Anders Magnusson | 1 Dec 19:01 2010
Picon

Status update 2010-12-01

Overview of some of the things happened since last update:

- Fix a bunch of bugs related to floating point extremes (nan/inf)
   Also a ton of other ff/related bugs fixed.

- Make ssp not provide broken code sometimes.

- Implement TLS on amd64.  As of now always enabled.

- Fix boolean support, there were many places where it was broken.

+ many small bugfixes.

Also, much better support for Linux and FreeBSD platforms.

-- Ragge

Shin'ichi Ichikawa | 11 Dec 13:51 2010

Mac OS X 10.6 Fixed compilation can, but not use debug option(-g).

Hi, able to compile the attached patch !

$ pwd
/usr/local/lib/pcc/i386-apple-darwin10.5.0/0.9.9/lib

$ ls -lF
total 112
-rw-r--r--  1 root  wheel  20048 12 11 08:52 libpcc.a
-rw-r--r--  1 root  wheel  36320 12 11 08:52 libpccsoftfloat.a

$ sudo cp -p /usr/lib/crt1.* .
Password:

$ ls -lF
total 208
-rw-r--r--  1 root  wheel  13260  9 30  2009 crt1.10.5.o
-rw-r--r--  1 root  wheel  12984  9 30  2009 crt1.10.6.o
-rw-r--r--  1 root  wheel  15768  9 30  2009 crt1.o
-rw-r--r--  1 root  wheel  20048 12 11 08:52 libpcc.a
-rw-r--r--  1 root  wheel  36320 12 11 08:52 libpccsoftfloat.a

$ pcc pcc_test.c
$ cat pcc_test.c

#include <stdarg.h>
#include <stdio.h>

static int x(const int i, ...)
{
    static int x = 0;
(Continue reading)

Anders Magnusson | 11 Dec 15:32 2010
Picon

Re: Mac OS X 10.6 Fixed compilation can, but not use debug option(-g).

Hi,

On 12/11/2010 01:51 PM, Shin'ichi Ichikawa wrote:
> Hi, able to compile the attached patch !
>
Done, thanks!  ...with a few changes;
- use exname() to create symbols with a prepended _ instead of 
hand-crafting it.
- I assume .byte should be .space for zeroed area.

You may want to doublecheck my changes.

Thank you very much for your patches!

-- Ragge

> $ pwd
> /usr/local/lib/pcc/i386-apple-darwin10.5.0/0.9.9/lib
>
> $ ls -lF
> total 112
> -rw-r--r--  1 root  wheel  20048 12 11 08:52 libpcc.a
> -rw-r--r--  1 root  wheel  36320 12 11 08:52 libpccsoftfloat.a
>
> $ sudo cp -p /usr/lib/crt1.* .
> Password:
>
> $ ls -lF
> total 208
> -rw-r--r--  1 root  wheel  13260  9 30  2009 crt1.10.5.o
(Continue reading)

Thorsten Glaser | 11 Dec 15:45 2010
Picon

Re: Mac OS X 10.6 Fixed compilation can, but not use debug option(-g).

Shin'ichi Ichikawa dixit:

>+		loctbl[DATA] = "section .data.rel.rw,�"aw�"";

I doubt that.

bye,
//mirabilos
--

-- 
<ch> you introduced a merge commit        │<mika> % g rebase -i HEAD^^
<mika> sorry, no idea and rebasing just fscked │<mika> Segmentation
<ch> should have cloned into a clean repo      │  fault (core dumped)
<ch> if I rebase that now, it's really ugh     │<mika:#grml> wuahhhhhh

Anders Magnusson | 18 Dec 12:42 2010
Picon

Change to cpp expansion logic.

I have just checked in a change to how cpp does recursive expansions.
I hope this will fix a bunch of reported bugs but there may be 
unexpected side
effects, so please test cpp behavior.

-- Ragge

Shin'ichi Ichikawa | 29 Dec 12:01 2010

Re: Mac OS X 10.6 Fixed compilation can, but not use debug option(-g).

Hi, part do not understand well.

2010/12/11 Thorsten Glaser <tg <at> mirbsd.de>:
> Shin'ichi Ichikawa dixit:
>
>>+              loctbl[DATA] = "section .data.rel.rw,�"aw�"";
>
> I doubt that.

Shin'ichi Ichikawa | 29 Dec 11:59 2010

Re: Mac OS X 10.6 Fixed compilation can, but not use debug option(-g).

Hi, fixed typo and side effects of exname().

diff -rup pcc-101229/arch/amd64/code.c pcc/arch/amd64/code.c
--- pcc-101229/arch/amd64/code.c	2010-12-27 02:26:03.000000000 +0900
+++ pcc/arch/amd64/code.c	2010-12-29 19:48:45.000000000 +0900
 <at>  <at>  -397,8 +397,8  <at>  <at>  ejobcode(int flag )
 	if (flag)
 		return;

-#ifdef MACHOAPI
-#define PT(x)
+#ifdef MACHOABI
+#define PT(x) printf("_")
 #else
 #define	PT(x) printf(".type __pcc_" x ", <at> function\n")
 #endif

2010/12/11 Anders Magnusson <ragge <at> ludd.ltu.se>:
> Hi,
>
> On 12/11/2010 01:51 PM, Shin'ichi Ichikawa wrote:
>>
>> Hi, able to compile the attached patch !
>>
> Done, thanks!  ...with a few changes;
> - use exname() to create symbols with a prepended _ instead of hand-crafting
> it.
> - I assume .byte should be .space for zeroed area.
>
> You may want to doublecheck my changes.
(Continue reading)

Thorsten Glaser | 29 Dec 14:21 2010
Picon

Re: Mac OS X 10.6 Fixed compilation can, but not use debug option(-g).

Shin'ichi Ichikawa dixit:

>2010/12/11 Thorsten Glaser <tg <at> mirbsd.de>:
>> Shin'ichi Ichikawa dixit:
>>
>>>+              loctbl[DATA] = "section .data.rel.rw,�"aw�"";
>>
>> I doubt that.

>Hi, part do not understand well.

I believe, 「"section .data.rel.rw,�"aw�"";」 will not work ☺
maybe this was a ¥ confused with backslash?

When doing development, you really should work on ASCII (or,
at most, UTF-8).

bye,
//mirabilos
--

-- 
<ch> you introduced a merge commit        │<mika> % g rebase -i HEAD^^
<mika> sorry, no idea and rebasing just fscked │<mika> Segmentation
<ch> should have cloned into a clean repo      │  fault (core dumped)
<ch> if I rebase that now, it's really ugh     │<mika:#grml> wuahhhhhh

Shin'ichi Ichikawa | 29 Dec 14:42 2010

Re: Mac OS X 10.6 Fixed compilation can, but not use debug option(-g).

Yes, was a ¥ confused with backslash.
Problem of Japanese mail environment.
Sorry.

2010/12/29 Thorsten Glaser <tg <at> mirbsd.de>:
> Shin'ichi Ichikawa dixit:
>
>>2010/12/11 Thorsten Glaser <tg <at> mirbsd.de>:
>>> Shin'ichi Ichikawa dixit:
>>>
>>>>+              loctbl[DATA] = "section .data.rel.rw,�"aw�"";
>>>
>>> I doubt that.
>
>>Hi, part do not understand well.
>
> I believe, 「"section .data.rel.rw,�"aw�"";」 will not work ☺
> maybe this was a ¥ confused with backslash?
>
> When doing development, you really should work on ASCII (or,
> at most, UTF-8).
>
> bye,
> //mirabilos
> --
> <ch> you introduced a merge commit        │<mika> % g rebase -i HEAD^^
> <mika> sorry, no idea and rebasing just fscked │<mika> Segmentation
> <ch> should have cloned into a clean repo      │  fault (core dumped)
> <ch> if I rebase that now, it's really ugh     │<mika:#grml> wuahhhhhh
>
(Continue reading)

Shin'ichi Ichikawa | 29 Dec 15:17 2010

Re: Mac OS X 10.6 Fixed compilation can, but not use debug option(-g).

Following the problem of debugging information.

$ pcc -g  pcc_test.c
ld: warning: can't find atom for N_GSYM stabs ___stdinp:G15=*1 in
/tmp/ctm.6FCtvW
ld: warning: can't find atom for N_GSYM stabs ___stdoutp:G15 in /tmp/ctm.6FCtvW
ld: warning: can't find atom for N_GSYM stabs ___stderrp:G15 in /tmp/ctm.6FCtvW
ld: warning: can't find atom for N_GSYM stabs _sys_nerr:G1 in /tmp/ctm.6FCtvW
ld: warning: can't find atom for N_GSYM stabs
_sys_errlist:G16=ar1;0;-10202;17=*2 in /tmp/ctm.6FCtvW

$ pcc -g -S pcc_test.c
$ cat pcc_test.s

.LL82:
        .stabs "___stdinp:G15=*1",32,0,8,0
        .stabs "___stdoutp:G15",32,0,8,0
        .stabs "___stderrp:G15",32,0,8,0
        .stabs  "/usr/include//stdio.h",132,0,0,.LL83
.LL83:
        .stabs  "/usr/include//stdio.h",132,0,0,.LL84
.LL84:
        .stabs  "/usr/include//stdio.h",132,0,0,.LL85
.LL85:
        .stabs "_sys_nerr:G1",32,0,4,0
        .stabs "_sys_errlist:G16=ar1;0;-10202;17=*2",32,0,0,0
        .stabs  "/usr/include//stdio.h",132,0,0,.LL86

2010/12/29 Shin'ichi Ichikawa <chan <at> vivi-vine.org>:
> Hi, fixed typo and side effects of exname().
(Continue reading)


Gmane