valgrind error in subscription code

From: Andres Freund <andres(at)anarazel(dot)de>
To: pgsql-hackers(at)postgresql(dot)org, Petr Jelinek <petr(at)2ndquadrant(dot)com>, Peter Eisentraut <peter_e(at)gmx(dot)net>
Subject: valgrind error in subscription code
Date: 2017-04-22 18:31:23
Message-ID: 20170422183123.w2jgiuxtts7qrqaq@alap3.anarazel.de
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi,

I enabled skink / the valgrind animal to run the tap tests too (hugely
increasing the test time :(), and that paid of:
https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=skink&dt=2017-04-22%2004%3A52%3A13

==606== VALGRINDERROR-BEGIN
==606== Conditional jump or move depends on uninitialised value(s)
==606== at 0x46A207: logicalrep_rel_open (relation.c:361)
==606== by 0x472D12: copy_table (tablesync.c:669)
==606== by 0x473186: LogicalRepSyncTableStart (tablesync.c:803)
==606== by 0x475287: ApplyWorkerMain (worker.c:1530)
==606== by 0x440AFD: StartBackgroundWorker (bgworker.c:835)
==606== by 0x44E48A: do_start_bgworker (postmaster.c:5577)
==606== by 0x44E59F: maybe_start_bgworker (postmaster.c:5761)
==606== by 0x44F11D: sigusr1_handler (postmaster.c:5015)
==606== by 0x4E430BF: ??? (in /lib/x86_64-linux-gnu/libpthread-2.24.so)
==606== by 0x6FB8212: __select_nocancel (syscall-template.S:84)
==606== by 0x44F868: ServerLoop (postmaster.c:1693)
==606== by 0x450C53: PostmasterMain (postmaster.c:1337)
==606== Uninitialised value was created by a heap allocation
==606== at 0x605382: MemoryContextAlloc (mcxt.c:729)
==606== by 0x5E4E6A: DynaHashAlloc (dynahash.c:266)
==606== by 0x5E4EEE: element_alloc (dynahash.c:1637)
==606== by 0x5E5018: get_hash_entry (dynahash.c:1248)
==606== by 0x5E5898: hash_search_with_hash_value (dynahash.c:1033)
==606== by 0x5E5A0D: hash_search (dynahash.c:890)
==606== by 0x469D38: logicalrep_relmap_update (relation.c:179)
==606== by 0x472D05: copy_table (tablesync.c:666)
==606== by 0x473186: LogicalRepSyncTableStart (tablesync.c:803)
==606== by 0x475287: ApplyWorkerMain (worker.c:1530)
==606== by 0x440AFD: StartBackgroundWorker (bgworker.c:835)
==606== by 0x44E48A: do_start_bgworker (postmaster.c:5577)
==606==
==606== VALGRINDERROR-END

there's also another bleat:
==2161== VALGRINDERROR-BEGIN
==2161== Syscall param write(buf) points to uninitialised byte(s)
==2161== at 0x4E42190: __write_nocancel (syscall-template.S:84)
==2161== by 0x4676C7: CheckPointReplicationOrigin (origin.c:574)
==2161== by 0x25DB99: CheckPointGuts (xlog.c:9007)
==2161== by 0x2647CD: CreateCheckPoint (xlog.c:8792)
==2161== by 0x26734F: StartupXLOG (xlog.c:7576)
==2161== by 0x4510FA: StartupProcessMain (startup.c:217)
==2161== by 0x273832: AuxiliaryProcessMain (bootstrap.c:426)
==2161== by 0x44D8A0: StartChildProcess (postmaster.c:5256)
==2161== by 0x44DF33: PostmasterStateMachine (postmaster.c:3791)
==2161== by 0x44EE25: reaper (postmaster.c:3038)
==2161== by 0x4E430BF: ??? (in /lib/x86_64-linux-gnu/libpthread-2.24.so)
==2161== by 0x6FB8212: __select_nocancel (syscall-template.S:84)
==2161== Location 0xffeffda42 is 2 bytes inside local var "disk_state"
==2161== declared at origin.c:555, in frame #1 of thread 1
==2161== Uninitialised value was created by a stack allocation
==2161== at 0x467421: CheckPointReplicationOrigin (origin.c:505)
==2161==
==2161== VALGRINDERROR-END
==2161== VALGRINDERROR-BEGIN
==2161== Syscall param write(buf) points to uninitialised byte(s)
==2161== at 0x4E42190: __write_nocancel (syscall-template.S:84)
==2161== by 0x467712: CheckPointReplicationOrigin (origin.c:591)
==2161== by 0x25DB99: CheckPointGuts (xlog.c:9007)
==2161== by 0x2647CD: CreateCheckPoint (xlog.c:8792)
==2161== by 0x26734F: StartupXLOG (xlog.c:7576)
==2161== by 0x4510FA: StartupProcessMain (startup.c:217)
==2161== by 0x273832: AuxiliaryProcessMain (bootstrap.c:426)
==2161== by 0x44D8A0: StartChildProcess (postmaster.c:5256)
==2161== by 0x44DF33: PostmasterStateMachine (postmaster.c:3791)
==2161== by 0x44EE25: reaper (postmaster.c:3038)
==2161== by 0x4E430BF: ??? (in /lib/x86_64-linux-gnu/libpthread-2.24.so)
==2161== by 0x6FB8212: __select_nocancel (syscall-template.S:84)
==2161== Location 0xffeffda58 is 0 bytes inside local var "crc"
==2161== declared at origin.c:511, in frame #1 of thread 1
==2161== Uninitialised value was created by a stack allocation
==2161== at 0x467421: CheckPointReplicationOrigin (origin.c:505)
==2161==
==2161== VALGRINDERROR-END

but that looks more like something that needs to be silenced, based on
the 3s look I gave it.

Greetings,

Andres Freund

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2017-04-22 18:40:28 Re: [COMMITTERS] pgsql: Replication lag tracking for walsenders
Previous Message Sandeep Thakkar 2017-04-22 17:58:49 Re: Patch - Tcl 8.6 version support for PostgreSQL