Horton

RE: Thread-unsafe coding in ecpg

Tsunakawa, Takayuki <tsunakawa.takay@jp.fujitsu.com>

View thread View on postgr.es
From: "Tsunakawa, Takayuki" <tsunakawa.takay@jp.fujitsu.com>
To: 'Tom Lane' <tgl@sss.pgh.pa.us>
Cc: Michael Meskes <meskes@postgresql.org>, "pgsql-hackers@lists.postgresql.org" <pgsql-hackers@lists.postgresql.org>
Date: 2019-01-22T06:57:22Z
Lists: pgsql-hackers 
From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
> BTW, I found another spot in descriptor.c where ecpglib is using
> setlocale() for the same purpose.  Perhaps that one's not reachable
> in threaded apps, but I didn't see any obvious reason to think so,
> so I changed it too.

Ouch, thanks.  And I'm sorry to annoy you by pointing out a trivial thing: in v3 patch, _configthreadlocale() is not called to restore the original value when setlocale() or ecpg_strdup() fails.  I hope this is fixed in v4.

+ #ifdef WIN32
+ 	stmt->oldthreadlocale = _configthreadlocale(_ENABLE_PER_THREAD_LOCALE);
+ 	if (stmt->oldthreadlocale == -1)
+ 	{
+ 		ecpg_do_epilogue(stmt);
+ 		return false;
+ 	}
+ #endif
  	stmt->oldlocale = ecpg_strdup(setlocale(LC_NUMERIC, NULL), lineno);
  	if (stmt->oldlocale == NULL)
  	{

Regards
Takayuki Tsunakawa

Commits

  1. Remove _configthreadlocale() calls in ecpg test suite.

  2. Remove infinite-loop hazards in ecpg test suite.

  3. Blind attempt to fix _configthreadlocale() failures on MinGW.

  4. Avoid thread-safety problem in ecpglib.

  5. Second try at fixing ecpglib thread-safety problem.