Tarantool development patches archive
 help / color / mirror / Atom feed
From: Sergey Bronnikov via Tarantool-patches <tarantool-patches@dev.tarantool.org>
To: Sergey Kaplun <skaplun@tarantool.org>,
	Maxim Kokryashkin <m.kokryashkin@tarantool.org>
Cc: tarantool-patches@dev.tarantool.org
Subject: Re: [Tarantool-patches] [PATCH luajit] Respect jit.off() on pending trace exit.
Date: Tue, 23 Jan 2024 11:33:41 +0300	[thread overview]
Message-ID: <793b7d4e-7d5f-4f77-bb82-c4496a94702a@tarantool.org> (raw)
In-Reply-To: <20240122093234.14407-1-skaplun@tarantool.org>

Hi, Sergey!


thanks for the patch! LGTM with a minor comment below


Sergey


On 1/22/24 12:32, Sergey Kaplun wrote:
> From: Mike Pall <mike>
>
> Thanks to Sergey Kaplun.
>
> (cherry-picked from commit 7dbe545933485849977d50384f2f20f2cccf0cf9)
>
> Before this commit, the JIT engine didn't check the status of JIT flags
> when compiling the side trace. Hence, after calling `jit.off()` the side
> traces are still recorded. This patch adds the aforementioned check.
>
> Sergey Kaplun:
> * added the description and the test for the problem
>
> Part of tarantool/tarantool#9595
> ---
>
> Branch: https://github.com/tarantool/luajit/tree/skaplun/lj-1134-hotside-jit-off
> Tarantool PR: https://github.com/tarantool/tarantool/pull/9607
> Related issues:
> * https://github.com/tarantool/tarantool/issues/9595
> * https://github.com/LuaJIT/LuaJIT/issues/1134
>
>   src/lj_trace.c                                |  2 +-
>   .../lj-1134-hotside-jit-off.test.lua          | 44 +++++++++++++++++++
>   2 files changed, 45 insertions(+), 1 deletion(-)
>   create mode 100644 test/tarantool-tests/lj-1134-hotside-jit-off.test.lua
>
> diff --git a/src/lj_trace.c b/src/lj_trace.c
> index 236e06a0..20014ecb 100644
> --- a/src/lj_trace.c
> +++ b/src/lj_trace.c
> @@ -917,7 +917,7 @@ int LJ_FASTCALL lj_trace_exit(jit_State *J, void *exptr)
>     } else if (G(L)->gc.state == GCSatomic || G(L)->gc.state == GCSfinalize) {
>       if (!(G(L)->hookmask & HOOK_GC))
>         lj_gc_step(L);  /* Exited because of GC: drive GC forward. */
> -  } else {
> +  } else if ((J->flags & JIT_F_ON)) {
>       trace_hotside(J, pc);
>     }
>     if (bc_op(*pc) == BC_JLOOP) {
> diff --git a/test/tarantool-tests/lj-1134-hotside-jit-off.test.lua b/test/tarantool-tests/lj-1134-hotside-jit-off.test.lua
> new file mode 100644
> index 00000000..cdee3eb2
> --- /dev/null
> +++ b/test/tarantool-tests/lj-1134-hotside-jit-off.test.lua
> @@ -0,0 +1,44 @@
> +local tap = require('tap')
> +
> +-- Test file to demonstrate the JIT misbehaviour, when the side
> +-- trace is compiled after `jit.off()`.
> +-- See also: https://github.com/LuaJIT/LuaJIT/issue/1134.
> +
> +local test = tap.test('lj-1134-hotside-jit-off'):skipcond({
> +  ['Test requires JIT enabled'] = not jit.status(),
> +  ['Disabled on *BSD due to #4819'] = jit.os == 'BSD',
> +})
> +
> +local traceinfo = require('jit.util').traceinfo
> +
> +test:plan(1)
> +
> +local take_side
> +local function trace()
> +  -- luacheck: ignore
> +  -- Branch for the side exit.
> +  if take_side then end
> +end
> +
> +-- Flush all possible traces.
> +jit.flush()
> +
> +jit.opt.start('hotloop=1', 'hotexit=1')
> +
> +trace()
> +trace()
> +
> +assert(traceinfo(1), 'root trace not compiled')

It is not clear what "1" means (here and below). As I got it right, it 
is a trace number.

I would left a comment about magic number or replace constant with 
variable with self-explained name.

> +
> +-- Force trace exit.
> +take_side = true
> +
> +jit.off()
> +
> +-- Attempt to compile a side trace.
> +trace()
> +trace()
> +
> +test:ok(not traceinfo(2), 'no side trace compiled')
> +
> +test:done(true)

  reply	other threads:[~2024-01-23  8:33 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-01-22  9:32 Sergey Kaplun via Tarantool-patches
2024-01-23  8:33 ` Sergey Bronnikov via Tarantool-patches [this message]
2024-01-23 13:14   ` Sergey Kaplun via Tarantool-patches
2024-01-25  7:49     ` Sergey Bronnikov via Tarantool-patches
2024-01-29 11:39     ` Maxim Kokryashkin via Tarantool-patches
2024-02-15 13:43 ` Igor Munkin via Tarantool-patches

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=793b7d4e-7d5f-4f77-bb82-c4496a94702a@tarantool.org \
    --to=tarantool-patches@dev.tarantool.org \
    --cc=m.kokryashkin@tarantool.org \
    --cc=sergeyb@tarantool.org \
    --cc=skaplun@tarantool.org \
    --subject='Re: [Tarantool-patches] [PATCH luajit] Respect jit.off() on pending trace exit.' \
    /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

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox