Re: [PATCH] Add a fast version of fstat to cygwin port

From: Johannes Sixt <j.sixt@viscovery.net>
Date: Mar 19, 2009, 14:54 UTC
Message-ID: <49C25CAE.6000003@viscovery.net>
In-Reply-To: <81b0412b0903190730s40589291iea9a861ddeedcc0@mail.gmail.com>

Alex Riesen schrieb:

> Besides, the output of the fast stat and lstat is not compatible
> with cygwin's fstat with regard to uid, gid and ctime fields.

Why do you need this? I don't think that fstat() is used in a critical path. Do you see problems with the incompatible fields?

> This is not strictly related to the other stat patches. The fstat
> code is shamelessly stolen from the mingw port, sorry.

I wouldn't call it "stolen", but "copied". Because if you copy it, it becomes your responsibility and all copied bugs are yours ;)

> BTW, why do we have to #undef fstat, but not stat/lstat?

Because stat and lstat are #defined with an argument list, but in those instances where the cygwin version of stat/lstat is meant, they are used *without* argument list (see cygwin_stat/lstat_stub), and no macro expansion happens, and therefore we don't need to #undef the macro.

OTOH, do_fstat calls into cygwin's fstat() if the file handle is not a file and uses an argument list that would cause a macro expansion if it were not #undef'd:

Show 13 quoted lines

> +#undef fstat
> +static int cygwin_fstat(int fd, struct stat *buf)
> +{
> +	HANDLE fh = (HANDLE)_get_osfhandle(fd);
> +	BY_HANDLE_FILE_INFORMATION fdata;
> +
> +	if (fh == INVALID_HANDLE_VALUE) {
> +		errno = EBADF;
> +		return -1;
> +	}
> +	/* direct non-file handles to cygwin's fstat() */
> +	if (GetFileType(fh) != FILE_TYPE_DISK)
> +		return fstat(fd, buf);

We never do that for stat/lstat.

-- Hannes