Tarantool development patches archive
 help / color / mirror / Atom feed
From: Olga Arkhangelskaia <arkholga@tarantool.org>
To: Nikita Pettik <korablev@tarantool.org>
Cc: tarantool-patches@dev.tarantool.org
Subject: Re: [Tarantool-patches] [PATCH] json: fix silent change of global json settings
Date: Tue, 11 Feb 2020 11:46:15 +0300	[thread overview]
Message-ID: <d080f92e-b466-8b56-37ba-dcfab49a6eac@tarantool.org> (raw)
In-Reply-To: <20200210130823.GF1110@tarantool.org>

Hi Nikita! Thanks for the review!

I fixed all issues that you have pointed out.

And pushed branch.

10.02.2020 16:08, Nikita Pettik пишет:
> On 10 Feb 10:57, Olga Arkhangelskaia wrote:
>> When json.decode is used with 2 arguments, 2nd argument seeps out to global
>> json settings. Morover,
> Nit: Moreover.
>
>> due to current serialier.cfg implementation it
> -> serializer
>
>> remains invisible while checking settings by json.cfg. To prevent sucj
> -> such
>
>> behaviour we stop writing to global serializer struct and use local one,
>> to get one-time action.
>> As was mention before json.cfg can not be trusted in this case, so to check that
>> everything remained unchanged we call decode twice with and without 2nd
>> argument.
>>
>> Closes #4761
> Note that there's no 'Closes #4761' label on your actual branch.
> I guess you simply forgot to push updated branch.
>
>> ---
>> +    --
>> +    -- gh-4761 json.decode silently changes global settings of json when called
>> +    -- with 2d parameter
>> +    --
>> +    test:ok(pcall(serializer.decode,'{"1":{"b":{"c":1,"d":null}},"a":1}'))
>>   
>>       --
>>       -- gh-3514: fix parsing integers with exponent in json
>> diff --git a/third_party/lua-cjson/lua_cjson.c b/third_party/lua-cjson/lua_cjson.c
>> index 3d25814f3..f855cbd80 100644
>> --- a/third_party/lua-cjson/lua_cjson.c
>> +++ b/third_party/lua-cjson/lua_cjson.c
>> @@ -1004,13 +1004,13 @@ static int json_decode(lua_State *l)
>>       luaL_argcheck(l, lua_gettop(l) == 2 || lua_gettop(l) == 1, 1,
>>                     "expected 1 or 2 arguments");
>>   
>> +    struct luaL_serializer *cfg = luaL_checkserializer(l);
> Nit: I'd add a brief comment here (to avoid any confusions concerning
> copying object on the stack):
>
> diff --git a/third_party/lua-cjson/lua_cjson.c b/third_party/lua-cjson/lua_cjson.c
> index f855cbd80..c9c987c8c 100644
> --- a/third_party/lua-cjson/lua_cjson.c
> +++ b/third_party/lua-cjson/lua_cjson.c
> @@ -1006,6 +1006,14 @@ static int json_decode(lua_State *l)
>   
>       struct luaL_serializer *cfg = luaL_checkserializer(l);
>       struct luaL_serializer user_cfg = *cfg;
> +    /*
> +     * user_cfg is per-call local version of global cfg: it is
> +     * used if user passes custom options to :decode() method
> +     * as a separate arguments. In this case it is required
> +     * to avoid modifying global parameters. Life span of
> +     * user_cfg is restricted by the scope of :decode() so it
> +     * is enough to allocate it on the stack.
> +     */
>       json.cfg = cfg;
>       if (lua_gettop(l) == 2) {
>
>> +    struct luaL_serializer user_cfg = *cfg;
>> +    json.cfg = cfg;
> What is more, you can avoid premature copying on the stack:
>
> @@ -1005,9 +1005,10 @@ static int json_decode(lua_State *l)
>                     "expected 1 or 2 arguments");
>   
>       struct luaL_serializer *cfg = luaL_checkserializer(l);
> -    struct luaL_serializer user_cfg = *cfg;
> +    struct luaL_serializer user_cfg;
>       json.cfg = cfg;
>       if (lua_gettop(l) == 2) {
> +        user_cfg = *cfg;
>           luaL_serializer_parse_options(l, &user_cfg);
>           lua_pop(l, 1);
>           json.cfg = &user_cfg;
>
>>       if (lua_gettop(l) == 2) {
>> -        struct luaL_serializer *user_cfg = luaL_checkserializer(l);
>> -        luaL_serializer_parse_options(l, user_cfg);
>> +        luaL_serializer_parse_options(l, &user_cfg);

Here is new diff

diff --git a/third_party/lua-cjson/lua_cjson.c 
b/third_party/lua-cjson/lua_cjson.c
index f855cbd80..5925e7e6f 100644
--- a/third_party/lua-cjson/lua_cjson.c
+++ b/third_party/lua-cjson/lua_cjson.c
@@ -1005,9 +1005,18 @@ static int json_decode(lua_State *l)
                    "expected 1 or 2 arguments");

      struct luaL_serializer *cfg = luaL_checkserializer(l);
-    struct luaL_serializer user_cfg = *cfg;
+    struct luaL_serializer user_cfg;
+    /*
+     * user_cfg is per-call local version of global cfg: it is
+     * used if user passes custom options to :decode() method
+     * as a separate arguments. In this case it is required
+     * to avoid modifying global parameters. Life span of
+     * user_cfg is restricted by the scope of :decode() so it
+     * is enough to allocate it on the stack.
+     */
      json.cfg = cfg;
      if (lua_gettop(l) == 2) {
+        user_cfg = *cfg;
          luaL_serializer_parse_options(l, &user_cfg);
          lua_pop(l, 1);
          json.cfg = &user_cfg;

  reply	other threads:[~2020-02-11  8:46 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-02-10  7:57 Olga Arkhangelskaia
2020-02-10 13:08 ` Nikita Pettik
2020-02-11  8:46   ` Olga Arkhangelskaia [this message]
2020-02-11 12:36     ` Nikita Pettik
2020-02-13 23:17 ` Alexander Turenko
2020-02-14 15:55   ` Olga Arkhangelskaia
2020-02-16  0:06     ` Alexander Turenko
  -- strict thread matches above, loose matches on Subject: below --
2020-02-05 12:28 Olga Arkhangelskaia
2020-02-10  7:57 ` Olga Arkhangelskaia

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=d080f92e-b466-8b56-37ba-dcfab49a6eac@tarantool.org \
    --to=arkholga@tarantool.org \
    --cc=korablev@tarantool.org \
    --cc=tarantool-patches@dev.tarantool.org \
    --subject='Re: [Tarantool-patches] [PATCH] json: fix silent change of global json settings' \
    /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