Date: Mon, 21 Jun 2010 13:33:40 -0400 From: Alexander Kabaev <kabaev@gmail.com> To: Bruce Evans <brde@optusnet.com.au> Cc: svn-src-head@freebsd.org, svn-src-all@freebsd.org, Marcel Moolenaar <marcel@freebsd.org>, src-committers@freebsd.org Subject: Re: svn commit: r209358 - head/cddl/contrib/opensolaris/lib/libdtrace/common Message-ID: <20100621133340.7501713b@kan.dnsalias.net> In-Reply-To: <20100622024652.C43995@delplex.bde.org> References: <201006200034.o5K0Y6xl041024@svn.freebsd.org> <20100622024652.C43995@delplex.bde.org>
next in thread | previous in thread | raw e-mail | index | archive | help
--Sig_/QdXti8V3Uvqn36DS029H5JY
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable
On Tue, 22 Jun 2010 03:22:40 +1000 (EST)
Bruce Evans <brde@optusnet.com.au> wrote:
> On Sun, 20 Jun 2010, Marcel Moolenaar wrote:
>=20
> > Log:
> > Unbreak platforms with char unsigned by default. Oddly enough, GCC
> > isn't satisfied with a simple cast to int in the check against EOF,
> > so the fix is a bit involved by actually having to go through a
> > temporary variable.
>=20
> Perhaps that is because gcc can see that the cast has no effect, so
> the comparison can never be true if `c' is an unsigned char (unless
> unsigned char has the same number of bits as signed int), but it
> cannot see that the conversion to the temporary variable has the same
> null effect.
>=20
> > Modified:
> > head/cddl/contrib/opensolaris/lib/libdtrace/common/dt_lex.l
> > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D
> > ---
> > head/cddl/contrib/opensolaris/lib/libdtrace/common/dt_lex.l
> > Sat Jun 19 22:13:40 2010 (r209357) +++
> > head/cddl/contrib/opensolaris/lib/libdtrace/common/dt_lex.l
> > Sun Jun 20 00:34:06 2010 (r209358) @@ -67,8 +67,12 @@
> > * for all subsequent invocations, which is the effect desired.
> > */
> > #undef unput
> > -#define unput(c) \
> > - if (c !=3D EOF) yyunput( c, yytext_ptr )
> > +#define unput(c) \
> > + do { \
> > + int _c =3D c; \
> > + if (_c !=3D EOF) \
> > + yyunput(_c, yytext_ptr); \
> > + } while(0)
> > #endif
> >
> > static int id_or_type(const char *);
> >
>=20
> This remains broken, especially on platforms with chars unsigned. No
> one should try to unput EOF, so c should never equal EOF, but if c is
> a negative character it may equal EOF and thus
>=20
> On platforms with chars unsigned (except exotic ones where chars have
> the same size as ints), if c is a char then it is >=3D 0 and thus cannot
> equal EOF (which is < 0). Since the platform is non-exotic, (int)c
> and "int _c =3D c;" equal c and are this still >=3D 0 and thus cannot
> equal EOF. Thus the comparison with EOF has no effect, and c is
> always unput.
>=20
> On platforms with chars signed, some chars may equal EOF. It is an
> error to unput almost any value held in a variable of type char,
> since its value might equal EOF and thus be rejected by unput(), but
> unput() should be able to handle any character in the character set.
>=20
> This problem is handled by ungetc() by always converting the value to
> unsigned char. Thus the value can never equal EOF, and the character
> set is effectively represented by unsigned char's, not the plain chars
> that stdio returns in some other interfaces (but not getc()).
>=20
> There seems to be no reason to break the warning about this instead of
> using the same approach as stdio. This depends on yyunput() not
> having similar bugs (it must take an arg of type int and convert to
> an unsigned cgar like ungetc()):
>=20
> #define unput(c) yyunput((unsigned char)(c), yytext_ptr)
>=20
> This also fixes the missing parantheses for 'c' and some style bugs.
>=20
> Bruce
DTrace _does_ try to unput EOF though and apparently gets away with it
on Solaris, so while yor version is correct, it is also useless.
--=20
Alexander Kabaev
--Sig_/QdXti8V3Uvqn36DS029H5JY
Content-Type: application/pgp-signature; name=signature.asc
Content-Disposition: attachment; filename=signature.asc
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.14 (FreeBSD)
iD8DBQFMH6J4Q6z1jMm+XZYRAggMAJ4qyCWuH+8bQAXQl5sbEEXq09Q2zwCfa7Op
NMCDlp149XmRXWob9zGeNts=
=ibEu
-----END PGP SIGNATURE-----
--Sig_/QdXti8V3Uvqn36DS029H5JY--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20100621133340.7501713b>