Michael Paquier - PostgreSQL committer

Here are a couple of things to know when programming with Postgres.

Enforce wait

When debugging something that happens at session startup, here are some settings that can be used to delay startup for a couple of seconds, enough to attach breakpoints to the process tested.

PGOPTIONS="-W n"

Compiling test programs

Here is a single commands to compile code file into an executable:

gcc hello_world.c -o hello_world_exec

To generate a .so file:

gcc -c -fPIC hello_world.c -o hello_world.o
gcc hello_world.o -shared -o libhello.so

Or in one step:

gcc -g -Wall -shared -o libhello.so -fPIC hello.c

When generating a library for a test, LD_PRELOAD can be useful to load this test library before any other:

LD_PRELOAD=/path/to/library/libhello.so my_command $ARGS

valgrind

valgrind is useful to find memory problems in Postgres, try to use with something like the attached.

valgrind --suppress
valgrind \
    --suppressions=$PG_SOURCE/src/tools/valgrind.supp \
    --trace-children=yes --track-origins=yes --read-var-info=yes \
    postgres -D REST_OF_ARGS

Corrupting pages

A way to manually corrupt data may be to use dd like that, notrunc being really important to not truncate the relation file once a block is changed:

dd if=/dev/random bs=8192 count=1 \
    seek=$BLOCK_ID of=base/$DBOID/$RELFILENODE
    conv=notrunc

Encoding

Here is a small trick to write directly in UTF-8 using raw data. For example for sequence c2 a2 (cent sign):

=# SELECT E'\xc2\xa2' AS "char";
 char
------
 c
(1 row)

Data structures

pahole, which is part of the dwarf utilities, is useful to see the size of structures using compiled files.

pahole source.o class_name