On Sun, Sep 25, 2011 at 2:36 PM, Joern Rennecke <amyl...@spamcop.net> wrote:
> This patch has not been reviewed for eight weeks.
>
> ----- Forwarded message from amyl...@spamcop.net -----
> Date: Mon, 01 Aug 2011 01:01:13 -0400
> From: Joern Rennecke <amyl...@spamcop.net>
> Reply-To: Joern Rennecke <amyl...@spamcop.net>
> Subject: RFA: Testsuite fixes (2/3): struct padding / alignment
> To: gcc-patches@gcc.gnu.org
>
> I am working with a target port that adds alignment and/or padding to
> certain
> structures. This causes some tests to fail that assume such padding does
> not exist.
>
> Fixed by adding a packed attribute to the structures in question.
I think you should instead dg-skip the tests on your target. At least ...
> Regression tested with x86_64-unknown-linux-gnu X sh-elf .
> No change there, as there is no such added alignment / padding by default
> on that target. (Although there is the -mpadstruct option; but you'd need
> to build appropriate mutilibs to test that.)
>
>
> ----- End forwarded message -----
>
>
> 2011-07-27 Joern Rennecke <joern.renne...@embecosm.com>
>
> * g++.dg/ext/strncpy-chk1.C (struct A): Add packed attribute.
>
> 2011-07-21 Joern Rennecke <joern.renne...@embecosm.com>
>
> * gcc.dg/builtin-object-size-10.c (drone_source_packet): Make packed.
>
> * gcc.dg/builtin-stringop-chk-1.c (t): Make packed.
>
> Index: gcc.dg/builtin-object-size-10.c
> ===================================================================
> --- gcc.dg/builtin-object-size-10.c
> (.../GNU/fsf-gcc/trunk/gcc/testsuite) (revision 2013)
> +++ gcc.dg/builtin-object-size-10.c
> (.../Customer-Projects/X/branches/Y-20110628/toolchain/gcc-4.7.0-20110701/gcc/testsuite)
> (revision 2013)
> @@ -8,7 +8,7 @@
> typedef struct {
> char type_str[16];
> char channel_hop;
> -} drone_source_packet;
> +} __attribute__((packed)) drone_source_packet;
> drone_packet *
> foo(char *x)
> {
> Index: gcc.dg/builtin-stringop-chk-1.c
> ===================================================================
> --- gcc.dg/builtin-stringop-chk-1.c
> (.../GNU/fsf-gcc/trunk/gcc/testsuite) (revision 2013)
> +++ gcc.dg/builtin-stringop-chk-1.c
> (.../Customer-Projects/X/branches/Y-20110628/toolchain/gcc-4.7.0-20110701/gcc/testsuite)
> (revision 2013)
> @@ -104,7 +104,7 @@ test2 (const H h)
> S *s[3];
> memset (s, 0, sizeof (S) * 3); /* { dg-warning "will always overflow"
> "memset" } */
>
> - struct T { char a[8]; char b[4]; char c[10]; } t;
> + struct T { char a[8]; char b[4]; char c[10]; } __attribute__((packed)) t;
> stpcpy (t.c,"Testing..."); /* { dg-warning "will always overflow" "stpcpy"
... for this case I don't see how padding should affect
builtin-object-size. If it does,
it looks like a bug in the object-size pass.
Richard.
> } */
>
> char b1[7];
> Index: g++.dg/ext/strncpy-chk1.C
> ===================================================================
> --- g++.dg/ext/strncpy-chk1.C (.../GNU/fsf-gcc/trunk/gcc/testsuite)
> (revision 2013)
> +++ g++.dg/ext/strncpy-chk1.C
> (.../Customer-Projects/X/branches/Y-20110628/toolchain/gcc-4.7.0-20110701/gcc/testsuite)
> (revision 2013)
> @@ -2,7 +2,7 @@
> // { dg-do compile }
> // { dg-options "-O2" }
>
> -struct A { char x[12], y[35]; };
> +struct A { char x[12], y[35]; } __attribute__((packed));
> struct B { char z[50]; };
>
> inline void
>
>
