Here's a patch which streamlines the dts contents a bit:
- Every "thing" should have an "enable;" line (or am I missing some
reason why they shouldn't?)
- Every "thing" should have the "enable;" as first entry, if it has one.
- Constructors should come before other stuff, but after the "enable;".
Comments?
Does the order in which variables are defined (e.g. ide0_enable) and in
which the constructors are called matter? I.e. is there a semantic
difference between
{
ide0_enable = "0";
ide1_enable = "0";
constructor = "i82371eb_constructors";
};
and
{
constructor = "i82371eb_constructors";
ide0_enable = "0";
ide1_enable = "0";
};
?
See patch for my proposal, but please note that it may be incorrent,
that's why I'm asking...
Uwe.
--
http://www.hermann-uwe.de | http://www.holsham-traders.dehttp://www.crazy-hacks.org | http://www.unmaintained-free-software.org
On Fri, Jun 29, 2007 at 04:36:03PM +0200, svn(a)openbios.org wrote:
> +#define ARRAY_SIZE(a) (sizeof(a) / sizeof((a)[0]))
Need extra () around a or no?
sizeof((a)) / sizeof(((a))[0])) ?
//Peter
* fix copyright headers to v2 only as suggested by Uwe
* drop silly code when not using SMP
* fix device/device.c
* drop spinlock_t, use struct spinlock instead.
Signed-off-by: Stefan Reinauer <stepan(a)coresystems.de>
Index: include/spinlock.h
===================================================================
--- include/spinlock.h (revision 427)
+++ include/spinlock.h (working copy)
@@ -2,11 +2,11 @@
* This file is part of the LinuxBIOS project.
*
* Copyright (C) 2001 Linux Networx
+ * Copyright (C) 2007 coresystems GmbH
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
+ * the Free Software Foundation; version 2 of the License.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
@@ -22,23 +22,17 @@
#define SPINLOCK_H
#if defined(CONFIG_SMP) && CONFIG_SMP == 1
+
#include <arch/spinlock.h>
+
#else /* !CONFIG_SMP */
-/* Most GCC versions have a nasty bug with empty initializers */
-#if (__GNUC__ > 2)
-typedef struct { } spinlock_t;
-#define SPIN_LOCK_UNLOCKED (spinlock_t) { }
-#else
-typedef struct { int gcc_is_buggy; } spinlock_t;
-#define SPIN_LOCK_UNLOCKED (spinlock_t) { 0 }
-#endif
-
#define barrier() do {} while(0)
#define spin_is_locked(lock) 0
#define spin_unlock_wait(lock) do {} while(0)
#define spin_lock(lock) do {} while(0)
#define spin_unlock(lock) do {} while(0)
+#define spin_define(lock) /* empty */
#endif
#endif /* SPINLOCK_H */
Index: include/arch/x86/arch/spinlock.h
===================================================================
--- include/arch/x86/arch/spinlock.h (revision 427)
+++ include/arch/x86/arch/spinlock.h (working copy)
@@ -2,11 +2,11 @@
* This file is part of the LinuxBIOS project.
*
* Copyright (C) 2001 Linux Networx
+ * Copyright (C) 2007 coresystems GmbH
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
+ * the Free Software Foundation; version 2 of the License.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
@@ -25,12 +25,12 @@
* Your basic SMP spinlocks, allowing only a single CPU anywhere
*/
-typedef struct {
+struct spinlock {
volatile unsigned int lock;
-} spinlock_t;
+};
-#define SPIN_LOCK_UNLOCKED (spinlock_t) { 1 }
+#define SPIN_LOCK_UNLOCKED (struct spinlock) { 1 }
/*
* Simple spin lock operations. There are two variants, one clears IRQ's
@@ -60,18 +60,21 @@
#define spin_unlock_string \
"movb $1,%0"
-static inline __attribute__((always_inline)) void spin_lock(spinlock_t *lock)
+static inline __attribute__((always_inline)) void spin_lock(struct spinlock *lock)
{
__asm__ __volatile__(
spin_lock_string
:"=m" (lock->lock) : : "memory");
}
-static inline __attribute__((always_inline)) void spin_unlock(spinlock_t *lock)
+static inline __attribute__((always_inline)) void spin_unlock(struct spinlock *lock)
{
__asm__ __volatile__(
spin_unlock_string
:"=m" (lock->lock) : : "memory");
}
+#define spin_define(spin) static struct spinlock spin = SPIN_LOCK_UNLOCKED
+
+
#endif /* ARCH_SPINLOCK_H */
Index: device/device.c
===================================================================
--- device/device.c (revision 427)
+++ device/device.c (working copy)
@@ -183,7 +183,9 @@
* @return Pointer to the newly created device structure.
* @see device_path
*/
-// static spinlock_t dev_lock = SPIN_LOCK_UNLOCKED;
+
+spin_define(dev_lock);
+
struct device *alloc_dev(struct bus *parent, struct device_path *path,
struct device_id *devid)
{
--
coresystems GmbH • Brahmsstr. 16 • D-79104 Freiburg i. Br.
Tel.: +49 761 7668825 • Fax: +49 761 7664613
Email: info(a)coresystems.de • http://www.coresystems.de/
Registergericht: Amtsgericht Freiburg • HRB 7656
Geschäftsführer: Stefan Reinauer • Ust-IdNr.: DE245674866
Hi,
this patch puts the lar size in the bootblock and reads it from there.
Why? This way we don't need to recompile the image when the size of the
LinuxBIOS image changes. This alows building images for 50 motherboards
and equipping each with 10 payloads, resulting in 500 images while you
only have to build each payload once and each motherboard, too.
There's also a small "fix" allowing the compression type to be case
insensitive. Not really relevant I guess.
Signed-off-by: Stefan Reinauer <stepan(a)coresystems.de>
Index: util/lar/bootblock.c
===================================================================
--- util/lar/bootblock.c (revision 426)
+++ util/lar/bootblock.c (working copy)
@@ -74,16 +74,16 @@
int fixup_bootblock(void)
{
- /* Per definition the bootblock starts with 256 empty bytes.
- * These are utilized to make the bootblock part of a lar file,
- * and store the image size.
- *
- * We will also calculate a checksum here.
- */
+ int i;
+ uint32_t *size_pos;
- /* first try. Clear out ugly left-over from ld hack */
- bootblock_code[bootblock_len - 13] = '\0';
- bootblock_code[bootblock_len - 12] = '\0';
+ /* This cleans out the area after the reset vector */
+ for(i=13; i>0; i--)
+ bootblock_code[bootblock_len - i] = '\0';
+
+ /* add lar size to image */
+ size_pos=(uint32_t *)(bootblock_code+bootblock_len-12);
+ size_pos[0] = get_larsize();
return 0;
}
Index: util/lar/lar.c
===================================================================
--- util/lar/lar.c (revision 426)
+++ util/lar/lar.c (working copy)
@@ -96,10 +96,10 @@
larmode = CREATE;
break;
case 'C':
- if (strcmp("lzma", optarg) == 0) {
+ if (strcasecmp("lzma", optarg) == 0) {
algo = lzma;
}
- if (strcmp("nrv2b", optarg) == 0) {
+ if (strcasecmp("nrv2b", optarg) == 0) {
algo = nrv2b;
}
break;
Index: arch/x86/stage1.c
===================================================================
--- arch/x86/stage1.c (revision 426)
+++ arch/x86/stage1.c (working copy)
@@ -101,8 +101,12 @@
// FIXME this should be defined in the VPD area
// but NOT IN THE CODE.
- archive.len=(CONFIG_LINUXBIOS_ROMSIZE_KB)*1024;
- archive.start=(void *)(0UL-(CONFIG_LINUXBIOS_ROMSIZE_KB*1024));
+ /* The len field starts behind the reset vector on x86.
+ * The start is not correct for all platforms. sc520 will
+ * need some hands on here.
+ */
+ archive.len = *(u32 *)0xfffffff4;
+ archive.start =(void *)(0UL-archive.len);
// FIXME check integrity
--
coresystems GmbH • Brahmsstr. 16 • D-79104 Freiburg i. Br.
Tel.: +49 761 7668825 • Fax: +49 761 7664613
Email: info(a)coresystems.de • http://www.coresystems.de/
Hi,
I got quite far with adding decompression to lar (lzma'd data runs fine,
nrv2b decompression seems to hang), but ran into issues that need
discussion.
Currently, two files in the lar archive are parsed directly out of rom
(using the find_file function which returns a pointer to there).
One of them is "option_table", a 900 byte file (which compresses to 350
or so), where I could live with it staying uncompressed in image.
The other one is the payload.
My current work around, is to just define a place in memory where to
decode it to and then load the ELF image from there.
This complicates ELF loading as now there is a risk that the raw ELF
file and the ELF image overlap. (Right now, I just use an address that I
know is beyond the ELF's range)
I propose a new API for the ELF loader that hides this issue and ensures
that decompression works fine. It would be used like this:
copy_elf(&archive, "normal/payload", &entrypoint, &last_address)
This function looks up "normal/payload" in the archive starting at
archive, and returns success in the return value. entrypoint could then
be used to jump into the image. last_address contains the last byte used
by the ELF image.
The rationale for returning the entrypoint and last_address is to allow
further loads, so multipart payloads can be used, which provides a
cleaner and simpler way to allow modular payloads.
Specifically, I'd like to see this for the grub2 support I work on, as
it's very trivial to just push the modules as "payload.data" into the
lar archive, load it right behind the ELF image and have grub2 find its
modules at _end.
Regards,
Patrick Georgi
On Fri, Jun 29, 2007 at 06:57:23PM +0200, svn(a)openbios.org wrote:
> * start using arch/foo.h again instead of archfoo.h (trivial)
> * make constructor an initializer.
I don't understand. What's an initializer in this context?
> + if(c && c->ops && c->ops->constructor)
> + c->ops->constructor(dev, c);
> + else
> + printk(BIOS_INFO, "No constructor called for %s.\n",
> + dev_id_string(&c->id));
BIOS_DEBUG?
> Copied: LinuxBIOSv3/include/arch/x86/arch/elf.h (from rev 417, LinuxBIOSv3/include/arch/x86/archelf.h)
> ===================================================================
> --- LinuxBIOSv3/include/arch/x86/arch/elf.h (rev 0)
> +++ LinuxBIOSv3/include/arch/x86/arch/elf.h 2007-06-29 16:57:23 UTC (rev 418)
Missing license header. Trivial file, so feel free to add yourself as
copyright owner.
> @@ -0,0 +1,8 @@
> +#ifndef ARCH_ELF_H
> +#define ARCH_ELF_H
Following our current guidelines this should be ARCH_X86_ARCH_ELF_H,
which looks a bit stupid, but I think it's necessary.
ARCH_ELF_H doesn't work as we'd have another ARCH_ELF_H if we add
another architecture...
> +
> +#define ELF_CLASS ELFCLASS32
> +#define ELF_DATA ELFDATA2LSB
> +#define ELF_ARCH EM_386
> +
> +#endif /* ARCH_ELF_H */
>
> Added: LinuxBIOSv3/include/arch/x86/arch/spinlock.h
> ===================================================================
> --- LinuxBIOSv3/include/arch/x86/arch/spinlock.h (rev 0)
> +++ LinuxBIOSv3/include/arch/x86/arch/spinlock.h 2007-06-29 16:57:23 UTC (rev 418)
> @@ -0,0 +1,83 @@
> +/*
> + * This file is part of the LinuxBIOS project.
> + *
> + * Copyright (C) 2001 Linux Networx
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
This should be GPLv2, I think. We know from email conversation with
Eric Biederman / Linux Networx that their code is GPLv2 only.
But this file is based on Linux code anyway, thus definately GPLv2,
I'll update the header and HACKING accordingly (later).
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program; if not, write to the Free Software
> + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
> + */
> +
> +#ifndef ARCH_SPINLOCK_H
> +#define ARCH_SPINLOCK_H
See above.
> +
> +/*
> + * Your basic SMP spinlocks, allowing only a single CPU anywhere
> + */
> +
> +typedef struct {
> + volatile unsigned int lock;
Why volatile? I think the Linux file doesn't have that.
> +} spinlock_t;
Drop spinlock_t and use 'struct spinlock' as we do for other structs?
> +
> +
> +#define SPIN_LOCK_UNLOCKED (spinlock_t) { 1 }
> +
> +/*
> + * Simple spin lock operations. There are two variants, one clears IRQ's
> + * on the local processor, one does not.
> + *
> + * We make no fairness assumptions. They have a cost.
> + */
> +#define barrier() __asm__ __volatile__("": : :"memory")
#define barrier() (__asm__ __volatile__("": : :"memory"))
(better safe than sorry)
> +#define spin_is_locked(x) (*(volatile char *)(&(x)->lock) <= 0)
> +#define spin_unlock_wait(x) do { barrier(); } while(spin_is_locked(x))
#define spin_unlock_wait(x) (do { barrier(); } while(spin_is_locked(x)))
> +
> +#define spin_lock_string \
> + "\n1:\t" \
> + "lock ; decb %0\n\t" \
> + "js 2f\n" \
> + ".section .text.lock,\"ax\"\n" \
> + "2:\t" \
> + "cmpb $0,%0\n\t" \
> + "rep;nop\n\t" \
> + "jle 2b\n\t" \
> + "jmp 1b\n" \
> + ".previous"
> +
> +/*
> + * This works. Despite all the confusion.
> + */
> +#define spin_unlock_string \
> + "movb $1,%0"
Shall we drop these #defines? They're only used in one place and a small
comment which says what the asm does should be enough.
> +static inline __attribute__((always_inline)) void spin_lock(spinlock_t *lock)
> +{
> + __asm__ __volatile__(
> + spin_lock_string
> + :"=m" (lock->lock) : : "memory");
> +}
> +
> +static inline __attribute__((always_inline)) void spin_unlock(spinlock_t *lock)
> +{
> + __asm__ __volatile__(
> + spin_unlock_string
> + :"=m" (lock->lock) : : "memory");
> +}
> Modified: LinuxBIOSv3/util/dtc/dtc-parser.y
> ===================================================================
> --- LinuxBIOSv3/util/dtc/dtc-parser.y 2007-06-29 15:32:19 UTC (rev 417)
> +++ LinuxBIOSv3/util/dtc/dtc-parser.y 2007-06-29 16:57:23 UTC (rev 418)
> @@ -142,7 +142,7 @@
> }
> ')' ';' {
> /* convention: first property is labeled with path */
> - $6->label = strdup($3.val);
> + $6->label = strdup((char *)$3.val);
Is this the correct fix for the warning? I didn't notice this warning
before so it must have been introduced by some recent change?
Uwe.
--
http://www.hermann-uwe.de | http://www.holsham-traders.dehttp://www.crazy-hacks.org | http://www.unmaintained-free-software.org
Author: stepan
Date: 2007-07-01 00:55:10 +0200 (Sun, 01 Jul 2007)
New Revision: 429
Modified:
LinuxBIOSv3/device/device.c
LinuxBIOSv3/include/arch/x86/arch/spinlock.h
LinuxBIOSv3/include/spinlock.h
Log:
* fix copyright headers to v2 only as suggested by Uwe
* drop silly code when not using SMP
* fix device/device.c
* drop spinlock_t, use struct spinlock instead.
Signed-off-by: Stefan Reinauer <stepan(a)coresystems.de>
Acked-by: Ronald G. Minnich <rminnich(a)gmail.com>
one kitten died already! ;-)
Modified: LinuxBIOSv3/device/device.c
===================================================================
--- LinuxBIOSv3/device/device.c 2007-06-30 22:09:57 UTC (rev 428)
+++ LinuxBIOSv3/device/device.c 2007-06-30 22:55:10 UTC (rev 429)
@@ -183,7 +183,9 @@
* @return Pointer to the newly created device structure.
* @see device_path
*/
-// static spinlock_t dev_lock = SPIN_LOCK_UNLOCKED;
+
+spin_define(dev_lock);
+
struct device *alloc_dev(struct bus *parent, struct device_path *path,
struct device_id *devid)
{
Modified: LinuxBIOSv3/include/arch/x86/arch/spinlock.h
===================================================================
--- LinuxBIOSv3/include/arch/x86/arch/spinlock.h 2007-06-30 22:09:57 UTC (rev 428)
+++ LinuxBIOSv3/include/arch/x86/arch/spinlock.h 2007-06-30 22:55:10 UTC (rev 429)
@@ -2,11 +2,11 @@
* This file is part of the LinuxBIOS project.
*
* Copyright (C) 2001 Linux Networx
+ * Copyright (C) 2007 coresystems GmbH
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
+ * the Free Software Foundation; version 2 of the License.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
@@ -25,12 +25,12 @@
* Your basic SMP spinlocks, allowing only a single CPU anywhere
*/
-typedef struct {
+struct spinlock {
volatile unsigned int lock;
-} spinlock_t;
+};
-#define SPIN_LOCK_UNLOCKED (spinlock_t) { 1 }
+#define SPIN_LOCK_UNLOCKED (struct spinlock) { 1 }
/*
* Simple spin lock operations. There are two variants, one clears IRQ's
@@ -60,18 +60,21 @@
#define spin_unlock_string \
"movb $1,%0"
-static inline __attribute__((always_inline)) void spin_lock(spinlock_t *lock)
+static inline __attribute__((always_inline)) void spin_lock(struct spinlock *lock)
{
__asm__ __volatile__(
spin_lock_string
:"=m" (lock->lock) : : "memory");
}
-static inline __attribute__((always_inline)) void spin_unlock(spinlock_t *lock)
+static inline __attribute__((always_inline)) void spin_unlock(struct spinlock *lock)
{
__asm__ __volatile__(
spin_unlock_string
:"=m" (lock->lock) : : "memory");
}
+#define spin_define(spin) static struct spinlock spin = SPIN_LOCK_UNLOCKED
+
+
#endif /* ARCH_SPINLOCK_H */
Modified: LinuxBIOSv3/include/spinlock.h
===================================================================
--- LinuxBIOSv3/include/spinlock.h 2007-06-30 22:09:57 UTC (rev 428)
+++ LinuxBIOSv3/include/spinlock.h 2007-06-30 22:55:10 UTC (rev 429)
@@ -2,11 +2,11 @@
* This file is part of the LinuxBIOS project.
*
* Copyright (C) 2001 Linux Networx
+ * Copyright (C) 2007 coresystems GmbH
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
+ * the Free Software Foundation; version 2 of the License.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
@@ -22,23 +22,17 @@
#define SPINLOCK_H
#if defined(CONFIG_SMP) && CONFIG_SMP == 1
+
#include <arch/spinlock.h>
+
#else /* !CONFIG_SMP */
-/* Most GCC versions have a nasty bug with empty initializers */
-#if (__GNUC__ > 2)
-typedef struct { } spinlock_t;
-#define SPIN_LOCK_UNLOCKED (spinlock_t) { }
-#else
-typedef struct { int gcc_is_buggy; } spinlock_t;
-#define SPIN_LOCK_UNLOCKED (spinlock_t) { 0 }
-#endif
-
#define barrier() do {} while(0)
#define spin_is_locked(lock) 0
#define spin_unlock_wait(lock) do {} while(0)
#define spin_lock(lock) do {} while(0)
#define spin_unlock(lock) do {} while(0)
+#define spin_define(lock) /* empty */
#endif
#endif /* SPINLOCK_H */
On Sun, Jul 01, 2007 at 12:03:39AM +0200, svn(a)openbios.org wrote:
> Author: stepan
> Date: 2007-07-01 00:03:39 +0200 (Sun, 01 Jul 2007)
> New Revision: 427
Patch sent: Sat, 30 Jun 2007 23:58:15 +0200
ACK: Sat, 30 Jun 2007 15:00:18 -0700
Commit: Sun, 1 Jul 2007 00:03:39 +0200
2:03 minutes til ack, 3:21 til commit.
*cough*
My theory: you guys are stuck in a room with lots of kittens and a
terrorist. Every time a patch doesn't get committed in less than
5 minutes the terrorist kills a kitten.
You must love kittens quite a lot :)
Uwe.
--
http://www.hermann-uwe.de | http://www.holsham-traders.dehttp://www.crazy-hacks.org | http://www.unmaintained-free-software.org
