Thomas Trepl wrote:
Am Sonntag, den 04.02.2018, 13:58 +0100 schrieb Thomas Trepl:
Am Sonntag, den 04.02.2018, 12:18 +0100 schrieb Armin K.:
On Sun, 2018-02-04 at 11:42 +0100, Thomas Trepl wrote:

got a abort message when compiling postgresql:

D_GNU_SOURCE   -c -o copy_fetch.o copy_fetch.c
copy_fetch.c:159:1: Fehler: In Konflikt stehende Typen für
  copy_file_range(const char *path, off_t begin, off_t end, bool
In file included from copy_fetch.c:15:0:
/usr/include/unistd.h:1110:9: Anmerkung: Vorherige Deklaration
»copy_file_range« war hier
  ssize_t copy_file_range (int __infd, __off64_t *__pinoff,

To me it looks like a clash to the standard library function
copy_file_range. It's easy to fix as postgresql's copy_file_range
static function. I did following sed

sed -e "s/copy_file_range/_&/" \
     -i src/bin/pg_rewind/copy_fetch.c

with which compilation was successfull. Donno whether it is
by gcc-7.3 and/or glibc-2.27 (on i686).

Have you seen that issue, too or is it something wired here


 From what I've seen, glibc-2.27 introduced copy_file_range. You
try and swap any instance of copy_file_range in postgresql with,
pg_copy_file_range to avoid conflicts (or try and delete
copy_file_range in pgsql entirely, and let it use the one from

Yes, now found the statement in glibc's changelog

"* The copy_file_range function was added."

I'll add a small sed to the postgresql instructions as it is quite
to fix (and occurs only in one file).

Should I open/fix/close a ticket for that?

Upstream allready caught that issue and they said it will be fixed in
"next week's round of minor releases" (10.2).  So I think we can just
wait a few days and problem will went away...
For urgent fix, just do

sed -e "s/copy_file_range/rewind_&/" -i src/bin/pg_rewind/copy_fetch.c

This is what upstream did.

Please go ahead and put that into the book for now. We always check sed and patch instructions when a new version of a package is released.

  -- Bruce

Unsubscribe: See the above information page

Reply via email to