Received-SPF: pass (google.com: domain of linux-kernel+bounces-150885-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Message-ID: <6b6bd108-817c-4a58-8b69-6c2dde436575@quicinc.com>
Date: Thu, 18 Apr 2024 17:30:42 -0700
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [RFC PATCH bpf-next v4 2/2] net: Add additional bit to support
 clockid_t timestamp type
Content-Language: en-US
To: Martin KaFai Lau <martin.lau@linux.dev>,
        Willem de Bruijn
	<willemdebruijn.kernel@gmail.com>
CC: "David S. Miller" <davem@davemloft.net>,
        Eric Dumazet
	<edumazet@google.com>, Jakub Kicinski <kuba@kernel.org>,
        Paolo Abeni
	<pabeni@redhat.com>, <netdev@vger.kernel.org>,
        <linux-kernel@vger.kernel.org>, Andrew Halaney <ahalaney@redhat.com>,
        "Martin
 KaFai Lau" <martin.lau@kernel.org>,
        Daniel Borkmann <daniel@iogearbox.net>, bpf <bpf@vger.kernel.org>,
        <kernel@quicinc.com>
References: <20240418004308.1009262-1-quic_abchauha@quicinc.com>
 <20240418004308.1009262-3-quic_abchauha@quicinc.com>
 <66216f3ec638b_f648a294ec@willemb.c.googlers.com.notmuch>
 <cb922600-783e-4741-be85-260d1ded5bdb@quicinc.com>
 <c6f33a36-1fac-4738-8a4f-c930b544ba62@linux.dev>
From: "Abhishek Chauhan (ABC)" <quic_abchauha@quicinc.com>
In-Reply-To: <c6f33a36-1fac-4738-8a4f-c930b544ba62@linux.dev>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 8bit


On 4/18/2024 2:57 PM, Martin KaFai Lau wrote:
> On 4/18/24 1:10 PM, Abhishek Chauhan (ABC) wrote:
>>>>   #ifdef CONFIG_NET_XGRESS
>>>>       __u8            tc_at_ingress:1;    /* See TC_AT_INGRESS_MASK */
>>>>       __u8            tc_skip_classify:1;
>>>> @@ -1096,10 +1100,12 @@ struct sk_buff {
>>>>    */
>>>>   #ifdef __BIG_ENDIAN_BITFIELD
>>>>   #define SKB_MONO_DELIVERY_TIME_MASK    (1 << 7)
>>>> -#define TC_AT_INGRESS_MASK        (1 << 6)
>>>> +#define SKB_TAI_DELIVERY_TIME_MASK    (1 << 6)
>>>
>>> SKB_TSTAMP_TYPE_BIT2_MASK?
> 
> nit. Shorten it to just SKB_TSTAMP_TYPE_MASK?
> 
Okay i will do the same. Noted!
> #ifdef __BIG_ENDIAN_BITFIELD
> #define SKB_TSTAMP_TYPE_MASK    (3 << 6)
> #define SKB_TSTAMP_TYPE_RSH    (6)    /* more on this later */
> #else
> #define SKB_TSTAMP_TYPE_MASK    (3)
> #endif
> 
>>>
>> I was thinking to keep it as TAI because it will confuse developers. I hope thats okay.
> 
> I think it is not very useful to distinguish each bit since it is an enum value now. It becomes more like the "pkt_type:3" and its PKT_TYPE_MAX.
>I see what you are saying.
>>>> +#define TC_AT_INGRESS_MASK        (1 << 5)
>>>>   #else
>>>>   #define SKB_MONO_DELIVERY_TIME_MASK    (1 << 0)
>>>> -#define TC_AT_INGRESS_MASK        (1 << 1)
>>>> +#define SKB_TAI_DELIVERY_TIME_MASK    (1 << 1)
>>>> +#define TC_AT_INGRESS_MASK        (1 << 2)
>>>>   #endif
>>>>   #define SKB_BF_MONO_TC_OFFSET        offsetof(struct sk_buff, __mono_tc_offset)
>>>>   @@ -4206,6 +4212,11 @@ static inline void skb_set_delivery_time(struct sk_buff *skb, ktime_t kt,
>>>>       case CLOCK_MONOTONIC:
>>>>           skb->tstamp_type = SKB_CLOCK_MONO;
>>>>           break;
>>>> +    case CLOCK_TAI:
>>>> +        skb->tstamp_type = SKB_CLOCK_TAI;
>>>> +        break;
>>>> +    default:
>>>> +        WARN_ONCE(true, "clockid %d not supported", tstamp_type);
>>>
>>> and set to 0 and default tstamp_type?
>>> Actually thinking about it. I feel if its unsupported just fall back to default is the correct thing. I will take care of this.
>>>>       }
>>>>   }
>>>
>>>>   >
>>>   @@ -9372,10 +9378,16 @@ static struct bpf_insn *bpf_convert_tstamp_type_read(const struct bpf_insn *si,
>>>>       *insn++ = BPF_LDX_MEM(BPF_B, tmp_reg, skb_reg,
>>>>                     SKB_BF_MONO_TC_OFFSET);
>>>>       *insn++ = BPF_JMP32_IMM(BPF_JSET, tmp_reg,
>>>> -                SKB_MONO_DELIVERY_TIME_MASK, 2);
>>>> +                SKB_MONO_DELIVERY_TIME_MASK | SKB_TAI_DELIVERY_TIME_MASK, 2);
>>>> +    *insn++ = BPF_JMP32_IMM(BPF_JSET, tmp_reg,
>>>> +                SKB_MONO_DELIVERY_TIME_MASK, 3);
>>>> +    *insn++ = BPF_JMP32_IMM(BPF_JSET, tmp_reg,
>>>> +                SKB_TAI_DELIVERY_TIME_MASK, 4);
>>>>       *insn++ = BPF_MOV32_IMM(value_reg, BPF_SKB_TSTAMP_UNSPEC);
>>>>       *insn++ = BPF_JMP_A(1);
>>>>       *insn++ = BPF_MOV32_IMM(value_reg, BPF_SKB_TSTAMP_DELIVERY_MONO);
>>>> +    *insn++ = BPF_JMP_A(1);
>>>> +    *insn++ = BPF_MOV32_IMM(value_reg, BPF_SKB_TSTAMP_DELIVERY_TAI);
> 
> With SKB_TSTAMP_TYPE_MASK defined like above, this could be simplified like this (untested):
> 
Let me think this through and raise it as part of the next rfc patch. 
> static struct bpf_insn *bpf_convert_tstamp_type_read(const struct bpf_insn *si,
>                                                      struct bpf_insn *insn)
> {
>     __u8 value_reg = si->dst_reg;
>     __u8 skb_reg = si->src_reg;
> 
>     BUILD_BUG_ON(__SKB_CLOCK_MAX != BPF_SKB_TSTAMP_DELIVERY_TAI);
>     *insn++ = BPF_LDX_MEM(BPF_B, value_reg, skb_reg, SKB_BF_MONO_TC_OFFSET);
>     *insn++ = BPF_ALU32_IMM(BPF_AND, value_reg, SKB_TSTAMP_TYPE_MASK);
> #ifdef __BIG_ENDIAN_BITFIELD
>     *insn++ = BPF_ALU32_IMM(BPF_RSH, value_reg, SKB_TSTAMP_TYPE_RSH);
> #else
>     BUILD_BUG_ON(!(SKB_TSTAMP_TYPE_MASK & 0x1));
> #endif
> 
>     return insn;
> }
> 
>>>>         return insn;
>>>>   }
>>>> @@ -9418,10 +9430,26 @@ static struct bpf_insn *bpf_convert_tstamp_read(const struct bpf_prog *prog,
>>>>           __u8 tmp_reg = BPF_REG_AX;
>>>>             *insn++ = BPF_LDX_MEM(BPF_B, tmp_reg, skb_reg, SKB_BF_MONO_TC_OFFSET);
>>>> +        /*check if all three bits are set*/
>>>>           *insn++ = BPF_ALU32_IMM(BPF_AND, tmp_reg,
>>>> -                    TC_AT_INGRESS_MASK | SKB_MONO_DELIVERY_TIME_MASK);
>>>> -        *insn++ = BPF_JMP32_IMM(BPF_JNE, tmp_reg,
>>>> -                    TC_AT_INGRESS_MASK | SKB_MONO_DELIVERY_TIME_MASK, 2);
>>>> +                    TC_AT_INGRESS_MASK | SKB_MONO_DELIVERY_TIME_MASK |
>>>> +                    SKB_TAI_DELIVERY_TIME_MASK);
>>>> +        /*if all 3 bits are set jump 3 instructions and clear the register */
>>>> +        *insn++ = BPF_JMP32_IMM(BPF_JEQ, tmp_reg,
>>>> +                    TC_AT_INGRESS_MASK | SKB_MONO_DELIVERY_TIME_MASK |
>>>> +                    SKB_TAI_DELIVERY_TIME_MASK, 4);
>>>> +        /*Now check Mono is set with ingress mask if so clear */
>>>> +        *insn++ = BPF_JMP32_IMM(BPF_JEQ, tmp_reg,
>>>> +                    TC_AT_INGRESS_MASK | SKB_MONO_DELIVERY_TIME_MASK, 3);
>>>> +        /*Now Check tai is set with ingress mask if so clear */
>>>> +        *insn++ = BPF_JMP32_IMM(BPF_JEQ, tmp_reg,
>>>> +                    TC_AT_INGRESS_MASK | SKB_TAI_DELIVERY_TIME_MASK, 2);
>>>> +        /*Now Check tai and mono are set if so clear */
>>>> +        *insn++ = BPF_JMP32_IMM(BPF_JEQ, tmp_reg,
>>>> +                    SKB_MONO_DELIVERY_TIME_MASK |
>>>> +                    SKB_TAI_DELIVERY_TIME_MASK, 1);
> 
> Same as the bpf_convert_tstamp_type_read, this could be simplified with SKB_TSTAMP_TYPE_MASK.
> 
>>>
>>> This looks as if all JEQ result in "if so clear"?
>>>
>>> Is the goal to only do something different for the two bits being 0x1,
>>> can we have a single test with a two-bit mask, rather than four tests?
>>>
>> I think Martin wanted to take care of TAI as well. I will wait for his comment here
>>
>> My Goal was to take care of invalid combos which does not hold valid
>> 1. If all 3 bits are set => invalid combo (Test case written is Insane)
>> 2. If 2 bits are set (tai+mono)(Test case written is Insane) => this cannot happen (because clock base can only be one in skb)
>> 3. If 2 bit are set (ingress + tai/mono) => This is existing logic + tai being added (clear tstamp in ingress)
>> 4. For all other cases go ahead and fill in the tstamp in the dest register.
> 
> If it is to ensure no new type is added without adding BPF_SKB_TSTAMP_DELIVERY_XYZ, I would simplify this runtime bpf insns here and use a BUILD_BUG_ON to catch it at compile time. Something like,
> 
> enum skb_tstamp_type {
>         SKB_CLOCK_REAL, /* Time base is skb is REALTIME */
>         SKB_CLOCK_MONO, /* Time base is skb is MONOTONIC */
>      SKB_CLOCK_TAI,  /* Time base in skb is TAI */
>         __SKB_CLOCK_MAX = SKB_CLOCK_TAI,
> };
> 
> /* Same one used in the bpf_convert_tstamp_type_read() above */
> BUILD_BUG_ON(__SKB_CLOCK_MAX != BPF_SKB_TSTAMP_DELIVERY_TAI);
> 
> Another thing is, the UDP test in test_tc_dtime.c probably needs to be adjusted, the userspace is using the CLOCK_TAI in SO_TXTIME and it is getting forwarded now.
Noted ! Let me check and evalute this as well.