[PATCH 0/2] x86 minor fixes and questions

[PATCH 0/2] x86 minor fixes and questions

[Thomas Petazzoni]

Hello,

Here are two minor fixes for the x86 support. The first patch avoids
linking problems on recent gcc versions, and the second one fixes the
documentation.

Also, I've tested the x86 support under Qemu, it starts fine. Thanks
to the bios disk driver, it detects one hard disk drive. However,
since the driver doesn't set num_blocks, the disk size is zero, and no
partition can be detected. Is this bios disk driver supposed to work
to read the entire hard drive, access a filesystem and load the kernel
image, or is the x86 support generally too limited for this at the
moment?

Thanks,

Thomas

Thomas Petazzoni (2):
  scripts/setupmbr: fix documentation
  arch/x86: fix link with recent gcc

 arch/x86/Makefile           | 3 ++-
 scripts/setupmbr/setupmbr.c | 4 ++--
 2 files changed, 4 insertions(+), 3 deletions(-)

-- 
1.8.1.2


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

[PATCH 1/2] scripts/setupmbr: fix documentation

[Thomas Petazzoni]

The documentation says setupmbr should be used with the image named
'barebox', but it should instead be done with the 'barebox.bin'
image. Attempting to use setupmbr on the ELF barebox image simply
fails:

$ ./scripts/setupmbr/setupmbr -s 32 -m ./barebox -d disk.img
No MBR signature found
barebox image seems not valid: Bad MBR signature

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
---
 scripts/setupmbr/setupmbr.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/scripts/setupmbr/setupmbr.c b/scripts/setupmbr/setupmbr.c
index 918fbe3..0060c2e 100644
--- a/scripts/setupmbr/setupmbr.c
+++ b/scripts/setupmbr/setupmbr.c
@@ -666,10 +666,10 @@ and populating the root filesystem for example) to make it useful.
 In the next step, @a barebox gets installed to this boot media:
 
 @verbatim
-[jb@host]~> scripts/setupmbr/setupmbr -s 32 -m ./barebox -d /dev/sda
+[jb@host]~> scripts/setupmbr/setupmbr -s 32 -m ./barebox.bin -d /dev/sda
 @endverbatim
 
-This command writes the @a barebox image file './barebox' onto the device
+This command writes the @a barebox image file './barebox.bin' onto the device
 @p /dev/sda.
 
 The @p -s option will keep the persistant storage sectors free and untouched
-- 
1.8.1.2


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

[PATCH 2/2] arch/x86: fix link with recent gcc

[Thomas Petazzoni]

Due to unwind tables being generated by recent versions of gcc by
default, the x86 Barebox link fails with:

ld: section .eh_frame loaded at [00000000000197c4,000000000001f31f]
overlaps section .barebox_initcalls loaded at [00000000000197c4,0000000000019833]

Passing -fno-unwind-tables -fno-asynchronous-unwind-tables avoids this
problem.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
---
 arch/x86/Makefile | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/arch/x86/Makefile b/arch/x86/Makefile
index 518b37f..a78fa22 100644
--- a/arch/x86/Makefile
+++ b/arch/x86/Makefile
@@ -5,7 +5,8 @@ machine-y := i386
 
 TEXT_BASE = $(CONFIG_TEXT_BASE)
 
-CPPFLAGS += -march=i386 -m32 -DTEXT_BASE=$(TEXT_BASE)
+CPPFLAGS += -march=i386 -m32 -DTEXT_BASE=$(TEXT_BASE) \
+	-fno-unwind-tables -fno-asynchronous-unwind-tables
 LDFLAGS += -m elf_i386
 
 ifndef CONFIG_MODULES
-- 
1.8.1.2


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 0/2] x86 minor fixes and questions

[Sascha Hauer]

Hi Thomas,

On Fri, Nov 08, 2013 at 07:30:37PM +0100, Thomas Petazzoni wrote:
> Hello,
> 
> Here are two minor fixes for the x86 support. The first patch avoids
> linking problems on recent gcc versions, and the second one fixes the
> documentation.
> 
> Also, I've tested the x86 support under Qemu, it starts fine. Thanks
> to the bios disk driver, it detects one hard disk drive. However,
> since the driver doesn't set num_blocks, the disk size is zero, and no
> partition can be detected. Is this bios disk driver supposed to work
> to read the entire hard drive, access a filesystem and load the kernel
> image, or is the x86 support generally too limited for this at the
> moment?

Does your disk contain a valid partition table? There's no way to
properly detect the disk size, so barebox guesses the size based on the
last partition on the disk. This fails of course on an unpartitioned
device. See disk_guess_size().

But indeed, the x86 support is quite limited.

> Thomas Petazzoni (2):
>   scripts/setupmbr: fix documentation
>   arch/x86: fix link with recent gcc

Applied, thanks

Sascha

-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 0/2] x86 minor fixes and questions

[Thomas Petazzoni]

Dear Sascha Hauer,

On Mon, 11 Nov 2013 08:57:38 +0100, Sascha Hauer wrote:

> > Also, I've tested the x86 support under Qemu, it starts fine. Thanks
> > to the bios disk driver, it detects one hard disk drive. However,
> > since the driver doesn't set num_blocks, the disk size is zero, and no
> > partition can be detected. Is this bios disk driver supposed to work
> > to read the entire hard drive, access a filesystem and load the kernel
> > image, or is the x86 support generally too limited for this at the
> > moment?
> 
> Does your disk contain a valid partition table? There's no way to
> properly detect the disk size, so barebox guesses the size based on the
> last partition on the disk. This fails of course on an unpartitioned
> device. See disk_guess_size().

Yes, my disk had a partition table. However, I see that
disk_guess_size() is only called if you have a DOS partition table, and
not a GPT type of partition table, and I don't remember which type I've
used during my testing.

I'll test again, and with the info that disk_guess_size() is the
important part, I'll have a closer look.

> But indeed, the x86 support is quite limited.

Yeah, no problem. I'm not really planning on making any real use of it,
was just curious to see whether it worked, and if yes, how :-)

> > Thomas Petazzoni (2):
> >   scripts/setupmbr: fix documentation
> >   arch/x86: fix link with recent gcc
> 
> Applied, thanks

Thanks!

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox