Michael LeMay f85654a82f x86, galileo: Add UEFI support

This patch adds support for optionally building EFI binaries in
addition to Multiboot ELF binaries.  It includes a script,
build_uefi.sh, that downloads tool and library sources from the EDK II
project, builds the GenFw tool that is used to create UEFI binaries,
and creates a makefile that is included from the main x86 common
makefile and enables UEFI support in the Contiki build system.  If the
script is not run prior to building Contiki, then an informational
message will be displayed with instructions for running build_uefi.sh
if UEFI support is desired.  This patch also adds the path to the
auto-generated makefile to .gitignore.

This patch modifies the linker script for the Intel Quark X1000 to
account for the output file section offsets and alignment expectations
of the EDK II GenFw project.

This patch also adds a newlib patch to remove the weak symbol
attribute from floating point stdio support routines.  See
<newlib>/newlib/README for an explanation of how the newlib developers
intended for _printf_float and _scanf_float to be linked.  Newlib
declares them as weak symbols with the intention that developers would
force them to be linked only when needed using a linker command line
option.  However, some but not all Contiki programs require them, so
we cannot simply always include or exclude them.  Instead, we remove
the weak symbol attributes and rely on the linker to automatically
determine whether or not they should be linked.  This avoids an issue
in which weak symbols were undefined in the intermediate DLL generated
as part of the UEFI build process.  That resulted in the GenFw program
emitting "ERROR 3000" messages when it encountered relocations
referencing such an undefined symbol.

Finally, this patch updates README.md to both make some revisions to
account for the UART support introduced in previous patches as well as
to provide instructions for using the UEFI support.

2015-12-21 08:06:14 -02:00

5.7 KiB

Raw Blame History

Intel Galileo Board

This README file contains general information about the Intel Galileo board support. In the following lines you will find information about supported features as well as instructions on how to build, run and debug applications for this platform. The instructions were only test in Linux environment.

Requirements

In order to build and debug the following packages must be installed in your system:

gcc
gdb
openocd

Moreover, in order to debug via JTAG or serial console, you will some extra devices as described in [1] and [2].

Features

This section presents the features currently supported (e.g. device drivers and Contiki APIs) by the Galileo port.

Device drivers:

Programmable Interrupt Controller (PIC)
Programmable Intergal Timer (PIT)
Real-Time Clock (RTC)
UART
Ethernet

Contiki APIs:

Clock module
Timer, Stimer, Etimer, Ctimer, and Rtimer libraries

Standard APIs:

Stdio library (stdout and stderr only). Console output through UART 1 device (connected to Galileo Gen2 FTDI header)

Building

To build applications for this platform you should first build newlib (in case it wasn't already built). To build newlib you can run the following command:

$ ./platform/galileo/bsp/libc/build_newlib.sh

Once newlib is built, you are ready to build applications. By default, the following steps will use gcc as the C compiler and to invoke the linker. To use LLVM clang instead, change the values for both the CC and LD variables in cpu/x86/Makefile.x86_common to 'clang'.

To build applications for the Galileo platform you should set the TARGET variable to 'galileo'. For instance, building the hello-world application should look like this:

$ cd examples/hello-world/ && make TARGET=galileo

This will generate the 'hello-world.galileo' file which is a multiboot- compliant [3] ELF image. This image contains debugging information and it should be used in your daily development.

You can also build a "Release" image by setting the BUILD_RELEASE variable to

This will generate a Contiki stripped-image optimized for size.

$ cd examples/hello-world/ && make TARGET=galileo BUILD_RELEASE=1

To also generate an '.galileo.efi' file which is a UEFI [4] image, you can run the following command prior to building applications:

$ cpu/x86/uefi/build_uefi.sh

Running

In order to boot the Contiki image, you will need a multiboot-compliant bootloader. In the bsp directory, we provide a helper script which builds the Grub bootloader with multiboot support. To build the bootloader, just run the following command:

$ platform/galileo/bsp/grub/build_grub.sh

Once Grub is built, we have three main steps to run Contiki applications: prepare SDcard, connect to console, and boot image. Below follows detailed instructions.

Prepare SDcard

Mount the sdcard in directory /mnt/sdcard.

Approach for Multiboot-compliant ELF Image

Copy Contiki binary image to sdcard

$ cp examples/hello-world/hello-world.galileo /mnt/sdcard

Copy grub binary to sdcard

$ cp platform/galileo/bsp/grub/bin/grub.efi /mnt/sdcard

Approach for UEFI Image

Copy Contiki binary image to sdcard

$ cp examples/hello-world/hello-world.galileo.efi /mnt/sdcard

Connect to the console output

Connect the serial cable to your computer as shown in [2].

Choose a terminal emulator such as PuTTY. Make sure you use the SCO keyboard mode (on PuTTY that option is at Terminal -> Keyboard, on the left menu). Connect to the appropriate serial port using a baud rate of 115200.

Boot Contiki Image

Turn on your board. After a few seconds you should see the following text in the screen:

Press [Enter] to directly boot.
Press [F7]    to show boot menu options.

Press and select the option "UEFI Internal Shell" within the menu.

Boot Multiboot-compliant ELF Image

Once you have a shell, run the following commands to run grub application:

$ fs0:
$ grub.efi

You'll reach the grub shell. Now run the following commands to boot Contiki image:

$ multiboot /hello-world.galileo
$ boot

Boot UEFI Image

Once you have a shell, run the following commands to boot Contiki image:

$ fs0:
$ hello-world.galileo.efi

Verify that Contiki is Running

This should boot the Contiki image, resulting in the following messages being sent to the serial console:

Starting Contiki
Hello World

Debugging

This section describes how to debug Contiki via JTAG. The following instructions consider you have the devices: Flyswatter2 and ARM-JTAG-20-10 adapter (see [1]).

Attach the Flyswatter2 to your host computer with an USB cable. Connect the Flyswatter2 and ARM-JTAG-20-10 adapter using the 20-pins head. Connect the ARM-JTAG-20-10 adapter to Galileo Gen2 JTAG port using the 10-pins head.

Once everything is connected, run Contiki as described in "Running" section, but right after loading Contiki image (multiboot command), run the following command:

$ make TARGET=galileo debug

The 'debug' rule will run OpenOCD and gdb with the right parameters. OpenOCD will run in background and its output will be redirected to a log file in the application's path called LOG_OPENOCD. Once gdb client is detached, OpenOCD is terminated.

If you use a gdb front-end, you can define the "GDB" environment variable and your gdb front-end will be used instead of default gdb. For instance, if you want to use cgdb front-end, just run the command: