Initialization for the Winbond W631GG6KB part using Synopsys
DDR uMCTL and DDR Phy.
This code adds a separate function for DDR3 initialization
and moves all the necessary defines in a separate header file.
The programming procedure that is executed at power up to bring
up the uMCTL, PHY and memories into a state where reads and
writes to the memory can be performed is the following:
1. uPCTL (Universal DDR protocol controller) initialization
The timining registers TOGCNT1U, TINIT, TOGCNT100N and TRSTH
needed for driving the memory power-up sequence are programmed
as a function of the internal timers clock frequency.
Organization (memory chip specific) values are set
(column/bank/row address width and number of ranks), together
with other static values (latency, timing, power up configuration).
All these values are static, provided by the datasheet,
being determined by the memory type, size and frequency.
2. PHY initialization
The PHY is programmed with datasheet provided values,
specifying the initialization values for it to send to the
external memory (timing parameters).
Also, delay lines (DLL) and strength of drive pads are
calibrated (based on external conditions: temperature,
voltage, noise) and locked. After that, the PHY goes
through a trainig process (also dependent on the
current conditions at boot time) to establish precise
timing configuration between the DDR clock and DQS (data strobe)
and between DQS and DQ (data).
3. Memory power up
4. Switch from configuration state to access state.
It was tested on Pistachio bring up board where DDR was initialized
properly and ramstage executed correctly
Change-Id: I3bcbce2044327a22fce09b184d85ee11228a6b2b
Signed-off-by: Ionela Voinescu <ionela.voinescu@imgtec.com>
Reviewed-on: http://review.coreboot.org/10529
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Switching from active windowing DQS gating scheme to
passive windowing mode resolves boot stability issues
on chips found to have memory corruption issues during
boot or memory tests.
It was tested on Pistachio bring up board where DDR is
initialized properly and ramstage executed correctly;
We have cycled units over 12,000 times with no boot errors.
This option was chosen over the alternative of using
passive windowing mode for DQS training and after switching
back to active mode, as this option was recommended by
Synopsys. Using the alternative would give different
timing values during training that were not longer accurate
during normal activity.
Change-Id: Ie604eddc0a9a982b2f89198f44deb88a01b7b322
Signed-off-by: Ionela Voinescu <ionela.voinescu@imgtec.com>
Reviewed-on: http://review.coreboot.org/10528
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Place included header files in alphabetical order.
Change-Id: Ice23178d1f07e2cb0178efbc7ce487d54bf3f708
Signed-off-by: Ionela Voinescu <ionela.voinescu@imgtec.com>
Reviewed-on: http://review.coreboot.org/10459
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
As per discussion with lawyers[tm], it's not a good idea to
shorten the license header too much - not for legal reasons
but because there are tools that look for them, and giving
them a standard pattern simplifies things.
However, we got confirmation that we don't have to update
every file ever added to coreboot whenever the FSF gets a
new lease, but can drop the address instead.
util/kconfig is excluded because that's imported code that
we may want to synchronize every now and then.
$ find * -type f -exec sed -i "s:Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, *MA[, ]*02110-1301[, ]*USA:Foundation, Inc.:" {} +
$ find * -type f -exec sed -i "s:Foundation, Inc., 51 Franklin Street, Suite 500, Boston, MA 02110-1335, USA:Foundation, Inc.:" {} +
$ find * -type f -exec sed -i "s:Foundation, Inc., 59 Temple Place[-, ]*Suite 330, Boston, MA *02111-1307[, ]*USA:Foundation, Inc.:" {} +
$ find * -type f -exec sed -i "s:Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.:Foundation, Inc.:" {} +
$ find * -type f
-a \! -name \*.patch \
-a \! -name \*_shipped \
-a \! -name LICENSE_GPL \
-a \! -name LGPL.txt \
-a \! -name COPYING \
-a \! -name DISCLAIMER \
-exec sed -i "/Foundation, Inc./ N;s:Foundation, Inc.* USA\.* *:Foundation, Inc. :;s:Foundation, Inc. $:Foundation, Inc.:" {} +
Change-Id: Icc968a5a5f3a5df8d32b940f9cdb35350654bef9
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Reviewed-on: http://review.coreboot.org/9233
Tested-by: build bot (Jenkins)
Reviewed-by: Vladimir Serbinenko <phcoder@gmail.com>
When the PHY is compiled to run in HDR(half data rate),
then either NOBUB or FXDAT must be set to 1 in the DDR
system general configuration register. NOBUB specifies
that reads should be returned to the controller with
no bubbles and this is felt preferable to the fixed
latency option (FXDAT). Both of them inrease read
latency.
BRANCH=none
BUG=chrome-os-partner:37087
TEST=tested on Pistachio bring up board -> DDR initialized
properly and ramstage executed correctly
Change-Id: Iee530ba5bb0acc889fba447dc2ee5cb965ba6926
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Original-Commit-Id: e7944b4af45d9504098f8b4af44d0f5abafea42c
Original-Change-Id: I9ced76bd670fc4efa7441d57e15f97871b046ae9
Original-Signed-off-by: Ionela Voinescu <ionela.voinescu@imgtec.com>
Original-Reviewed-on: https://chromium-review.googlesource.com/264341
Original-Reviewed-by: James Hartley <james.hartley@imgtec.com>
Original-Reviewed-by: David Hendricks <dhendrix@chromium.org>
Reviewed-on: http://review.coreboot.org/9917
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The DRAM configuration register, apart from holding the
device density and width also has a rudimentary address
mapping scheme. Currently this is set to the default
Bank/Row/Column. This means that the memory is segmented
into 8 chunks, each with a page detector. If all the
activity is in one section of memory then the other 7
page detectors could be idle.
Changing this to Row/Bank/Column would concatenate the
page detectors meaning that all 8 could be used by a
single initiator. This may not gain anything in a
synthetic bandwidth test but could yield extra performance
in a real world application or benchmark.
BRANCH=none
BUG=chrome-os-partner:37087
TEST=tested on Pistachio bring up board -> DDR initialized
properly; all access to DDR works properly in
Coreboot ramstage, Depthcharge and Linux;
no performance tests were ran so far.
Change-Id: I22d86bf3b679ed63884d7436d9d7bbaf1726f640
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Original-Commit-Id: e852ed42afcdc2062a0037144bab723227cb1f1f
Original-Change-Id: If90b0cf5ce86db5e3d6d362873d22d4269e3a49f
Original-Signed-off-by: Ionela Voinescu <ionela.voinescu@imgtec.com>
Original-Reviewed-on: https://chromium-review.googlesource.com/264340
Original-Reviewed-by: James Hartley <james.hartley@imgtec.com>
Original-Reviewed-by: David Hendricks <dhendrix@chromium.org>
Reviewed-on: http://review.coreboot.org/9916
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The existing DDR setup configures the burst length to be 8. However
the DDR controller can only be given sufficient data per clock to
satisfy a burst length of 4, hence the bursts are only half
populated. This results in a 50% drop of efficiency.
Fix this by configuring the burst size to 4.
BUG=chrome-os-partner:31438, chrome-os-partner:37087
TEST=tested on Pistachio bring up board -> DDR initialized
properly and ramstage executed correctly
BRANCH=none
Change-Id: I761ba73a04688841ca39a370b7cb99b6e0b22964
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Original-Commit-Id: 0e590ab8387dbbccef45dc84d1eeafee2abc9e2e
Original-Change-Id: I585385b65e330624ad70292349e50c6695eeeb6c
Original-Signed-off-by: Ionela Voinescu <ionela.voinescu@imgtec.com>
Original-Reviewed-on: https://chromium-review.googlesource.com/256305
Original-Reviewed-by: David Hendricks <dhendrix@chromium.org>
Reviewed-on: http://review.coreboot.org/9847
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The DDR On Die Termination was incorrectly configured at 75R,
where as the data sheet suggests for DDR2-800 it should be
set to 50R.
Correct this by adjusting the ODT setting in the EMR register.
BUG=chrome-os-partner:31438, chrome-os-partner:37087
TEST=tested on Pistachio bring up board -> DDR initialized
properly and ramstage executed correctly
BRANCH=none
Change-Id: I2f0242c422b1cb3d1f64ce3dd17b62fef5e7e155
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Original-Commit-Id: ac081ac59c0dc3d16a7b540cd379fb870b6cfe40
Original-Change-Id: If7951812033c4e88f4be3c143fb49526eddba142
Original-Signed-off-by: Ionela Voinescu <ionela.voinescu@imgtec.com>
Original-Reviewed-on: https://chromium-review.googlesource.com/256304
Original-Reviewed-by: David Hendricks <dhendrix@chromium.org>
Reviewed-on: http://review.coreboot.org/9846
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The proper way to initialize DDR2 is for the PHY to
automatically establish precise timing configuration
through the training process. The alternative (used
initially for testing) is no longer needed.
This change determined the removal of some local
variables as they ended up being used in one location only.
BUG=chrome-os-partner:31438, chrome-os-partner:37087
TEST=tested on Pistachio bring up board -> DDR initialized
properly and ramstage executed correctly.
BRANCH=none
Change-Id: I31e9a8975d176a04061f9c84fe06cce850bb53b9
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Original-Commit-Id: e28f3ef9a22436bb0fa949df6f5a5c6a67002dfd
Original-Change-Id: Ifb9c1bb6e0b71af72340381bd2349850d1b4af2d
Original-Signed-off-by: Ionela Voinescu <ionela.voinescu@imgtec.com>
Original-Reviewed-on: https://chromium-review.googlesource.com/256303
Original-Reviewed-by: David Hendricks <dhendrix@chromium.org>
Reviewed-on: http://review.coreboot.org/9845
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
This is the intialization code specific to the Winbond
W972GG6JB-25 part using Synopsys DDR uMCTL and DDR Phy.
This is DDR2 initialization code only (currently present
on the bring up board). DDR3 initialization code will follow
for boards having DDR3 memory.
The programming procedure that is executed at power up to bring
up the uMCTL, PHY and memories into a state where reads and
writes to the memory can be performed is the following:
1. uPCTL (Universal DDR protocol controller) initialization
The timining registers TOGCNT1U, TINIT, TOGCNT100N and TRSTH
needed for driving the memory power-up sequence are programmed
as a function of the internal timers clock frequency.
Organization (memory chip specific) values are set
(column/bank/row address width and number of ranks), together
with other static values (latency, timing, power up configuration).
All these values are static, provided by the datasheet,
being determined by the memory type, size and frequency.
2. PHY initialization
The PHY is programmed with datasheet provided values,
specifying the initialization values for it to send to the
external memory (timing parameters).
Also, delay lines (DLL) and strength of drive pads are
calibrated (based on external conditions: temperature,
voltage, noise) and locked. After that, the PHY goes
through a trainig process (also dependent on the
current conditions at boot time) to establish precise
timing configuration between the DDR clock and DQS (data strobe)
and between DQS and DQ (data).
3. Memory power up
4. Switch from configuration state to access state.
BUG=chrome-os-partner:31438, chrome-os-partner:37087
TEST=tested on Pistachio bring up board -> DDR initialized
properly and ramstage executed correctly
DDR2 is also tested during chip sort.
Corner cases (performace of DDR in different conditions)
will be tested after the chip reaches a stable state.
BRANCH=none
Change-Id: I0093dc175d064aad03052d5281679b008c1bf012
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Original-Commit-Id: 3d0bacea0fd5bd3b12008b47e80de8398f447785
Original-Change-Id: I8437db6c84d77c4c51a3ee2b09cd3d14913c0d16
Original-Signed-off-by: Ionela Voinescu <ionela.voinescu@imgtec.com>
Original-Reviewed-on: https://chromium-review.googlesource.com/241424
Original-Reviewed-by: Vadim Bendebury <vbendeb@chromium.org>
Original-Commit-Queue: Vadim Bendebury <vbendeb@chromium.org>
Original-Tested-by: Vadim Bendebury <vbendeb@chromium.org>
Reviewed-on: http://review.coreboot.org/9769
Tested-by: build bot (Jenkins)
Reviewed-by: Edward O'Callaghan <edward.ocallaghan@koparo.com>
