This way we can depend on it during build.
Change-Id: I7e773c6a029e376e3d70d0a8c9e96ffe0c2cf82e
Signed-off-by: Patrick Georgi <Patrick.Georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/845
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The socket mPGA604 is for P4 Xeon which to my knowledge is always
HT-enabled. I assume the existing usage of car/cache_as_ram.inc
on socket_mPGA604, namely the Tyan S2735, as broken.
Existing car/cache_as_ram.inc has invalid SIPI vector and it does
not initialise AP CPU's to activate L2 cache.
Other mPGA604 boards are not affected, as they have not been
converted to CAR.
Change-Id: I7320589695c7f6a695b313a8d0b01b6b1cafbb04
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/607
Tested-by: build bot (Jenkins)
Reviewed-by: Idwer Vollering <vidwer@gmail.com>
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
some blank changing is integrated into the previous patches, which hold
the unsplitted diff hunk.
Change-Id: If9e5066927c5e27fee7ac8422dbfbf2cbeac7df5
Signed-off-by: Zheng Bao <zheng.bao@amd.com>
Signed-off-by: zbao <fishbaozi@gmail.com>
Reviewed-on: http://review.coreboot.org/625
Tested-by: build bot (Jenkins)
Reviewed-by: Marc Jones <marcj303@gmail.com>
It is for S3, storing the recovring data in the nonvolatile storage,
i.e., flash.
Change-Id: Ie9e4f42a80c93d92d2e442f0e833ce06d88294f9
Signed-off-by: Zheng Bao <zheng.bao@amd.com>
Signed-off-by: zbao <fishbaozi@gmail.com>
Reviewed-on: http://review.coreboot.org/620
Tested-by: build bot (Jenkins)
Reviewed-by: Marc Jones <marcj303@gmail.com>
The Option ROM might mess with the EFLAGS register and break assumptions
the C part of coreboot implicitly has, e.g. the state of the direction
flag.
Prevent Option ROMs from confusing coreboot by restoring the old EFLAGS
value after the Option ROMs has finished and always clear the direction
flag before calling the C part of the interrupt handler.
Change-Id: I84663be6681b17f95f48d93f0b730e443336b4a8
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Reviewed-on: http://review.coreboot.org/837
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
ChromeOS features two different modes: normal mode and developer mode
(aka jailbreak mode). In developer mode, we need to display a warning
screen for security reasons.
However, in normal mode we want to boot blazingly fast. Therefore we
don't run (VGA) option ROMs, unless we have to print something on the
screen before the kernel is loaded.
Change-Id: I37f63d0b082a48e037e65bde2b380f9b8743ed29
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/829
Tested-by: build bot (Jenkins)
Reviewed-by: Mathias Krause <minipli@googlemail.com>
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
... and drop duplicate definition in via/epia-n code.
Change-Id: Id79daaaa35c4d412c8c1f621a3638d129681d331
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/820
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Google's ChromeOS can be booted super fast and safely
using coreboot. This adds the ChromeOS specific code that
is required by all ChromeBooks to do this.
Change-Id: Ic03ff090a569a27acbd798ce1e5f89a34897a2f2
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/817
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
This change partially addresses the problem with attempting to
generate coreboot image out of tree. The configuration step fails when
in cheroot, if the destination directory is placed in /tmp.
The problem is that the mconf package tries renaming the temporary
file created in the local directory into the destination config file.
If the destination root and the local directory are located on
different file systems, the rename operation fails.
The proper fix (still upcoming) would be to identify all places where
mconf creates temp files, and make sure that all temp files get
created in the destination tree.
This change modifies just one location, which prevents building out of
tree in the most common case.
Test:
run the following in the coreboot directory in chroot:
(coreboot) cp config.lumpy .config
(coreboot) /bin/rm -rf /tmp/cb
(coreboot) CROSS_COMPILE=i686-pc-linux-gnu- make obj=/tmp/cb oldconfig
(coreboot) CROSS_COMPILE=i686-pc-linux-gnu- make obj=/tmp/cb
Observe the build succeed (it was failing during the config phase
before this change)
Change-Id: If4506e984b8afc192a1689c7b0aa956dd35f66c6
Signed-off-by: Vadim Bendebury <vbendeb@chromium.org>
Reviewed-on: http://review.coreboot.org/815
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
This command removes the first file it finds with the given name by changing
its type to CBFS_COMPONENT_NULL and setting the first character of its name to
a null terminator. If the "files" immediately before or after the target file
are already marked as empty, they're all merged together into one large file.
Change-Id: Idc6b2a4c355c3f039c2ccae81866e3ed6035539b
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-by: Ronald G. Minnich <rminnich@google.com>
Reviewed-on: http://review.coreboot.org/814
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
The x86 memcpy() implementation did not mention its implicit output
registers ESI, EDI and ECX which might make this code miscompile when
the compiler uses the value of EDI for the return value *after* the 'rep
movsb' has completed. That would break the API of memcpy as this would
return 'dst+len' instead of 'dst'.
Fix this possible bug by removing the wrong comment and listing all
output registers as such (using dummy stack variables that get optimized
away).
Also the leading 'cld' is superflous as the ABI mandates the direction
flag to be cleared all the time when we're in C (see
<http://gcc.gnu.org/gcc-4.3/changes.html>) and we have no ASM call sites
that might require it to be cleared explicitly (SMM might come to mind,
but it clears the DF itself before passing control to the C part of the
SMI handler).
Last but not least fix the prototype to match the one from <string.h>.
Change-Id: I106422d41180c4ed876078cabb26b45e49f3fa93
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Reviewed-on: http://review.coreboot.org/836
Tested-by: build bot (Jenkins)
Reviewed-by: Peter Stuge <peter@stuge.se>
This adds detection of x86_64 gcc toolchain (which buildgcc can build
if provided the option).
Change-Id: I8b12f3e705157741279c7347f4847fb50ccc2b0e
Signed-off-by: Marc Jones <marc.jones@se-eng.com>
Reviewed-on: http://review.coreboot.org/673
Reviewed-by: Rudolf Marek <r.marek@assembler.cz>
Tested-by: build bot (Jenkins)
At least one of the console drivers, coreboot fb, uses information in the
sysinfo structure to set itself up. If that structure hasn't been populated,
the driver decides that there is no framebuffer and disables itself. Reversing
the order these are set up fixes that problem.
Change-Id: Idd8b5518980dfdd82fd4359dd0133ab7736fc428
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: http://review.coreboot.org/816
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
... and remove some dead code.
Change-Id: Id959bdf57af09db2a1f5742555c2dcabca38ac9a
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/818
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
It doesnt make sense to delete cscope.out when make
distclean. Distclean is done all the time, and cscope database is also
needed all the time. If we need to delete all the untracked files, we
can use git-clean.
Change-Id: Ic248ccd602ddc88d0b98d5d7f6cbbf530cd82e87
Signed-off-by: Zheng Bao <zheng.bao@amd.com>
Signed-off-by: zbao <fishbaozi@gmail.com>
Reviewed-on: http://review.coreboot.org/831
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Use CPUID to get MAXPHYADDR and set MTRR masks correctly.
Also only BSP CPU clears MTRRs and initializes its Local APIC.
Change-Id: I89ee765a17ec7c041284ed402f21d9a969d699bd
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/686
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Reviewed-by: Idwer Vollering <vidwer@gmail.com>
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
This improvement of CAR code starts the sibling CPU processors and
clears their cache disable bits (CR0.CD) in case a hyper-threading
CPU is detected.
Change-Id: Ieabb86a7c47afb3e178cc75bb89dee3efe0c3d18
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/604
Tested-by: build bot (Jenkins)
Reviewed-by: Idwer Vollering <vidwer@gmail.com>
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Most or many Xeons have no MSR 0x11e.
I have previously tested that a HT-enabled P4 (model f25) can
execute this but will not have cache-as-ram enabled. Should work
for non-HT P4.
Change-Id: I28cbfa68858df45a69aa0d5b050cd829d070ad66
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/644
Tested-by: build bot (Jenkins)
Reviewed-by: Idwer Vollering <vidwer@gmail.com>
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
For a hyper-threading processor, enabling cache requires that both the
BSP and AP CPU clear CR0.CD (Cache Disable) bit. For a Cache-As-Ram
implementation, partial multi-processor initialisation precedes
raminit and AP CPUs' 16bit entry must be run from ROM.
The AP CPU can only start execute real-mode code at a 4kB aligned
address below 1MB. The protected mode entry code for AP is identical
with the BSP code, which is already located at the top of bootblock.
This patch takes the simplest approach and aligns the bootblock
16 bit entry at highest possible 4kB boundary below 1MB.
The symbol ap_sipi_vector is tested to match CONFIG_AP_SIPI_VECTOR
used by the CAR code in romstage. Adress is not expected to ever
change, but if it does, link will fail.
Change-Id: I82e4edbf208c9ba863f51a64e50cd92871c528ef
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/454
Tested-by: build bot (Jenkins)
Reviewed-by: Idwer Vollering <vidwer@gmail.com>
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Delete dead CAR code and whitespace fixes.
Replace cryptic 32bit hex values with existing LAPIC definitions.
Do not assume state of direction flag before "rep" instruction.
Do not load immediate values on temporary registers when not needed.
Parameter pushed on stack was not popped (or flushed) after returning
from call. This is a sort-of memory leak if multiple call's are
implemented the same way.
Change-Id: Ibb93e889b3a0af87b89345c462e331881e78686a
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/643
Tested-by: build bot (Jenkins)
Reviewed-by: Idwer Vollering <vidwer@gmail.com>
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Cache was enabled for the last 4 MB below 4 GB when ramstage is
loaded. This does not cover the case of a 8 MB Flash and could
overlap with some system device placed at high memory.
Use the actual device size for the cache region. Mainboard
may override this with Kconfig CACHE_ROM_SIZE if necessary.
Change-Id: I622223b1e2af0b3c1831f3570b74eacfde7189dc
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/641
Tested-by: build bot (Jenkins)
Reviewed-by: Idwer Vollering <vidwer@gmail.com>
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Copy model_6ex CAR as car/cache_as_ram_ht.inc to be extended
with hyper-threading CPU support.
Change-Id: I09619363e714b1ebf813932b0b22123c1d89010e
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/606
Tested-by: build bot (Jenkins)
Reviewed-by: Idwer Vollering <vidwer@gmail.com>
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
After change it is more clear how romstage is linked twice and with
what scripts. Also with the change, it is easier to add some
object of static size that need to be re-compiled for the 2nd link.
One such object could be md5sum of executable.
Change-Id: Ib34d1876071a51345c5c7319a0ed937868817fd1
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/803
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
There were cases where output file was generated and modified within
a recipe. If make was interrupted, it could exit with an output file
that appears as up-to-date, but was generated with incomplete recipe.
The output file should be created only when successful, in an atomic
operation. There could be other places in the make system which
require a similar fix, this needs to be investigated further.
Change-Id: I25c8ee23577a460eace196fd28c23cc67aa72a9a
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/830
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
- Fix some poor programming practice (breaks of strict aliasing as well
as not checking the return value of read)
- Use PRIx64 instead of %llx to prevent compilation warnings with both
32bit and 64bit compilers
- Use same compiler command options when linking inteltool and when
detecting libpci for inteltool
Change-Id: I08b2e8d1bbc908f6b1f26d25cb3a4b03d818e124
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/752
Tested-by: build bot (Jenkins)
Reviewed-by: Mathias Krause <minipli@googlemail.com>
It's not significantly faster, but easier to read and smaller.
Change-Id: Ibab0b478873912d67bf1f07743f628586353368a
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/755
Reviewed-by: Mathias Krause <minipli@googlemail.com>
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
It's still on by default.
Change-Id: I8b6539eaf2f8d6a4fa975deb14789a00f2090d34
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/756
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
instead of having to edit the source code of tpm.c
Change-Id: I519d9ada14dd383e668a2da4219e5373a24c7c3d
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/757
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
This affects the algorithm when determining when to
transform a range into a larger range with a hole.
It is needed when for when I switch on an 8MB TSEG
and cause the memory maps to go crazy.
Also add header defines for the SMRR.
Change-Id: I1a06ccc28ef139cc79f655a8b19fd3533aca0401
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: http://review.coreboot.org/765
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
With >= 4GB memory installed we get a memory map split in the middle
due to remap that has boundaries that are inconveniently aligned for
MTRRs due to the various UMA regions.
0000MB-2780MB 2780MB RAM (writeback)
2780MB-2782MB 2MB TSEG (uncached/SMRR)
2782MB-2784MB 2MB GFX GTT (uncached)
2784MB-2816MB 32MB GFX UMA (uncached)
2816MB-4096MB 1280MB EMPTY (N/A)
4096MB-5368MB 1272MB RAM (writeback)
5368MB-5376MB 8MB ME UMA (uncached)
The default MTRR allocation method of trying to cover everything
with one MTRR and then carve out a single uncached region does
not work for the GPU aperture which needs write-combining type,
and it also has issues trying to cover the uneven boundaries
in the avaiable variable MTRRs.
My goal was to make a minimal set of changes and avoid modifying
behavior on existing systems with an algorithm that is not always
optimal for a typical memory layout. So the flag 'above4gb=2'
will change these allocation behaviors:
1) Detect the number of available variable MTRRs rather than
limiting to hardcoded value. We need every last MTRR.
2) Don't try to cover all RAM with one MTRR, instead let each
RAM region get covered independently.
3) Don't assume uma_memory_base is part of the last region
and increase the size of that region. In this case the UMA
region is carved out from the lower memory region and it is
already declared as part of the ram region.
4) If a memory region can't be covered with MTRRs >= 16MB then
instead make a larger region and trim it with uncached MTRRs.
Change-Id: I5a60a44ab6d3ae2f46ea6ffa9e3677aaad2485eb
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: http://review.coreboot.org/761
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
I forgot to implement this the first time around.
It does not seem to cause noticeable problems but
in heavy suspend/resume testing I saw a suspicious
crash in the kernel when trying to bring one of the
CPUs back online.
Change-Id: I950ac260f251e2683693d9bd20a0dd5e041aa26e
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: http://review.coreboot.org/770
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Since we do not run option roms in normal mode nothing was
initializing the BDA/EBDA and yet Linux depends very much
on it having sane values here. For the most part the kernel
tries to work around this not being initialized, but every
once in awhile (1/300 boots or so) it would end up reading
something that looked sane from BDA but was not and then
it would panic.
In this change the EBDA is unconditionally setup before devices
are initialized. I'm not set on the location in dev_initialize()
but there does not seem to be another place to hook it in so
that it runs just once for ALL platforms regardless of whether
they use option roms or not. (possibly hardwaremain?)
The EBDA setup code has been moved into its own location in
arch/x86/lib/ebda.c so it can be compiled in even if the option
rom code is not.
The low memory size is still set to 1MB which is enough to make
linux happy without having to hook into each mainboard to get a
more appropriate value. The setup_ebda() function takes inputs
so it could be changed for a mainboard if needed.
OLD/BROKEN would read garbage. Examples from different boots:
ebda_addr=0x75e80 lowmem=0x1553400
ebda_addr=0x5e080 lowmem=0x3e51400
ebda_addr=0x7aa80 lowmem=0x2f8a800
NEW/FIXED now reads consistent values:
ebda_addr=0xf6000 lowmem=0x100000
Change-Id: I6cb79f0e3e43cc65f7e5fe98b6cad1a557ccd949
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: http://review.coreboot.org/769
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
idftool was failing to add the ME blobs into the output image in case
the blob size does not exactly match the size allocated for it in the
flashrom structure.
It is difficult to set the field in the structure to exactly match the
size (for some reason Intel flash tool fails to insert the correct
size even when given the exact ME blob). On the other hand there is no
harm in using am ME blob smaller than the allocated size, this change
modifies the tool building the image to allow for smaller components.
Change-Id: I1b04f90051b91157391943c9bad0eb06dd297431
Signed-off-by: Vadim Bendebury <vbendeb@chromium.org>
Reviewed-on: http://review.coreboot.org/751
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Future CPUs will require TSEG use for SMM
Change-Id: I1432569ece4371d6e12c997e90d66c175fa54c5c
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: http://review.coreboot.org/766
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
This avoids using EBX and instead uses EDI where possible,
and ESI when necessary to get the EBX value out.
This allows me to enable -fpic for SMM TSEG code.
Also add a new CPUID extended function to query with ECX set.
Change-Id: I10dbded3f3ad98a39ba7b53da59af6ca3145e2e5
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: http://review.coreboot.org/764
Tested-by: build bot (Jenkins)
Reviewed-by: Mathias Krause <minipli@googlemail.com>
This change makes significant changes to cbmem.py to make it use the
coreboot tables to find the memory console and timestamp areas instead
of looking for the in memory table TOC structure. That appears to be
more robust and gets cbmem.py working again after some unrelated
changes that affected memory layout.
It also introduces some small infrastructure to make accessing C style
structures in physical memory easier and more transparent.
Change-Id: I51833055a50c2d76423520ba6e059bf8fc50adea
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: http://review.coreboot.org/762
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
It's not used on any board supported by coreboot but has been
detected at run time since ages. No new boards (since 2000?)
are using the CONF2 method, so it is unlikely we ever have to
turn this on for a board.
Change-Id: I17df94a8a77b9338fde10a6b114b44d393776e66
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/758
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
This adds a number of timestamps in ramstage and romstage
so we can figure out where execution time goes.
Change-Id: Iea17c08774e623fc1ca3fa4505b70523ba4cbf01
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/749
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
It looks like the cbfstool utility generates the output file even when
it fails to generate it properly. This causes make, if started second
time in a row, after cbfstool failure, to continue beyond the point of
failure (as the corrupted output file is present in the output tree,
the second make invocation presumes that it is valid, as it is newer
than the dependencies).
The output file should be created only when successful, in an atomic
operation. There could be other places in the make system which
require a similar fix, this needs to be investigated further.
Change-Id: I7c17f033ee5937eb712b1a594122430cee5c9146
Signed-off-by: Vadim Bendebury <vbendeb@chromium.org>
Reviewed-on: http://review.coreboot.org/750
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Tested-by: build bot (Jenkins)
Patrick Georgi