Fix handling of copy_file_range() return value

Treat copy_file_range() return value of zero as an error: it indicates
that no bytes could be copied (perhaps the source file is shorter than
expected), and the existing retry loop would otherwise spin forever
since nwritten would never reach BLCKSZ.

The other uses of copy_file_range() in the tree don't have this
problem.

Reviewed-by: Nazir Bilal Yavuz <byavuz81@gmail.com>
Reviewed-by: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
Reviewed-by: Yingying Chen <cyy9255@gmail.com>
Discussion: https://www.postgresql.org/message-id/flat/3208cf7a-c7f3-41eb-92f6-33cbeff4df40%40eisentraut.org
This commit is contained in:
Peter Eisentraut 2026-06-29 11:49:11 +02:00
parent b7e4e3e7fa
commit 994f770a0f

View file

@ -705,6 +705,9 @@ write_reconstructed_file(char *input_filename,
if (wb < 0)
pg_fatal("error while copying file range from \"%s\" to \"%s\": %m",
input_filename, output_filename);
else if (wb == 0)
pg_fatal("unexpected end of file while copying file range from \"%s\" to \"%s\"",
input_filename, output_filename);
nwritten += wb;