This patch implements a full SDRAM buffer strength programming algorithm in
set_dram_buffer_strength(), checked against my P2B-LS factory BIOS. With this
in place, I now have 133MHz (!) stability with three 256MB PC133 modules, and
can boot Fedora 11 all the way to the init daemon (actually upstart, but that's
another story). Not to login prompt yet. We'll find out why later.
This again assumes a 4-DIMM board because that's all I have. I need someone
with a 3-DIMM board to test it.
As a bonus, there's a big comment block within that illustrates the algorithm.
:-)
Signed-off-by: Keith Hui <buurin@gmail.com>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5238 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
util/x86emu is the only part of coreboot that is linked into coreboot
itself that lives in util/.
It's not a utility and it does not really belong where it lives.
---> svn mv util/x86emu src/devices/oprom
plus necessary Makefile changes to get it building again
Signed-off-by: Stefan Reinauer <stepan@coresystems.de>
Acked-by: Ronald G. Minnich <rminnich@gmail.com>
Acked-by: Peter Stuge <peter@stuge.se>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5228 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
already provides this.
Thanks Myles for catching this.
Signed-off-by: Patrick Georgi <patrick.georgi@coresystems.de>
Acked-by: Patrick Georgi <patrick.georgi@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5227 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Macro-ify stripping quotes
Signed-off-by: Patrick Georgi <patrick.georgi@coresystems.de>
Acked-by: Patrick Georgi <patrick.georgi@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5226 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Improve handling of problems while building coreboot.rom
Signed-off-by: Patrick Georgi <patrick.georgi@coresystems.de>
Acked-by: Myles Watson <mylesgw@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5225 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
- Single instance of the CC build rule in Makefile, instantiated as
necessary
- Remove manual static.o and option_table.o rules, they're now covered
by those instances of the CC build rule
- Normalize object file paths, so it can be $(obj)/option_table.o
instead of $(obj)/arch/i386/../../option_table.o now
- Add -pipe to compiler flags. It might be detrimental on rare scenarios
(building with extremly high disk bandwidth, eg. RAM disk), but it
significantly helps on win32 (which seems to cache less aggressively
than most unix-alikes)
- Silence stderr on hostname and domainname invocations (cosmetic fix
for cygwin)
- Test for -Wa,--divide functionality of the target compiler (taken from
abuild). It might be possible to remove most patches in crossgcc with that.
- Report build of failover.inc and romstage.inc
Signed-off-by: Patrick Georgi <patrick.georgi@coresystems.de
Acked-by: Peter Stuge <peter@stuge.se>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5224 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Signed-off-by: Zheng Bao <zheng.bao@amd.com>
Acked-by: Marc Jones <marcj303@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5223 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Makefile and Kconfig.
Signed-off-by: Zheng Bao <zheng.bao@amd.com>
Acked-by: Marc Jones <marcj303@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5222 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
made by AMD. Its major features are:
CPU:
* AMD AM2+
* AMD Athlon 64 x2
* AMD Athlon 64 FX
* AMD Athlon 64
* AMD Sempron CPUs
System Chipset:
* RS780E
* SB700
On Board Chipset:
* BIOS - SPI
* Azalia CODEC - Realtek ALC888
* LPC SuperIO - ITE8718F(GX).
* LAN - REALTEK 8111C
* TPM - SLB9635TT1.2
Main Memory:
* DDR II * 4 (Max 4GB)
Expansion Slots:
* PCI Express X16 slot*2 (PCI-E X8 Bus)
* PCI Express X4 Slot*1
Intersil PWM:
* Controller - Intersil 6323
Note:
1. The only difference to mahogany is the CPU is changed to K8 family 10.
2. The main structure of the code is based on
serengeti_cheetah_fam10. I am a rookie to fam10. I am still
confused about CONFIG_HT_CHAIN_UNITID_BASE and
CONFIG_HT_CHAIN_END_UNITID_BASE. I set them as the way dbm690t
does. And I have to modify the some fam10 code (see the patch
ht_chain_unitid_base.patch). I dont know how to solve this. Please
help.
Updated findings: In h3finit.c, the code goes the AMD_CB_ManualBUIDSwapList().
The swaplist is {0x00, 0x00, 0xFF, 0x00, 0xFF}. I am trying to find the meaning
of the list. The amdht wrapper needs to modify definitely.
3. With fam10 processor, the HT link can work in HT3.
4. The ACPI _PSS table is set staticly. The auto configuaration
process doesnt seem to work correctly.
5. Currently the fam10 code in coreboot doesn't support DDR3. If you
happen to get a board with DDR3 and you don't have the patience to wait,
please find another board with DDR2.
6. It will take "Uncompressing image to RAM" about 1 minute. I know it is a
issue for a long time. I disable the compressing currently. When the problem
is fixed, we can re-enable it.
Signed-off-by: Zheng Bao <zheng.bao@amd.com>
Acked-by: Marc Jones <marcj303@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5221 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
The code can run on the Mahogany board, which is one of sample boards
made by AMD. Its major features are:
CPU (only K8 system):
* AMD AM2+
* AMD Athlon 64 x2
* AMD Athlon 64 FX
* AMD Athlon 64
* AMD Sempron CPUs
System Chipset:
* RS780E
* SB700
On Board Chipset:
* BIOS - SPI
* Azalia CODEC - Realtek ALC888
* LPC SuperIO - ITE8718F(GX).
* LAN - REALTEK 8111C
* TPM - SLB9635TT1.2
Main Memory:
* DDR II * 4 (Max 4GB)
Expansion Slots:
* PCI Express X16 slot*2 (PCI-E X8 Bus)
* PCI Express X4 Slot*1
Intersil PWM:
* Controller - Intersil 6323
2. The ACPI feature is already added. I suggest that firstly we can test
the board without the ACPI by setting the HAVE_ACPI_TABLE as 0.
With Rev F processor, the HT link can only work in HT1, whose max
frequency is 1GHz.
Signed-off-by: Zheng Bao <zheng.bao@amd.com>
Acked-by: Marc Jones <marcj303@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5220 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
* PCIe initialization.
* Internal Graphics initialization.
* HT Link initialization. It works in HT1 or HT3 mode.
Note:
1. I tried to add the description of every step to the code. For example,
if it is made based on rpr, section 2.4.5, I will pasted the words
from 2.4.5 to the c code. But the document I worked with might be
different with the most updated one. A new section has been added and
the 2.4.5 might be changed to 2.5.5. That migh lead to confusing. I
correct every comment if I met one. But I have to confess that I am so
reluctant to find out everyone. I believe it will be correct in the long
run.
2. The interanl graphics part is done by Libo Feng <libo.feng@amd.com>.
3. There is a conflict between RPR and our CIM code. Please see the comment in
switching_gppsb_configurations in rs780_pcie.c.
Signed-off-by: Zheng Bao <zheng.bao@amd.com>
Acked-by: Marc Jones <marcj303@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5219 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
* SATA initialization.
* USB initialization.
* HDA initialization.
* LPC initialization.
* IDE initialization.
* SMBUS initialization.
Note:
1. I tried to add the description of every step to the code. For example,
if it is made based on rpr, section 2.4.5, I will pasted the words
from 2.4.5 to the c code. But the document I worked with might be
different with the most updated one. A new section has been added and
the 2.4.5 might be changed to 2.5.5. That migh lead to confusing. I
correct every comment if I met one. But I have to confess that I am so
reluctant to find out everyone. I believe it will be correct in the long
run.
2. I only test the SATA port 0-3. The ports 4, 5 are "PATA emulations".
I am confused about it.
3. This patch is not only about SB700. Actually it should be
SB7x0. But I dont think it is nice to change everything to
SB7x0. It is ugly, isn't it. As far as I know, they all use the
same code with revision checking. If you guys think it is
appropriate, please modify it to sb7x0.
Signed-off-by: Zheng Bao <zheng.bao@amd.com>
Acked-by: Marc Jones <marcj303@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5218 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
seperatedly because both RS780 and SB700 will modify the pci_ids.h. It
maybe will cause conflict if the sequence the patches are applied is
different with the one they are created.
2. Dev 0-10 of RS780 has AMD's Vendor ID. So we think it is better to
define the Device ID as XXX_AMD_RS780_XXX. Does anyone think it is
better to move this definition to the AMD zone? That will split the
RS780 into two parts. Is it inappropriate?
Signed-off-by: Zheng Bao <zheng.bao@amd.com
Acked-by: Marc Jones <marcj303@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5217 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Signed-off-by: Patrick Georgi <patrick.georgi@coresystems.de>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5216 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
- call va_* directly if coreboot is running on GCC so we don't need
to maintain hacks to get to stdarg.h
- only define LIBGCC_FILE_NAME if it's an absolute path. GCC and LLVM
just print "libgcc.a" if the file is not there.
Signed-off-by: Stefan Reinauer <stepan@coresystems.de>
Acked-by: Ronald G. Minnich <rminnich@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5215 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Signed-off-by: Zheng Bao <zheng.bao@amd.com>
Acked-by: Patrick Georgi <patrick.georgi@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5211 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
so it's clear that they should be gone.
More can (and should) be added, but this is a start.
Of course, eliminating the uses of the flags (and then
the flags themselves) that are in Kconfig.deprecated_options
is a noble task for the future :-)
Signed-off-by: Patrick Georgi <patrick.georgi@coresystems.de>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5209 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
When configured in Kconfig, just running "make"
calls scan-build as appropriate (however, it does not
check for the presence of scan-build)
The target directory for the scan-build report is configurable
and defaults to the scan-build default of /tmp/scan-build-$date-$num
abuild is adapted to properly run scanbuild when ran
with the -sb option.
Signed-off-by: Patrick Georgi <patrick.georgi@coresystems.de>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5208 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
fix some PCI IDs, enable USB bus mastering, add some license headers, ...
LPC code needs another look, but I think we're getting there.
Signed-off-by: Stefan Reinauer <stepan@coresystems.de>
Acked-by: Joseph Smith <joe@settoplinux.org>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5207 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Signed-off-by: Stefan Reinauer <stepan@coresystems.de>
Acked-by: Patrick Georgi <patrick.georgi@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5206 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
This fixes a longstanding TODO item.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5205 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Each Intel 440BX board should select this option if it has 4 DIMM
slots on the PCB, and _not_ select it (it defaults to 'n') if it
has 3 DIMMs on the PCB.
Signed-off-by: Keith Hui <buurin@gmail.com>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5204 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Also configure GPIOs so the power LED is working.
Some whitespace cleanups (but by no means comprehensive)
Signed-off-by: Knut Kujat <knuku@gap.upv.es>
Acked-by: Myles Watson <mylesgw@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5203 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Replace set_init_ram_access with the call to set_var_mtrr.
Remove unused #include statments.
Signed-off-by: Myles Watson <mylesgw@gmail.com>
Acked-by: Patrick Georgi <patrick.georgi@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5201 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
support for coreboot. I have not updated MAX_CPUS for all fam10
mainboards, but it might make sense to multiply those by 1.5.
Signed-off-by: Arne Georg Gleditsch <arne.gleditsch@numascale.com>
I assume the line
pci_write_config32(NODE_PCI(nodeid, 0), 0x168, dword);
should be put outside the loop.
Everything seems to be fine. I don't have Istanbul to test. I have
read every changes and they all look good.
Acked-by: Zheng Bao <zheng.bao@amd.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5200 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Previously the sb600 setup was equivalent to clear_ioapic(), so that is what we will do for now.
Signed-off-by: Marc Jones <marcj303@gmail.com>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5199 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
WAIT_BEFORE_CPUS_INIT. It's disabled by default
(see src/cpu/x86/Kconfig)
Signed-off-by: Patrick Georgi <patrick.georgi@coresystems.de>
Acked-by: Myles Watson <mylesgw@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5198 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
and increase MAX_CPUS to aid support 6-core CPUs.
Signed-off-by: Arne Georg Gleditsch <arne.gleditsch@numascale.com>
Acked-by: Patrick Georgi <patrick.georgi@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5197 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
The code currently assumes a 4-DIMM-slots board, this will be fixed soon.
Signed-off-by: Keith Hui <buurin@gmail.com>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5194 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
2. Use time.h for get_time() and move tb_freq into functions.c
3. Move read_io and write_io to io.c and make them static
4. Make a couple of functions static in interrupt.c
5. Refactor a cast from char[] to u64 to get rid of potential alignment problems and a warning
Signed-off-by: Myles Watson <mylesgw@gmail.com>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5191 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
- don't skip the reset on S4 violations. Specs ask us to do this so we do it
- hlt on waiting for reset instead of hot looping.
Signed-off-by: Stefan Reinauer <stepan@coresystems.de>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5190 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
According to some GCC folks -Os should be considered a buggy and unreliable
code path, so at least keep -O2 working. coreboot_ram is only 4KB bigger.
Signed-off-by: Stefan Reinauer <stepan@coresystems.de>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5189 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Convert all DEBUG_SMBUS, DEBUG_SMI, and DEBUG_RAM_SETUP custom and
local #defines into globally configurable kconfig options (and Options.lb
options for as long as newconfig still exists) which can be enabled
by the user in the "Debugging" menu.
The respective menu items only appear if a board is selected where the
chipset code actually provides such additional DEBUG output.
All three variables default to 0 / off for now.
Also, drop a small chunk of dead/useless code in the
src/northbridge/via/cn700/raminit.c file, which would otherwise break
compilation.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Reworked to still apply to trunk, added X86EMU_DEBUG (and make the x86emu/yabel
code only work printf instead of a redefined version of printk and
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5185 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Also, fix another typo in the ms9652 board name.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5184 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
- Fix whitespace, alignment, and indentation in a few places.
- Some more consistency fixes in license headers.
- Fix incomplete license header: src/mainboard/msi/ms9652_fam10/devicetree.cb.
- Fix typo for LIMIT_HT_SPEED_1800: s/1.6GHz/1.8GHz/.
- Fix typo in src/mainboard/msi/ms9652_fam10/Kconfig: s/MS-9256/MS-9252/.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5182 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
- Check that this doesn't run into vga/oprom/bios area at link time
- Avoid overly complicated and not well understood hack which avoids that
area by leaving a hole in the stack area.
- Adapt technexion/tim5690 to put ramstage at 1MB
Signed-off-by: Patrick Georgi <patrick.georgi@coresystems.de>
Acked-by: Myles Watson <mylesgw@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5181 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Updated Timothy's patch to match recent changes in the tree. It's build tested.
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Acked-by: Ronald G. Minnich <rminnich@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5180 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
less than optimal PCB designs can still work reliably
with reduced clock.
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Acked-by: Patrick Georgi <patrick.georgi@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5179 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Stefan Reinauer