Re: git segfaults on older Solaris releases
- From
Jeff King <peff@peff.net>
- Date
- Apr 7, 2016, 19:07 UTC
- Message-ID
- <20160407190709.GC4478@sigill.intra.peff.net>
- In-Reply-To
- <xmqqk2k9z20p.fsf@gitster.mtv.corp.google.com>
On Thu, Apr 07, 2016 at 11:50:46AM -0700, Junio C Hamano wrote:
Show 9 quoted lines
> Junio C Hamano <gitster@pobox.com> writes: > > > "Tom G. Christensen" <tgc@jupiterrise.com> writes: > > > >> The reason for the crash is simple, a null value was passed to the 's' > >> format for the *printf family of functions. > >> ... > >> Passing a null value to the 's' format is explicitly documented as > >> giving undefined results on Solaris, even on Solaris 11(2).
Thanks, TIL (though it is not really surprising, I guess, since some memcpy implementations have the same problem).
> So, I've looked at places where we use "%.*s" with "prefix" nearby, > and it seems that this is the only place.
Thank you for digging; I obviously didn't think about this issue at all when doing the mass conversions recently.
Show 23 quoted lines
> The "prefix" being a NULL is a perfectly valid state throughout the
> system and means a different thing than it being an empty string, so
> it is valid for callers of prefix_path() and prefix_path_gently() to
> pass prefix=NULL as long as they pass len=0.
>
> So perhaps this is all we need to fix your box.
>
> setup.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/setup.c b/setup.c
> index 3439ec6..b6c8aab 100644
> --- a/setup.c
> +++ b/setup.c
> @@ -103,7 +103,7 @@ char *prefix_path_gently(const char *prefix, int len,
> return NULL;
> }
> } else {
> - sanitized = xstrfmt("%.*s%s", len, prefix, path);
> + sanitized = xstrfmt("%.*s%s", len, prefix ? prefix : "", path);
> if (remaining_prefix)
> *remaining_prefix = len;
> if (normalize_path_copy_len(sanitized, sanitized, remaining_prefix)) {The original pre-75faa45ae0230b321bf72027b2274315d7e14e34 version checked "if (len)", but I think this should be equally right.
-Peff