linux-kernel - Re: [RFC PATCH RESEND bpf-next 1/4] bpf: Rollback to text_poke when arch not supported ftrace direct call

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <1c35ce46-27ff-f311-44f5-b3add388c045@huaweicloud.com>
Date:   Fri, 6 Jan 2023 10:35:57 +0800
From:   Pu Lehui <pulehui@...weicloud.com>
To:     Björn Töpel <bjorn@...nel.org>,
        Xu Kuohai <xukuohai@...wei.com>, bpf@...r.kernel.org,
        linux-riscv@...ts.infradead.org, linux-kernel@...r.kernel.org
Cc:     Luke Nelson <luke.r.nels@...il.com>, Xi Wang <xi.wang@...il.com>,
        Alexei Starovoitov <ast@...nel.org>,
        Daniel Borkmann <daniel@...earbox.net>,
        Andrii Nakryiko <andrii@...nel.org>,
        Martin KaFai Lau <martin.lau@...ux.dev>,
        Song Liu <song@...nel.org>, Yonghong Song <yhs@...com>,
        John Fastabend <john.fastabend@...il.com>,
        KP Singh <kpsingh@...nel.org>,
        Stanislav Fomichev <sdf@...gle.com>,
        Hao Luo <haoluo@...gle.com>, Jiri Olsa <jolsa@...nel.org>,
        Paul Walmsley <paul.walmsley@...ive.com>,
        Palmer Dabbelt <palmer@...belt.com>,
        Albert Ou <aou@...s.berkeley.edu>,
        Pu Lehui <pulehui@...wei.com>
Subject: Re: [RFC PATCH RESEND bpf-next 1/4] bpf: Rollback to text_poke when
 arch not supported ftrace direct call



On 2023/1/3 20:05, Björn Töpel wrote:
> Pu Lehui <pulehui@...weicloud.com> writes:
> 
>> On 2022/12/20 10:32, Xu Kuohai wrote:
>>> On 12/20/2022 10:13 AM, Pu Lehui wrote:
>>>> From: Pu Lehui <pulehui@...wei.com>
>>>>
>>>> The current bpf trampoline attach to kernel functions via ftrace direct
>>>> call API, while text_poke is applied for bpf2bpf attach and tail call
>>>> optimization. For architectures that do not support ftrace direct call,
>>>> text_poke is still able to attach bpf trampoline to kernel functions.
>>>> Let's relax it by rollback to text_poke when architecture not supported.
>>>>
>>>> Signed-off-by: Pu Lehui <pulehui@...wei.com>
>>>> ---
>>>>    kernel/bpf/trampoline.c | 8 ++------
>>>>    1 file changed, 2 insertions(+), 6 deletions(-)
>>>>
>>>> diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
>>>> index d6395215b849..386197a7952c 100644
>>>> --- a/kernel/bpf/trampoline.c
>>>> +++ b/kernel/bpf/trampoline.c
>>>> @@ -228,15 +228,11 @@ static int modify_fentry(struct bpf_trampoline
>>>> *tr, void *old_addr, void *new_ad
>>>>    static int register_fentry(struct bpf_trampoline *tr, void *new_addr)
>>>>    {
>>>>        void *ip = tr->func.addr;
>>>> -    unsigned long faddr;
>>>>        int ret;
>>>> -    faddr = ftrace_location((unsigned long)ip);
>>>> -    if (faddr) {
>>>> -        if (!tr->fops)
>>>> -            return -ENOTSUPP;
>>>> +    if (IS_ENABLED(CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS) &&
>>>> +        !!ftrace_location((unsigned long)ip))
>>>>            tr->func.ftrace_managed = true;
>>>> -    }
>>>>
>>>
>>> After this patch, a kernel function with true trace_location will be
>>> patched
>>> by bpf when CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS is disabled, which
>>> means
>>> that a kernel function may be patched by both bpf and ftrace in a mutually
>>> unaware way. This will cause ftrace and bpf_arch_text_poke to fail in a
>>> somewhat random way if the function to be patched was already patched
>>> by the other.
>>
>> Thanks for your review. And yes, this is a backward compatible solution
>> for architectures that not support DYNAMIC_FTRACE_WITH_DIRECT_CALLS.
> 
> It's not "backward compatible". Reiterating what Kuohai said; The BPF
> trampoline must be aware of ftrace's state -- with this patch, the
> trampoline can't blindly poke the text managed my ftrace.
> 
> I'd recommend to remove this patch from the series.
> 

After deep consideration, Kuohai's catching is much more reasonable. 
Will remove it in the next.

In the meantime, I found that song and guoren have worked on supporting 
riscv ftrace with direct call [0], so we can concentrate on making 
bpf_arch_text_poke specifically for the bpf context.

However, riscv ftrace base framework will change because [0] uses t0 as 
the link register of traced function. We should consider the generality 
of riscv bpf trampoline for kernel function and bpf context. It's not 
clear if [0] will be upstreamed, so maybe we should wait for it?

[0] 
https://lore.kernel.org/linux-riscv/20221208091244.203407-7-guoren@kernel.org

Anyway, thanks both of you for the review.

> 
> Björn