The host bridge PCI device ID can be changed by the firmware. There
is no documentation about it, though. There's 'official' IDs, which
appear in spec updates and Windows drivers, and 'mysterious' IDs,
which Intel doesn't want OSes to know about and thus are not listed.
The current coreboot code seems to be able to change the device ID
of the host bridge, but it seems to be missing a warm reset so that
the device ID changes. Account for the 'mysterious' device IDs in
the northbridge driver, so that booting an OS has a chance to work.
For the sake of completeness, add the PCI device IDs for Clarkdale.
Although only Arrandale is known to work, both of them are Ironlake.
It is possible that the Management Engine handles changing the PCI
device ID, which would not happen when using a broken ME firmware.
Change-Id: I93c9c47e2b0bf13d80c986c7d66b6cdf0e192b22
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45562
Reviewed-by: Nico Huber <nico.h@gmx.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The code is known to work on processors other than just i7's. Also, use
the northbridge's name (Ironlake) in place of the CPU's (Arrandale).
Change-Id: Ia33fa285b4bacd652932d2187384ca1814c9528a
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/46664
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
Move all files with register definitions into a `registers` subfolder.
Subsequent commits will move the remaining registers into this folder.
Tested with BUILD_TIMELESS=1, Packard Bell MS2290 remains identical.
Change-Id: I872269ca3c7fbbcffe83327a20bcf8d98b356beb
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45381
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Several registers have been copy-pasted from i945 and do not exist on
Ironlake. Moreover, other register definitions were missing. Use the
newly-added definitions in existing code, in place of numerical offsets.
Tested with BUILD_TIMELESS=1, Packard Bell MS2290 remains identical.
Change-Id: I8ac99166a8029dcdbb59028b4a7ee297249de5db
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45380
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
The only raminfo field it needs is `memory_reserved_for_heci_mb`. So,
pass in that value directly. As it's read-only, make it const as well.
Change-Id: Ib5d4604e6c1c9bc77df9adfead93b6028d536a3d
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45365
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
There's no need to have it in raminfo. Also, bump MRC_CACHE_VERSION.
Change-Id: Ida48ec4f50c880fe48d88d016acd3737a0650f80
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45364
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
It has been read twice already, so don't read it a third time.
Change-Id: I56ec3a10246f6ebe8074e7b8c164bda6b90eee87
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45363
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Some of the HECI functions do not need raminfo at all.
Change-Id: If0720fa87e5e18820db77a1b61bcdb42ecc538fb
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45362
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
This field is only written to, never read. Drop it from raminfo.
Also, bump MRC_CACHE_VERSION as the saved data layout has changed.
Change-Id: I83d6e69addff996e2f18472d3e1d4f7b9ba974fd
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45361
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
This macro is unused, and RCBA is located in the PCH. Drop it.
Change-Id: Id7c095496360bbe96dc2a36dcc557a1481c02c31
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45347
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This fixes commit e1d1fe454c
initialize 'reply.command'.
The compiler now optimized away the final condition, that checks
the result of heci message, resulting in a binary that always
calls die().
Fix that behaviour by using volatile.
Tested on Lenovo T410: Boots again into Linux.
Change-Id: I63cffc8812bd22695c01bf57283ca593b12e3d87
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45174
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
There may be a gap between TSEG and the graphics stolen memory due to
the alignment done in `raminit.c`. If we allocate MMIO resources in
this range, it misbehaves unpredictably, so reserve it.
TEST=Booted Thinkpad X201s, allocated resources are above TOLUD.
Change-Id: If305e9751ebf4edc945cf038ed72698f3696e52d
Signed-off-by: Nico Huber <nico.h@gmx.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45325
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The PRM does not describe the relevant bits, but Linux's i915 driver
handles these bits the same way for both Ironlake and Sandy Bridge.
Change-Id: Ice7412e335752bd7e297ad50f685effcefbd41d2
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45036
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Create two new functions to fetch mrc_cache data (replacing
mrc_cache_get_current):
- mrc_cache_load_current: fetches the mrc_cache data and drops it into
the given buffer. This is useful for ARM platforms where the mmap
operation is very expensive.
- mrc_cache_mmap_leak: fetch the mrc_cache data and puts it into a
given buffer. This is useful for platforms where the mmap operation
is a no-op (like x86 platforms). As the name mentions, we are not
freeing the memory that we allocated with the mmap, so it is the
caller's responsibility to do so.
Additionally, we are replacing mrc_cache_latest with
mrc_cache_get_latest_slot_info, which does not check the validity of
the data when retrieving the current mrc_cache slot. This allows the
caller some flexibility in deciding where they want the mrc_cache data
stored (either in an mmaped region or at a given address).
BUG=b:150502246
BRANCH=None
TEST=Testing on a nami (x86) device:
reboot from ec console. Make sure memory training happens.
reboot from ec console. Make sure that we don't do training again.
Signed-off-by: Shelley Chen <shchen@google.com>
Change-Id: I259dd4f550719d821bbafa2d445cbae6ea22e988
Reviewed-on: https://review.coreboot.org/c/coreboot/+/44006
Reviewed-by: Furquan Shaikh <furquan@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Other northbridges have a `decode_pcie_bar` function. Since it's not
needed anywhere else, keep it as a static function for now.
Change-Id: Ide42ffcebb73c3e683e0ccaf0ab3aeae805d1123
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/44146
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Like the QPI Link device, there can be more of these devices on
multi-socket platforms. So, name it Physical Layer 0.
Tested with BUILD_TIMELESS=1, Packard Bell MS2290 remains identical.
Change-Id: Ia5f6e42a742bc69237de38f1833e56c8da7c4f7e
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/43737
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
On multi-socket platforms, there can be two QPI buses, each with its own
PCI device. We only have one QPI link on Arrandale, though. In case
support for multi-socket processors ever gets added, name it Link 0.
Tested with BUILD_TIMELESS=1, Packard Bell MS2290 does not change.
Change-Id: I6481154a2d1cc1c84c1f167a374a62af3b2cf3d8
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/43735
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
This register resides within the SAD's config space, and is 64-bit.
Change-Id: I19458f7c6be6b1a5fcd47ac93ee0597f1251a770
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/43733
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Let's hope this cheers up the poor System Address Decoder device.
Tested with BUILD_TIMELESS=1, Packard Bell MS2290 does not change.
Change-Id: Ia62c05abb07216dc1ba449c3a17f8d53050b5af1
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/43732
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Only some registers have such a prefix. Drop it for consistency.
Change-Id: I1ef7307d10a06db8f3c1a05bd9184f21fceb9d90
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/43731
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Uppercase variable names can be confused with register definitions. Use
lowercase names instead, conforming to the coding style guidelines.
Tested with BUILD_TIMELESS=1, Packard Bell MS2290 remains identical.
Change-Id: I61a28bf964ea8c2c662539825ae9f2c88348bdba
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/43730
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
This is the only instance of `BETTER_MEMORY_MAP` in the tree.
Change-Id: I118e5b5a0f10da56e2335828477caed81c5bf855
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/43729
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
This register does not seem to exist on Ironlake.
Change-Id: I3fba6a3fd443f2c9eab874e1d1b8f081f58b1536
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/43728
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Looks like some registers are defined twice. Also, group some QPI
registers together. They were scattered around and mixed with the host
bridge registers, probably because other northbridges have such
registers in the host bridge's PCI config space. But not Ironlake.
Tested with BUILD_TIMELESS=1, Packard Bell MS2290 remains identical.
Change-Id: I6e60f7fcb1467f302618eeab1b0d995920a98569
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/43726
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
This to silent a bug found using gcc-10.
src/northbridge/intel/ironlake/raminit.c: In function 'setup_heci_uma':
src/northbridge/intel/ironlake/raminit.c:1805:11: error: 'reply.command' may be used uninitialized in this function [-Werror=maybe-uninitialized]
1805 | if (reply.command != (MKHI_SET_UMA | (1 << 7)))
| ~~~~~^~~~~~~~
cc1: all warnings being treated as errors
Change-Id: I0d13de549b6d428ac3675ee3f91eb5e42aeb25e8
Signed-off-by: Elyes HAOUAS <ehaouas@noos.fr>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/42461
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
There's no need to set up the southbridge in the northbridge code.
Change-Id: I0f80c92aca885812c27a8803c2745844d8dfb939
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/43689
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
This code is not even being build-tested. Drop it before it grows moss.
Change-Id: I36500c1f0eb3c37d08c691d22382ceca732d1355
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/43231
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Michael Niewöhner
Reflow lines, correct coding style and align struct members, among
other things. As raminit is very large, handle it on a follow-up.
Tested with BUILD_TIMELESS=1, packardbell/ms2290 does not change.
Change-Id: I343edf1bc2a5ac20ff0aa6de4486e685ce430737
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/42701
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Just call the called function directly.
Change-Id: I0c997a63cbbd2b1029f94c23685847df910f8a0e
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/42696
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
