public inbox for bpf@vger.kernel.org
 help / color / mirror / Atom feed
From: Yonghong Song <yonghong.song@linux.dev>
To: Eduard Zingerman <eddyz87@gmail.com>, bpf@vger.kernel.org
Cc: Alexei Starovoitov <ast@kernel.org>,
	Andrii Nakryiko <andrii@kernel.org>,
	Daniel Borkmann <daniel@iogearbox.net>,
	kernel-team@fb.com, Martin KaFai Lau <martin.lau@kernel.org>
Subject: Re: [PATCH bpf-next v3 1/2] bpf: Do not include stack ptr register in precision backtracking bookkeeping
Date: Wed, 21 May 2025 13:34:50 -0700	[thread overview]
Message-ID: <2c0fa9ee-f9dd-4cde-b4fb-6f28ebefc619@linux.dev> (raw)
In-Reply-To: <45e399c6-74ad-4e58-bfda-06b392d1d28d@gmail.com>



On 5/21/25 11:55 AM, Eduard Zingerman wrote:
> [...]
>
>> diff --git a/include/linux/bpf_verifier.h b/include/linux/bpf_verifier.h
>> index 78c97e12ea4e..e73a910e4ece 100644
>> --- a/include/linux/bpf_verifier.h
>> +++ b/include/linux/bpf_verifier.h
>> @@ -357,6 +357,10 @@ enum {
>>       INSN_F_SPI_SHIFT = 3, /* shifted 3 bits to the left */
>>         INSN_F_STACK_ACCESS = BIT(9), /* we need 10 bits total */
>> +
>> +    INSN_F_DST_REG_STACK = BIT(10), /* dst_reg is PTR_TO_STACK */
>> +    INSN_F_SRC_REG_STACK = BIT(11), /* src_reg is PTR_TO_STACK */
>
> INSN_F_STACK_ACCESS can be inferred from INSN_F_DST_REG_STACK
> and INSN_F_SRC_REG_STACK if insn_stack_access_flags() is adjusted
> to track these flags instead. So, can be one less flag/bit.

You are correct, we could have BIT(9) for both INSN_F_STACK_ACCESS and INSN_F_DST_REG_STACK,
and BIT(10) for INSN_F_SRC_REG_STACK. But it makes code a little bit
complicated. I am okay with this if Andrii also thinks it is
worthwhile to do this.

>
>> +    /* total 12 bits are used now. */
>>   };
>>     static_assert(INSN_F_FRAMENO_MASK + 1 >= MAX_CALL_FRAMES);
>
> [...]
>
>> @@ -4402,6 +4418,8 @@ static int backtrack_insn(struct 
>> bpf_verifier_env *env, int idx, int subseq_idx,
>>                */
>>               return 0;
>>           } else if (BPF_SRC(insn->code) == BPF_X) {
>> +            bool dreg_precise, sreg_precise;
>> +
>>               if (!bt_is_reg_set(bt, dreg) && !bt_is_reg_set(bt, sreg))
>>                   return 0;
>>               /* dreg <cond> sreg
>> @@ -4410,8 +4428,16 @@ static int backtrack_insn(struct 
>> bpf_verifier_env *env, int idx, int subseq_idx,
>>                * before it would be equally necessary to
>>                * propagate it to dreg.
>>                */
>> -            bt_set_reg(bt, dreg);
>> -            bt_set_reg(bt, sreg);
>> +            if (!hist)
>> +                return 0;
>> +            dreg_precise = !insn_dreg_stack_ptr(hist->flags);
>> +            sreg_precise = !insn_sreg_stack_ptr(hist->flags);
>> +            if (!dreg_precise && !sreg_precise)
>> +                return 0;
>> +            if (dreg_precise)
>> +                bt_set_reg(bt, dreg);
>> +            if (sreg_precise)
>> +                bt_set_reg(bt, sreg);
>
> This check can be done in a generic way at backtrack_insn() callsite:
> check which register is pointer to stack and remove it from set 
> registers.

Looks like other cases in backtrack_insn() has been handled properly,
so it may still be worthwhile to put the code here.

>
>>           } else if (BPF_SRC(insn->code) == BPF_K) {
>>                /* dreg <cond> K
>>                 * Only dreg still needs precision before
>> @@ -16397,6 +16423,29 @@ static void sync_linked_regs(struct 
>> bpf_verifier_state *vstate, struct bpf_reg_s
>>       }
>>   }
>>   +static int push_cond_jmp_history(struct bpf_verifier_env *env, 
>> struct bpf_verifier_state *state,
>> +                 struct bpf_reg_state *dst_reg, struct bpf_reg_state 
>> *src_reg,
>> +                 u64 linked_regs)
>> +{
>> +    bool dreg_stack_ptr, sreg_stack_ptr;
>> +    int insn_flags;
>> +
>> +    if (!src_reg) {
>> +        if (linked_regs)
>> +            return push_insn_history(env, state, 0, linked_regs);
>> +        return 0;
>> +    }
>
> Nit: this 'if' is not needed, src_reg is always set (it might point to 
> a fake register,
>      but in that case it is a scalar without id).
>
Here, there is a bug here. Thanks for pointing this out. I need to check
BPF_SRC(insn->code) != BPF_X instead of "!src_reg". Basically passing one
more parameter (e.g., faked_sreg) to decide whether src_reg is faked or not.

>
>> +
>> +    dreg_stack_ptr = dst_reg->type == PTR_TO_STACK;
>> +    sreg_stack_ptr = src_reg->type == PTR_TO_STACK;
>> +
>> +    if (!dreg_stack_ptr && !sreg_stack_ptr && !linked_regs)
>> +        return 0;
>> +
>> +    insn_flags = insn_reg_access_flags(dreg_stack_ptr, sreg_stack_ptr);
>> +    return push_insn_history(env, state, insn_flags, linked_regs);
>> +}
>> +
>
> [...]
>


  reply	other threads:[~2025-05-21 20:34 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-05-21 17:04 [PATCH bpf-next v3 1/2] bpf: Do not include stack ptr register in precision backtracking bookkeeping Yonghong Song
2025-05-21 17:04 ` [PATCH bpf-next v3 2/2] selftests/bpf: Add tests with stack ptr register in conditional jmp Yonghong Song
2025-05-21 19:02   ` Eduard Zingerman
2025-05-21 20:57     ` Yonghong Song
2025-05-21 21:03       ` Eduard Zingerman
2025-05-21 18:55 ` [PATCH bpf-next v3 1/2] bpf: Do not include stack ptr register in precision backtracking bookkeeping Eduard Zingerman
2025-05-21 20:34   ` Yonghong Song [this message]
2025-05-21 20:58     ` Eduard Zingerman
2025-05-21 21:35       ` Yonghong Song
2025-05-21 21:59         ` Eduard Zingerman
2025-05-21 22:04           ` Eduard Zingerman
2025-05-21 22:38     ` Andrii Nakryiko
2025-05-21 22:49       ` Eduard Zingerman
2025-05-22  2:53         ` Yonghong Song
2025-05-22 20:05         ` Andrii Nakryiko

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=2c0fa9ee-f9dd-4cde-b4fb-6f28ebefc619@linux.dev \
    --to=yonghong.song@linux.dev \
    --cc=andrii@kernel.org \
    --cc=ast@kernel.org \
    --cc=bpf@vger.kernel.org \
    --cc=daniel@iogearbox.net \
    --cc=eddyz87@gmail.com \
    --cc=kernel-team@fb.com \
    --cc=martin.lau@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox