This gets rid of the somewhat unstructured placement of AMD's
sysinfo structure in CAR.
We used to carve out some CAR space using a Kconfig variable,
and then put sysinfo there manually (by "virtue" of pointer magic).
Now it's a variable with the CAR_GLOBAL qualifier, and build
system magic.
For this, the following steps were done (but must happen together
since the intermediates won't build):
- Add new CAR_GLOBAL sysinfo_car
- point all sysinfo pointers to sysinfo_car instead of GLOBAL_VAR
- remove DCACHE_RAM_GLOBAL_VAR_SIZE
- from CAR setup (no need to reserve the space)
- commented out code (that was commented out for years)
- only copy sizeof(sysinfo) into RAM after ram init, where
before it copied the whole GLOBAL_VAR area.
- from Kconfig
Change-Id: I3cbcccd883ca6751326c8e32afde2eb0c91229ed
Signed-off-by: Patrick Georgi <patrick@georgi-clan.de>
Reviewed-on: http://review.coreboot.org/1887
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-by: Marc Jones <marcj303@gmail.com>
This broke because those components were not yet
committed when the patch to drop the driver class
was made.
Change-Id: I29948223503a6c4b196eafa169c064cd26da1be1
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1934
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Tested-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
- Optionally override FSB clock detection in generic
LAPIC code with constant value.
- Override on AMD Model fxx, 10xxx, agesa CPUs with 200MHz
- compile LAPIC code for romstage, too
- Remove #include ".../apic_timer.c" in AMD based mainboards
- Remove custom udelay implementation from intel northbridges' romstages
Future work:
- remove the compile time special case
(requires some cpuid based switching)
- drop northbridge udelay implementations (i945, i5000) if
not required anymore (eg. can SMM use the LAPIC timer?)
Change-Id: I25bacaa2163f5e96ab7f3eaf1994ab6899eff054
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/1618
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Tested-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
When dropping ramstage.a, unused functions with unresolved
symbols are not silently dropped anymore. This makes the
tim5690 compilation fail.
This fix makes sure we don't compile in the int15 handler code
when we don't set CONFIG_VGA_ROM_RUN
Change-Id: If6872c983d9fd811eb33259421f94b551f3b9b34
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1929
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Tested-by: build bot (Jenkins)
Change-Id: Ib4401897570f9e4d31c18d05144b5deb6f4523bc
Signed-off-by: Dave Frodin <dave.frodin@se-eng.com>
Reviewed-on: http://review.coreboot.org/1873
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-by: Marc Jones <marcj303@gmail.com>
When dropping ramstage.a duplicate symbols in ramstage
will start breaking the build. Hence drop all the duplicate
functions implemented by mainboards that have those functions
in generic or component code already.
Change-Id: I5cf8245c67b6f0f348388db54256d28f47017a61
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1865
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Tested-by: build bot (Jenkins)
c_start.o has a special case in the build system, which we can
eliminate, somewhat simplifying the build.
To ensure that the entry point is at the beginning, introduce a
new section .textfirst that is placed appropriately. In principle
the ENTRY() definition in the linker script should be enough, but
better be safe.
Change-Id: I9737f7f5731e12ceb2119eb432b0e09832bc53fa
Signed-off-by: Patrick Georgi <patrick@georgi-clan.de>
Reviewed-on: http://review.coreboot.org/1909
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
The use of ramstage.a required the build system to handle some
object files in a special way, which were put in the drivers
class.
These object files didn't provide any symbols that were used
directly (but only via linker magic), and so the linker never
considered them for inclusion.
With ramstage.a gone, we can drop this special class, too.
Change-Id: I6f1369e08d7d12266b506a5597c3a139c5c41a55
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/1872
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
ramstage.a has two issues:
1. duplicate source filenames don't survive the ar(1)
treatment properly (so files aren't considered)
2. ld doesn't resolve symbols if it isn't forced to, in
particular no overrides of weak symbols
Downside: The resulting binaries get slightly larger.
Link time optimizations should fix that, as would tighter
rules in the build system (to not compile unused code in
the first place).
Change-Id: Iaae771ec8f92b42069237acd3b79c14e5bf9c03d
Signed-off-by: Patrick Georgi <patrick@georgi-clan.de>
Reviewed-on: http://review.coreboot.org/1566
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
sconfig creates empty defaults for all chip_ops, which can be overridden
by drivers simply by providing a concrete implementation.
Change-Id: Ib37515f0b0747bdbf4da780d28690a1e719944b2
Signed-off-by: Patrick Georgi <patrick@georgi-clan.de>
Reviewed-on: http://review.coreboot.org/1567
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The calculation of COMB's irq reading was wrong by the 4-bit shift.
Also, the asl compiler warned about the splitting in lo/hi bytes which
seems unnecessary.
Change-Id: Ia5101d5a19f68c2da827d7e37a18922f959604c7
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1923
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Forgot to update the rk9 for the unified VGABIOS handling.
This applies to rk9 what is done for other boards in commits
3c84261e84d5d340695b
Change-Id: I892b7d81927e277778c1c5251d27416fa79c9868
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/1924
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
For SST chips, the Write-Status-Register instruction must be
executed immediately after the execution of the
Enable-Write-Status-Register instruction, instead of Write-Enable.
Change-Id: I4b3473cd671829def3bd1641ececcf8d9dad4a56
Signed-off-by: Zheng Bao <zheng.bao@amd.com>
Signed-off-by: zbao <fishbaozi@gmail.com>
Reviewed-on: http://review.coreboot.org/1919
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Roda RK9 is a notebook based on the GM45/ICH9 platform using DDR3 memory.
http://roda-computer.com/products/notebooks/rk9/
Tested with various Linux versions, known to work:
- 2x4GB RAM
- IGD
- HD Audio
- UHCI, EHCI
- AHCI
- NIC
- PCI
- PS/2 keyboard
- serial console
- ACPI lid switch
- ACPI battery/AC events
- power off, reboot
Change-Id: I7299dccbff2eea3544363fdd4f49f05aa3dae7bc
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1691
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Add support for ICH9 southbridge
Change-Id: I70612431101bf48d9dcc96ee1b37d257c9ad2ee2
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1690
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
The code supports DDR3 boards only. RAM init for DDR2 is sufficiently
different that it requires separate code, and we have no boards to
test that.
Change-Id: I9076546faf8a2033c89eb95f5eec524439ab9fe1
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1689
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Add an init() function to the chip_operations which will be called
before bus enumeration. This allows to disable unused devices before
they get enumerated.
Change-Id: I63dd9cbfc7b5995ccafb7bf7a81dc71fc67906a0
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1623
Tested-by: build bot (Jenkins)
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-by: Marc Jones <marcj303@gmail.com>
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
This will omit the mc1468181rtc header if it is not needed. Currently
it contains a lot of inlined functions which depend on architecture-
specific IO.
Change-Id: I4ef1bc1362c159e0c780c3eade01af04f029f949
Signed-off-by: David Hendricks <dhendrix@chromium.org>
Reviewed-on: http://review.coreboot.org/1916
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Tested-by: build bot (Jenkins)
This adds more configurability to POST codes. The current assumption
is that POST codes should be written to an IO port (e.g. LPC) if POST
codes are enabled. This changes the assumption so that POST codes can
be written to the serial console without being written to an IO port.
This enables POST codes by default using "default y" to avoid
changing current behavior.
Change-Id: I3db91c358ccb1557096983c4d07f70b2e872c4b3
Signed-off-by: David Hendricks <dhendrix@chromium.org>
Reviewed-on: http://review.coreboot.org/1685
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Some boards have two instances of the int15 handler that supports
the onboard VGA BIOS, for YABEL and realmode.
These are now similar enough that they can be deduplicated.
Due to minor differences this requires manual effort.
Change-Id: I03ae314cb90dd65d96591ce448504aa961cbeb88
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/1893
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Makes it more similar to what realmode looks like.
Change-Id: I4407431f2d979c43dd186114d67ed11845907afe
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/1892
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
By using the (global) register file as defined by x86emu,
we can use the same register access for YABEL and realmode
interrupt handlers.
- the x86 realmode interrupt handlers changed in signature
- to access registers, use X86_$REGNAME now (eg. X86_EAX)
- x86_exception_handler still uses struct eregs *regs to
avoid spilling the x86emu register file stuff everywhere
Coccinelle script that handled most of this commit:
@ inthandler @
identifier FUNC, regs;
@@
int FUNC(
-struct eregs *regs
+void
)
{ ... }
@ depends on inthandler @
identifier regs;
@@
-regs->eax
+X86_EAX
@ depends on inthandler @
identifier regs;
@@
-regs->ebx
+X86_EBX
@ depends on inthandler @
identifier regs;
@@
-regs->ecx
+X86_ECX
@ depends on inthandler @
identifier regs;
@@
-regs->edx
+X86_EDX
@ depends on inthandler @
identifier regs;
@@
-regs->esi
+X86_ESI
@ depends on inthandler @
identifier regs;
@@
-regs->edi
+X86_EDI
@ depends on inthandler @
identifier regs;
@@
-regs->eflags
+X86_EFLAGS
@ depends on inthandler @
identifier regs;
@@
-regs->vector
+M.x86.intno
Change-Id: I60cc2c36646fe4b7f97457b1e297e3df086daa36
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/1891
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
realmode int handlers must return the same codes as the YABEL
int handlers now: 1 for "interrupt handled", 0 for "not handled"
(ie. error).
Change-Id: Idc01cf64e2c97150fc4643671a0bc4cca2ae6668
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/1890
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
It's really a feature in parallel to YABEL/x86emu. Reflect this in
the directory structure.
Change-Id: Ie88e4fa6bfef13d23c55b2db3faacbd90f8cc30b
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/1889
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Change-Id: I4128af7912bec090bbd48acc1b20d0452e7a4a28
Signed-off-by: Dave Frodin <dave.frodin@se-eng.com>
Reviewed-on: http://review.coreboot.org/1876
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-by: Zheng Bao <zheng.bao@amd.com>
Reviewed-by: Marc Jones <marcj303@gmail.com>
e.g.
-#if CONFIG_LOGICAL_CPUS == 1
+#if CONFIG_LOGICAL_CPUS
This will make it easier to switch over to use the config_enabled()
macro later on.
Change-Id: I0bcf223669318a7b1105534087c7675a74c1dd8a
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1874
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
This unused code was not silently dropped as before.
Change-Id: Ic76c58e233869a60c3a8a27c2efc2182b3a4442d
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1863
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
The Agesa wrapper and UDELAY_TIMER2 define their own timer functions,
so don't shove in UDELAY_IO
Change-Id: Ibe3345e825e0c074d5f531dba1198cd6e7b0a42d
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1864
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
This code used a special case for checksum calculation to
prevent the century byte from messing things up, since
writes "sometimes" didn't happen.
That should be stable now, so the special case isn't necessary.
Downside: On century rollovers (ie. 1999-12-31, 2099-12-31)
CMOS will be reset to the defaults.
Change-Id: Ibe589a1ec953b7b3ba39be30cebd9fc2b27326ae
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/1870
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
This ensures that there's only one disable/enable cycle for
the entire rewrite instead for every single byte.
Change-Id: Ic06e6dcb08976d158ff784660838c0fbad875176
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/1869
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
In principle this isn't necessary. However there's a byte (or several)
outside the first 14 bytes that are part of the RTC, and require
locking (century/altCentury).
Since their location is mostly unknown, guard writes properly.
Change-Id: I847cd4efa92722e8504d29feaf7dbfa5c5244b4e
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/1868
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
intel_irq_routing_table is a local structure that should not be used
globally, because it might not be there on all mainboards.
Instead, the API has to be corrected to allow passing a PIRQ table in
where needed.
Change-Id: Icf08928b67727a366639b648bf6aac8e1a87e765
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1862
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Some boards selected GENERATE_ instead of HAVE_
Change-Id: I450c22d7b044f0c88c21692246d452d516a68a83
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1841
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
- move VGA handling options into devices/Kconfig
- make Devices a top level menu
- move some options "closer" to the code they control
Change-Id: Ia79541d18b2b0d9b89a8b154255e312060627c48
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1840
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Both remaining options, DRIVERS_PS2_KEYBOARD and ID_SECTION_OFFSET
are not likely to go away any time soon, so let's not keep them
in Kconfig.deprecated_options but move them close to the code they
control.
Change-Id: I310b877c5b3d5a3444056641c4aee07a48c4c4be
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1839
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
It's only mentioned in Kconfig and never set nor used.
Change-Id: Icc0ac56ae7b325a9e93ed5cdce9dc4b7bab43140
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1838
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
hard_reset was indeed consolidated and moved into the southbridge
code a while ago, but the config variable was still kept alife, with
some duplicate code.
Change-Id: I60d4a87de916667f6e89353dfbe1a7b9eca380f7
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1837
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
- Add mainboard_smi.c from arch/x86/Makefile if it's there
- Add mainboard's chromeos.c from the chromeos Makefile
Change-Id: I3f80e2cb368f88d2a38036895a19f3576dd9553b
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1835
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
This patch is the beginning of a Kconfig cleanup series
- drop CONFIG_HAVE_BUS_CONFIG and add get_bus_conf.c if it
exists in the mainboard directory
- drop duplicate ACPI_SSDTX_NUM from mainboard Kconfig
if it only defines the defaul value of 0
- Add mptable.c, fadt.c, reset.c and ssdtX.asl when they
exist, not based on some Kconfig magic
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Change-Id: Ia14a7116dad6a724af7e531920fee9a51fd0b200
Reviewed-on: http://review.coreboot.org/1832
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The board incorrectly overrides the southbridge hook, so use the
new mainboard hook instead. This change also activates the actual
southbridge hook to enable decode of complete 4 MB flash memory region.
Change-Id: I02c6fe89ae9ad4a7403f024fac875ebd88a8e142
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/1831
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
It's a bool, not a number
Change-Id: I70d52c6af6703101dbd534970ec65275902a283d
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1842
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
The board was broken for use with CONFIG_PCI_OPTION_ROM_RUN_YABEL.
Change-Id: Ia57d630143386fe637af83b9e7345d0d3750b089
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1854
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
YABEL's version of mainboard_interrupt_handlers() was hidden behind an
inline stub. This fixes it.
Change-Id: Ie53424a8ce074e93a720c0ef94cb39994cacd023
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1853
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Some time ago our CMOS checksum algorithm was changed under the topic:
Fix our CMOS checksum algorithm so it matches what /dev/nvram expects
Here is another copy of the algorithm that had to be updated.
Change-Id: I58659c7b8a89c89c76efdff405ee0620e7302277
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1852
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
The VIA chipsets CX700, VT8623 and VX800 required to be
configured with real mode option rom code enabled. This
patch fixes the issue and drops some unneeded header files.
Change-Id: I0d8a3f8f99c2eacec7666f08f85b99f09c06af84
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1833
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
We moved GNVS to it's own section, but forgot to tell the cbmem code
about it. This is purely cosmetical, but add it anyways.
Change-Id: Icb3788c0325ea79cc1efff4a876412d07da7936e
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/1782
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Patrick Georgi