Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 16 Aug 2009 13:54:47 +0200
From:      Andreas Tobler <andreast-list@fgznet.ch>
To:        Joe Marcus Clarke <marcus@marcuscom.com>
Cc:        gecko@freebsd.org
Subject:   Re: firefox3 under powerpc not building
Message-ID:  <4A87F387.507@fgznet.ch>
In-Reply-To: <1250359762.23396.28.camel@shumai.marcuscom.com>
References:  <4A7F1AB0.9040605@fgznet.ch> <4A807B1B.20605@fgznet.ch> <1250359762.23396.28.camel@shumai.marcuscom.com>

next in thread | previous in thread | raw e-mail | index | archive | help
This is a multi-part message in MIME format.
--------------080205010802050808030005
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Joe Marcus Clarke wrote:
> On Mon, 2009-08-10 at 21:55 +0200, Andreas Tobler wrote:
>> Hi again,
>>
>> replying to myself.
>>
>>
>> Andreas Tobler wrote:
>>
>>> I can not build firefox3 (3.0.13 and up) under FreeBSD -CURRENT. As far 
>>> as I tracked down it is an issue in the firefox sources itself.
>>>
>>> The last working build I had was 3.0.10.
>>>
>>> Here in this area it breaks:
>>>
>>> http://mxr.mozilla.org/mozilla1.9.1/diff?file=/security/nss/lib/freebl/Makefile&diffvar=mozilla1.8.0
>>>
>>> The below snippet does somehow not work.
>>> -----
>>> # some code wants _X86_ defined for intel CPUs.
>>> # coreconf does this for windows, but not for Linux, FreeBSD, etc.
>>> ifeq (,$(filter-out x86 x86_64,$(CPU_ARCH)))
>>> ifneq (,$(filter-out WIN%,$(OS_TARGET)))
>>>          DEFINES += -D_X86_
>>> endif
>>> endif
>>> -----
>> It does not work on powerpc since the CPU_ARCH is set to x86 in 
>> security/coreconf/FreeBSD.mk, gah!
>>
>>> On powerpc I get the -D_X86_ passed to the compile options, very bad. 
>>> But I do not understand why. Might be that the first filter-out does not 
>>> work as expected?
>>>
>>> I compared with an x86 build and with an x86_64 build. On the x86 it 
>>> works as expected. But also under der x86_64 it does not do what I 
>>> expect. Under x86_64 I'd expect the -D_X86_ in the compile options, right?
>> The x86_64 case is also clear now, the CPU_ARCH is set to amd64, so the 
>> -D_X86_ is not set. We might need a patch to 
>> security/nss/lib/freebl/Makefile which also includes the amd64 to the 
>> first filter-out above? Like this one:
> 
> Thanks for these patches.  Can you resend them as attachments (or post
> them) so they can be easily extracted?

Here it is, the powerpc one. It is against 3.0.13 source base. The amd64 
I did not follow up since it does build, with or without -D_X86_ on amd64.

Andreas

--------------080205010802050808030005
Content-Type: text/plain; x-mac-type="54455854"; x-mac-creator="74657874";
	name="patch-security-coreconf-FreeBSD.mk"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
 filename="patch-security-coreconf-FreeBSD.mk"

--- security/coreconf/FreeBSD.mk.orig	2008-07-12 16:28:59.000000000 +0200
+++ security/coreconf/FreeBSD.mk	2009-08-10 22:30:54.000000000 +0200
@@ -45,8 +45,15 @@
 ifeq ($(OS_TEST),alpha)
 CPU_ARCH		= alpha
 else
+ifeq ($(OS_TEST),amd64)
+CPU_ARCH		= amd64
+else
+ifeq ($(OS_TEST),powerpc)
+CPU_ARCH		= powerpc
+else
 CPU_ARCH		= x86
 endif
+endif
 
 OS_CFLAGS		= $(DSO_CFLAGS) -ansi -Wall -Wno-switch -DFREEBSD -DHAVE_STRERROR -DHAVE_BSD_FLOCK
 
@@ -73,7 +80,7 @@
 DLL_SUFFIX		= so.1.0
 endif
 
-MKSHLIB			= $(CC) $(DSO_LDOPTS)
+MKSHLIB			= $(CC) -Wl,-Bsymbolic -lc $(DSO_LDOPTS) -o $@
 ifdef MAPFILE
 	MKSHLIB += -Wl,--version-script,$(MAPFILE)
 endif
@@ -82,4 +89,4 @@
 
 G++INCLUDES		= -I/usr/include/g++
 
-INCLUDES		+= -I/usr/X11R6/include
+#INCLUDES		+= -I/usr/local/include

--------------080205010802050808030005--



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4A87F387.507>