Re: possible bug in fseek of buffered files shared between processes (example uses stdin)

[RVP <rvp%SDF.ORG@localhost>]

On Tue, 2 Nov 2021, Carlo Arenas wrote:

> while troubleshooting a test failure for pcre2[0] in OpenBSD, tracked
> down the problem to code that was inherited from NetBSD for their
> fseek implementation in libc.

This isn't actually an fseek issue. Glibc and musl do an explicit lseek()
when closing a file if there is unread data in the buffer:

$ cc -o h head.c
$ (strace -o x.log ./h 1; echo .; ./h 2; echo .; cat) < head.c
#include <stdio.h>
.
#include <stdlib.h>

.
int
main(int argc, char* argv[])
{
        char line[BUFSIZ];
        int n;

        if (argc != 2)
                return EXIT_FAILURE;

        n = atoi(argv[1]);
        while (n-- && fgets(line, sizeof line, stdin))
                printf("%s", line);
        return EXIT_SUCCESS;
}
$ tail x.log
munmap(0x7f0f0e89d000, 78789)           = 0
fstat(0, {st_mode=S_IFREG|0600, st_size=258, ...}) = 0
brk(NULL)                               = 0x5564cc789000
brk(0x5564cc7aa000)                     = 0x5564cc7aa000
read(0, "#include <stdio.h>\n#include <std"..., 4096) = 258
fstat(1, {st_mode=S_IFCHR|0600, st_rdev=makedev(136, 2), ...}) = 0
write(1, "#include <stdio.h>\n", 19)    = 19
lseek(0, -239, SEEK_CUR)                = 19
exit_group(0)                           = ?
+++ exited with 0 +++
$

This looks like a useful behaviour to add to the libc close routines.

-RVP