| 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
| ||
|
Message-ID: <13cd161b-43a2-ce66-6a27-6662fc36e063@huawei.com>
Date: Sun, 24 Apr 2022 13:05:43 +0800
From: Xu Kuohai <xukuohai@...wei.com>
To: Jakub Sitnicki <jakub@...udflare.com>
CC: <bpf@...r.kernel.org>, <linux-arm-kernel@...ts.infradead.org>,
<linux-kernel@...r.kernel.org>, <netdev@...r.kernel.org>,
<linux-kselftest@...r.kernel.org>,
Catalin Marinas <catalin.marinas@....com>,
Will Deacon <will@...nel.org>,
Steven Rostedt <rostedt@...dmis.org>,
Ingo Molnar <mingo@...hat.com>,
Daniel Borkmann <daniel@...earbox.net>,
Alexei Starovoitov <ast@...nel.org>,
Zi Shen Lim <zlim.lnx@...il.com>,
Andrii Nakryiko <andrii@...nel.org>,
Martin KaFai Lau <kafai@...com>,
Song Liu <songliubraving@...com>, Yonghong Song <yhs@...com>,
John Fastabend <john.fastabend@...il.com>,
KP Singh <kpsingh@...nel.org>,
"David S . Miller" <davem@...emloft.net>,
Hideaki YOSHIFUJI <yoshfuji@...ux-ipv6.org>,
David Ahern <dsahern@...nel.org>,
Thomas Gleixner <tglx@...utronix.de>,
Borislav Petkov <bp@...en8.de>,
Dave Hansen <dave.hansen@...ux.intel.com>, <x86@...nel.org>,
<hpa@...or.com>, Shuah Khan <shuah@...nel.org>,
Mark Rutland <mark.rutland@....com>,
Ard Biesheuvel <ardb@...nel.org>,
Pasha Tatashin <pasha.tatashin@...een.com>,
Peter Collingbourne <pcc@...gle.com>,
Daniel Kiss <daniel.kiss@....com>,
Sudeep Holla <sudeep.holla@....com>,
Steven Price <steven.price@....com>,
Marc Zyngier <maz@...nel.org>, Mark Brown <broonie@...nel.org>,
Kumar Kartikeya Dwivedi <memxor@...il.com>,
Delyan Kratunov <delyank@...com>, <kernel-team@...udflare.com>
Subject: Re: [PATCH bpf-next v2 4/6] bpf, arm64: Impelment
bpf_arch_text_poke() for arm64
On 4/22/2022 6:54 PM, Jakub Sitnicki wrote:
> Hi Xu,
>
> Thanks for working on this.
>
> We are also looking forward to using fentry hooks on arm64.
> In particular, attaching to entry/exit into/from XDP progs.
>
> On Thu, Apr 14, 2022 at 12:22 PM -04, Xu Kuohai wrote:
>> Impelment bpf_arch_text_poke() for arm64, so bpf trampoline code can use
>> it to replace nop with jump, or replace jump with nop.
>>
>> Signed-off-by: Xu Kuohai <xukuohai@...wei.com>
>> Acked-by: Song Liu <songliubraving@...com>
>> ---
>> arch/arm64/net/bpf_jit_comp.c | 52 +++++++++++++++++++++++++++++++++++
>> 1 file changed, 52 insertions(+)
>>
>> diff --git a/arch/arm64/net/bpf_jit_comp.c b/arch/arm64/net/bpf_jit_comp.c
>> index 8ab4035dea27..1a1c3ea75ee2 100644
>> --- a/arch/arm64/net/bpf_jit_comp.c
>> +++ b/arch/arm64/net/bpf_jit_comp.c
>> @@ -9,6 +9,7 @@
>>
>> #include <linux/bitfield.h>
>> #include <linux/bpf.h>
>> +#include <linux/memory.h>
>> #include <linux/filter.h>
>> #include <linux/printk.h>
>> #include <linux/slab.h>
>> @@ -18,6 +19,7 @@
>> #include <asm/cacheflush.h>
>> #include <asm/debug-monitors.h>
>> #include <asm/insn.h>
>> +#include <asm/patching.h>
>> #include <asm/set_memory.h>
>>
>> #include "bpf_jit.h"
>> @@ -1529,3 +1531,53 @@ void bpf_jit_free_exec(void *addr)
>> {
>> return vfree(addr);
>> }
>> +
>> +static int gen_branch_or_nop(enum aarch64_insn_branch_type type, void *ip,
>> + void *addr, u32 *insn)
>> +{
>> + if (!addr)
>> + *insn = aarch64_insn_gen_nop();
>> + else
>> + *insn = aarch64_insn_gen_branch_imm((unsigned long)ip,
>> + (unsigned long)addr,
>> + type);
>> +
>> + return *insn != AARCH64_BREAK_FAULT ? 0 : -EFAULT;
>> +}
>> +
>> +int bpf_arch_text_poke(void *ip, enum bpf_text_poke_type poke_type,
>> + void *old_addr, void *new_addr)
>> +{
>> + int ret;
>> + u32 old_insn;
>> + u32 new_insn;
>> + u32 replaced;
>> + enum aarch64_insn_branch_type branch_type;
>> +
>> + if (poke_type == BPF_MOD_CALL)
>> + branch_type = AARCH64_INSN_BRANCH_LINK;
>
> This path, bpf_arch_text_poke(<ip>, BPF_MOD_CALL, ...), is what we hit
> when attaching a BPF program entry. It is exercised by selftest #232
> xdp_bpf2bpf.
>
> However, with this patchset alone it will not work because we don't
> emit, yet, the ftrace patch (MOV X9, LR; NOP) as a part of BPF prog
> prologue, like ftrace_init_nop() does. So patching attempt will fail.
>
> I think that is what you mentioned to in your reply to Hou [1]
>
> So my question is - is support for attaching to BPF progs in scope for
> this patchset?
>
> If no, then perhaps it would be better for now to fail early with
> something like -EOPNOTSUPP when poke_type is BPF_MOD_CALL, rather then
> attempt to patch the code.
>
> If you plan to enable it as a part of this patchset, then I've given it
> a quick try, and it seems that not a lot is needed get fentry to BPF
> attachment to work.
>
> I'm including the diff for my quick and dirty attempt below. With that
> patch on top, the xdp_bpf2bpf tests pass:
>
> #232 xdp_bpf2bpf:OK
>
> [1] https://lore.kernel.org/bpf/d8c4f1fb-a020-9457-44e2-dc63982a9213@huawei.com/
>
Hi Jakub,
Thanks for your testing and suggestion! I added bpf2bpf poking to this
series and rebased it to [2] a few days ago, so there are some conflicts
with the bpf-next branch. I'll rebase it to bpf-next and send v3.
[2] https://lore.kernel.org/bpf/20220416042940.656344-1-kuifeng@fb.com/
>> + else
>> + branch_type = AARCH64_INSN_BRANCH_NOLINK;
>> +
>> + if (gen_branch_or_nop(branch_type, ip, old_addr, &old_insn) < 0)
>> + return -EFAULT;
>> +
>> + if (gen_branch_or_nop(branch_type, ip, new_addr, &new_insn) < 0)
>> + return -EFAULT;
>> +
>> + mutex_lock(&text_mutex);
>> + if (aarch64_insn_read(ip, &replaced)) {
>> + ret = -EFAULT;
>> + goto out;
>> + }
>> +
>> + if (replaced != old_insn) {
>> + ret = -EFAULT;
>> + goto out;
>> + }
>> +
>> + ret = aarch64_insn_patch_text_nosync((void *)ip, new_insn);
>> +out:
>> + mutex_unlock(&text_mutex);
>
> The body of this critical section is identical as ftrace_modify_code().
> Perhaps we could export it and reuse?
>
ftrace_modify_code() is defined in the arch code, and the prototypes are
not consistent across archs, so it doesn't seem appropriate to export
ftrace_modify_code() as a public interface.
>> + return ret;
>> +}
>
> ---
> diff --git a/arch/arm64/net/bpf_jit_comp.c b/arch/arm64/net/bpf_jit_comp.c
> index 5f6bd755050f..94d8251500ab 100644
> --- a/arch/arm64/net/bpf_jit_comp.c
> +++ b/arch/arm64/net/bpf_jit_comp.c
> @@ -240,9 +240,9 @@ static bool is_lsi_offset(int offset, int scale)
> /* Tail call offset to jump into */
> #if IS_ENABLED(CONFIG_ARM64_BTI_KERNEL) || \
> IS_ENABLED(CONFIG_ARM64_PTR_AUTH_KERNEL)
> -#define PROLOGUE_OFFSET 9
> +#define PROLOGUE_OFFSET 11
> #else
> -#define PROLOGUE_OFFSET 8
> +#define PROLOGUE_OFFSET 10
> #endif
>
> static int build_prologue(struct jit_ctx *ctx, bool ebpf_from_cbpf)
> @@ -281,6 +281,10 @@ static int build_prologue(struct jit_ctx *ctx, bool ebpf_from_cbpf)
> *
> */
>
> + /* Set up ftrace patch (initially in disabled state) */
> + emit(A64_MOV(1, A64_R(9), A64_LR), ctx);
> + emit(A64_NOP, ctx);
> > /* Sign lr */
> if (IS_ENABLED(CONFIG_ARM64_PTR_AUTH_KERNEL))
> emit(A64_PACIASP, ctx);
> @@ -1888,10 +1892,16 @@ int bpf_arch_text_poke(void *ip, enum bpf_text_poke_type poke_type,
> u32 replaced;
> enum aarch64_insn_branch_type branch_type;
>
> - if (poke_type == BPF_MOD_CALL)
> + if (poke_type == BPF_MOD_CALL) {
> branch_type = AARCH64_INSN_BRANCH_LINK;
> - else
> + /*
> + * Adjust addr to point at the BL in the callsite.
> + * See ftrace_init_nop() for the callsite sequence.
> + */
> + ip = (void *)((unsigned long)ip + AARCH64_INSN_SIZE);
> + } else {
> branch_type = AARCH64_INSN_BRANCH_NOLINK;
> + }
>
> if (gen_branch_or_nop(branch_type, ip, old_addr, &old_insn) < 0)
> return -EFAULT;
> .
Powered by blists - more mailing lists