Thread: Re: pgsql: Properly unregister OpenSSL callbacks when libpq is done with
Magnus Hagander wrote: > Log Message: > ----------- > Properly unregister OpenSSL callbacks when libpq is done with > it's connection. This is required for applications that unload > the libpq library (such as PHP) in which case we'd otherwise > have pointers to these functions when they no longer exist. Breaks the build with --enable-thread-safety and --with-openssl because of this typo. Kris Jurka *** a/src/interfaces/libpq/fe-secure.c --- b/src/interfaces/libpq/fe-secure.c *************** *** 918,925 **** destroy_ssl_system(void) * This means we leak a little memory on repeated load/unload * of the library. */ ! free(pqlockarray); ! pqlockarray = NULL; } } --- 918,925 ---- * This means we leak a little memory on repeated load/unload * of the library. */ ! free(pq_lockarray); ! pq_lockarray = NULL; } }
Re: Re: pgsql: Properly unregister OpenSSL callbacks when libpq is done with
From
Bruce Momjian
Date:
Kris Jurka wrote: > Magnus Hagander wrote: > > Log Message: > > ----------- > > Properly unregister OpenSSL callbacks when libpq is done with > > it's connection. This is required for applications that unload > > the libpq library (such as PHP) in which case we'd otherwise > > have pointers to these functions when they no longer exist. > > Breaks the build with --enable-thread-safety and --with-openssl because > of this typo. Thanks, applied. -- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB http://enterprisedb.com + If your life is a hard drive, Christ can be your backup. +
Re: Re: pgsql: Properly unregister OpenSSL callbacks when libpq is done with
From
Magnus Hagander
Date:
Bruce Momjian wrote: > Kris Jurka wrote: >> Magnus Hagander wrote: >>> Log Message: >>> ----------- >>> Properly unregister OpenSSL callbacks when libpq is done with >>> it's connection. This is required for applications that unload >>> the libpq library (such as PHP) in which case we'd otherwise >>> have pointers to these functions when they no longer exist. >> Breaks the build with --enable-thread-safety and --with-openssl because >> of this typo. > > Thanks, applied. That fix is wrong. The comment clearly says the code *shouldn't* free the lockarray, so the proper fix is to remove those two lines. I have applied a patch that does this. //Magnus