From: Alexei Starovoitov <alexei.starovoitov@gmail.com>
To: Dave Thaler <dthaler@microsoft.com>,
Daniel Borkmann <daniel@iogearbox.net>
Cc: "bpf@vger.kernel.org" <bpf@vger.kernel.org>,
Johan Almbladh <johan.almbladh@anyfinetworks.com>
Subject: div_k. Was: [PATCH 07/15] ebpf-docs: Fix modulo zero, division by zero, overflow, and underflow
Date: Tue, 4 Oct 2022 10:24:57 -0700 [thread overview]
Message-ID: <CAADnVQKYb9fB2-AcPB0sMbE_pT2cQhOiKQw0q5rPxXHZ-6eXTw@mail.gmail.com> (raw)
In-Reply-To: <DM4PR21MB34409A021A6658DDAAB3B5AEA35A9@DM4PR21MB3440.namprd21.prod.outlook.com>
On Tue, Oct 4, 2022 at 9:36 AM Dave Thaler <dthaler@microsoft.com> wrote:
> > Those differences are in signed div/mod only, right?
> > Unsigned div/mod doesn't have it, right?
> > bpf has only unsigned div/mod.
>
> Ah right, will replace. However since imm is a signed integer, that leaves
> an ambiguity that is important to clarify.
>
> What is the expected value for the following 64-bit BPF_DIV operation:
> r0 = 0xFFFFFFFFFFFFFFFF
> r0 /= -10
> Is it 0x1 or 0x10000000a? i.e., is the -10 sign extended to
> 0xFFFFFFFFFFFFFFF6 or treated as 0xFFFFFFF6 when doing the unsigned
> division?
x86 and arm64 JITs treat it as imm32 is zero extended.
But looking at the interpreter:
ALU64_DIV_K:
DST = div64_u64(DST, IMM);
it looks like we have a bug there.
But we have a bunch of div_k tests in lib/test_bpf.c
including negative imm32. Hmm.
next prev parent reply other threads:[~2022-10-04 17:25 UTC|newest]
Thread overview: 50+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-09-27 18:59 [PATCH 01/15] ebpf-docs: Move legacy packet instructions to a separate file dthaler1968
2022-09-27 18:59 ` [PATCH 02/15] ebpf-docs: Linux byteswap note dthaler1968
2022-09-27 18:59 ` [PATCH 03/15] ebpf-docs: Move Clang notes to a separate file dthaler1968
2022-09-27 18:59 ` [PATCH 04/15] ebpf-docs: Add Clang note about BPF_ALU dthaler1968
2022-09-27 18:59 ` [PATCH 05/15] ebpf-docs: Add TOC and fix formatting dthaler1968
2022-09-27 18:59 ` [PATCH 06/15] ebpf-docs: Use standard type convention in standard doc dthaler1968
2022-09-30 20:49 ` Alexei Starovoitov
2022-09-27 18:59 ` [PATCH 07/15] ebpf-docs: Fix modulo zero, division by zero, overflow, and underflow dthaler1968
2022-09-30 20:52 ` Alexei Starovoitov
2022-09-30 21:54 ` Dave Thaler
2022-09-30 21:59 ` Alexei Starovoitov
2022-09-30 22:41 ` Dave Thaler
2022-09-30 23:41 ` Alexei Starovoitov
2022-10-04 16:36 ` Dave Thaler
2022-10-04 17:24 ` Alexei Starovoitov [this message]
2022-10-04 18:23 ` div_k. Was: " Dave Thaler
2022-10-04 18:34 ` Alexei Starovoitov
2023-09-29 21:03 ` Signed modulo operations Dave Thaler
2023-09-29 21:03 ` [Bpf] " Dave Thaler
2023-09-30 6:07 ` Carsten Bormann
2023-09-30 6:07 ` Carsten Bormann
2023-09-30 14:48 ` Alexei Starovoitov
2023-09-30 14:48 ` Alexei Starovoitov
2023-10-02 13:19 ` Eduard Zingerman
2022-09-27 18:59 ` [PATCH 08/15] ebpf-docs: Use consistent names for the same field dthaler1968
2022-09-30 20:57 ` Alexei Starovoitov
2022-10-04 14:44 ` Dave Thaler
2022-09-27 18:59 ` [PATCH 09/15] ebpf-docs: Explain helper functions dthaler1968
2022-09-30 22:01 ` Alexei Starovoitov
2022-09-30 23:01 ` Dave Thaler
2022-09-27 18:59 ` [PATCH 10/15] ebpf-docs: Add appendix of all opcodes in order dthaler1968
2022-09-30 22:02 ` Alexei Starovoitov
2022-09-30 22:43 ` Dave Thaler
2022-09-27 18:59 ` [PATCH 11/15] ebpf-docs: Improve English readability dthaler1968
2022-09-30 22:16 ` Alexei Starovoitov
2022-10-04 14:32 ` Dave Thaler
2022-10-04 15:38 ` Alexei Starovoitov
2022-10-04 15:55 ` Dave Thaler
2022-10-04 15:56 ` Dave Thaler
2022-10-04 16:19 ` Alexei Starovoitov
2022-10-04 16:41 ` Dave Thaler
2022-10-04 16:54 ` Dave Thaler
2022-10-06 20:44 ` Jim Harris
2022-09-27 18:59 ` [PATCH 12/15] ebpf-docs: Add Linux note about register calling convention dthaler1968
2022-09-30 22:17 ` Alexei Starovoitov
2022-09-27 18:59 ` [PATCH 13/15] ebpf-docs: Add extended 64-bit immediate instructions dthaler1968
2022-09-27 18:59 ` [PATCH 14/15] ebpf-docs: Add extended call instructions dthaler1968
2022-09-27 18:59 ` [PATCH 15/15] ebpf-docs: Add note about invalid instruction dthaler1968
2022-09-30 22:21 ` Alexei Starovoitov
2022-09-30 20:50 ` [PATCH 01/15] ebpf-docs: Move legacy packet instructions to a separate file patchwork-bot+netdevbpf
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=CAADnVQKYb9fB2-AcPB0sMbE_pT2cQhOiKQw0q5rPxXHZ-6eXTw@mail.gmail.com \
--to=alexei.starovoitov@gmail.com \
--cc=bpf@vger.kernel.org \
--cc=daniel@iogearbox.net \
--cc=dthaler@microsoft.com \
--cc=johan.almbladh@anyfinetworks.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.