From: Yonghong Song <yonghong.song@linux.dev>
To: Eduard Zingerman <eddyz87@gmail.com>,
bpf@vger.kernel.org, ast@kernel.org
Cc: andrii@kernel.org, daniel@iogearbox.net, martin.lau@linux.dev,
kernel-team@fb.com, hffilwlqm@gmail.com
Subject: Re: [PATCH bpf-next 2/4] selftests/bpf: utility function to get program disassembly after jit
Date: Tue, 13 Aug 2024 09:05:20 -0700 [thread overview]
Message-ID: <59186574-984c-4ccc-9861-27a9db15d2e6@linux.dev> (raw)
In-Reply-To: <20240809010518.1137758-3-eddyz87@gmail.com>
On 8/8/24 6:05 PM, Eduard Zingerman wrote:
> int get_jited_program_text(int fd, char *text, size_t text_sz)
>
> Loads and disassembles jited instructions for program pointed to by fd.
> Much like 'bpftool prog dump jited ...'.
>
> The code and makefile changes are inspired by jit_disasm.c from bpftool.
> Use llvm libraries to disassemble BPF program instead of libbfd to avoid
> issues with disassembly output stability pointed out in [1].
>
> Selftests makefile uses Makefile.feature to detect if LLVM libraries
> are available. If that is not the case selftests build proceeds but
> the function returns -ENOTSUP at runtime.
>
> [1] commit eb9d1acf634b ("bpftool: Add LLVM as default library for disassembling JIT-ed programs")
>
> Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
> ---
> tools/testing/selftests/bpf/.gitignore | 1 +
> tools/testing/selftests/bpf/Makefile | 51 +++-
> .../selftests/bpf/jit_disasm_helpers.c | 228 ++++++++++++++++++
> .../selftests/bpf/jit_disasm_helpers.h | 10 +
> 4 files changed, 288 insertions(+), 2 deletions(-)
> create mode 100644 tools/testing/selftests/bpf/jit_disasm_helpers.c
> create mode 100644 tools/testing/selftests/bpf/jit_disasm_helpers.h
>
[...]
> +static int disasm_one_func(FILE *text_out, uint8_t *image, __u32 len)
> +{
> + char *label, *colon, *triple = NULL;
> + LLVMDisasmContextRef ctx = NULL;
> + struct local_labels labels = {};
> + __u32 *label_pc, pc;
> + int i, cnt, err = 0;
> + char buf[256];
> +
> + triple = LLVMGetDefaultTargetTriple();
> + ctx = LLVMCreateDisasm(triple, &labels, 0, NULL, lookup_symbol);
> + if (!ASSERT_OK_PTR(ctx, "LLVMCreateDisasm")) {
> + err = -EINVAL;
> + goto out;
> + }
> +
> + cnt = LLVMSetDisasmOptions(ctx, LLVMDisassembler_Option_PrintImmHex);
> + if (!ASSERT_EQ(cnt, 1, "LLVMSetDisasmOptions")) {
> + err = -EINVAL;
> + goto out;
> + }
> +
> + /* discover labels */
> + labels.prog_len = len;
> + pc = 0;
> + while (pc < len) {
> + cnt = disasm_insn(ctx, image, len, pc, buf, 1);
> + if (cnt < 0) {
> + err = cnt;
> + goto out;
> + }
> + pc += cnt;
> + }
> + qsort(labels.pcs, labels.cnt, sizeof(*labels.pcs), cmp_u32);
> + /* GCC can't figure max bound for i and thus reports possible truncation */
> +#pragma GCC diagnostic push
> +#pragma GCC diagnostic ignored "-Wformat-truncation"
> + for (i = 0; i < labels.cnt; ++i)
> + snprintf(labels.names[i], sizeof(labels.names[i]), "L%d", i);
> +#pragma GCC diagnostic pop
"-Wformat-truncation" is only available for llvm >= 18. One of my build with llvm15
has the following warning/error:
jit_disasm_helpers.c:113:32: error: unknown warning group '-Wformat-truncation', ignored [-Werror,-Wunknown-warning-option]
#pragma GCC diagnostic ignored "-Wformat-truncation"
Maybe you want to guard with proper clang version?
Not sure on gcc side when "-Wformat-truncation" is supported.
> +
> + /* now print with labels */
> + labels.print_phase = true;
> + pc = 0;
> + while (pc < len) {
> + cnt = disasm_insn(ctx, image, len, pc, buf, sizeof(buf));
> + if (cnt < 0) {
> + err = cnt;
> + goto out;
> + }
> + label_pc = bsearch(&pc, labels.pcs, labels.cnt, sizeof(*labels.pcs), cmp_u32);
> + label = "";
> + colon = "";
> + if (label_pc) {
> + label = labels.names[label_pc - labels.pcs];
> + colon = ":";
> + }
> + fprintf(text_out, "%x:\t", pc);
> + for (i = 0; i < cnt; ++i)
> + fprintf(text_out, "%02x ", image[pc + i]);
> + for (i = cnt * 3; i < 12 * 3; ++i)
> + fputc(' ', text_out);
> + fprintf(text_out, "%s%s%s\n", label, colon, buf);
> + pc += cnt;
> + }
> +
> +out:
> + if (triple)
> + LLVMDisposeMessage(triple);
> + if (ctx)
> + LLVMDisasmDispose(ctx);
> + return err;
> +}
> +
[...]
next prev parent reply other threads:[~2024-08-13 16:05 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-08-09 1:05 [PATCH bpf-next 0/4] __jited_x86 test tag to check x86 assembly after jit Eduard Zingerman
2024-08-09 1:05 ` [PATCH bpf-next 1/4] selftests/bpf: less spam in the log for message matching Eduard Zingerman
2024-08-09 1:05 ` [PATCH bpf-next 2/4] selftests/bpf: utility function to get program disassembly after jit Eduard Zingerman
2024-08-13 16:05 ` Yonghong Song [this message]
2024-08-13 22:01 ` Eduard Zingerman
2024-08-15 19:27 ` Yonghong Song
2024-08-15 19:34 ` Eduard Zingerman
2024-08-15 21:06 ` Andrii Nakryiko
2024-08-15 21:50 ` Eduard Zingerman
2024-08-15 22:04 ` Andrii Nakryiko
2024-08-19 19:45 ` Eduard Zingerman
2024-08-19 21:05 ` Andrii Nakryiko
2024-08-09 1:05 ` [PATCH bpf-next 3/4] selftests/bpf: __jited_x86 test tag to check x86 assembly " Eduard Zingerman
2024-08-15 21:11 ` Andrii Nakryiko
2024-08-15 21:48 ` Eduard Zingerman
2024-08-09 1:05 ` [PATCH bpf-next 4/4] selftests/bpf: validate jit behaviour for tail calls Eduard Zingerman
2024-08-15 21:15 ` Andrii Nakryiko
2024-08-15 21:42 ` Eduard Zingerman
2024-08-15 22:07 ` Andrii Nakryiko
2024-08-15 22:10 ` Eduard Zingerman
2024-08-15 22:14 ` Andrii Nakryiko
2024-08-15 22:19 ` Eduard Zingerman
2024-08-15 21:32 ` Yonghong Song
2024-08-15 21:47 ` Eduard Zingerman
2024-08-15 22:09 ` Andrii Nakryiko
2024-08-15 22:16 ` Eduard Zingerman
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=59186574-984c-4ccc-9861-27a9db15d2e6@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=hffilwlqm@gmail.com \
--cc=kernel-team@fb.com \
--cc=martin.lau@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