From: Vineet Gupta <vineet.gupta@linux.dev>
To: Alexei Starovoitov <alexei.starovoitov@gmail.com>, bpf@vger.kernel.org
Cc: daniel@iogearbox.net, andrii@kernel.org, eddyz87@gmail.com,
memxor@gmail.com, martin.lau@linux.dev, song@kernel.org,
yonghong.song@linux.dev, jolsa@kernel.org, emil@etsalapatis.com,
ihor.solodrai@linux.dev, john.fastabend@gmail.com,
shuah@kernel.org, linux-kernel@vger.kernel.org,
linux-kselftest@vger.kernel.org, jose.marchesi@oracle.com,
david.faust@oss.qualcomm.com
Subject: Re: [PATCH bpf-next v3 4/8] selftests/bpf: make arena_base() opaque to GCC's -Warray-bounds
Date: Mon, 2 Nov 2026 16:15:55 -0800 [thread overview]
Message-ID: <7c603b2b-6fcc-41d7-84e3-31e8af005776@linux.dev> (raw)
In-Reply-To: <DML32WAB5JZD.1LR3JA1OYY6A5@gmail.com>
On 11/2/26 2:38 PM, Alexei Starovoitov wrote:
> On Mon, Nov 02, 2026 at 02:07 PM Vineet Gupta <vineet.gupta@linux.dev> wrote:
>> The barrier is behind #ifdef __clang__ so clang keeps the plain macro.
>> It does not need it, and the barrier is not free: it blocks reuse of
>> the map address, so each call site reloads it. Across the five progs
>> and 20 call sites that is +35 instructions under GCC
>> (2528 -> 2563), about 1.4% on those objects and nothing elsewhere.
>> Clang output is unchanged, bit for bit.
> stream.c already has barrier_var() for both compilers.
>
> How about this one instead:
Yeah definitely much better. Just to be clear the intent of series is to
not make current compiler look better o anything. As I mentioned in the
cover, this was during in my in-works arena compiler testing, ....
Thx for the quick feedback.
-Vineet
> diff --git a/tools/testing/selftests/bpf/libarena/include/bpf_arena_common.h b/tools/testing/selftests/bpf/libarena/include/bpf_arena_common.h
> index 82aafe8..0eba529 100644
> --- a/tools/testing/selftests/bpf/libarena/include/bpf_arena_common.h
> +++ b/tools/testing/selftests/bpf/libarena/include/bpf_arena_common.h
> @@ -49,7 +49,16 @@ void __arena* bpf_arena_alloc_pages(void *map, void __arena *addr, __u32 page_cn
> int bpf_arena_reserve_pages(void *map, void __arena *addr, __u32 page_cnt) __ksym __weak;
> void bpf_arena_free_pages(void *map, void __arena *ptr, __u32 page_cnt) __ksym __weak;
>
> -#define arena_base(map) ((void __arena *)((struct bpf_arena *)(map))->user_vm_start)
> +/*
> + * The map definition is smaller than struct bpf_arena.
> + * barrier_var() hides that from -Warray-bounds of GCC.
> + */
> +#define arena_base(map) \
> + ({ \
> + struct bpf_arena *___arena = (void *)(map); \
> + barrier_var(___arena); \
> + (void __arena *)___arena->user_vm_start; \
> + })
>
> #else /* when compiled as user space code */
>
> diff --git a/tools/testing/selftests/bpf/progs/stream.c b/tools/testing/selftests/bpf/progs/stream.c
> index 1ae015c..b4d88d0 100644
> --- a/tools/testing/selftests/bpf/progs/stream.c
> +++ b/tools/testing/selftests/bpf/progs/stream.c
> @@ -178,16 +178,8 @@ __stderr("Call trace:\n"
> "|[ \t]+[^\n]+\n)*}}")
> int stream_arena_write_fault(void *ctx)
> {
> - struct bpf_arena *ptr = (void *)&arena;
> - u64 user_vm_start;
> + u64 user_vm_start = (u64)arena_base(&arena);
>
> - /* Prevent GCC bounds warning: casting &arena to struct bpf_arena *
> - * triggers bounds checking since the map definition is smaller than struct
> - * bpf_arena. barrier_var() makes the pointer opaque to GCC, preventing the
> - * bounds analysis
> - */
> - barrier_var(ptr);
> - user_vm_start = ptr->user_vm_start;
> fault_addr = user_vm_start + 0x7fff;
> bpf_addr_space_cast(user_vm_start, 0, 1);
> asm volatile (
> @@ -212,16 +204,8 @@ __stderr("Call trace:\n"
> "|[ \t]+[^\n]+\n)*}}")
> int stream_arena_read_fault(void *ctx)
> {
> - struct bpf_arena *ptr = (void *)&arena;
> - u64 user_vm_start;
> + u64 user_vm_start = (u64)arena_base(&arena);
>
> - /* Prevent GCC bounds warning: casting &arena to struct bpf_arena *
> - * triggers bounds checking since the map definition is smaller than struct
> - * bpf_arena. barrier_var() makes the pointer opaque to GCC, preventing the
> - * bounds analysis
> - */
> - barrier_var(ptr);
> - user_vm_start = ptr->user_vm_start;
> fault_addr = user_vm_start + 0x7fff;
> bpf_addr_space_cast(user_vm_start, 0, 1);
> asm volatile (
> @@ -252,17 +236,9 @@ int stream_arena_load_acquire_fault(void *ctx)
> .off = 0x7fff,
> .imm = 0x100, /* BPF_LOAD_ACQ */
> };
> - struct bpf_arena *ptr = (void *)&arena;
> - u64 user_vm_start, val;
> + u64 user_vm_start = (u64)arena_base(&arena);
> + u64 val;
>
> - /*
> - * Prevent GCC bounds warning: casting &arena to struct bpf_arena *
> - * triggers bounds checking since the map definition is smaller than
> - * struct bpf_arena. barrier_var() makes the pointer opaque to GCC,
> - * preventing the bounds analysis.
> - */
> - barrier_var(ptr);
> - user_vm_start = ptr->user_vm_start;
> fault_addr = user_vm_start + 0x7fff;
> bpf_addr_space_cast(user_vm_start, 0, 1);
> asm volatile (
> @@ -296,17 +272,9 @@ int stream_arena_xchg_fault(void *ctx)
> .off = 0x7fff,
> .imm = 0xe1, /* BPF_XCHG */
> };
> - struct bpf_arena *ptr = (void *)&arena;
> - u64 user_vm_start, val;
> + u64 user_vm_start = (u64)arena_base(&arena);
> + u64 val;
>
> - /*
> - * Prevent GCC bounds warning: casting &arena to struct bpf_arena *
> - * triggers bounds checking since the map definition is smaller than
> - * struct bpf_arena. barrier_var() makes the pointer opaque to GCC,
> - * preventing the bounds analysis.
> - */
> - barrier_var(ptr);
> - user_vm_start = ptr->user_vm_start;
> fault_addr = user_vm_start + 0x7fff;
> bpf_addr_space_cast(user_vm_start, 0, 1);
> /*
> @@ -346,17 +314,9 @@ int stream_arena_cmpxchg_fault(void *ctx)
> .off = 0x7fff,
> .imm = 0xf1, /* BPF_CMPXCHG */
> };
> - struct bpf_arena *ptr = (void *)&arena;
> - u64 user_vm_start, val;
> + u64 user_vm_start = (u64)arena_base(&arena);
> + u64 val;
>
> - /*
> - * Prevent GCC bounds warning: casting &arena to struct bpf_arena *
> - * triggers bounds checking since the map definition is smaller than
> - * struct bpf_arena. barrier_var() makes the pointer opaque to GCC,
> - * preventing the bounds analysis.
> - */
> - barrier_var(ptr);
> - user_vm_start = ptr->user_vm_start;
> fault_addr = user_vm_start + 0x7fff;
> bpf_addr_space_cast(user_vm_start, 0, 1);
> /*
next prev parent reply other threads:[~2026-11-03 0:16 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-02 22:07 [PATCH bpf-next v3 0/8] selftests/bpf: cleanup bpf-gcc runs Vineet Gupta
2026-11-02 22:07 ` [PATCH bpf-next v3 1/8] selftests/bpf: Build callx tests with GCC that defines __BPF_FEATURE_CALLX Vineet Gupta
2026-11-02 22:07 ` [PATCH bpf-next v3 2/8] selftests/bpf: don't pass clang-only warning flags to GCC Vineet Gupta
2026-11-02 22:07 ` [PATCH bpf-next v3 3/8] selftests/bpf: report SKIP for arena tests a compiler cannot build Vineet Gupta
2026-11-02 22:36 ` Alexei Starovoitov
2026-11-03 0:24 ` Vineet Gupta
2026-11-02 22:07 ` [PATCH bpf-next v3 4/8] selftests/bpf: make arena_base() opaque to GCC's -Warray-bounds Vineet Gupta
2026-11-02 22:38 ` Alexei Starovoitov
2026-11-03 0:15 ` Vineet Gupta [this message]
2026-11-02 22:07 ` [PATCH bpf-next v3 5/8] selftests/bpf: don't report arena tests as passing with an empty body Vineet Gupta
2026-11-02 22:07 ` [PATCH bpf-next v3 6/8] selftests/bpf: cast explicitly in the arena subprog and callback faults Vineet Gupta
2026-11-02 22:07 ` [PATCH bpf-next v3 7/8] selftests/bpf: write the mixed arena/scalar ALU tests in asm Vineet Gupta
2026-11-02 22:07 ` [PATCH bpf-next v3 8/8] selftests/bpf: skip the struct_ops arena tests without compiler support Vineet Gupta
2026-11-02 22:39 ` Alexei Starovoitov
2026-11-03 0:17 ` Vineet Gupta
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=7c603b2b-6fcc-41d7-84e3-31e8af005776@linux.dev \
--to=vineet.gupta@linux.dev \
--cc=alexei.starovoitov@gmail.com \
--cc=andrii@kernel.org \
--cc=bpf@vger.kernel.org \
--cc=daniel@iogearbox.net \
--cc=david.faust@oss.qualcomm.com \
--cc=eddyz87@gmail.com \
--cc=emil@etsalapatis.com \
--cc=ihor.solodrai@linux.dev \
--cc=john.fastabend@gmail.com \
--cc=jolsa@kernel.org \
--cc=jose.marchesi@oracle.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-kselftest@vger.kernel.org \
--cc=martin.lau@linux.dev \
--cc=memxor@gmail.com \
--cc=shuah@kernel.org \
--cc=song@kernel.org \
--cc=yonghong.song@linux.dev \
/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
all inboxes | Powered by JetHome®