Message-ID: <20110106075553.GC15914@linux-sh.org>
Date:	Thu, 6 Jan 2011 16:55:53 +0900
From:	Paul Mundt <lethal@...ux-sh.org>
To:	Guan Xuetao <guanxuetao@...c.pku.edu.cn>
Cc:	linux-arch@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCHv1 01/12] unicore32 core architecture: build infrastructure

On Sun, Dec 26, 2010 at 02:41:58AM +0800, Guan Xuetao wrote:
> diff --git a/arch/unicore32/Kconfig b/arch/unicore32/Kconfig
> new file mode 100644
> index 0000000..da59420
> --- /dev/null
> +++ b/arch/unicore32/Kconfig
> @@ -0,0 +1,258 @@
> +config UNICORE32
> +	bool
> +	default y
> +	select HAVE_MEMBLOCK
> +	select HAVE_GENERIC_DMA_COHERENT
> +	select HAVE_DMA_ATTRS
> +	select HAVE_KERNEL_GZIP
> +	select HAVE_KERNEL_BZIP2
> +	select HAVE_KERNEL_LZO
> +	select HAVE_KERNEL_LZMA
> +	select GENERIC_FIND_FIRST_BIT
> +	select ARCH_WANT_FRAME_POINTERS

You want to select HAVE_GENERIC_HARDIRQS here.

> +config GENERIC_HARDIRQS
> +	bool
> +	default y
> +
> +config GENERIC_HARDIRQS_NO__DO_IRQ
> +	def_bool y
> +

Then you can get rid of these, and source kernel/irq/Kconfig.

> +# arch/unicore32/Makefile
> +#
> +# This file is included by the global makefile so that you can add your own
> +# architecture-specific flags and dependencies.
> +#
> +# This file is subject to the terms and conditions of the GNU General Public
> +# License.  See the file "COPYING" in the main directory of this archive
> +# for more details.
> +#
> +# Copyright (C) 2002~2010 by Guan Xue-tao
> +LOCALVERSION	:= -uc32
> +
> +CROSS_COMPILE	:= /usr/unicore/gnu-toolchain-unicore/uc4/bin/unicore32-linux-
> +
This isn't terribly portable. You can elect to tie in cc-cross-prefix
like other architectures, or just drop it entirely and store it in your
.config (see the top-level Makefile for generic CROSS_COMPILE handling
options).

> +INSTALL_MOD_PATH := $(obj)/usr
> +
This and the LOCALVERSION specifications are pretty nasty surprises for
users. This is all handled generically for a reason, so please kill off
of this off.

> +LDFLAGS_vmlinux	:=-p --no-undefined -X
> +
> +OBJCOPYFLAGS	:=-O binary -R .note -R .note.gnu.build-id -R .comment -S
> +GZFLAGS		:=-9
> +#KBUILD_CFLAGS	+=-pipe
> +# Explicitly specifiy 32-bit UniCore ISA:
> +KBUILD_CFLAGS	+=$(call cc-option,-municore,)
> +
> +# Never generate .eh_frame
> +KBUILD_CFLAGS	+= $(call cc-option,-fno-dwarf2-cfi-asm)
> +
> +ifeq ($(CONFIG_FRAME_POINTER),y)
> +KBUILD_CFLAGS	+=-fno-omit-frame-pointer -mno-sched-prolog
> +endif
> +
-fno-omit-frame-pointer is handled generically, so there is no need for
it here.

> +KBUILD_CPPFLAGS	+= -mlittle-endian
> +AS		+= -EL
> +LD		+= -EL
> +
No. If you need to set AS and LD flags for endianness, then do so
explicitly. See how every other architecture is doing this. You may also
wish to read through Documentation/kbuild/makefiles.txt carefully.

> +comma = ,
> +
This is unused?

> +# Need -Uunicore32 for gcc < 3.x
> +# delete -mstructure-size-boundary=32, and default being 8
> +# delete -funsigned-char, and default being signed-char
> +KBUILD_CFLAGS	+=$(call cc-option,-mshort-load-bytes,$(call cc-option,-malignment-traps,)) -msoft-float -Uunicore32
> +KBUILD_AFLAGS	+=-msoft-float
> +
Is it realistic to even support gcc 2.x versions on a recent kernel?

> +drivers-$(CONFIG_ARCH_PUV3)	+= drivers/staging/puv3/
> +
> +libs-y				+= arch/unicore32/lib/
> +libs-y				+= $(shell $(CC) $(KBUILD_CFLAGS) -print-file-name=libc.a)
> +libs-y				+= $(shell $(CC) $(KBUILD_CFLAGS) -print-file-name=libgcc.a)
> +
The libgcc thing is not too surprising, but you do have
-print-libgcc-file-name for this. That libc.a thing however needs some
explaining.

> diff --git a/arch/unicore32/kernel/asm-offsets.c b/arch/unicore32/kernel/asm-offsets.c
> new file mode 100644
> index 0000000..ffcbe75
> --- /dev/null
> +++ b/arch/unicore32/kernel/asm-offsets.c
> +/*
> + * GCC 3.0, 3.1: general bad code generation.
> + * GCC 3.2.0: incorrect function argument offset calculation.
> + * GCC 3.2.x: miscompiles NEW_AUX_ENT in fs/binfmt_elf.c
> + *	(http://gcc.gnu.org/PR8896) and incorrect structure
> + *		initialisation in fs/jffs2/erase.c
> + */
> +#if (__GNUC__ < 4)
> +#error Your compiler should upgrade to uc4
> +#error	Known good compilers: 4.2.2
> +#endif
> +
If your compiler situation is this screwed up, then you need to be
erroring out in the Makefile, and you can most certainly kill off that
gcc 2.x cruft.

> +++ b/arch/unicore32/kernel/vmlinux.lds.S
> @@ -0,0 +1,120 @@
> +/*
> + * linux/arch/unicore32/kernel/vmlinux.lds.S
> + *
> + * Code specific to PKUnity SoC and UniCore ISA
> + *
> + * Copyright (C) 2001-2010 GUAN Xue-tao
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 as
> + * published by the Free Software Foundation.
> + */
> +
> +#include <asm-generic/vmlinux.lds.h>
> +#include <asm/thread_info.h>
> +#include <asm/memory.h>
> +#include <asm/page.h>
> +
> +OUTPUT_ARCH(unicore32)
> +ENTRY(stext)
> +
> +jiffies = jiffies_64;
> +
> +SECTIONS
> +{
> +	. = PAGE_OFFSET + TEXT_OFFSET;
> +
> +	.init : {			/* Init code and data		*/
> +		_stext = .;
> +		_sinittext = .;
> +			HEAD_TEXT
> +			INIT_TEXT
> +		_einittext = .;
> +
> +		INIT_SETUP(16)
> +
> +		INIT_CALLS
> +		CON_INITCALL
> +		SECURITY_INITCALL
> +		INIT_RAM_FS
> +
> +		__init_begin = _stext;
> +		INIT_DATA
> +	}
> +
This looks really broken, please take a look at asm-generic/vmlinux.lds.h
and use that as a model. Any reasons you have for needing to deviate from
that should be thoroughly explained.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists