Hi,
David Turner wrote:
> From: René Scharfe <l.s.r@xxxxxx>
>
> POSIX limits the length of host names to HOST_NAME_MAX. Export the
> fallback definition from daemon.c and use this constant to make all
> buffers used with gethostname(2) big enough for any possible result
> and a terminating NUL.
Since some platforms do not define HOST_NAME_MAX and we provide a
fallback, this is not actually big enough for any possible result.
For example, the Hurd allows arbitrarily long hostnames.
Nevertheless this patch seems like the right thing to do.
> Inspired-by: David Turner <dturner@xxxxxxxxxxxx>
> Signed-off-by: Rene Scharfe <l.s.r@xxxxxx>
> Signed-off-by: David Turner <dturner@xxxxxxxxxxxx>
> ---
> builtin/gc.c | 10 +++++++---
> builtin/receive-pack.c | 2 +-
> daemon.c | 4 ----
> fetch-pack.c | 2 +-
> git-compat-util.h | 4 ++++
> ident.c | 2 +-
> 6 files changed, 14 insertions(+), 10 deletions(-)
Thanks for picking this up.
[...]
> +++ b/builtin/gc.c
[...]
> @@ -257,8 +257,12 @@ static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
> fd = hold_lock_file_for_update(&lock, pidfile_path,
> LOCK_DIE_ON_ERROR);
> if (!force) {
> - static char locking_host[128];
> + static char locking_host[HOST_NAME_MAX + 1];
> + static char *scan_fmt;
> int should_exit;
> +
> + if (!scan_fmt)
> + scan_fmt = xstrfmt("%s %%%dc", "%"SCNuMAX, HOST_NAME_MAX);
> fp = fopen(pidfile_path, "r");
> memset(locking_host, 0, sizeof(locking_host));
> should_exit =
> @@ -274,7 +278,7 @@ static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
> * running.
> */
> time(NULL) - st.st_mtime <= 12 * 3600 &&
> - fscanf(fp, "%"SCNuMAX" %127c", &pid, locking_host) == 2 &&
> + fscanf(fp, scan_fmt, &pid, locking_host) == 2 &&
I hoped this could be simplified since HOST_NAME_MAX is a numeric literal,
using the double-expansion trick:
#define STR_(s) # s
#define STR(s) STR_(s)
fscanf(fp, "%" SCNuMAX " %" STR(HOST_NAME_MAX) "c",
&pid, locking_host);
Unfortunately, I don't think there's anything stopping a platform from
defining
#define HOST_NAME_MAX 0x100
which would break that.
So this run-time calculation appears to be necessary.
Reviewed-by: Jonathan Nieder <jrnieder@xxxxxxxxx>
Thanks.