From: Andrii Nakryiko <andrii.nakryiko@gmail.com>
To: Alexei Starovoitov <alexei.starovoitov@gmail.com>
Cc: "David S. Miller" <davem@davemloft.net>,
Daniel Borkmann <daniel@iogearbox.net>,
Andrii Nakryiko <andrii@kernel.org>,
john fastabend <john.fastabend@gmail.com>,
bpf <bpf@vger.kernel.org>, Kernel Team <kernel-team@fb.com>
Subject: Re: [PATCH v4 bpf-next 16/22] libbpf: Cleanup temp FDs when intermediate sys_bpf fails.
Date: Tue, 11 May 2021 16:34:05 -0700 [thread overview]
Message-ID: <CAEf4BzYPHBeK3vfwFm7oUru5Qb2MFq+sfnFV+=J-duevxXeryA@mail.gmail.com> (raw)
In-Reply-To: <20210508034837.64585-17-alexei.starovoitov@gmail.com>
On Fri, May 7, 2021 at 8:49 PM Alexei Starovoitov
<alexei.starovoitov@gmail.com> wrote:
>
> From: Alexei Starovoitov <ast@kernel.org>
>
> Fix loader program to close temporary FDs when intermediate
> sys_bpf command fails.
>
> Signed-off-by: Alexei Starovoitov <ast@kernel.org>
> ---
Looks good, but curious about 2 jumps vs 1 jump for cleanup
Acked-by: Andrii Nakryiko <andrii@kernel.org>
> tools/lib/bpf/bpf_gen_internal.h | 1 +
> tools/lib/bpf/gen_loader.c | 38 ++++++++++++++++++++++++++++----
> 2 files changed, 35 insertions(+), 4 deletions(-)
>
> diff --git a/tools/lib/bpf/bpf_gen_internal.h b/tools/lib/bpf/bpf_gen_internal.h
> index f42a55efd559..da2c026a3f31 100644
> --- a/tools/lib/bpf/bpf_gen_internal.h
> +++ b/tools/lib/bpf/bpf_gen_internal.h
> @@ -15,6 +15,7 @@ struct bpf_gen {
> void *data_cur;
> void *insn_start;
> void *insn_cur;
> + size_t cleanup_label;
> __u32 nr_progs;
> __u32 nr_maps;
> int log_level;
> diff --git a/tools/lib/bpf/gen_loader.c b/tools/lib/bpf/gen_loader.c
> index 585c672cc53e..b1709421ba90 100644
> --- a/tools/lib/bpf/gen_loader.c
> +++ b/tools/lib/bpf/gen_loader.c
> @@ -97,8 +97,36 @@ static void bpf_gen__emit2(struct bpf_gen *gen, struct bpf_insn insn1, struct bp
>
> void bpf_gen__init(struct bpf_gen *gen, int log_level)
> {
> + size_t stack_sz = sizeof(struct loader_stack);
> + int i;
> +
> gen->log_level = log_level;
> + /* save ctx pointer into R6 */
> bpf_gen__emit(gen, BPF_MOV64_REG(BPF_REG_6, BPF_REG_1));
> +
> + /* bzero stack */
> + bpf_gen__emit(gen, BPF_MOV64_REG(BPF_REG_1, BPF_REG_10));
> + bpf_gen__emit(gen, BPF_ALU64_IMM(BPF_ADD, BPF_REG_1, -stack_sz));
> + bpf_gen__emit(gen, BPF_MOV64_IMM(BPF_REG_2, stack_sz));
> + bpf_gen__emit(gen, BPF_MOV64_IMM(BPF_REG_3, 0));
> + bpf_gen__emit(gen, BPF_EMIT_CALL(BPF_FUNC_probe_read_kernel));
> +
> + /* jump over cleanup code */
> + bpf_gen__emit(gen, BPF_JMP_IMM(BPF_JA, 0, 0,
> + /* size of cleanup code below */
> + (stack_sz / 4) * 3 + 2));
> +
> + /* remember the label where all error branches will jump to */
> + gen->cleanup_label = gen->insn_cur - gen->insn_start;
> + /* emit cleanup code: close all temp FDs */
> + for (i = 0; i < stack_sz; i+= 4) {
nit: checkpatch complains about missing space before +=
> + bpf_gen__emit(gen, BPF_LDX_MEM(BPF_W, BPF_REG_1, BPF_REG_10, -stack_sz + i));
> + bpf_gen__emit(gen, BPF_JMP_IMM(BPF_JSLE, BPF_REG_1, 0, 1));
> + bpf_gen__emit(gen, BPF_EMIT_CALL(BPF_FUNC_sys_close));
> + }
> + /* R7 contains the error code from sys_bpf. Copy it into R0 and exit. */
> + bpf_gen__emit(gen, BPF_MOV64_REG(BPF_REG_0, BPF_REG_7));
> + bpf_gen__emit(gen, BPF_EXIT_INSN());
> }
>
> static int bpf_gen__add_data(struct bpf_gen *gen, const void *data, __u32 size)
> @@ -179,10 +207,12 @@ static void bpf_gen__emit_sys_bpf(struct bpf_gen *gen, int cmd, int attr, int at
>
> static void bpf_gen__emit_check_err(struct bpf_gen *gen)
> {
> - bpf_gen__emit(gen, BPF_JMP_IMM(BPF_JSGE, BPF_REG_7, 0, 2));
> - bpf_gen__emit(gen, BPF_MOV64_REG(BPF_REG_0, BPF_REG_7));
> - /* TODO: close intermediate FDs in case of error */
> - bpf_gen__emit(gen, BPF_EXIT_INSN());
> + /* R7 contains result of last sys_bpf command.
> + * if (R7 < 0) goto cleanup;
> + */
> + bpf_gen__emit(gen, BPF_JMP_IMM(BPF_JSGE, BPF_REG_7, 0, 1));
> + bpf_gen__emit(gen, BPF_JMP_IMM(BPF_JA, 0, 0,
> + -(gen->insn_cur - gen->insn_start - gen->cleanup_label) / 8 - 1));
Just curious, why not a single BPF_JSLT straight to the cleanup label?
> }
>
> /* reg1 and reg2 should not be R1 - R5. They can be R0, R6 - R10 */
> --
> 2.30.2
>
next prev parent reply other threads:[~2021-05-11 23:34 UTC|newest]
Thread overview: 49+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-05-08 3:48 [PATCH v4 bpf-next 00/22] bpf: syscall program, FD array, loader program, light skeleton Alexei Starovoitov
2021-05-08 3:48 ` [PATCH v4 bpf-next 01/22] bpf: Introduce bpf_sys_bpf() helper and program type Alexei Starovoitov
2021-05-08 3:48 ` [PATCH v4 bpf-next 02/22] bpf: Introduce bpfptr_t user/kernel pointer Alexei Starovoitov
2021-05-11 22:23 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 03/22] bpf: Prepare bpf syscall to be used from kernel and user space Alexei Starovoitov
2021-05-11 22:31 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 04/22] libbpf: Support for syscall program type Alexei Starovoitov
2021-05-08 3:48 ` [PATCH v4 bpf-next 05/22] selftests/bpf: Test " Alexei Starovoitov
2021-05-11 22:36 ` Andrii Nakryiko
2021-05-11 22:44 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 06/22] bpf: Make btf_load command to be bpfptr_t compatible Alexei Starovoitov
2021-05-11 22:41 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 07/22] selftests/bpf: Test for btf_load command Alexei Starovoitov
2021-05-11 22:45 ` Andrii Nakryiko
2021-05-12 4:04 ` Alexei Starovoitov
2021-05-12 18:00 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 08/22] bpf: Introduce fd_idx Alexei Starovoitov
2021-05-11 22:47 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 09/22] libbpf: Support for fd_idx Alexei Starovoitov
2021-05-11 22:57 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 10/22] bpf: Add bpf_btf_find_by_name_kind() helper Alexei Starovoitov
2021-05-11 23:02 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 11/22] bpf: Add bpf_sys_close() helper Alexei Starovoitov
2021-05-08 3:48 ` [PATCH v4 bpf-next 12/22] libbpf: Change the order of data and text relocations Alexei Starovoitov
2021-05-11 23:06 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 13/22] libbpf: Add bpf_object pointer to kernel_supports() Alexei Starovoitov
2021-05-08 3:48 ` [PATCH v4 bpf-next 14/22] libbpf: Generate loader program out of BPF ELF file Alexei Starovoitov
2021-05-11 23:22 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 15/22] libbpf: Use fd_array only with gen_loader Alexei Starovoitov
2021-05-11 23:24 ` Andrii Nakryiko
2021-05-12 4:17 ` Alexei Starovoitov
2021-05-12 18:11 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 16/22] libbpf: Cleanup temp FDs when intermediate sys_bpf fails Alexei Starovoitov
2021-05-11 23:34 ` Andrii Nakryiko [this message]
2021-05-12 4:33 ` Alexei Starovoitov
2021-05-08 3:48 ` [PATCH v4 bpf-next 17/22] libbpf: Introduce bpf_map__get_initial_value() Alexei Starovoitov
2021-05-11 23:39 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 18/22] bpftool: Use syscall/loader program in "prog load" and "gen skeleton" command Alexei Starovoitov
2021-05-12 4:17 ` Andrii Nakryiko
2021-05-12 18:43 ` Alexei Starovoitov
2021-05-12 18:55 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 19/22] selftests/bpf: Convert few tests to light skeleton Alexei Starovoitov
2021-05-12 4:19 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 20/22] selftests/bpf: Convert atomics test " Alexei Starovoitov
2021-05-12 4:21 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 21/22] selftests/bpf: Convert test printk to use rodata Alexei Starovoitov
2021-05-12 4:23 ` Andrii Nakryiko
2021-05-08 3:48 ` [PATCH v4 bpf-next 22/22] selftests/bpf: Convert test trace_printk to lskel Alexei Starovoitov
2021-05-12 4:24 ` Andrii Nakryiko
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='CAEf4BzYPHBeK3vfwFm7oUru5Qb2MFq+sfnFV+=J-duevxXeryA@mail.gmail.com' \
--to=andrii.nakryiko@gmail.com \
--cc=alexei.starovoitov@gmail.com \
--cc=andrii@kernel.org \
--cc=bpf@vger.kernel.org \
--cc=daniel@iogearbox.net \
--cc=davem@davemloft.net \
--cc=john.fastabend@gmail.com \
--cc=kernel-team@fb.com \
/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;
as well as URLs for NNTP newsgroup(s).