From: Jiri Olsa <olsajiri@gmail.com>
To: Oleg Nesterov <oleg@redhat.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>,
Peter Zijlstra <peterz@infradead.org>,
Andrii Nakryiko <andrii@kernel.org>,
bpf@vger.kernel.org, linux-kernel@vger.kernel.org,
linux-trace-kernel@vger.kernel.org, x86@kernel.org,
Song Liu <songliubraving@fb.com>, Yonghong Song <yhs@fb.com>,
John Fastabend <john.fastabend@gmail.com>,
Steven Rostedt <rostedt@goodmis.org>,
Ingo Molnar <mingo@kernel.org>,
David Laight <David.Laight@aculab.com>
Subject: Re: [RFC PATCH 5/8] uprobe/x86: Add support to optimize on top of emulated instructions
Date: Wed, 26 Nov 2025 08:54:18 +0100 [thread overview]
Message-ID: <aSayKtsTNkuyu0TP@krava> (raw)
In-Reply-To: <aSSdavSy_unRaEgF@redhat.com>
On Mon, Nov 24, 2025 at 07:01:14PM +0100, Oleg Nesterov wrote:
> Hi Jiri,
>
> I am trying to understand this series, will try to read it more carefully
> later...
>
> (damn why do you always send the patches when I am on PTO? ;)
it's more fun that way ;-) thanks for checking on it
>
> On 11/17, Jiri Olsa wrote:
> >
> > struct arch_uprobe {
> > union {
> > - u8 insn[MAX_UINSN_BYTES];
> > + u8 insn[5*MAX_UINSN_BYTES];
>
> Hmm. OK, this matches the "for (i = 0; i < 5; i++)" loop in
> opt_setup_xol_ops(), but do we really need this change? Please see
> the question at the end.
>
> > +static int opt_setup_xol_ops(struct arch_uprobe *auprobe, struct insn *insn)
> > +{
> > + unsigned long offset = insn->length;
> > + struct insn insnX;
> > + int i, ret;
> > +
> > + if (test_bit(ARCH_UPROBE_FLAG_CAN_OPTIMIZE, &auprobe->flags))
> > + return -ENOSYS;
>
> I think this logic needs some cleanups... If ARCH_UPROBE_FLAG_CAN_OPTIMIZE
> is set by the caller, the it doesn't make sense to call xxx_setup_xol_ops(),
> right? But lets forget it for now.
>
> > + ret = opt_setup_xol_insns(auprobe, &auprobe->opt.xol[0], insn);
>
> I think this should go into the main loop, see below
>
> > + for (i = 1; i < 5; i++) {
> > + ret = uprobe_init_insn_offset(auprobe, offset, &insnX, true);
> > + if (ret)
> > + break;
> > + ret = opt_setup_xol_insns(auprobe, &auprobe->opt.xol[i], &insnX);
> > + if (ret)
> > + break;
> > + offset += insnX.length;
> > + auprobe->opt.cnt++;
> > + if (offset >= 5)
> > + goto optimize;
> > + }
> > +
> > + return -ENOSYS;
>
> I don't think -ENOSYS makes sense if opt_setup_xol_insns() succeeds at least once.
> IOW, how about
>
> static int opt_setup_xol_ops(struct arch_uprobe *auprobe, struct insn *insn)
> {
> unsigned long offset = 0;
> struct insn insnX;
> int i, ret;
>
> if (test_bit(ARCH_UPROBE_FLAG_CAN_OPTIMIZE, &auprobe->flags))
> return -ENOSYS;
>
> for (i = 0; i < 5; i++) {
> ret = opt_setup_xol_insns(auprobe, &auprobe->opt.xol[i], insn);
> if (ret)
> break;
> offset += insn->length;
> if (offset >= 5)
> break;
>
> insn = &insnX;
> ret = uprobe_init_insn_offset(auprobe, offset, insn, true);
> if (ret)
> break;
> }
>
> if (!offset)
> return -ENOSYS;
>
> if (offset >= 5) {
> auprobe->opt.cnt = i + 1;
> auprobe->xol.ops = &opt_xol_ops;
> set_bit(ARCH_UPROBE_FLAG_CAN_OPTIMIZE, &auprobe->flags);
> set_bit(ARCH_UPROBE_FLAG_OPTIMIZE_EMULATE, &auprobe->flags);
> }
>
> return 0;
> }
>
> ?
>
> This way the caller, arch_uprobe_analyze_insn(), doesn't need to call
> push/mov/sub/_setup_xol_ops(), and the code looks a bit simpler to me.
ah nice, will try that
>
> No?
>
> > + * TODO perhaps we could 'emulate' nop, so there would be no need for
> > + * ARCH_UPROBE_FLAG_OPTIMIZE_EMULATE flag, because we would emulate
> > + * allways.
>
> Agreed... and this connects to "this logic needs some cleanups" above.
> I guess we need nop_setup_xol_ops() extracted from branch_setup_xol_ops()
> but again, lets forget it for now.
ok, it will hopefully make the code simpler, will check on that
>
> -------------------------------------------------------------------------------
> Now the main question. What if we avoid this change
>
> - u8 insn[MAX_UINSN_BYTES];
> + u8 insn[5*MAX_UINSN_BYTES];
>
> mentioned above, and change opt_setup_xol_ops() to just do
>
> - for (i = 0; i < 5; i++)
> + for (i = 0;; i++)
>
> ?
>
> The main loop stops when offset >= 5 anyway.
>
> And. if auprobe->insn[offset:MAX_UINSN_BYTES] doesn't contain a full/valid
> insn at the start, then uprobe_init_insn_offset()->insn_decode() should fail?
>
> Most probably I missed something, but I can't understand this part.
no, I think you're right, I did not realize we fit under MAX_UINSN_BYTES
anyway, call instruction needs only 5 bytes
thanks,
jirka
next prev parent reply other threads:[~2025-11-26 7:54 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-11-17 12:40 [RFC PATCH 0/8] uprobe/x86: Add support to optimize prologue Jiri Olsa
2025-11-17 12:40 ` [RFC PATCH 1/8] uprobe/x86: Introduce struct arch_uprobe_xol object Jiri Olsa
2025-11-17 12:40 ` [RFC PATCH 2/8] uprobe/x86: Use struct arch_uprobe_xol in emulate callback Jiri Olsa
2025-11-17 12:40 ` [RFC PATCH 3/8] uprobe/x86: Add support to emulate mov reg,reg instructions Jiri Olsa
2025-11-17 12:40 ` [RFC PATCH 4/8] uprobe/x86: Add support to emulate sub imm,reg instructions Jiri Olsa
2025-11-17 12:40 ` [RFC PATCH 5/8] uprobe/x86: Add support to optimize on top of emulated instructions Jiri Olsa
2025-11-24 18:01 ` Oleg Nesterov
2025-11-26 7:54 ` Jiri Olsa [this message]
2025-11-17 12:40 ` [RFC PATCH 6/8] selftests/bpf: Add test for mov and sub emulation Jiri Olsa
2025-11-17 12:40 ` [RFC PATCH 7/8] selftests/bpf: Add test for uprobe prologue optimization Jiri Olsa
2025-11-17 12:40 ` [RFC PATCH 8/8] selftests/bpf: Add race test for uprobe proglog optimization Jiri Olsa
2025-11-24 18:12 ` [RFC PATCH 0/8] uprobe/x86: Add support to optimize prologue Oleg Nesterov
2025-12-08 6:30 ` Masami Hiramatsu
2025-12-08 10:29 ` Oleg Nesterov
2025-12-07 22:23 ` Jiri Olsa
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=aSayKtsTNkuyu0TP@krava \
--to=olsajiri@gmail.com \
--cc=David.Laight@aculab.com \
--cc=andrii@kernel.org \
--cc=bpf@vger.kernel.org \
--cc=john.fastabend@gmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-trace-kernel@vger.kernel.org \
--cc=mhiramat@kernel.org \
--cc=mingo@kernel.org \
--cc=oleg@redhat.com \
--cc=peterz@infradead.org \
--cc=rostedt@goodmis.org \
--cc=songliubraving@fb.com \
--cc=x86@kernel.org \
--cc=yhs@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).