diff mbox

[net-next,v3,1/3] net: filter: simplify label names from jump-table

Message ID 1398960996-10469-2-git-send-email-dborkman@redhat.com
State Superseded, archived
Delegated to: David Miller
Headers show

Commit Message

Daniel Borkmann May 1, 2014, 4:16 p.m. UTC 
This patch simplifies label naming for the BPF jump-table.
When we define labels via DL(), we just concatenate/textify
the combination of instruction opcode which consists of the
class, subclass, word size, target register and so on. Each
time we leave BPF_ prefix intact, so that e.g. the preprocessor
generates a label BPF_ALU_BPF_ADD_BPF_X for DL(BPF_ALU, BPF_ADD,
BPF_X) whereas a label name of ALU_ADD_X is much more easy
to grasp. Pure cleanup only.

Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Alexei Starovoitov <ast@plumgrid.com>
---
 include/linux/filter.h |   3 +
 net/core/filter.c      | 308 ++++++++++++++++++++++++-------------------------
 2 files changed, 157 insertions(+), 154 deletions(-)

Comments

Eric Dumazet May 1, 2014, 5:20 p.m. UTC | #1 
On Thu, 2014-05-01 at 18:16 +0200, Daniel Borkmann wrote:
> This patch simplifies label naming for the BPF jump-table.
> When we define labels via DL(), we just concatenate/textify
> the combination of instruction opcode which consists of the
> class, subclass, word size, target register and so on. Each
> time we leave BPF_ prefix intact, so that e.g. the preprocessor
> generates a label BPF_ALU_BPF_ADD_BPF_X for DL(BPF_ALU, BPF_ADD,
> BPF_X) whereas a label name of ALU_ADD_X is much more easy
> to grasp. Pure cleanup only.
> 
> Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
> Acked-by: Alexei Starovoitov <ast@plumgrid.com>
> ---

This exactly makes harder to use grep and future code reviews.

# git grep -n BPF_XADD
include/linux/filter.h:21:#define BPF_XADD      0xc0    /* exclusive add */
net/core/filter.c:224:          DL(BPF_STX, BPF_XADD, BPF_W),
net/core/filter.c:225:          DL(BPF_STX, BPF_XADD, BPF_DW),
net/core/filter.c:481:  BPF_STX_BPF_XADD_BPF_W: /* lock xadd *(u32 *)(A + insn->off) += X */
net/core/filter.c:485:  BPF_STX_BPF_XADD_BPF_DW: /* lock xadd *(u64 *)(A + insn->off) += X */

Compare now with :

# git grep -n XADD

So I am quite against this patch.



--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Daniel Borkmann May 1, 2014, 5:39 p.m. UTC | #2 
On 05/01/2014 07:20 PM, Eric Dumazet wrote:
> On Thu, 2014-05-01 at 18:16 +0200, Daniel Borkmann wrote:
>> This patch simplifies label naming for the BPF jump-table.
>> When we define labels via DL(), we just concatenate/textify
>> the combination of instruction opcode which consists of the
>> class, subclass, word size, target register and so on. Each
>> time we leave BPF_ prefix intact, so that e.g. the preprocessor
>> generates a label BPF_ALU_BPF_ADD_BPF_X for DL(BPF_ALU, BPF_ADD,
>> BPF_X) whereas a label name of ALU_ADD_X is much more easy
>> to grasp. Pure cleanup only.
>>
>> Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
>> Acked-by: Alexei Starovoitov <ast@plumgrid.com>
>> ---
>
> This exactly makes harder to use grep and future code reviews.
>
> # git grep -n BPF_XADD
> include/linux/filter.h:21:#define BPF_XADD      0xc0    /* exclusive add */
> net/core/filter.c:224:          DL(BPF_STX, BPF_XADD, BPF_W),
> net/core/filter.c:225:          DL(BPF_STX, BPF_XADD, BPF_DW),
> net/core/filter.c:481:  BPF_STX_BPF_XADD_BPF_W: /* lock xadd *(u32 *)(A + insn->off) += X */
> net/core/filter.c:485:  BPF_STX_BPF_XADD_BPF_DW: /* lock xadd *(u64 *)(A + insn->off) += X */
>
> Compare now with :
>
> # git grep -n XADD
>
> So I am quite against this patch.

Well, if you review the code itself in filter.c it makes it hard
to read and review, with this patch, you'll immediately get the
label name and what it actually does, so I think it's quite convenient
and way more readable by itself. For grepping, you can always add
subdirectories you're searching for, besides that I don't think
that everything in the kernel needs to have unique names only so
that one can grep for it among the whole tree.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Eric Dumazet May 1, 2014, 6:50 p.m. UTC | #3 
On Thu, 2014-05-01 at 19:39 +0200, Daniel Borkmann wrote:

> Well, if you review the code itself in filter.c it makes it hard
> to read and review, with this patch, you'll immediately get the
> label name and what it actually does, so I think it's quite convenient
> and way more readable by itself. For grepping, you can always add
> subdirectories you're searching for, besides that I don't think
> that everything in the kernel needs to have unique names only so
> that one can grep for it among the whole tree.



Considering that Alexei claimed :

+Internal BPF is a general purpose RISC instruction set. Not every register and
+every instruction are used during translation from original BPF to new format.
+For example, socket filters are not using 'exclusive add' instruction, but
+tracing filters may do to maintain counters of events, for example. Register R9
+is not used by socket filters either, but more complex filters may be running
+out of registers and would have to resort to spill/fill to stack.

I tried to verify this claim and I could not.

After your patch, it will be really this :

"We, Daniel and Alexei, hereby certify that this code is 100% safe.

 If you Eric, or anyone else, want to check, you'll have to grep in
appropriate places, and use your brain as we did."

I find very worrying all this stuff. This looks like you guys make
everything possible to make it absolutely unmaintainable.



--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Alexei Starovoitov May 1, 2014, 7:47 p.m. UTC | #4 
On Thu, May 1, 2014 at 11:50 AM, Eric Dumazet <eric.dumazet@gmail.com> wrote:
> On Thu, 2014-05-01 at 19:39 +0200, Daniel Borkmann wrote:
>
>> Well, if you review the code itself in filter.c it makes it hard
>> to read and review, with this patch, you'll immediately get the
>> label name and what it actually does, so I think it's quite convenient
>> and way more readable by itself. For grepping, you can always add
>> subdirectories you're searching for, besides that I don't think
>> that everything in the kernel needs to have unique names only so
>> that one can grep for it among the whole tree.
>
>
>
> Considering that Alexei claimed :
>
> +Internal BPF is a general purpose RISC instruction set. Not every register and
> +every instruction are used during translation from original BPF to new format.
> +For example, socket filters are not using 'exclusive add' instruction, but
> +tracing filters may do to maintain counters of events, for example. Register R9
> +is not used by socket filters either, but more complex filters may be running
> +out of registers and would have to resort to spill/fill to stack.
>
> I tried to verify this claim and I could not.

I'm not sure what part are you referring to.
That xadd is not used right now? It's not. I think we agreed on it
last week.
What else do you mean?

> After your patch, it will be really this :
>
> "We, Daniel and Alexei, hereby certify that this code is 100% safe.

I don't see where we say this.
Few paragraphs below the doc clearly says that it's a RISC instruction set
and safety on instruction set comes from verifier which is TBD.
Instruction set by itself is just an instruction set. Just like original BPF.

>  If you Eric, or anyone else, want to check, you'll have to grep in
> appropriate places, and use your brain as we did."
>
> I find very worrying all this stuff. This looks like you guys make
> everything possible to make it absolutely unmaintainable.

imo that's a harsh and unfair statement.
Calling it 'unmaintainable' just because you don't see how xadd
can be used?
I'm working on tracing filters that will demonstrate that in a bit.

This particular Daniel's patch I think is a nice cleanup.
It indeed makes interpreter code shorter, less verbose and easier to read.

I suspect you think of this internal BPF format as it's solving only
socket filters use case and all filters will be coming from user space.
That's very narrow view. In case of tracing filters there is no BPF
to send to kernel. User to kernel interface is a character string
like "a==1 || b == 2". That is the filter. I'm parsing the string
and converting it inside the kernel to internal bpf format,
so I would like to use full power of underlying cpu.
Think of internal BPF as 'uasm' that mips guys are using to construct tlb
handler on the fly.
But this 'uasm' is generic for all architectures.
Eventually we may want to expose internal bpf to user space, only then
verifier will be must have and we'll be discussing '100% safety'.
Right now everything is internal and I'm pretty sure you understood
very well how original bpf was translated to internal bpf, so I don't follow
your concern about 'unmaintainable'.

At the same time it's impossible for me to follow what you guys are
doing with tcp stack, but I'm not calling it unmaintainable just because
I don't understand it.
If you feel more documentation is needed, sure. Please let us know
what pieces are not explained well.
When I look at the doc by myself, it's hard to see what else is missing
or what pieces are difficult to grasp.

Regards,
Alexei
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/include/linux/filter.h b/include/linux/filter.h
index 759abf7..b042d1d 100644
--- a/include/linux/filter.h
+++ b/include/linux/filter.h
@@ -37,6 +37,9 @@ 
 #define BPF_CALL	0x80	/* function call */
 #define BPF_EXIT	0x90	/* function return */
 
+/* Placeholder/dummy for 0 */
+#define BPF_0		0
+
 /* BPF has 10 general purpose 64-bit registers and stack frame. */
 #define MAX_BPF_REG	11
 
diff --git a/net/core/filter.c b/net/core/filter.c
index 7c4db3d..a1784e9 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -156,94 +156,94 @@  unsigned int __sk_run_filter(void *ctx, const struct sock_filter_int *insn)
 	static const void *jumptable[256] = {
 		[0 ... 255] = &&default_label,
 		/* Now overwrite non-defaults ... */
-#define DL(A, B, C)	[A|B|C] = &&A##_##B##_##C
-		DL(BPF_ALU, BPF_ADD, BPF_X),
-		DL(BPF_ALU, BPF_ADD, BPF_K),
-		DL(BPF_ALU, BPF_SUB, BPF_X),
-		DL(BPF_ALU, BPF_SUB, BPF_K),
-		DL(BPF_ALU, BPF_AND, BPF_X),
-		DL(BPF_ALU, BPF_AND, BPF_K),
-		DL(BPF_ALU, BPF_OR, BPF_X),
-		DL(BPF_ALU, BPF_OR, BPF_K),
-		DL(BPF_ALU, BPF_LSH, BPF_X),
-		DL(BPF_ALU, BPF_LSH, BPF_K),
-		DL(BPF_ALU, BPF_RSH, BPF_X),
-		DL(BPF_ALU, BPF_RSH, BPF_K),
-		DL(BPF_ALU, BPF_XOR, BPF_X),
-		DL(BPF_ALU, BPF_XOR, BPF_K),
-		DL(BPF_ALU, BPF_MUL, BPF_X),
-		DL(BPF_ALU, BPF_MUL, BPF_K),
-		DL(BPF_ALU, BPF_MOV, BPF_X),
-		DL(BPF_ALU, BPF_MOV, BPF_K),
-		DL(BPF_ALU, BPF_DIV, BPF_X),
-		DL(BPF_ALU, BPF_DIV, BPF_K),
-		DL(BPF_ALU, BPF_MOD, BPF_X),
-		DL(BPF_ALU, BPF_MOD, BPF_K),
-		DL(BPF_ALU, BPF_NEG, 0),
-		DL(BPF_ALU, BPF_END, BPF_TO_BE),
-		DL(BPF_ALU, BPF_END, BPF_TO_LE),
-		DL(BPF_ALU64, BPF_ADD, BPF_X),
-		DL(BPF_ALU64, BPF_ADD, BPF_K),
-		DL(BPF_ALU64, BPF_SUB, BPF_X),
-		DL(BPF_ALU64, BPF_SUB, BPF_K),
-		DL(BPF_ALU64, BPF_AND, BPF_X),
-		DL(BPF_ALU64, BPF_AND, BPF_K),
-		DL(BPF_ALU64, BPF_OR, BPF_X),
-		DL(BPF_ALU64, BPF_OR, BPF_K),
-		DL(BPF_ALU64, BPF_LSH, BPF_X),
-		DL(BPF_ALU64, BPF_LSH, BPF_K),
-		DL(BPF_ALU64, BPF_RSH, BPF_X),
-		DL(BPF_ALU64, BPF_RSH, BPF_K),
-		DL(BPF_ALU64, BPF_XOR, BPF_X),
-		DL(BPF_ALU64, BPF_XOR, BPF_K),
-		DL(BPF_ALU64, BPF_MUL, BPF_X),
-		DL(BPF_ALU64, BPF_MUL, BPF_K),
-		DL(BPF_ALU64, BPF_MOV, BPF_X),
-		DL(BPF_ALU64, BPF_MOV, BPF_K),
-		DL(BPF_ALU64, BPF_ARSH, BPF_X),
-		DL(BPF_ALU64, BPF_ARSH, BPF_K),
-		DL(BPF_ALU64, BPF_DIV, BPF_X),
-		DL(BPF_ALU64, BPF_DIV, BPF_K),
-		DL(BPF_ALU64, BPF_MOD, BPF_X),
-		DL(BPF_ALU64, BPF_MOD, BPF_K),
-		DL(BPF_ALU64, BPF_NEG, 0),
-		DL(BPF_JMP, BPF_CALL, 0),
-		DL(BPF_JMP, BPF_JA, 0),
-		DL(BPF_JMP, BPF_JEQ, BPF_X),
-		DL(BPF_JMP, BPF_JEQ, BPF_K),
-		DL(BPF_JMP, BPF_JNE, BPF_X),
-		DL(BPF_JMP, BPF_JNE, BPF_K),
-		DL(BPF_JMP, BPF_JGT, BPF_X),
-		DL(BPF_JMP, BPF_JGT, BPF_K),
-		DL(BPF_JMP, BPF_JGE, BPF_X),
-		DL(BPF_JMP, BPF_JGE, BPF_K),
-		DL(BPF_JMP, BPF_JSGT, BPF_X),
-		DL(BPF_JMP, BPF_JSGT, BPF_K),
-		DL(BPF_JMP, BPF_JSGE, BPF_X),
-		DL(BPF_JMP, BPF_JSGE, BPF_K),
-		DL(BPF_JMP, BPF_JSET, BPF_X),
-		DL(BPF_JMP, BPF_JSET, BPF_K),
-		DL(BPF_JMP, BPF_EXIT, 0),
-		DL(BPF_STX, BPF_MEM, BPF_B),
-		DL(BPF_STX, BPF_MEM, BPF_H),
-		DL(BPF_STX, BPF_MEM, BPF_W),
-		DL(BPF_STX, BPF_MEM, BPF_DW),
-		DL(BPF_STX, BPF_XADD, BPF_W),
-		DL(BPF_STX, BPF_XADD, BPF_DW),
-		DL(BPF_ST, BPF_MEM, BPF_B),
-		DL(BPF_ST, BPF_MEM, BPF_H),
-		DL(BPF_ST, BPF_MEM, BPF_W),
-		DL(BPF_ST, BPF_MEM, BPF_DW),
-		DL(BPF_LDX, BPF_MEM, BPF_B),
-		DL(BPF_LDX, BPF_MEM, BPF_H),
-		DL(BPF_LDX, BPF_MEM, BPF_W),
-		DL(BPF_LDX, BPF_MEM, BPF_DW),
-		DL(BPF_LD, BPF_ABS, BPF_W),
-		DL(BPF_LD, BPF_ABS, BPF_H),
-		DL(BPF_LD, BPF_ABS, BPF_B),
-		DL(BPF_LD, BPF_IND, BPF_W),
-		DL(BPF_LD, BPF_IND, BPF_H),
-		DL(BPF_LD, BPF_IND, BPF_B),
+#define DL(A, B, C)	[BPF_##A|BPF_##B|BPF_##C] = &&A##_##B##_##C
+		DL(ALU, ADD, X),
+		DL(ALU, ADD, K),
+		DL(ALU, SUB, X),
+		DL(ALU, SUB, K),
+		DL(ALU, AND, X),
+		DL(ALU, AND, K),
+		DL(ALU, OR, X),
+		DL(ALU, OR, K),
+		DL(ALU, LSH, X),
+		DL(ALU, LSH, K),
+		DL(ALU, RSH, X),
+		DL(ALU, RSH, K),
+		DL(ALU, XOR, X),
+		DL(ALU, XOR, K),
+		DL(ALU, MUL, X),
+		DL(ALU, MUL, K),
+		DL(ALU, MOV, X),
+		DL(ALU, MOV, K),
+		DL(ALU, DIV, X),
+		DL(ALU, DIV, K),
+		DL(ALU, MOD, X),
+		DL(ALU, MOD, K),
+		DL(ALU, NEG, 0),
+		DL(ALU, END, TO_BE),
+		DL(ALU, END, TO_LE),
+		DL(ALU64, ADD, X),
+		DL(ALU64, ADD, K),
+		DL(ALU64, SUB, X),
+		DL(ALU64, SUB, K),
+		DL(ALU64, AND, X),
+		DL(ALU64, AND, K),
+		DL(ALU64, OR, X),
+		DL(ALU64, OR, K),
+		DL(ALU64, LSH, X),
+		DL(ALU64, LSH, K),
+		DL(ALU64, RSH, X),
+		DL(ALU64, RSH, K),
+		DL(ALU64, XOR, X),
+		DL(ALU64, XOR, K),
+		DL(ALU64, MUL, X),
+		DL(ALU64, MUL, K),
+		DL(ALU64, MOV, X),
+		DL(ALU64, MOV, K),
+		DL(ALU64, ARSH, X),
+		DL(ALU64, ARSH, K),
+		DL(ALU64, DIV, X),
+		DL(ALU64, DIV, K),
+		DL(ALU64, MOD, X),
+		DL(ALU64, MOD, K),
+		DL(ALU64, NEG, 0),
+		DL(JMP, CALL, 0),
+		DL(JMP, JA, 0),
+		DL(JMP, JEQ, X),
+		DL(JMP, JEQ, K),
+		DL(JMP, JNE, X),
+		DL(JMP, JNE, K),
+		DL(JMP, JGT, X),
+		DL(JMP, JGT, K),
+		DL(JMP, JGE, X),
+		DL(JMP, JGE, K),
+		DL(JMP, JSGT, X),
+		DL(JMP, JSGT, K),
+		DL(JMP, JSGE, X),
+		DL(JMP, JSGE, K),
+		DL(JMP, JSET, X),
+		DL(JMP, JSET, K),
+		DL(JMP, EXIT, 0),
+		DL(STX, MEM, B),
+		DL(STX, MEM, H),
+		DL(STX, MEM, W),
+		DL(STX, MEM, DW),
+		DL(STX, XADD, W),
+		DL(STX, XADD, DW),
+		DL(ST, MEM, B),
+		DL(ST, MEM, H),
+		DL(ST, MEM, W),
+		DL(ST, MEM, DW),
+		DL(LDX, MEM, B),
+		DL(LDX, MEM, H),
+		DL(LDX, MEM, W),
+		DL(LDX, MEM, DW),
+		DL(LD, ABS, W),
+		DL(LD, ABS, H),
+		DL(LD, ABS, B),
+		DL(LD, IND, W),
+		DL(LD, IND, H),
+		DL(LD, IND, B),
 #undef DL
 	};
 
@@ -257,93 +257,93 @@  select_insn:
 
 	/* ALU */
 #define ALU(OPCODE, OP)			\
-	BPF_ALU64_##OPCODE##_BPF_X:	\
+	ALU64_##OPCODE##_X:		\
 		A = A OP X;		\
 		CONT;			\
-	BPF_ALU_##OPCODE##_BPF_X:	\
+	ALU_##OPCODE##_X:		\
 		A = (u32) A OP (u32) X;	\
 		CONT;			\
-	BPF_ALU64_##OPCODE##_BPF_K:	\
+	ALU64_##OPCODE##_K:		\
 		A = A OP K;		\
 		CONT;			\
-	BPF_ALU_##OPCODE##_BPF_K:	\
+	ALU_##OPCODE##_K:		\
 		A = (u32) A OP (u32) K;	\
 		CONT;
 
-	ALU(BPF_ADD,  +)
-	ALU(BPF_SUB,  -)
-	ALU(BPF_AND,  &)
-	ALU(BPF_OR,   |)
-	ALU(BPF_LSH, <<)
-	ALU(BPF_RSH, >>)
-	ALU(BPF_XOR,  ^)
-	ALU(BPF_MUL,  *)
+	ALU(ADD,  +)
+	ALU(SUB,  -)
+	ALU(AND,  &)
+	ALU(OR,   |)
+	ALU(LSH, <<)
+	ALU(RSH, >>)
+	ALU(XOR,  ^)
+	ALU(MUL,  *)
 #undef ALU
-	BPF_ALU_BPF_NEG_0:
+	ALU_NEG_0:
 		A = (u32) -A;
 		CONT;
-	BPF_ALU64_BPF_NEG_0:
+	ALU64_NEG_0:
 		A = -A;
 		CONT;
-	BPF_ALU_BPF_MOV_BPF_X:
+	ALU_MOV_X:
 		A = (u32) X;
 		CONT;
-	BPF_ALU_BPF_MOV_BPF_K:
+	ALU_MOV_K:
 		A = (u32) K;
 		CONT;
-	BPF_ALU64_BPF_MOV_BPF_X:
+	ALU64_MOV_X:
 		A = X;
 		CONT;
-	BPF_ALU64_BPF_MOV_BPF_K:
+	ALU64_MOV_K:
 		A = K;
 		CONT;
-	BPF_ALU64_BPF_ARSH_BPF_X:
+	ALU64_ARSH_X:
 		(*(s64 *) &A) >>= X;
 		CONT;
-	BPF_ALU64_BPF_ARSH_BPF_K:
+	ALU64_ARSH_K:
 		(*(s64 *) &A) >>= K;
 		CONT;
-	BPF_ALU64_BPF_MOD_BPF_X:
+	ALU64_MOD_X:
 		if (unlikely(X == 0))
 			return 0;
 		tmp = A;
 		A = do_div(tmp, X);
 		CONT;
-	BPF_ALU_BPF_MOD_BPF_X:
+	ALU_MOD_X:
 		if (unlikely(X == 0))
 			return 0;
 		tmp = (u32) A;
 		A = do_div(tmp, (u32) X);
 		CONT;
-	BPF_ALU64_BPF_MOD_BPF_K:
+	ALU64_MOD_K:
 		tmp = A;
 		A = do_div(tmp, K);
 		CONT;
-	BPF_ALU_BPF_MOD_BPF_K:
+	ALU_MOD_K:
 		tmp = (u32) A;
 		A = do_div(tmp, (u32) K);
 		CONT;
-	BPF_ALU64_BPF_DIV_BPF_X:
+	ALU64_DIV_X:
 		if (unlikely(X == 0))
 			return 0;
 		do_div(A, X);
 		CONT;
-	BPF_ALU_BPF_DIV_BPF_X:
+	ALU_DIV_X:
 		if (unlikely(X == 0))
 			return 0;
 		tmp = (u32) A;
 		do_div(tmp, (u32) X);
 		A = (u32) tmp;
 		CONT;
-	BPF_ALU64_BPF_DIV_BPF_K:
+	ALU64_DIV_K:
 		do_div(A, K);
 		CONT;
-	BPF_ALU_BPF_DIV_BPF_K:
+	ALU_DIV_K:
 		tmp = (u32) A;
 		do_div(tmp, (u32) K);
 		A = (u32) tmp;
 		CONT;
-	BPF_ALU_BPF_END_BPF_TO_BE:
+	ALU_END_TO_BE:
 		switch (K) {
 		case 16:
 			A = (__force u16) cpu_to_be16(A);
@@ -356,7 +356,7 @@  select_insn:
 			break;
 		}
 		CONT;
-	BPF_ALU_BPF_END_BPF_TO_LE:
+	ALU_END_TO_LE:
 		switch (K) {
 		case 16:
 			A = (__force u16) cpu_to_le16(A);
@@ -371,7 +371,7 @@  select_insn:
 		CONT;
 
 	/* CALL */
-	BPF_JMP_BPF_CALL_0:
+	JMP_CALL_0:
 		/* Function call scratches R1-R5 registers, preserves R6-R9,
 		 * and stores return value into R0.
 		 */
@@ -380,122 +380,122 @@  select_insn:
 		CONT;
 
 	/* JMP */
-	BPF_JMP_BPF_JA_0:
+	JMP_JA_0:
 		insn += insn->off;
 		CONT;
-	BPF_JMP_BPF_JEQ_BPF_X:
+	JMP_JEQ_X:
 		if (A == X) {
 			insn += insn->off;
 			CONT_JMP;
 		}
 		CONT;
-	BPF_JMP_BPF_JEQ_BPF_K:
+	JMP_JEQ_K:
 		if (A == K) {
 			insn += insn->off;
 			CONT_JMP;
 		}
 		CONT;
-	BPF_JMP_BPF_JNE_BPF_X:
+	JMP_JNE_X:
 		if (A != X) {
 			insn += insn->off;
 			CONT_JMP;
 		}
 		CONT;
-	BPF_JMP_BPF_JNE_BPF_K:
+	JMP_JNE_K:
 		if (A != K) {
 			insn += insn->off;
 			CONT_JMP;
 		}
 		CONT;
-	BPF_JMP_BPF_JGT_BPF_X:
+	JMP_JGT_X:
 		if (A > X) {
 			insn += insn->off;
 			CONT_JMP;
 		}
 		CONT;
-	BPF_JMP_BPF_JGT_BPF_K:
+	JMP_JGT_K:
 		if (A > K) {
 			insn += insn->off;
 			CONT_JMP;
 		}
 		CONT;
-	BPF_JMP_BPF_JGE_BPF_X:
+	JMP_JGE_X:
 		if (A >= X) {
 			insn += insn->off;
 			CONT_JMP;
 		}
 		CONT;
-	BPF_JMP_BPF_JGE_BPF_K:
+	JMP_JGE_K:
 		if (A >= K) {
 			insn += insn->off;
 			CONT_JMP;
 		}
 		CONT;
-	BPF_JMP_BPF_JSGT_BPF_X:
-		if (((s64)A) > ((s64)X)) {
+	JMP_JSGT_X:
+		if (((s64) A) > ((s64) X)) {
 			insn += insn->off;
 			CONT_JMP;
 		}
 		CONT;
-	BPF_JMP_BPF_JSGT_BPF_K:
-		if (((s64)A) > ((s64)K)) {
+	JMP_JSGT_K:
+		if (((s64) A) > ((s64) K)) {
 			insn += insn->off;
 			CONT_JMP;
 		}
 		CONT;
-	BPF_JMP_BPF_JSGE_BPF_X:
-		if (((s64)A) >= ((s64)X)) {
+	JMP_JSGE_X:
+		if (((s64) A) >= ((s64) X)) {
 			insn += insn->off;
 			CONT_JMP;
 		}
 		CONT;
-	BPF_JMP_BPF_JSGE_BPF_K:
-		if (((s64)A) >= ((s64)K)) {
+	JMP_JSGE_K:
+		if (((s64) A) >= ((s64) K)) {
 			insn += insn->off;
 			CONT_JMP;
 		}
 		CONT;
-	BPF_JMP_BPF_JSET_BPF_X:
+	JMP_JSET_X:
 		if (A & X) {
 			insn += insn->off;
 			CONT_JMP;
 		}
 		CONT;
-	BPF_JMP_BPF_JSET_BPF_K:
+	JMP_JSET_K:
 		if (A & K) {
 			insn += insn->off;
 			CONT_JMP;
 		}
 		CONT;
-	BPF_JMP_BPF_EXIT_0:
+	JMP_EXIT_0:
 		return R0;
 
 	/* STX and ST and LDX*/
 #define LDST(SIZEOP, SIZE)					\
-	BPF_STX_BPF_MEM_##SIZEOP:				\
+	STX_MEM_##SIZEOP:					\
 		*(SIZE *)(unsigned long) (A + insn->off) = X;	\
 		CONT;						\
-	BPF_ST_BPF_MEM_##SIZEOP:				\
+	ST_MEM_##SIZEOP:					\
 		*(SIZE *)(unsigned long) (A + insn->off) = K;	\
 		CONT;						\
-	BPF_LDX_BPF_MEM_##SIZEOP:				\
+	LDX_MEM_##SIZEOP:					\
 		A = *(SIZE *)(unsigned long) (X + insn->off);	\
 		CONT;
 
-	LDST(BPF_B,   u8)
-	LDST(BPF_H,  u16)
-	LDST(BPF_W,  u32)
-	LDST(BPF_DW, u64)
+	LDST(B,   u8)
+	LDST(H,  u16)
+	LDST(W,  u32)
+	LDST(DW, u64)
 #undef LDST
-	BPF_STX_BPF_XADD_BPF_W: /* lock xadd *(u32 *)(A + insn->off) += X */
+	STX_XADD_W: /* lock xadd *(u32 *)(A + insn->off) += X */
 		atomic_add((u32) X, (atomic_t *)(unsigned long)
 			   (A + insn->off));
 		CONT;
-	BPF_STX_BPF_XADD_BPF_DW: /* lock xadd *(u64 *)(A + insn->off) += X */
+	STX_XADD_DW: /* lock xadd *(u64 *)(A + insn->off) += X */
 		atomic64_add((u64) X, (atomic64_t *)(unsigned long)
 			     (A + insn->off));
 		CONT;
-	BPF_LD_BPF_ABS_BPF_W: /* R0 = ntohl(*(u32 *) (skb->data + K)) */
+	LD_ABS_W: /* R0 = ntohl(*(u32 *) (skb->data + K)) */
 		off = K;
 load_word:
 		/* BPF_LD + BPD_ABS and BPF_LD + BPF_IND insns are only
@@ -524,7 +524,7 @@  load_word:
 			CONT;
 		}
 		return 0;
-	BPF_LD_BPF_ABS_BPF_H: /* R0 = ntohs(*(u16 *) (skb->data + K)) */
+	LD_ABS_H: /* R0 = ntohs(*(u16 *) (skb->data + K)) */
 		off = K;
 load_half:
 		ptr = load_pointer((struct sk_buff *) ctx, off, 2, &tmp);
@@ -533,7 +533,7 @@  load_half:
 			CONT;
 		}
 		return 0;
-	BPF_LD_BPF_ABS_BPF_B: /* R0 = *(u8 *) (ctx + K) */
+	LD_ABS_B: /* R0 = *(u8 *) (ctx + K) */
 		off = K;
 load_byte:
 		ptr = load_pointer((struct sk_buff *) ctx, off, 1, &tmp);
@@ -542,13 +542,13 @@  load_byte:
 			CONT;
 		}
 		return 0;
-	BPF_LD_BPF_IND_BPF_W: /* R0 = ntohl(*(u32 *) (skb->data + X + K)) */
+	LD_IND_W: /* R0 = ntohl(*(u32 *) (skb->data + X + K)) */
 		off = K + X;
 		goto load_word;
-	BPF_LD_BPF_IND_BPF_H: /* R0 = ntohs(*(u16 *) (skb->data + X + K)) */
+	LD_IND_H: /* R0 = ntohs(*(u16 *) (skb->data + X + K)) */
 		off = K + X;
 		goto load_half;
-	BPF_LD_BPF_IND_BPF_B: /* R0 = *(u8 *) (skb->data + X + K) */
+	LD_IND_B: /* R0 = *(u8 *) (skb->data + X + K) */
 		off = K + X;
 		goto load_byte;