set_size should zero-fill new data region after extending data region

[artemyarulin]

set_size should zero-fill new data region after extending data region. Interestingly this one only fails when --net is provided as then we use another implementation for the filesystem

#include <assert.h>
#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <wasi/api_wasi.h>

int main(void) {
    int fd = open("fd_filestat_set_size_data", O_CREAT | O_TRUNC | O_RDWR, 0644);
    write(fd, "aaaaa", 5);

    __wasi_fd_filestat_set_size(fd, 0);

    // BUG: set_size should zero-fill new data region after extending data region.
    // Repro requires --net because it selects the overlay/COW filesystem path.
    __wasi_fd_filestat_set_size(fd, 5);

    lseek(fd, 0, SEEK_SET);
    char zeros[5];
    read(fd, zeros, sizeof(zeros));
    if (zeros[0] != 0 || zeros[1] != 0 || zeros[2] != 0 || zeros[3] != 0 || zeros[4] != 0) {
        printf("Unexpected data: %02x %02x %02x %02x %02x\n",
               (unsigned char)zeros[0],
               (unsigned char)zeros[1],
               (unsigned char)zeros[2],
               (unsigned char)zeros[3],
               (unsigned char)zeros[4]);
        exit(1);
    }
}

[code3002]

Hi @marxin, I’d like to work on this issue if it’s still available. Could you please assign it to me or let me know if there are any specific guidelines I should follow before getting started?

[marxin]

Thank you @rahul3002 for your interest. The question is more for @artemyarulin - do you want to work on it itself or not?

[artemyarulin]

Yeah, pull requests are very welcome! I don't have capacity to work on it during this/next week probably, so @rahul3002 - you can take it. Just ensure you'll extend the tests in c-wasi-tests with repro and then fix it in wasix.