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: <CAEf4BzY6vKQ_or+EDq3zKb=HbUu+C6Ti4pTphOeqsNoQhL5m9w@mail.gmail.com>
Date: Wed, 20 Aug 2025 16:24:47 -0700
From: Andrii Nakryiko <andrii.nakryiko@...il.com>
To: Jiawei Zhao <phoenix500526@....com>
Cc: ast@...nel.org, daniel@...earbox.net, andrii@...nel.org, 
	yonghong.song@...ux.dev, bpf@...r.kernel.org, linux-kselftest@...r.kernel.org, 
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH bpf-next v11 2/3] selftests/bpf: Add an usdt_o1 test case
 in selftests to cover SIB handling logic

On Mon, Aug 18, 2025 at 12:46 AM Jiawei Zhao <phoenix500526@....com> wrote:
>
> When using GCC on x86-64 to compile an usdt prog with -O1 or higher
> optimization, the compiler will generate SIB addressing mode for global
> array and PC-relative addressing mode for global variable,
> e.g. "1@-96(%rbp,%rax,8)" and "-1@...1(%rip)".
>
> In this patch:
> - add usdt_o1 test case to cover SIB addressing usdt argument spec
>   handling logic
>
> Signed-off-by: Jiawei Zhao <phoenix500526@....com>
> ---
>  .../selftests/bpf/prog_tests/usdt_o1.c        | 70 +++++++++++++++++++
>  .../selftests/bpf/progs/test_usdt_o1.c        | 37 ++++++++++
>  2 files changed, 107 insertions(+)
>  create mode 100644 tools/testing/selftests/bpf/prog_tests/usdt_o1.c
>  create mode 100644 tools/testing/selftests/bpf/progs/test_usdt_o1.c
>
> diff --git a/tools/testing/selftests/bpf/prog_tests/usdt_o1.c b/tools/testing/selftests/bpf/prog_tests/usdt_o1.c
> new file mode 100644
> index 000000000000..706168e804cb
> --- /dev/null
> +++ b/tools/testing/selftests/bpf/prog_tests/usdt_o1.c
> @@ -0,0 +1,70 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/* Copyright (c) 2025 Jiawei Zhao <phoenix500526@....com>. */
> +#include <test_progs.h>
> +
> +#include "../sdt.h"
> +#include "test_usdt_o1.skel.h"
> +
> +#if (defined(__GNUC__) && !defined(__clang__))
> +#pragma GCC optimize("O1")

that optimize attribute is a function attribute, while your earlier
version just randomly put it here. Have you tried using that optimize
attribute on a function properly?

One of the points of that attribute suggestion was to avoid the need
to add another file in prog_tests and keeping various USDT test cases
in one test, so can you give that attribute another try and see if it
works?

> +#else
> +#pragma message("non-gcc compiler: the correct probes might not be installed")

hm... non-gcc is expected, so I don't think we should pollute compiler
output with extra messages

> +#endif
> +
> +
> +#define test_value 0xFEDCBA9876543210ULL
> +#define SEC(name) __attribute__((section(name), used))
> +
> +int lets_test_this(int);
> +static volatile __u64 array[1] = {test_value};
> +
> +static __always_inline void trigger_func(void)
> +{
> +       /* Base address + offset + (index * scale) */
> +       for (volatile int i = 0; i <= 0; i++)
> +               STAP_PROBE1(test, usdt1, array[i]);

I think I saw a bunch of mentions that getting desired SIB form for
USDT arg was a challenge, right? Have you checked STAP_PROBE_ASM from
sdt.h? That will allow to define USDT from inside assembly block,
which should give you a lot of control over the shape of the argument.


[...]

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ