[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAEf4BzbLPxNo--P7mCxS_miagFHF4fyoec1VOkpL=uY=oNqpVg@mail.gmail.com>
Date: Thu, 20 May 2021 10:23:01 -0700
From: Andrii Nakryiko <andrii.nakryiko@...il.com>
To: Pavel Begunkov <asml.silence@...il.com>
Cc: io-uring@...r.kernel.org, Networking <netdev@...r.kernel.org>,
bpf <bpf@...r.kernel.org>,
open list <linux-kernel@...r.kernel.org>,
Jens Axboe <axboe@...nel.dk>,
Alexei Starovoitov <ast@...nel.org>,
Daniel Borkmann <daniel@...earbox.net>,
Andrii Nakryiko <andrii@...nel.org>,
Martin KaFai Lau <kafai@...com>,
Song Liu <songliubraving@...com>, Yonghong Song <yhs@...com>,
John Fastabend <john.fastabend@...il.com>,
KP Singh <kpsingh@...nel.org>,
Horst Schirmeier <horst.schirmeier@...dortmund.de>,
"Franz-B . Tuneke" <franz-bernhard.tuneke@...dortmund.de>,
Christian Dietrich <stettberger@...ucode.de>
Subject: Re: [PATCH 18/23] libbpf: support io_uring
On Thu, May 20, 2021 at 2:58 AM Pavel Begunkov <asml.silence@...il.com> wrote:
>
> On 5/19/21 6:38 PM, Andrii Nakryiko wrote:
> > On Wed, May 19, 2021 at 7:14 AM Pavel Begunkov <asml.silence@...il.com> wrote:
> >>
> >> Signed-off-by: Pavel Begunkov <asml.silence@...il.com>
> >> ---
> >> tools/lib/bpf/libbpf.c | 7 +++++++
> >> 1 file changed, 7 insertions(+)
> >>
> >> diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
> >> index 4181d178ee7b..de5d1508f58e 100644
> >> --- a/tools/lib/bpf/libbpf.c
> >> +++ b/tools/lib/bpf/libbpf.c
> >> @@ -13,6 +13,10 @@
> >> #ifndef _GNU_SOURCE
> >> #define _GNU_SOURCE
> >> #endif
> >> +
> >> +/* hack, use local headers instead of system-wide */
> >> +#include "../../../include/uapi/linux/bpf.h"
> >> +
> >
> > libbpf is already using the latest UAPI headers, so you don't need
> > this hack. You just haven't synced include/uapi/linux/bpf.h into
> > tools/include/uapi/linux/bpf.h
>
> It's more convenient to keep it local to me while RFC, surely will
> drop it later.
>
> btw, I had a problem with find_sec_def() successfully matching
> "iouring.s" string with "iouring", because section_defs[i].len
> doesn't include final \0 and so does a sort of prefix comparison.
> That's why "iouring/". Can we fix it? Are compatibility concerns?
If you put "iouring.s" before "iouring" it will be matched first,
libbpf matches them in order, so more specific prefix should go first.
It is currently always treated as a prefix, not exact match,
unfortunately. I have a work planned to revamp this logic quite a bit
for libbpf 1.0, so this should be improved as part of that work.
>
> >
> >> #include <stdlib.h>
> >> #include <stdio.h>
> >> #include <stdarg.h>
> >> @@ -8630,6 +8634,9 @@ static const struct bpf_sec_def section_defs[] = {
> >> BPF_PROG_SEC("struct_ops", BPF_PROG_TYPE_STRUCT_OPS),
> >> BPF_EAPROG_SEC("sk_lookup/", BPF_PROG_TYPE_SK_LOOKUP,
> >> BPF_SK_LOOKUP),
> >> + SEC_DEF("iouring/", IOURING),
> >> + SEC_DEF("iouring.s/", IOURING,
> >> + .is_sleepable = true),
> >> };
> >>
> >> #undef BPF_PROG_SEC_IMPL
> >> --
> >> 2.31.1
> >>
>
> --
> Pavel Begunkov
Powered by blists - more mailing lists