Обсуждение: thread-safety: gmtime_r(), localtime_r()
Here is a patch for using gmtime_r() and localtime_r() instead of gmtime() and localtime(), for thread-safety. There are a few affected calls in libpq and ecpg's libpgtypes, which are probably effectively bugs, because those libraries already claim to be thread-safe. There is one affected call in the backend. Most of the backend otherwise uses the custom functions pg_gmtime() and pg_localtime(), which are implemented differently. Some portability fun: gmtime_r() and localtime_r() are in POSIX but are not available on Windows. Windows has functions gmtime_s() and localtime_s() that can fulfill the same purpose, so we can add some small wrappers around them. (Note that these *_s() functions are also different from the *_s() functions in the bounds-checking extension of C11. We are not using those here.) MinGW exposes neither *_r() nor *_s() by default. You can get at the POSIX-style *_r() functions by defining _POSIX_C_SOURCE appropriately before including <time.h>. (There is apparently probably also a way to get at the Windows-style *_s() functions by supplying some additional options or defines. But we might as well just use the POSIX ones.)
Вложения
On Thu, Jun 27, 2024 at 1:42 AM Peter Eisentraut <peter@eisentraut.org> wrote:
Here is a patch for using gmtime_r() and localtime_r() instead of
gmtime() and localtime(), for thread-safety.
There are a few affected calls in libpq and ecpg's libpgtypes, which are
probably effectively bugs, because those libraries already claim to be
thread-safe.
There is one affected call in the backend. Most of the backend
otherwise uses the custom functions pg_gmtime() and pg_localtime(),
which are implemented differently.
Some portability fun: gmtime_r() and localtime_r() are in POSIX but are
not available on Windows. Windows has functions gmtime_s() and
localtime_s() that can fulfill the same purpose, so we can add some
small wrappers around them. (Note that these *_s() functions are also
different from the *_s() functions in the bounds-checking extension of
C11. We are not using those here.)
MinGW exposes neither *_r() nor *_s() by default. You can get at the
POSIX-style *_r() functions by defining _POSIX_C_SOURCE appropriately
before including <time.h>. (There is apparently probably also a way to
get at the Windows-style *_s() functions by supplying some additional
options or defines. But we might as well just use the POSIX ones.)
Hi! Looks good to me.
But why you don`t change localtime function at all places?
For example:
src/bin/pg_controldata/pg_controldata.c
src/bin/pg_dump/pg_backup_archiver.c
src/bin/initdb/findtimezone.c
Best regards, Stepan Neretin.
On 27.06.24 06:47, Stepan Neretin wrote: > Hi! Looks good to me. > But why you don`t change localtime function at all places? > For example: > src/bin/pg_controldata/pg_controldata.c > src/bin/pg_dump/pg_backup_archiver.c > src/bin/initdb/findtimezone.c At the moment, I am focusing on the components that are already meant to be thread-safe (libpq, ecpg libs) and the ones we are actively looking at maybe converting (backend). I don't intend at this point to convert all other code to use only thread-safe APIs.