[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20130708183517.GJ29800@brightrain.aerifal.cx>
Date: Mon, 8 Jul 2013 14:35:17 -0400
From: Rich Felker <dalias@...ifal.cx>
To: musl@...ts.openwall.com
Cc: ian@...hack.net
Subject: Re: Compile error doing cross build on arm
On Mon, Jul 08, 2013 at 07:51:52PM +0200, Harald Becker wrote:
> Hi Ian !
>
> 08-07-2013 11:58 Ian Denhardt <ian@...hack.net>:
>
> > So I just hit a compilation error trying to cross compile musl
> > for arm:
> > -fno-stack-protector -c -o src/stdio/vfwprintf.o
> > src/stdio/vfwprintf.c src/stdio/vfwprintf.c: In function
> > 'wprintf_core': src/stdio/vfwprintf.c:195:3: error: pointer
> > targets in passing argument 2 of 'wcsspn' differ in signedness
> > [-Werror=pointer-sign] In file included from
> > src/stdio/vfwprintf.c:7:0: ./include/wchar.h:62:8: note:
> > expected 'const wchar_t *' but argument is of type 'int *' cc1:
>
> > CFLAGS= -Os -pipe -fomit-frame-pointer -fno-unwind-tables
> > -fno-asynchronous-unwind-tables -Wa,--noexecstack
> > -falign-functions=1 -falign-labels=1 -falign-loops=1
> > -falign-jumps=1 -Werror=implicit-function-declaration
> > -Werror=implicit-int -Werror=pointer-sign -Werror=pointer-arith
> > -fno-stack-protector CFLAGS_C99FSE = -std=c99 -nostdinc
> > -ffreestanding -fexcess-precision=standard -frounding-math
>
> It is curious, but I had similar compile error on native ARM
> compile (gcc 4.6.4). They vanished after removing some of the
> option on my CFLAGS list. Didn't analyze this, but probable
> candidates the unwind-tables and math options. So play a bit with
> your CFLAGS list, try "CFLAGS=-Os" if problem vanishes.
The direct source of this error is that the compiler's wchar_t is
defined as int rather than unsigned. ARM EABI requires (stupidly) that
wchar_t be unsigned, so we follow that.
The indirect cause of the error, and the real problem, is almost
surely that the compiler was configured for old, pre-EABI ARM ABI.
This ABI is NOT supported by musl, and will lead to various major
problems that could otherwise go undetected for a long time, like
misaligned structures (the old ABI only had 4-byte alignment for
64-bit types, for example). So this issue needs to be checked and
fixed before moving ahead.
Rich
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
