From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bernhard Fischer Date: Thu, 3 Jul 2008 10:05:57 +0200 Subject: [Buildroot] uClibc missing limits.h In-Reply-To: <486C7A22.8000604@cetrtapot.si> References: <485F9581.2040204@cetrtapot.si> <486B4357.8060908@cetrtapot.si> <3E421C7434714D94B87F0D1F52CDEA8D@aeglos> <486C7A22.8000604@cetrtapot.si> Message-ID: <20080703080557.GA29943@mx.loc> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: buildroot@busybox.net On Thu, Jul 03, 2008 at 09:05:06AM +0200, Hinko Kocevar wrote: >Ulf Samuelsson wrote: >> >> >>> Hinko Kocevar wrote: >>>> Bernhard Fischer wrote: >>>>> On Mon, Jun 23, 2008 at 02:22:25PM +0200, Hinko Kocevar wrote: >>>>>> Hi, >>>>>> >>>>>> I've found a way for uClibc to find headers that are not in the >>>>>> $(shell $CC -print-file-name=include), but in $(shell $CC >>>>>> -print-file-name=include)-fixed. >>>>>> According to >>>>>> http://sourceware.org/ml/libc-alpha/2007-03/msg00017.html LIBC >>>>>> should also look in other gcc provided header directories (eg. >>>>>> include-fixed) instead only relying on 'include' to contain all >>>>>> correct headers. >>>>>> >>>>>> Below is uClibc patch for this (tested on cris). >>>>> Well, that's not really correct since the default search path already >>>>> should to be >>>>> $target_triple/$ver/include \ >>>>> $target_triple/$ver/include-fixed \ >>>>> >> >> Removing "-nostdinc" from CFLAGS in "uClibc-0.9.29/Rules.mak" >> allows the ARM compilation to continue, > >But is this OK? no. > >> but it again bombs since the function "testandset" has multiple >> definitions. >> Detected in libc_pthread_init.c and rpc_thread.c. >> It is "first defined" in "forward.c" > >I guess this would be the problem in uClibc-0.9.29/libpthread/linuxthreads.old/sysdeps/arm/pt-machine.h. > >Dirty hack (works on cris arch), change: ># define PT_EI extern inline > >to > ># define PT_EI static inline Already fixed in svn trunk.