Tarantool development patches archive
 help / color / mirror / Atom feed
From: Sergey Kaplun via Tarantool-patches <tarantool-patches@dev.tarantool.org>
To: Maxim Kokryashkin <max.kokryashkin@gmail.com>
Cc: tarantool-patches@dev.tarantool.org
Subject: Re: [Tarantool-patches] [PATCH luajit] sysprof: disable proto and trace dumps in default
Date: Thu, 16 Jun 2022 09:56:14 +0300	[thread overview]
Message-ID: <YqrUDsqZ0D/5w4d7@root> (raw)
In-Reply-To: <20220612185220.502537-1-m.kokryashkin@tarantool.org>

Hi, Maxim!

Thanks for the patch!
LGTM, except a few nits below.

On 12.06.22, Maxim Kokryashkin wrote:
> There is no need to dump proto or trace information for sysprof
> in the default mode. Moreover, attempts to do so will lead to
> segmentation fault due to uninitialized buffer. This commit
> disables proto and trace dumps in the default mode.
> 
> Resolves tarantool/tarantool#7264
> ---
> PR: https://github.com/tarantool/tarantool/pull/7265
> Branch: https://github.com/tarantool/luajit/tree/fckxorg/gh-7264-dump-proto-default-mode
> 
>  src/lj_sysprof.c                              |  4 +-
>  ...4-add-proto-trace-sysprof-default.test.lua | 58 +++++++++++++++++++
>  2 files changed, 60 insertions(+), 2 deletions(-)
>  create mode 100644 test/tarantool-tests/gh-7264-add-proto-trace-sysprof-default.test.lua
> 
> diff --git a/src/lj_sysprof.c b/src/lj_sysprof.c
> index a4a70146..2e9ed9b3 100644
> --- a/src/lj_sysprof.c
> +++ b/src/lj_sysprof.c

<snipped>

> diff --git a/test/tarantool-tests/gh-7264-add-proto-trace-sysprof-default.test.lua b/test/tarantool-tests/gh-7264-add-proto-trace-sysprof-default.test.lua
> new file mode 100644
> index 00000000..2d2a756a
> --- /dev/null
> +++ b/test/tarantool-tests/gh-7264-add-proto-trace-sysprof-default.test.lua
> @@ -0,0 +1,58 @@
> +-- Sysprof is implemented for x86 and x64 architectures only.
> +local ffi = require("ffi")
> +require("utils").skipcond(
> +  jit.arch ~= "x86" and jit.arch ~= "x64" or jit.os ~= "Linux"
> +    or ffi.abi("gc64"),
> +  jit.arch.." architecture or "..jit.os..
> +  " OS is NIY for sysprof"
> +)
> +
> +local tap = require("tap")
> +local jit = require("jit")
> +
> +local test = tap.test("gh-7264-add-proto-trace-sysprof-default.test.lua")
> +test:plan(2)
> +
> +jit.off()
> +jit.flush()
> +
> +local function allocate()
> +  local a = {}
> +  for _ = 1, 1e4 do table.insert(a, "teststring") end

Why do we need so many iterations?
We can manipulate JIT behaviour by using `jit.opt.start('hotloop=1')`.

> +  return a
> +end
> +
> +local chunk = [[
> +function lua_global_f()
> +  local a = 'teststring'

Nit: I suggest to use one type of quotes in the whole test.
(Also, it is better to use single quotes everywhere (its offline
agreement with Igor).

> +end
> +]]
> +
> +-- Trace creation during the sysprof runtime.
> +jit.on()
> +
> +local _ = nil

Why do we need this variable? (If for the `res, _ =` below, we can just
use the single `res = `).

> +local res, err = misc.sysprof.start({ mode = "D" })
> +assert(res, err)
> +allocate()
> +res, _ = misc.sysprof.stop()
> +test:ok(res)

Nit: please add name for the test.

> +
> +jit.off()
> +
> +-- Proto creation during the sysprof runtime.
> +res, err = misc.sysprof.start({ mode = "D" })
> +assert(res, err)
> +
> +res, err = loadstring(chunk)
> +assert(res, err)
> +res()
> +
> +-- luacheck: globals lua_global_f
> +lua_global_f()
> +collectgarbage()
> +res, _ = misc.sysprof.stop()
> +test:ok(res)

Ditto.

> +
> +jit.on()
> +os.exit(test:check() and 0 or 1)
> -- 
> 2.36.1
> 

-- 
Best regards,
Sergey Kaplun

  reply	other threads:[~2022-06-16  6:58 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-06-12 18:52 Maxim Kokryashkin via Tarantool-patches
2022-06-16  6:56 ` Sergey Kaplun via Tarantool-patches [this message]
2022-07-04 15:27   ` Maxim Kokryashkin via Tarantool-patches
2022-07-15 13:36 ` 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=YqrUDsqZ0D/5w4d7@root \
    --to=tarantool-patches@dev.tarantool.org \
    --cc=max.kokryashkin@gmail.com \
    --cc=skaplun@tarantool.org \
    --subject='Re: [Tarantool-patches] [PATCH luajit] sysprof: disable proto and trace dumps in default' \
    /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