to Song Liu
Quoted 26 lines · Song Liu

Changes v1 => v2:
1. Update arch_prepare_bpf_dispatcher to use a RO image and a RW buffer. (Alexei) Note: I haven't found an existing test to cover this part, so this part was tested manually (comparing the generated dispatcher is the same).

Jeff Layton reported CPA W^X warning linux-next [1]. It turns out to be W^X issue with bpf trampoline and bpf dispatcher. Fix these by:

1. Use bpf_prog_pack for bpf_dispatcher;
2. Set memory permission properly with bpf trampoline.

[1] https://lore.kernel.org/lkml/[email protected]/

Song Liu (2):
bpf: use bpf_prog_pack for bpf_dispatcher
bpf: Enforce W^X for bpf trampoline

 arch/x86/net/bpf_jit_comp.c | 16 ++++++++--------
 include/linux/bpf.h         |  4 ++--
 include/linux/filter.h      |  5 +++++
 kernel/bpf/core.c           |  9 +++++++--
 kernel/bpf/dispatcher.c     | 27 +++++++++++++++++++++------
 kernel/bpf/trampoline.c     | 22 +++++-----------------
 6 files changed, 48 insertions(+), 35 deletions(-)
Quoted 2 lines

--
2.30.2

Your patch seems to have fixed the issue. You can add:

Thanks!

#5 lore raw
In reply to Song Liu
Quoted 11 lines · Song Liu

Changes v1 => v2:
1. Update arch_prepare_bpf_dispatcher to use a RO image and a RW buffer. (Alexei) Note: I haven't found an existing test to cover this part, so this part was tested manually (comparing the generated dispatcher is the same).

Jeff Layton reported CPA W^X warning linux-next [1]. It turns out to be W^X issue with bpf trampoline and bpf dispatcher. Fix these by:

1. Use bpf_prog_pack for bpf_dispatcher;
2. Set memory permission properly with bpf trampoline.

Indirectly related to your patchset[0].
- TL;DR calling set_memory_x() have side-effects

We are getting reports that loading BPF-progs (jit stage) cause issues for RT in the form of triggering work on isolated CPUs. It looks like BTF JIT stage cause a TLB flush on all CPUs, including isolated CPUs.

The triggering function is set_memory_x() (see call-stack[2]).

We have noticed (and appreciate) you have previously improved the situation in this patchset[3]:
[3]
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=80123f0ac4a6

Is this patchset also part of improving the situation, or does it introduce more calls to set_memory_x() ?

Quoted 1 line

[1] https://lore.kernel.org/lkml/[email protected]/

[2] Call stack triggering issue:

        smp_call_function_many_cond+0x1
        smp_call_function+0x39
        on_each_cpu+0x2a
        cpa_flush+0x11a
        change_page_attr_set_clr+0x129
        set_memory_x+0x37
        bpf_int_jit_compile+0x36f
        bpf_prog_select_runtime+0xc6
        bpf_prepare_filter+0x523
        sk_attach_filter+0x13
        sock_setsockopt+0x920
        __sys_setsockopt+0x16a
        __x64_sys_setsockopt+0x20
        do_syscall_64+0x87
        entry_SYSCALL_64_after_hwframe+0x65

[0] https://lore.kernel.org/all/[email protected]/#r

--Jesper

#6 lore raw
Quoted 19 lines · Jesper Dangaard Brouer
Quoted 9 lines · Song Liu

Changes v1 => v2:
1. Update arch_prepare_bpf_dispatcher to use a RO image and a RW buffer. (Alexei) Note: I haven't found an existing test to cover this part, so this part was tested manually (comparing the generated dispatcher is the same).
Jeff Layton reported CPA W^X warning linux-next [1]. It turns out to be W^X issue with bpf trampoline and bpf dispatcher. Fix these by:
1. Use bpf_prog_pack for bpf_dispatcher;
2. Set memory permission properly with bpf trampoline.

Indirectly related to your patchset[0].
- TL;DR calling set_memory_x() have side-effects

Quoted 8 lines

We are getting reports that loading BPF-progs (jit stage) cause issues for RT in the form of triggering work on isolated CPUs. It looks like BTF JIT stage cause a TLB flush on all CPUs, including isolated CPUs.

The triggering function is set_memory_x() (see call-stack[2]).

We have noticed (and appreciate) you have previously improved the situation in this patchset[3]:
[3] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=80123f0ac4a6

Is this patchset also part of improving the situation, or does it introduce more calls to set_memory_x() ?

This set doesn't change numbers of set_memory_x() calls for trampolines. We plan to move trampolines to use bpf_prog_pack (or the new vmalloc_exec, if I am lucky) in 6.2. We will see fewer set_memory_x() calls after that.

Thanks,
Song

Quoted 28 lines
Quoted 3 lines

[1] https://lore.kernel.org/lkml/[email protected]/

[2] Call stack triggering issue:

       smp_call_function_many_cond+0x1
       smp_call_function+0x39
       on_each_cpu+0x2a
       cpa_flush+0x11a
       change_page_attr_set_clr+0x129
       set_memory_x+0x37
       bpf_int_jit_compile+0x36f
       bpf_prog_select_runtime+0xc6
       bpf_prepare_filter+0x523
       sk_attach_filter+0x13
       sock_setsockopt+0x920
       __sys_setsockopt+0x16a
       __x64_sys_setsockopt+0x20
       do_syscall_64+0x87
       entry_SYSCALL_64_after_hwframe+0x65

[0] https://lore.kernel.org/all/[email protected]/#r

--Jesper