Divisor is a function of requested baudrate, platform-specific
reference clock and amount of oversampling done on the UART reference.
Calculate this parameter with divisor rounded to nearest integer.
When building without option_table or when there is no entry for
baud_rate, CONFIG_TTYS0_BAUD is used for default baudrate.
For OxPCIe use of 4 MHz for reference was arbitrary giving correct
divisor for 115200 but somewhat inaccurate for lower baudrates.
Actual hardware is 62500000 with 16 times oversampling.
FIXME: Field for baudrate in lb_tables is still incorrect.
Change-Id: I68539738469af780fadd3392263dd9b3d5964d2d
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/5229
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
This option is used to make uart8250mem option visible in menuconfig.
Showing it for these ARMs is incorrect.
Change-Id: I2c28e1c3781df41c09c365355a5105c9fe4945ed
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/5259
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Do not guard the file by CONFIG_CONSOLE_SERIAL8250 or
CONFIG_CONSOLE_SERIAL8250MEM or CONFIG_CONSOLE_SERIAL.
Don't do indirect includes for <uart8250.h>.
The config-specific options are already properly guarded, and there
is no need to guard the register and bit definitions.
Change-Id: I7528b18cdc62bc5c22486f037e14002838a2176e
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/4585
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The miniPCIe ports hanging off 15.0 are infact x1, as are the two
onboard NIC's on 6.0 and 15.0.
Change-Id: I6247838f6b5823369543e338975a4c5c6fd00d7c
Signed-off-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-on: http://review.coreboot.org/5328
Tested-by: build bot (Jenkins)
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Provide ACPI table node so that the PS/2 keyboard/mouse port works
in GNU/Linux.
Change-Id: If73b8d37a81bb9066cbcc650b518d25e243b84e7
Signed-off-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-on: http://review.coreboot.org/5327
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Old video init just replayed the sequence.
This one actually computes the values.
Change-Id: Ic1fe7a2e90dc2cc36ac0d8bcea5cfabc583f09a3
Signed-off-by: Vladimir Serbinenko <phcoder@gmail.com>
Reviewed-on: http://review.coreboot.org/5270
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
SPI registers didnt change since ICH8. No need to have separate
files for them. Unify.
Change-Id: I4e2ac3221b419c007e135c9ee615fc3b84424cbc
Signed-off-by: Vladimir Serbinenko <phcoder@gmail.com>
Reviewed-on: http://review.coreboot.org/5254
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Without this memory decoding isn't activated which, in turn,
makes SeaBIOS crash.
Change-Id: I3dcc721b500ab7468e1082157eeeed38044462d0
Signed-off-by: Vladimir Serbinenko <phcoder@gmail.com>
Reviewed-on: http://review.coreboot.org/5326
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Tested-by: build bot (Jenkins)
No one is interrogating the write_tables() return value. Therefore,
drop it.
Change-Id: I97e707f071942239c9a0fa0914af3679ee7a9c3c
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/5301
Tested-by: build bot (Jenkins)
Reviewed-by: Vladimir Serbinenko <phcoder@gmail.com>
Not really used and conflicts with SSKPD from i915_regs.h
Change-Id: I1462457f656310df99e78aee8cbfe0206f6e2a1e
Signed-off-by: Vladimir Serbinenko <phcoder@gmail.com>
Reviewed-on: http://review.coreboot.org/5268
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Usefull to select between text mode which offers best compatibility with
payloads and gfx mode which makes the best-looking screen.
Also right now we have an unfortunate situation when qemu is in gfx mode
while most real systems use text mode.
Change-Id: Ifad7ba197875edfdd06eb932afeb5800229ef055
Signed-off-by: Vladimir Serbinenko <phcoder@gmail.com>
Reviewed-on: http://review.coreboot.org/5282
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
s_srcaddr is uninitialized in the BSS section, leading to a
garbage valued operand on the LHS of a '<' on line 383.
Change-Id: Ie4fec91b09c70fb1d91ad3918ac3f60653fa1d83
Signed-off-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-on: http://review.coreboot.org/5314
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@google.com>
The get_lb_mem() is no longer used. Therefore, remove it.
Change-Id: I2d8427c460cfbb2b7a9870dfd54f4a75738cfb88
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/5304
Tested-by: build bot (Jenkins)
Reviewed-by: Vladimir Serbinenko <phcoder@gmail.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Instead of packing and unpacking entries in lb_mem use
the bootmem infrastructure for performing sanity checks
during payload loading.
Change-Id: Ica2bee7ebb0f6bf9ded31deac8cb700aa387bc7a
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/5303
Tested-by: build bot (Jenkins)
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
The write_coreboot_table() in coreboot_table.c was already using
struct memrange for managing and building up the entries that
eventually go into the lb_memory table. Abstract that concept
out to a bootmem memory map. The bootmem concept can then be
used as a basis for loading payloads, for example.
Change-Id: I7edbbca6bbd0568f658fde39ca93b126cab88367
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/5302
Tested-by: build bot (Jenkins)
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Clock generator is mobo-specific. Don't touch it in raminit.
Change-Id: Ie114696b7fb13b8daee8dd1393d43bc609e149b3
Signed-off-by: Vladimir Serbinenko <phcoder@gmail.com>
Reviewed-on: http://review.coreboot.org/5265
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Tested-by: build bot (Jenkins)
The selfboot() function relied on global variables
within the selfboot.c compilation unit. Now that the
bounce buffer is a part of struct payload use a new
architecture-specific arch_payload_run() function
for jumping to the payload. selfboot() can then be
removed.
Change-Id: Icec74942e94599542148561b3311ce5096ac5ea5
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/5300
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
In order to break the dependency on selfboot for jumping to
payload the bounce buffer location needs to be communicated.
Therefore, add the bounce buffer to struct payload.
Change-Id: I9d9396e5c5bfba7a63940227ee0bdce6cba39578
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/5299
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
In order to encapsulate more data for self loading use struct
payload as the type. That way modifications to what is needed
for payload loading does not introduce more global variables.
Change-Id: I5b8facd7881e397ca7de1c04cec747fc1dce2d5f
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/5298
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
The selfboot() routine was perfoming most of the common teardown
and stack checking infrastructure. Move that code into
payload_run() to prepare removal of the selfboot() function.
Change-Id: I29f2a5cfcc692f7a0fe2656cb1cda18158c49c6e
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/5297
Tested-by: build bot (Jenkins)
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
A payload can be loaded either from a vboot region or from cbfs.
Provide a common place for choosing where the payload is loaded
from. Additionally, place the logic in the 'loaders' directory
similarly to the ramstage loader infrastructure.
Change-Id: I6b0034ea5ebd04a3d058151819ac77a126a6bfe2
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/5296
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
A comparison with a two's complement in gcccar.inc has dubious
GAS/AT&T notation. Clang miss-parses 0x-1 as an invalid hexadecimal
number.
Change-Id: I88baa5c2513f062ff309df05916a3832b9bd9bb1
Signed-off-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-on: http://review.coreboot.org/5277
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Remove a bunch of dead code which depends either on commented out
#defines, or compiler definitions. Use this opportunity to remove the
need for "-D_7ZIP_ST" in the compiler flags.
Change-Id: Ib6629002be7bf4cee6d95d7baa724893b5e8ba32
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-on: http://review.coreboot.org/5083
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
The PCI ids are taken from:
Intel® 6 Series Chipset and
Intel® C200 Series Chipset
Specification Update – NDA
October 2013
CDI / IBP#: 440377
Change-Id: Ib8418173fd36fd4109b3c4ec0d5543ca8e39ffa6
Signed-off-by: Christopher Douglass <cdouglass.orion@gmail.com>
Reviewed-on: http://review.coreboot.org/5226
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Rather than having it inside mainboard_enable.
Change-Id: Ie8bd25eb49b919b4e25c4628e3557fc66b2ba4d9
Signed-off-by: Vladimir Serbinenko <phcoder@gmail.com>
Reviewed-on: http://review.coreboot.org/4840
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
There are five firmware regions that are (currently) defined. This
was assumed throughout the ifdtool code with many literal 4s and
5s. This patch changes them to refer to a new #define NUM_REGIONS.
Change-Id: I523d3763942f875025ebc4b9ba8b2ccf1db5b2f5
Signed-off-by: Christopher Douglass <cdouglass.orion@gmail.com>
Reviewed-on: http://review.coreboot.org/5313
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The new option "--newlayout <file>" will read <file> in flashrom's
layout format and copy flash regions from the current flash image
file to a new flash image file.
If a region grows, the padding is added at the beginning of the target
region in the new file so that the data is "right-aligned" to the
end of the region.
If a region shrinks, a warning is given and the tail end of existing
data is copied to the target region in the new file.
Regions of zero or negative size are ignored. (In the example below
00fff000:00000fff regions are an artifact of the address encoding
in the register fields.)
Example Usage:
Given a flash image for a board with a Sandy Bridge processor and
Intel 6-Series chipset in the file vpx7654.bin
ifdtool --layout layout.txt vpx7564.bin
will yield the file layout.txt:
00000000:00000fff fd
00180000:003fffff bios
00001000:0017ffff me
00fff000:00000fff gbe
00fff000:00000fff pd
Notice that the "bios" portion extends to the end of the 4MB flash.
It may be edited to extend the bios portion to consume to the extent
of an 8MB flash. like layout2.txt:
00000000:00000fff fd
00180000:007fffff bios
00001000:0017ffff me
00fff000:00000fff gbe
00fff000:00000fff pd
ifdtool --newlayout layout.txt vpx7654.bin
will create a file vpx7654.bin.new that is 8MB.
Change-Id: I0e0925a725c40fa44d8c4b6e86552028779d0523
Signed-off-by: Christopher Douglass <cdouglass.orion@gmail.com>
Reviewed-on: http://review.coreboot.org/5312
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Dump the Intel Flash Descriptor map in the format expected
by flashrom's "layout" option.
Example usage:
Given a 4MB flash image vpx7654.bin that was generated by Intel's
FITC tool for a 6-Series chipset...
./ifdtool --layout l.txt vpx7654.bin
cat l.txt
00000000:00000fff fd
00180000:003fffff bios
00001000:0017ffff me
00fff000:00000fff gbe
00fff000:00000fff pd
Change-Id: Ib740178ed6935b5f6e1dba1be674303f9f980429
Signed-off-by: Christopher Douglass <cdouglass.orion@gmail.com>
Reviewed-on: http://review.coreboot.org/5306
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The same sequence is used regardless of the port
being read or written. Therefore, use the same
implementation for reading or writing to a port.
BUG=None
BRANCH=None
TEST=Built and booted through depthcharge. Dev and recovery
screens still work. Nothing bizarre in console output.
Change-Id: I1a64b54b50472fa7d601e199653eb4a76accf910
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/175441
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: http://review.coreboot.org/4922
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Tested-by: build bot (Jenkins)
The low power subsystem devices have a lot of their
configuration done in the IOSF sideband message space.
Add support for these access methods.
BUG=chrome-os-partner:23790
BRANCH=None
TEST=Built and booted through depthcharge.
Change-Id: I0dd52b952a16ef1280c29301164db041ee87f636
Signed-off-by: Aaron Durbin <adurbin@chromum.org>
Reviewed-on: https://chromium-review.googlesource.com/175440
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Tested-by: Aaron Durbin <adurbin@chromium.org>
Commit-Queue: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/4921
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Tested-by: build bot (Jenkins)
The elog boot counter in cmos was not being initialized
nor incremented. Start doing that in romstage. Since S3
resume is not detected yet the increment is unconditional.
BUG=None
BRANCH=None
TEST=Built and booted through depthcharge multiple times. Noted
output such as 'Boot Count incremented to 4'.
Change-Id: Ic585d4ad4b3af086e0067e28fe0f35c02979bbd2
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/174717
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: http://review.coreboot.org/4919
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Tested-by: build bot (Jenkins)
The ACPI code was previously complaining about not being able
to find the GNVS area: 'ACPI: Could not find CBMEM GNVS'. Fix
this by adding GNVS area early in start up. This is also the
appropriate place to set the acpi_slp_type variable to indicate
an S3 resume or not.
BUG=chrome-os-partner:22867
BUG=chrome-os-partner:23505
BRANCH=None
TEST=Built and booted through depthcharge. Noted cbmem has 'ACPI GNVS'
entry.
Change-Id: Ifbca3dd390ebe573730ee204ca4c2f19626dd6b1
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/174647
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: http://review.coreboot.org/4918
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Tested-by: build bot (Jenkins)
The callers of the following functions assume the storage
area provided by the pointers is initialized. That's not the
case as these were just place holders.
- void acpi_create_intel_hpet(acpi_hpet_t * hpet);
- void acpi_create_serialio_ssdt(acpi_header_t *ssdt);
To fix this properly initialize the hpet entry, and just remove
the serialio_ssdt function entirely.
BUG=chrome-os-partner:23505
BRANCH=None
TEST=Built and booted through depthcharge on rambi. Noted no more
ACPI errors relating to invalid length.
Change-Id: If56ab033562ef2d755e9c9de42f507c95d291aba
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/174716
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: http://review.coreboot.org/4917
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
The EC LPC init function needs to run to enable the internal keyboard.
I needed this to confirm that it is just USB keyboards that are causing
all sorts of issues.
BUG=chrome-os-partner:23635
BRANCH=rambi
TEST=boot to recovery screen and hit tab
Change-Id: Iea0fc66ba62ea7da71ef83c26e25ae32bef102bd
Signed-off-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/175207
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/4915
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Tested-by: build bot (Jenkins)
Enable first SATA port in Rambi device tree.
BUG=chrome-os-partner:23643
TEST=TEST=Manual, in dev mode. Verify on rambi that SATA disk is
detected, and kernel is found + booted.
Change-Id: Ic0cb5f9ff17ca0f6cc7941f203b9338df200811d
Signed-off-by: Shawn Nematbakhsh <shawnn@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/174916
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/4914
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Tested-by: build bot (Jenkins)
Add SATA driver for baytrail platform.
BUG=chrome-os-partner:23643
TEST=Manual, in dev mode. Verify on rambi that SATA disk is detected, and
kernel is found + booted.
Change-Id: I5c13e03203c8f26d233c7d10af8ff6812c460578
Signed-off-by: Shawn Nematbakhsh <shawnn@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/174914
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/4913
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Tested-by: build bot (Jenkins)
Add the on-board devices in the SoC to the device tree.
Also, disable the unused devices aside from TXE and HDA.
Those particular devices cause the system to shut down
when they are disabled.
BUG=chrome-os-partner:22871
BRANCH=None
TEST=Built and booted through depthcharge. Noted the calls to the
southcluster disable function.
Change-Id: I482c1c9609833054aeb2948144af54b57d3df086
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/174645
Reviewed-by: Shawn Nematbakhsh <shawnn@chromium.org>
Reviewed-on: http://review.coreboot.org/4912
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Tested-by: build bot (Jenkins)
When the southcluster pci devices are listed in the devicetree add
the ability to perform the proper disabling sequence for turning
off devices. This only turns off the pci device interface as well
as put the device into D3Hot. It is not yet known how to put the TXE
device into D3Hot so it's currently not possible to disable that
device.
Also, expose the southcluster_enable_dev() function so that other
devices can call this if they require doing specific things before
disabling the device. The southcluster_enable_dev() is only called
on devices found in the devicetree and if they currently have no
ops associated with them.
BUG=chrome-os-partner:22871
BRANCH=None
TEST=Built and booted through depthcharge. Interrogated
output to ensure devices were being properly disabled.
Change-Id: I537ddcb9379907af2fe012948542b6150a8bf7c5
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/174644
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: http://review.coreboot.org/4911
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Kyösti Mälkki