Do not allow non-identical DIMMs yet, but prepare the code.
Calculate tCL related settings per DIMM in a dual channel setup. The
check for compatibility will come in a later patch, but since DIMMs
still have to be identical, this does not hurt.
Factor out tRC calculation to prepare for per-DIMM calculation.
Add diagnostic messages to tRC code.
Test booted to FILO, behaviour is identical if you ignore the added
debug messages (which are switched off by default).
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Acked-by: Peter Stuge <peter@stuge.se>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3867 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
This will allow usage of compatible DIMMS in a dual channel setup
instead of requiring the DIMMS to be identical.
Code impact is minimal because a large chunk of code has been moved into
a separate function with almost no changes.
Tested, yields identical results and identical logs.
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Acked-by: Peter Stuge <peter@stuge.se>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3866 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
print_* in K8 RAM init does not make sense anymore. Convert almost all
print_* to printk_*. This improves readability a lot and makes the code
shorter.
Reorder the SPD equality checks in the dual channel DIMM compatibility
checking code. This is to make sure that we know if any other mismatches
are present in the DIMM. The new order eases debugging with the old
code.
Add a comment about false negatives in that code. This needs to be
implemented correctly, but that is hard to do in an efficient way.
Check if the DIMMS in a dual channel setup have any compatible CAS
latencies.
Add better comments to explain why wrong-at-first-glance SPD CL walking
code is actually correct.
Fix a few typos.
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Acked-by: Marc Jones <marcj303@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3865 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Trying to read the FIDVID register when the processor does not support FIDVID
control causes a GP Fault. This patch reads the startup FID from a different
MSR. I have verified this patch to work on the dbm690t platform.
Signed-off-by: Dan Lykowski <lykowdk@gmail.com>
Acked-by: Peter Stuge <peter@stuge.se>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3863 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
initialization.
This patch has helped immensely to track down a bug in 690G ncHT init.
It depends on my earlier patch which enables CONFIG_USE_PRINTK_IN_CAR
for all boards using HT. Of course that means ROMCC is not an option
anymore for those boards, but I don't think that's a big problem.
Another way to solve this would be #defining printk_spew to nothing.
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Marc says:
ROMCC doesn't make sense for k8 boards.
Acked-by: Marc Jones <marcj303@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3836 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
This is tested on hardware with four 128MB DIMMs and works ok, _iff_
you also fix additional registers (e.g. DRB, RPS, ...) for your setup.
This requirement will be eliminated in another upcoming patch (i.e. all
of the required settings will be auto-detected).
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Joseph Smith <joe@settoplinux.org>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3807 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Signed-off-by: Stefan Reinauer <stepan@coresystems.de>
Acked-by: Marc Jones <marcj303@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3802 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
memory controller.
Also, drop some unused '#if 0' code.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Ronald G. Minnich <rminnich@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3773 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
as well as most (all?) combinations thereof.
Drop some unused code, the unused row_offset variable, and obsolete comments.
Also, fix a typo (thanks to Stefan Reinauer for noticing).
This is tested on the MSI MS-6178 with a number of different DIMM
combinations and so far all of them worked fine.
Signed-off-by: Elia Yehuda <z4ziggy@gmail.com>
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Peter Stuge <peter@stuge.se>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3765 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Intel 810 chipset (and all boards using it). This isn't required for this
chipset as there's only one memory controller.
This also helps a lot with romcc register usage, you should see the dreaded
"too few registers" less often.
Build-tested with all three boards using the Intel 810 chipset.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Corey Osgood <corey.osgood@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3764 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
values for different DIMM configurations. This should be converted to a
table or code later on and actually be used for BUFF_SC.
Many thanks to Elia Yehuda <z4ziggy@gmail.com> for testing and collecting
the table entries.
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@3759 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Also, add missing C1DRA2 #define (as per public datasheet).
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@3738 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Build-tested with the kontron/986lcd-m target.
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@3705 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
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@3703 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
due to integer rounding errors. Previously, the formula was:
speed = 2 * (10000/spd_value)
For spd_value=60 this means speed = 2 * 166 = 332, which is less than
333 and coreboot died saying RAM was incompatible. The new formula is:
speed = 20000 / spd_value
For spd_value=60, speed=333, which is fine.
Signed-off-by: Jens Rottmann <JRottmann@LiPPERTEmbedded.de>
Acked-by: Marc Jones <marc.jones@amd.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3689 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
lib/debug.c and use that one.
Signed-off-by: Uwe Hermann <uwe@hermann-uwe.de>
Acked-by: Peter Stuge <peter@stuge.se>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3653 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
to v3 sooner or later we cleanup _now_, so we don't have to do it twice.
- Whitespace, coding style improvements.
- Fix a few typos.
- Add a missing #endif in raminit.h.
- Drop an unused variable.
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@3644 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
code to use it. That makes the code more readable and also less
error-prone.
Abuild tested.
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Acked-by: Peter Stuge <peter@stuge.se>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3624 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
A trivial fix to correct the address of the high byte in SDRAMC.
Thus the leadoff timing IPDLT will be correctly referenced.
Signed-off-by: Mats Erik Andersson <mats.andersson@gisladisker.se>
Acked-by: Peter Stuge <peter@stuge.se>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3620 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
ChannelB slot. Previously a DIMM could only be populated in ChannelB
if there was a DIMM already in ChannelA. This patch doesn't allow unmatched
DIMMs to be populate in ChannelA and ChannelB. In an A & B configuration
the DIMM must still be matched.
Signed-off-by: Marc Jones <marc.jones@amd.com>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3614 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
EP80579 Integrated Processor (codename "Tolapai"). The memory
controller code supports only 64-bit-wide DIMMs with x8 devices and
ECC. It has been tested on a development board using a single Micron
MT9HTF6472PY-667D2 DIMM. Your mileage will definitely vary with other
DIMMs.
Signed-off-by: Ed Swierk <eswierk@arastra.com>
Acked-by: Joseph Smith <joe@settoplinux.org>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3600 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Signed-off-by: Marc Jones <marc.jones@amd.com>
Acked-by: Marc Jones <marc.jones@amd.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3586 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Signed-off-by: Michael Xie Michael.Xie@amd.com
Acked-by: Ronald G. Minnich <rminnich@gmail.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3585 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
All devices work, no irq storms. Enjoy.
Signed-off-by: Bari Ari <bari@onelabs.com>
Acked-by: Peter Stuge <peter@stuge.se>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3556 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
LinuxNetworx own the copyright for the Intel e7520, e7525 and 3100
raminit code.
Signed-off-by: Ed Swierk <eswierk@arastra.com>
Acked-by: Ronald G. Minnich <rminnich@gmail.com>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3546 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
and renames some existing macros for clarity.
Signed-off-by: Ed Swierk <eswierk@arastra.com>
Acked-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3536 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Signed-off-by: Marc Jones <marc.jones@amd.com>
Acked-by: Ronald G. Minnich <rminnich@gmail.com>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3504 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
in a number of places.
Signed-off-by: Stefan Reinauer <stepan@coresystems.de>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3466 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Signed-off-by: Marc Jones <marc.jones@amd.com>
Acked-by: Marc Jones <marc.jones@amd.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3432 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Signed-off-by: Marc Jones <marc.jones@amd.com>
Acked-by: Peter Stuge <peter@stuge.se>
Acked-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3425 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Signed-off-by: Marc Jones <marc.jones@amd.com>
Acked-by: Marc Jones <marc.jones@amd.com>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3423 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Signed-off-by: Joseph Smith <joe@settoplinux.org>
Acked-by: Peter Stuge <peter@stuge.se>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3380 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
This also contains various improvements of the CN700 code in svn.
Signed-off-by: Aaron Lwe <aaron.lwe@gmail.com>
Acked-by: Uwe Hermann <uwe@hermann-uwe.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3334 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Signed-off-by: Joseph Smith <joe@settoplinux.org>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3322 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
This doesn't save any space for me but it is the right thing to allow GCC to
optimize.
Signed-off-by: Marc Jones <marc.jones@amd.com>
Acked-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3266 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Supports single and split plane systems. Set P0 on all cores for best performance.
All APs will be in hlt(C1).
The platform warm rest logic has been updated to alway reset for HT and FID/VID setup. It is not optional anymore.
Signed-off-by: Marc Jones <marc.jones@amd.com>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3251 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
This fixes many default setting as well as erratas.
Some CPU core functions were moved from the HT init and platform specific code to the generic Fam10 CPU code.
Signed-off-by: Marc Jones <marc.jones@amd.com>
Acked-by: Stefan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3248 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
recomendations.
Changes include the following:
fix > 4GB dqs tests
fix channel interleaving
ecc memory scrub updates
MC tristating updates
debug print changes
fix memory hoisting across nodes -
The DRAM Hole Address Register is set via devx in each node, but the Node
number <-> DRAM Base mapping and the Node number <-> DstNode mapping is
set in Node 0. The memmap is setup on node0 and copied to the other nodes
later. so dev, not devx. The bug was the hole was always being set on the
first node.
Signed-off-by: Marc Jones (marc.jones@amd.com)
Acked-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@3232 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
