This patch allows a new kind of compatible string for LPDDR parts in the
device tree bindings, in addition to the existing hardcoded
<vendor>,<part-number> strings. The new format contains manufacturer and
part (revision) information in numerical form, such as lpddr3-ff,0201
for an LPDDR3 part with manufacturer ID ff and revision ID 0201. This
helps cases where LPDDR parts are probed at runtime by boot firmware and
cannot be matched to hardcoded part numbers, such as the firmware on the
qcom/sc7280-herobrine boards does (which supports 4 different memory
configurations at the moment, and more are expected to be added later at
a point where the boot firmware can no longer be updated to specifically
accommodate them).
Signed-off-by: Julius Werner <jwerner@chromium.org>
Acked-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20220930220606.303395-2-jwerner@chromium.org
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
The bindings for different LPDDR versions mostly use the same kinds of
properties, so in order to reduce duplication when we're adding support
for more versions, this patch creates a new lpddr-props subschema that
can be referenced by the others to define these common parts. (This will
consider a few smaller I/O width and density numbers "legal" for LPDDR3
that are usually not used there, but this should be harmless.)
Signed-off-by: Julius Werner <jwerner@chromium.org>
Acked-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20220930220606.303395-1-jwerner@chromium.org
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Krzysztof Kozlowski's @canonical.com email stopped working, so switch to
generic @kernel.org account for all Devicetree bindings.
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Acked-by: Rob Herring <robh@kernel.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Link: https://lore.kernel.org/r/20220330074016.12896-2-krzysztof.kozlowski@linaro.org
The timings node maximum frequency was passed as an unit address, which
is actually a workaround. Such workaround and unit address are not
needed at all, because the device memory node (parent) can contain
multiple timing nodes without unit addresses but with suffix used for
nodenames, e.g. timings-1.
LPDDR2 bindings already use such version, so unify the LPDDR3 with them.
Suggested-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20220206135807.211767-7-krzysztof.kozlowski@canonical.com
The memory manufacturer should be described in vendor part of
compatible, so there is no need to duplicate it in a separate property.
Similarly is done in LPDDR2 bindings.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Reviewed-by: Alim Akhtar <alim.akhtar@samsung.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20220206135807.211767-6-krzysztof.kozlowski@canonical.com
According to JEDEC Standard No. 209-3 (table 3.4.1 "Mode Register
Assignment and Definition in LPDDR3 SDRAM"), the LPDDR3 supports only
16- and 32-bit IO width. Drop the unsupported others.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20220206135807.211767-5-krzysztof.kozlowski@canonical.com
