Hi Hackers,
The following script, when built with addresssanitizer, fails with a
DoubleFree error. reproduce on master (d3bba0415435)
psql postgres <<EOF
\pset display_false 'f'
SELECT 1 as one, 2 as two \g (display_false=csv csv_fieldsep='\t')
\pset display_false 'f'
EOF
Boolean false display is "f".
one | two
-----+-----
1 | 2
(1 строка)
=================================================================
==2263488==ERROR: AddressSanitizer: attempting double-free on
0x774e419e15d0 in thread T0:
#0 0x64c15cb777ab in free.part.0
(/pgpro/builds/master/inst_asan/bin/psql+0x23e7ab) (BuildId:
235e8c12978fc34235af5b00bd4a2feaab9cb794)
#1 0x64c15cbedfbd in do_pset
/pgpro/postgres/src/bin/psql/command.c:5310
#2 0x64c15cbef308 in exec_command_pset
/pgpro/postgres/src/bin/psql/command.c:2737
#3 0x64c15cbf10d3 in exec_command
/pgpro/postgres/src/bin/psql/command.c:431
#4 0x64c15cbf1579 in HandleSlashCmds
/pgpro/postgres/src/bin/psql/command.c:258
#5 0x64c15cc25ecd in MainLoop
/pgpro/postgres/src/bin/psql/mainloop.c:496
#6 0x64c15cbec791 in process_file
/pgpro/postgres/src/bin/psql/command.c:4977
#7 0x64c15cc48cda in main /pgpro/postgres/src/bin/psql/startup.c:424
#8 0x7b2e4322a574 in __libc_start_call_main
../sysdeps/nptl/libc_start_call_main.h:58
#9 0x7b2e4322a627 in __libc_start_main_impl ../csu/libc-start.c:360
#10 0x64c15ca91b84 in _start
(/pgpro/builds/master/inst_asan/bin/psql+0x158b84) (BuildId:
235e8c12978fc34235af5b00bd4a2feaab9cb794)
0x774e419e15d0 is located 0 bytes inside of 2-byte region
[0x774e419e15d0,0x774e419e15d2)
freed by thread T0 here:
#0 0x64c15cb777ab in free.part.0
(/pgpro/builds/master/inst_asan/bin/psql+0x23e7ab) (BuildId:
235e8c12978fc34235af5b00bd4a2feaab9cb794)
#1 0x64c15cbedfbd in do_pset
/pgpro/postgres/src/bin/psql/command.c:5310
#2 0x783e419e2980 (<unknown module>)
previously allocated by thread T0 here:
#0 0x64c15cb725c8 in strdup
(/pgpro/builds/master/inst_asan/bin/psql+0x2395c8) (BuildId:
235e8c12978fc34235af5b00bd4a2feaab9cb794)
#1 0x64c15cc92bd5 in pg_strdup
/pgpro/postgres/src/common/fe_memutils.c:95
SUMMARY: AddressSanitizer: double-free
(/pgpro/builds/master/inst_asan/bin/psql+0x23e7ab) (BuildId:
235e8c12978fc34235af5b00bd4a2feaab9cb794) in free.part.0
==2263488==ABORTING
645cb44c5490f70da4dca57b8ecca6562fb883a7 is the first bad commit
gcc --version
gcc (Ubuntu 15.2.0-4ubuntu4) 15.2.0
building CPPFLAGS="-Og -fsanitize=address -fsanitize=undefined
-fno-sanitize-recover=all -fno-sanitize=nonnull-attribute
-fstack-protector" LDFLAGS='-fsanitize=address -fsanitize=undefined
-static-libasan' \
./configure --prefix /pgpro/builds/master_simple && make world-bin -s
-j$(nproc) && make install-world-bin -s
04.11.2025 00:16, David G. Johnston пишет:
On Monday, November 3, 2025, Álvaro Herrera <[email protected]> wrote:
On 2025-Oct-21, Álvaro Herrera wrote:
> On 2025-Oct-20, David G. Johnston wrote:
> > Thank you. Seems good from a quick read. I’m regretting the
choice of the
> > display_ prefix; is there any technical limitation or other
opposition to
> > using just true and false?
> >
> > \pset true ‘true’
> > \pset false ‘false’
> Uhm. I don't know. [...] I'll gladly take a vote.
I got zero votes and lots of digression, so I have pushed with your
original choice of "display_true" and "display_false". The "true" and
"false" variable names sound too generic and I think they're more
likely
to cause confusion. I think "null" is not a great name either,
but it's
been there since forever so I'm not going to propose changing it.
Thank you.
David J.
