2022-08-08 04:21:20

by Xu Kuohai

[permalink] [raw]
Subject: [PATCH bpf] bpf, arm64: Fix bpf trampoline instruction endianness

The sparse tool complains as follows:

arch/arm64/net/bpf_jit_comp.c:1684:16:
warning: incorrect type in assignment (different base types)
arch/arm64/net/bpf_jit_comp.c:1684:16:
expected unsigned int [usertype] *branch
arch/arm64/net/bpf_jit_comp.c:1684:16:
got restricted __le32 [usertype] *
arch/arm64/net/bpf_jit_comp.c:1700:52:
error: subtraction of different types can't work (different base
types)
arch/arm64/net/bpf_jit_comp.c:1734:29:
warning: incorrect type in assignment (different base types)
arch/arm64/net/bpf_jit_comp.c:1734:29:
expected unsigned int [usertype] *
arch/arm64/net/bpf_jit_comp.c:1734:29:
got restricted __le32 [usertype] *
arch/arm64/net/bpf_jit_comp.c:1918:52:
error: subtraction of different types can't work (different base
types)

This is because the variable branch in function invoke_bpf_prog and the
variable branches in function prepare_trampoline are defined as type
u32 *, which conflicts with ctx->image's type __le32 *, so sparse complains
when assignment or arithmetic operation are performed on these two
variables and ctx->image.

Since arm64 instructions are always little-endian, change the type of
these two variables to __le32 * and call cpu_to_le32 to convert
instruction to little-endian before writing it to memory.

Reported-by: kernel test robot <[email protected]>
Fixes: efc9909fdce0 ("bpf, arm64: Add bpf trampoline for arm64")
Signed-off-by: Xu Kuohai <[email protected]>
---
arch/arm64/net/bpf_jit_comp.c | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/arch/arm64/net/bpf_jit_comp.c b/arch/arm64/net/bpf_jit_comp.c
index 7ca8779ae34f..29dc55da2476 100644
--- a/arch/arm64/net/bpf_jit_comp.c
+++ b/arch/arm64/net/bpf_jit_comp.c
@@ -1643,7 +1643,7 @@ static void invoke_bpf_prog(struct jit_ctx *ctx, struct bpf_tramp_link *l,
int args_off, int retval_off, int run_ctx_off,
bool save_ret)
{
- u32 *branch;
+ __le32 *branch;
u64 enter_prog;
u64 exit_prog;
struct bpf_prog *p = l->link.prog;
@@ -1698,7 +1698,7 @@ static void invoke_bpf_prog(struct jit_ctx *ctx, struct bpf_tramp_link *l,

if (ctx->image) {
int offset = &ctx->image[ctx->idx] - branch;
- *branch = A64_CBZ(1, A64_R(0), offset);
+ *branch = cpu_to_le32(A64_CBZ(1, A64_R(0), offset));
}

/* arg1: prog */
@@ -1713,7 +1713,7 @@ static void invoke_bpf_prog(struct jit_ctx *ctx, struct bpf_tramp_link *l,

static void invoke_bpf_mod_ret(struct jit_ctx *ctx, struct bpf_tramp_links *tl,
int args_off, int retval_off, int run_ctx_off,
- u32 **branches)
+ __le32 **branches)
{
int i;

@@ -1784,7 +1784,7 @@ static int prepare_trampoline(struct jit_ctx *ctx, struct bpf_tramp_image *im,
struct bpf_tramp_links *fexit = &tlinks[BPF_TRAMP_FEXIT];
struct bpf_tramp_links *fmod_ret = &tlinks[BPF_TRAMP_MODIFY_RETURN];
bool save_ret;
- u32 **branches = NULL;
+ __le32 **branches = NULL;

/* trampoline stack layout:
* [ parent ip ]
@@ -1892,7 +1892,7 @@ static int prepare_trampoline(struct jit_ctx *ctx, struct bpf_tramp_image *im,
flags & BPF_TRAMP_F_RET_FENTRY_RET);

if (fmod_ret->nr_links) {
- branches = kcalloc(fmod_ret->nr_links, sizeof(u32 *),
+ branches = kcalloc(fmod_ret->nr_links, sizeof(__le32 *),
GFP_KERNEL);
if (!branches)
return -ENOMEM;
@@ -1916,7 +1916,7 @@ static int prepare_trampoline(struct jit_ctx *ctx, struct bpf_tramp_image *im,
/* update the branches saved in invoke_bpf_mod_ret with cbnz */
for (i = 0; i < fmod_ret->nr_links && ctx->image != NULL; i++) {
int offset = &ctx->image[ctx->idx] - branches[i];
- *branches[i] = A64_CBNZ(1, A64_R(10), offset);
+ *branches[i] = cpu_to_le32(A64_CBNZ(1, A64_R(10), offset));
}

for (i = 0; i < fexit->nr_links; i++)
--
2.30.2


2022-08-09 10:56:03

by Jean-Philippe Brucker

[permalink] [raw]
Subject: Re: [PATCH bpf] bpf, arm64: Fix bpf trampoline instruction endianness

[+ arm64 maintainers]

On Mon, Aug 08, 2022 at 12:07:35AM -0400, Xu Kuohai wrote:
> The sparse tool complains as follows:
>
> arch/arm64/net/bpf_jit_comp.c:1684:16:
> warning: incorrect type in assignment (different base types)
> arch/arm64/net/bpf_jit_comp.c:1684:16:
> expected unsigned int [usertype] *branch
> arch/arm64/net/bpf_jit_comp.c:1684:16:
> got restricted __le32 [usertype] *
> arch/arm64/net/bpf_jit_comp.c:1700:52:
> error: subtraction of different types can't work (different base
> types)
> arch/arm64/net/bpf_jit_comp.c:1734:29:
> warning: incorrect type in assignment (different base types)
> arch/arm64/net/bpf_jit_comp.c:1734:29:
> expected unsigned int [usertype] *
> arch/arm64/net/bpf_jit_comp.c:1734:29:
> got restricted __le32 [usertype] *
> arch/arm64/net/bpf_jit_comp.c:1918:52:
> error: subtraction of different types can't work (different base
> types)
>
> This is because the variable branch in function invoke_bpf_prog and the
> variable branches in function prepare_trampoline are defined as type
> u32 *, which conflicts with ctx->image's type __le32 *, so sparse complains
> when assignment or arithmetic operation are performed on these two
> variables and ctx->image.
>
> Since arm64 instructions are always little-endian, change the type of
> these two variables to __le32 * and call cpu_to_le32 to convert
> instruction to little-endian before writing it to memory.
>
> Reported-by: kernel test robot <[email protected]>
> Fixes: efc9909fdce0 ("bpf, arm64: Add bpf trampoline for arm64")
> Signed-off-by: Xu Kuohai <[email protected]>

Reviewed-by: Jean-Philippe Brucker <[email protected]>

> ---
> arch/arm64/net/bpf_jit_comp.c | 12 ++++++------
> 1 file changed, 6 insertions(+), 6 deletions(-)
>
> diff --git a/arch/arm64/net/bpf_jit_comp.c b/arch/arm64/net/bpf_jit_comp.c
> index 7ca8779ae34f..29dc55da2476 100644
> --- a/arch/arm64/net/bpf_jit_comp.c
> +++ b/arch/arm64/net/bpf_jit_comp.c
> @@ -1643,7 +1643,7 @@ static void invoke_bpf_prog(struct jit_ctx *ctx, struct bpf_tramp_link *l,
> int args_off, int retval_off, int run_ctx_off,
> bool save_ret)
> {
> - u32 *branch;
> + __le32 *branch;
> u64 enter_prog;
> u64 exit_prog;
> struct bpf_prog *p = l->link.prog;
> @@ -1698,7 +1698,7 @@ static void invoke_bpf_prog(struct jit_ctx *ctx, struct bpf_tramp_link *l,
>
> if (ctx->image) {
> int offset = &ctx->image[ctx->idx] - branch;
> - *branch = A64_CBZ(1, A64_R(0), offset);
> + *branch = cpu_to_le32(A64_CBZ(1, A64_R(0), offset));
> }
>
> /* arg1: prog */
> @@ -1713,7 +1713,7 @@ static void invoke_bpf_prog(struct jit_ctx *ctx, struct bpf_tramp_link *l,
>
> static void invoke_bpf_mod_ret(struct jit_ctx *ctx, struct bpf_tramp_links *tl,
> int args_off, int retval_off, int run_ctx_off,
> - u32 **branches)
> + __le32 **branches)
> {
> int i;
>
> @@ -1784,7 +1784,7 @@ static int prepare_trampoline(struct jit_ctx *ctx, struct bpf_tramp_image *im,
> struct bpf_tramp_links *fexit = &tlinks[BPF_TRAMP_FEXIT];
> struct bpf_tramp_links *fmod_ret = &tlinks[BPF_TRAMP_MODIFY_RETURN];
> bool save_ret;
> - u32 **branches = NULL;
> + __le32 **branches = NULL;
>
> /* trampoline stack layout:
> * [ parent ip ]
> @@ -1892,7 +1892,7 @@ static int prepare_trampoline(struct jit_ctx *ctx, struct bpf_tramp_image *im,
> flags & BPF_TRAMP_F_RET_FENTRY_RET);
>
> if (fmod_ret->nr_links) {
> - branches = kcalloc(fmod_ret->nr_links, sizeof(u32 *),
> + branches = kcalloc(fmod_ret->nr_links, sizeof(__le32 *),
> GFP_KERNEL);
> if (!branches)
> return -ENOMEM;
> @@ -1916,7 +1916,7 @@ static int prepare_trampoline(struct jit_ctx *ctx, struct bpf_tramp_image *im,
> /* update the branches saved in invoke_bpf_mod_ret with cbnz */
> for (i = 0; i < fmod_ret->nr_links && ctx->image != NULL; i++) {
> int offset = &ctx->image[ctx->idx] - branches[i];
> - *branches[i] = A64_CBNZ(1, A64_R(10), offset);
> + *branches[i] = cpu_to_le32(A64_CBNZ(1, A64_R(10), offset));
> }
>
> for (i = 0; i < fexit->nr_links; i++)
> --
> 2.30.2
>

2022-08-10 15:38:50

by patchwork-bot+netdevbpf

[permalink] [raw]
Subject: Re: [PATCH bpf] bpf, arm64: Fix bpf trampoline instruction endianness

Hello:

This patch was applied to bpf/bpf.git (master)
by Daniel Borkmann <[email protected]>:

On Mon, 8 Aug 2022 00:07:35 -0400 you wrote:
> The sparse tool complains as follows:
>
> arch/arm64/net/bpf_jit_comp.c:1684:16:
> warning: incorrect type in assignment (different base types)
> arch/arm64/net/bpf_jit_comp.c:1684:16:
> expected unsigned int [usertype] *branch
> arch/arm64/net/bpf_jit_comp.c:1684:16:
> got restricted __le32 [usertype] *
> arch/arm64/net/bpf_jit_comp.c:1700:52:
> error: subtraction of different types can't work (different base
> types)
> arch/arm64/net/bpf_jit_comp.c:1734:29:
> warning: incorrect type in assignment (different base types)
> arch/arm64/net/bpf_jit_comp.c:1734:29:
> expected unsigned int [usertype] *
> arch/arm64/net/bpf_jit_comp.c:1734:29:
> got restricted __le32 [usertype] *
> arch/arm64/net/bpf_jit_comp.c:1918:52:
> error: subtraction of different types can't work (different base
> types)
>
> [...]

Here is the summary with links:
- [bpf] bpf, arm64: Fix bpf trampoline instruction endianness
https://git.kernel.org/bpf/bpf/c/aada47665546

You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html


2022-08-10 15:42:19

by Daniel Borkmann

[permalink] [raw]
Subject: Re: [PATCH bpf] bpf, arm64: Fix bpf trampoline instruction endianness

On 8/9/22 12:33 PM, Jean-Philippe Brucker wrote:
> [+ arm64 maintainers]
>
> On Mon, Aug 08, 2022 at 12:07:35AM -0400, Xu Kuohai wrote:
>> The sparse tool complains as follows:
>>
>> arch/arm64/net/bpf_jit_comp.c:1684:16:
>> warning: incorrect type in assignment (different base types)
>> arch/arm64/net/bpf_jit_comp.c:1684:16:
>> expected unsigned int [usertype] *branch
>> arch/arm64/net/bpf_jit_comp.c:1684:16:
>> got restricted __le32 [usertype] *
>> arch/arm64/net/bpf_jit_comp.c:1700:52:
>> error: subtraction of different types can't work (different base
>> types)
>> arch/arm64/net/bpf_jit_comp.c:1734:29:
>> warning: incorrect type in assignment (different base types)
>> arch/arm64/net/bpf_jit_comp.c:1734:29:
>> expected unsigned int [usertype] *
>> arch/arm64/net/bpf_jit_comp.c:1734:29:
>> got restricted __le32 [usertype] *
>> arch/arm64/net/bpf_jit_comp.c:1918:52:
>> error: subtraction of different types can't work (different base
>> types)
>>
>> This is because the variable branch in function invoke_bpf_prog and the
>> variable branches in function prepare_trampoline are defined as type
>> u32 *, which conflicts with ctx->image's type __le32 *, so sparse complains
>> when assignment or arithmetic operation are performed on these two
>> variables and ctx->image.
>>
>> Since arm64 instructions are always little-endian, change the type of
>> these two variables to __le32 * and call cpu_to_le32 to convert
>> instruction to little-endian before writing it to memory.
>>
>> Reported-by: kernel test robot <[email protected]>
>> Fixes: efc9909fdce0 ("bpf, arm64: Add bpf trampoline for arm64")
>> Signed-off-by: Xu Kuohai <[email protected]>
>
> Reviewed-by: Jean-Philippe Brucker <[email protected]>

Applied, thanks! Also added small note that this is in line with emit() as well.