Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1599202 > unrolled thread

Re: [PATCH] MIPS: BPF: Add support for SKF_AD_HATYPE

Started byJames Hogan <james.hogan@imgtec.com>
First post2017-03-13 12:00 +0100
Last post2017-03-13 17:50 +0100
Articles 2 — 2 participants

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.


Contents

  Re: [PATCH] MIPS: BPF: Add support for SKF_AD_HATYPE James Hogan <james.hogan@imgtec.com> - 2017-03-13 12:00 +0100
    Re: [PATCH] MIPS: BPF: Add support for SKF_AD_HATYPE David Daney <ddaney.cavm@gmail.com> - 2017-03-13 17:50 +0100

#1599202 — Re: [PATCH] MIPS: BPF: Add support for SKF_AD_HATYPE

FromJames Hogan <james.hogan@imgtec.com>
Date2017-03-13 12:00 +0100
SubjectRe: [PATCH] MIPS: BPF: Add support for SKF_AD_HATYPE
Message-ID<tkvRL-7Mz-3@gated-at.bofh.it>

[Multipart message — attachments visible in raw view] — view raw

On Fri, Mar 10, 2017 at 02:14:05PM -0800, David Daney wrote:
> This let's us pass some additional "modprobe test-bpf" tests.
> 
> Reuse the code for SKF_AD_IFINDEX, but substitute the offset and size
> of the "type" field.
> 
> Signed-off-by: David Daney <david.daney@cavium.com>

I think the BPF maintainers should probably be Cc'd on this patch.
Cc'ing now.

> ---
>  arch/mips/net/bpf_jit.c | 14 ++++++++++----
>  1 file changed, 10 insertions(+), 4 deletions(-)
> 
> diff --git a/arch/mips/net/bpf_jit.c b/arch/mips/net/bpf_jit.c
> index 49a2e22..f613708 100644
> --- a/arch/mips/net/bpf_jit.c
> +++ b/arch/mips/net/bpf_jit.c
> @@ -1111,6 +1111,7 @@ static int build_body(struct jit_ctx *ctx)
>  			emit_load(r_A, 28, off, ctx);
>  			break;
>  		case BPF_ANC | SKF_AD_IFINDEX:
> +		case BPF_ANC | SKF_AD_HATYPE:
>  			/* A = skb->dev->ifindex */

this comment should probably be updated.

>  			ctx->flags |= SEEN_SKB | SEEN_A;
>  			off = offsetof(struct sk_buff, dev);
> @@ -1120,10 +1121,15 @@ static int build_body(struct jit_ctx *ctx)
>  			emit_bcond(MIPS_COND_EQ, r_s0, r_zero,
>  				   b_imm(prog->len, ctx), ctx);
>  			emit_reg_move(r_ret, r_zero, ctx);
> -			BUILD_BUG_ON(FIELD_SIZEOF(struct net_device,
> -						  ifindex) != 4);
> -			off = offsetof(struct net_device, ifindex);
> -			emit_load(r_A, r_s0, off, ctx);
> +			if (code == (BPF_ANC | SKF_AD_IFINDEX)) {
> +				BUILD_BUG_ON(FIELD_SIZEOF(struct net_device, ifindex) != 4);
> +				off = offsetof(struct net_device, ifindex);
> +				emit_load(r_A, r_s0, off, ctx);
> +			} else { /* (code == (BPF_ANC | SKF_AD_HATYPE) */
> +				BUILD_BUG_ON(FIELD_SIZEOF(struct net_device, type) != 2);
> +				off = offsetof(struct net_device, type);
> +				emit_half_load(r_A, r_s0, off, ctx);

Technically net_device::type is unsigned, and emit_half_load uses LH
which sign extends. Does that matter in practice.

Cheers
James

[toc] | [next] | [standalone]


#1599598

FromDavid Daney <ddaney.cavm@gmail.com>
Date2017-03-13 17:50 +0100
Message-ID<tkBku-3fs-29@gated-at.bofh.it>
In reply to#1599202
On 03/13/2017 03:56 AM, James Hogan wrote:
> On Fri, Mar 10, 2017 at 02:14:05PM -0800, David Daney wrote:
>> This let's us pass some additional "modprobe test-bpf" tests.
>>
>> Reuse the code for SKF_AD_IFINDEX, but substitute the offset and size
>> of the "type" field.
>>
>> Signed-off-by: David Daney <david.daney@cavium.com>
>
> I think the BPF maintainers should probably be Cc'd on this patch.
> Cc'ing now.
>

Good point.

Since there are some corrections needed, I will send another version and 
CC the proper people.


>> ---
>>  arch/mips/net/bpf_jit.c | 14 ++++++++++----
>>  1 file changed, 10 insertions(+), 4 deletions(-)
>>
>> diff --git a/arch/mips/net/bpf_jit.c b/arch/mips/net/bpf_jit.c
>> index 49a2e22..f613708 100644
>> --- a/arch/mips/net/bpf_jit.c
>> +++ b/arch/mips/net/bpf_jit.c
>> @@ -1111,6 +1111,7 @@ static int build_body(struct jit_ctx *ctx)
>>  			emit_load(r_A, 28, off, ctx);
>>  			break;
>>  		case BPF_ANC | SKF_AD_IFINDEX:
>> +		case BPF_ANC | SKF_AD_HATYPE:
>>  			/* A = skb->dev->ifindex */
>
> this comment should probably be updated.

Right.

>
>>  			ctx->flags |= SEEN_SKB | SEEN_A;
>>  			off = offsetof(struct sk_buff, dev);
>> @@ -1120,10 +1121,15 @@ static int build_body(struct jit_ctx *ctx)
>>  			emit_bcond(MIPS_COND_EQ, r_s0, r_zero,
>>  				   b_imm(prog->len, ctx), ctx);
>>  			emit_reg_move(r_ret, r_zero, ctx);
>> -			BUILD_BUG_ON(FIELD_SIZEOF(struct net_device,
>> -						  ifindex) != 4);
>> -			off = offsetof(struct net_device, ifindex);
>> -			emit_load(r_A, r_s0, off, ctx);
>> +			if (code == (BPF_ANC | SKF_AD_IFINDEX)) {
>> +				BUILD_BUG_ON(FIELD_SIZEOF(struct net_device, ifindex) != 4);
>> +				off = offsetof(struct net_device, ifindex);
>> +				emit_load(r_A, r_s0, off, ctx);
>> +			} else { /* (code == (BPF_ANC | SKF_AD_HATYPE) */
>> +				BUILD_BUG_ON(FIELD_SIZEOF(struct net_device, type) != 2);
>> +				off = offsetof(struct net_device, type);
>> +				emit_half_load(r_A, r_s0, off, ctx);
>
> Technically net_device::type is unsigned, and emit_half_load uses LH
> which sign extends. Does that matter in practice.

The next version will use LHU.


>
> Cheers
> James
>

[toc] | [prev] | [standalone]


Back to top | Article view | linux.kernel


csiph-web