Those GPUs and CPUs are connected together through the coherent fabric
and the GPU memory controllers report errors through x86's MCA so EDAC
needs to support them. The amd64_edac driver supports now HBM (High
Bandwidth Memory) and thus such heterogeneous memory controller
systems
- Other small cleanups and improvements
-----BEGIN PGP SIGNATURE-----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=Vcif
-----END PGP SIGNATURE-----
Merge tag 'ras_core_for_v6.5' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull RAS updates from Borislav Petkov:
- Add initial support for RAS hardware found on AMD server GPUs (MI200).
Those GPUs and CPUs are connected together through the coherent
fabric and the GPU memory controllers report errors through x86's MCA
so EDAC needs to support them. The amd64_edac driver supports now HBM
(High Bandwidth Memory) and thus such heterogeneous memory controller
systems
- Other small cleanups and improvements
* tag 'ras_core_for_v6.5' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
EDAC/amd64: Cache and use GPU node map
EDAC/amd64: Add support for AMD heterogeneous Family 19h Model 30h-3Fh
EDAC/amd64: Document heterogeneous system enumeration
x86/MCE/AMD, EDAC/mce_amd: Decode UMC_V2 ECC errors
x86/amd_nb: Re-sort and re-indent PCI defines
x86/amd_nb: Add MI200 PCI IDs
ras/debugfs: Fix error checking for debugfs_create_dir()
x86/MCE: Check a hw error's address to determine proper recovery action
AMD systems have historically provided an "AMD Node ID" that is a unique
identifier for each die in a multi-die package. This was associated with
a unique instance of the AMD Northbridge on a legacy system. And now it
is associated with a unique instance of the AMD Data Fabric on modern
systems. Each instance is referred to as a "Node"; this is an
AMD-specific term not to be confused with NUMA nodes.
The data fabric provides a number of interfaces accessible through a set
of functions in a single PCI device. There is one PCI device per Data
Fabric (AMD Node), and multi-die systems will see multiple such PCI
devices. The AMD Node ID matches a Node's position in the PCI hierarchy.
For example, the Node 0 is accessed using the first PCI device, Node 1
is accessed using the second, and so on. A logical CPU can find its AMD
Node ID using CPUID. Furthermore, the AMD Node ID is used within the
hardware fabric, so it is not purely a logical value.
Heterogeneous AMD systems, with a CPU Data Fabric connected to GPU data
fabrics, follow a similar convention. Each CPU and GPU die has a unique
AMD Node ID value, and each Node ID corresponds to PCI devices in
sequential order.
However, there are two caveats:
1) GPUs are not x86, and they don't have CPUID to read their AMD Node ID
like on CPUs. This means the value is more implicit and based on PCI
enumeration and hardware-specifics.
2) There is a gap in the hardware values for AMD Node IDs. Values 0-7
are for CPUs and values 8-15 are for GPUs.
For example, a system with one CPU die and two GPUs dies will have the
following values:
CPU0 -> AMD Node 0
GPU0 -> AMD Node 8
GPU1 -> AMD Node 9
EDAC is the only subsystem where this has a practical effect. Memory
errors on AMD systems are commonly reported through MCA to a CPU on the
local AMD Node. The error information is passed along to EDAC where the
AMD EDAC modules use the AMD Node ID of reporting logical CPU to access
AMD Node information.
However, memory errors from a GPU die will be reported to the CPU die.
Therefore, the logical CPU's AMD Node ID can't be used since it won't
match the AMD Node ID of the GPU die. The AMD Node ID of the GPU die is
provided as part of the MCA information, and the value will match the
hardware enumeration (e.g. 8-15).
Handle this situation by discovering GPU dies the same way as CPU dies
in the AMD NB code. But do a "node id" fixup in AMD64 EDAC where it's
needed.
The GPU data fabrics provide a register with the base AMD Node ID for
their local "type", i.e. GPU data fabric. This value is the same for all
fabrics of the same type in a system.
Read and cache the base AMD Node ID from one of the GPU devices during
module initialization. Use this to fixup the "node id" when reporting
memory errors at runtime.
[ bp: Squash a fix making gpu_node_map static as reported by
Tom Rix <trix@redhat.com>.
Link: https://lore.kernel.org/r/20230610210930.174074-1-trix@redhat.com ]
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Co-developed-by: Muralidhara M K <muralidhara.mk@amd.com>
Signed-off-by: Muralidhara M K <muralidhara.mk@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230515113537.1052146-6-muralimk@amd.com
AMD Family 19h Model 30h-3Fh systems can be connected to AMD MI200
accelerator/GPU devices such that the CPU and GPU data fabrics are
connected together. In this configuration, the CPU manages error logging
and reporting for MCA banks located on the GPUs. This includes HBM memory
errors reported from Unified Memory Controllers (UMCs) on the GPUs.
The GPU memory errors are handled like CPU memory errors.
AMD CPU UMC support in EDAC can be re-used for GPU UMC support. However,
keeping them separate means drastic changes in one path (e.g. to support
newer products) should have less impact on the other path.
Also, simplify the "gpu_" helper functions where possible. GPU product
configuration, like memory type and channel count, is fixed compared to
CPU products.
GPU UMCs each have four physical connections (phys) connected to eight
channels. There is a single "chip select". This differs from CPUs where
each UMC has one physical connection connected to one channel, and each
channel has up to four "chip selects".
Enumerate each UMC "phy" as an EDAC CSROW, since there is only a single
chip select for each physical connection. This is similar to how a CPU
UMC "phy" is enumerated as an EDAC CHANNEL, since there is only a single
channel for each physical connection.
Signed-off-by: Muralidhara M K <muralidhara.mk@amd.com>
Co-developed-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Signed-off-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Co-developed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230515113537.1052146-5-muralimk@amd.com
The AMD64 EDAC module version information is not exposed through ABI
like MODULE_VERSION(). Instead it is printed during module init.
Version numbers can be confusing in cases where module updates are
partly backported resulting in a difference between upstream and
backported module versions.
Remove the AMD64 EDAC module version information to avoid user
confusion.
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230410190959.3367528-1-yazen.ghannam@amd.com
Combine all queued EDAC changes for submission into v6.4:
* ras/edac-drivers:
EDAC/i10nm: Add Intel Sierra Forest server support
EDAC/skx: Fix overflows on the DRAM row address mapping arrays
* ras/edac-amd64: (27 commits)
EDAC/amd64: Fix indentation in umc_determine_edac_cap()
EDAC/amd64: Add get_err_info() to pvt->ops
EDAC/amd64: Split dump_misc_regs() into dct/umc functions
EDAC/amd64: Split init_csrows() into dct/umc functions
EDAC/amd64: Split determine_edac_cap() into dct/umc functions
EDAC/amd64: Rename f17h_determine_edac_ctl_cap()
EDAC/amd64: Split setup_mci_misc_attrs() into dct/umc functions
EDAC/amd64: Split ecc_enabled() into dct/umc functions
EDAC/amd64: Split read_mc_regs() into dct/umc functions
EDAC/amd64: Split determine_memory_type() into dct/umc functions
EDAC/amd64: Split read_base_mask() into dct/umc functions
EDAC/amd64: Split prep_chip_selects() into dct/umc functions
EDAC/amd64: Rework hw_info_{get,put}
EDAC/amd64: Merge struct amd64_family_type into struct amd64_pvt
EDAC/amd64: Do not discover ECC symbol size for Family 17h and later
EDAC/amd64: Drop dbam_to_cs() for Family 17h and later
EDAC/amd64: Split get_csrow_nr_pages() into dct/umc functions
EDAC/amd64: Rename debug_display_dimm_sizes()
* ras/edac-misc:
EDAC/altera: Remove MODULE_LICENSE in non-module
EDAC: Sanitize MODULE_AUTHOR strings
EDAC/amd81[13]1: Remove trailing newline from MODULE_AUTHOR
EDAC/i5100: Fix typo in comment
EDAC/altera: Remove redundant error logging
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
GPU Nodes will use a different method to determine the chip select
and channel of an error. A function pointer should be used rather than
introduce another branching condition.
Prepare for this by adding get_err_info() to pvt->ops. This function is
only called from the modern code path, so a legacy function is not
defined.
Make sure to call this after MCA_STATUS[SyndV] is checked, since the
csrow value is found in MCA_SYND.
[ Yazen: rebased/reworked patch and reworded commit message. ]
Signed-off-by: Muralidhara M K <muralidhara.mk@amd.com>
Co-developed-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Signed-off-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Co-developed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-23-yazen.ghannam@amd.com
Call them from their respective setup_mci_misc_attrs() paths.
Also, drop the check for an "empty" device, i.e. one without memory.
This is redundant and already done in instance_has_memory() earlier in
the init path.
No functional change is intended.
[ Yazen: rebased/reworked patch and reworded commit message. ]
Signed-off-by: Muralidhara M K <muralidhara.mk@amd.com>
Co-developed-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Signed-off-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Co-developed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-21-yazen.ghannam@amd.com
The init_one_instance() path is shared between legacy and modern
systems. So add the new functions to a function pointer in pvt->ops.
No functional change is intended.
[ Yazen: Rebased/reworked patch and reworded commit message. ]
Signed-off-by: Muralidhara M K <muralidhara.mk@amd.com>
Co-developed-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Signed-off-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Co-developed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-18-yazen.ghannam@amd.com
Call them using a function pointer in pvt->ops. The "ECC enabled"
check is done outside of the hardware information gathering done in
hw_info_get(). So a high-level function pointer is needed to separate
the legacy and modern paths.
No functional change is intended.
[Yazen: rebased/reworked patch and reworded commit message. ]
Signed-off-by: Muralidhara M K <muralidhara.mk@amd.com>
Co-developed-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Signed-off-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Co-developed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-17-yazen.ghannam@amd.com
Call them from their respective hw_info_get() paths.
ECC symbol size is not needed on UMC systems, so determine_ecc_sym_sz()
is left out of the UMC path. Do not save TOP_MEM* values on modern
controllers because they're not needed there (read: they were used only
for debugging, if anything).
[ Yazen: rebased/reworked patch and reworded commit message. ]
Signed-off-by: Muralidhara M K <muralidhara.mk@amd.com>
Co-developed-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Signed-off-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Co-developed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-16-yazen.ghannam@amd.com
Call them from their respective hw_info_get() paths.
Call them after all other hardware registers have been saved, since the
memory type for a device will be determined based on the saved
information.
[ Yazen: rebased/reworked patch and reworded commit message. ]
Signed-off-by: Muralidhara M K <muralidhara.mk@amd.com>
Co-developed-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Signed-off-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Co-developed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-15-yazen.ghannam@amd.com
Call them from their respective hw_info_get() paths.
Call the new functions after the setting the chip select base and mask
counts, since those are need to read the correct number of chip select
base and mask registers. And call the new functions before the remaining
set up, because the base and mask register values will be needed later.
[Yazen: Rebased/reworked patch and reworded commit message. ]
Signed-off-by: Muralidhara M K <muralidhara.mk@amd.com>
Co-developed-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Signed-off-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Co-developed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-14-yazen.ghannam@amd.com
Call them from their respective hw_info_get() function. Avoid the
need for family/model-based function pointers.
Add the calls before reading hardware registers from the memory
controllers, since the number of chip select bases and masks needs to be
known first.
[ Yazen: rebased/reworked patch and reworded commit message. ]
Signed-off-by: Muralidhara M K <muralidhara.mk@amd.com>
Co-developed-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Signed-off-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Co-developed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-13-yazen.ghannam@amd.com
The bulk of system-specific information is gathered at init time with
hw_info_get(). This function calls a number of helper functions, and
many of these helper functions are split between a modern UMC/DF path
and a legacy DCT path.
Split hw_info_get() into legacy and modern versions. This creates two
separate code paths early on, and legacy and modern helper functions can
be called directly in the appropriate code path.
Also, simplify hw_info_put() and share it between legacy and modern
systems. NULL pointer checks are done in pci_dev_put() and kfree(), so
they can be called unconditionally.
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-12-yazen.ghannam@amd.com
Future AMD systems will support heterogeneous "AMD Node" types, e.g.
CPU and GPU types. Therefore, a global family type shared across all
AMD nodes is no longer appropriate.
Move struct low_ops routines and members of struct amd64_family_type
to struct amd64_pvt.
Currently, there are many code branches that split between "modern" and
"legacy" systems. Another code branch will be needed in order to cover
GPU cases. However, rather than introduce another branching case in
multiple functions, the current branching code should be switched to a
set of function pointers. This change makes the code more readable and
simplifies adding support for new families/models.
In order to reuse code, define two sets of function pointers. Use one
for modern systems (Family 17h and later). This will not change between
current CPU families. Use another set of function pointers for legacy
systems (before Family 17h). Use the Family 16h versions as default
for the legacy ops since these are the latest, and adjust the function
pointers as needed for older families.
[ Yazen: rebased/reworked patch and reworded commit message. ]
[ bp: Fix rev8 or later check. ]
Signed-off-by: Muralidhara M K <muralidhara.mk@amd.com>
Co-developed-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Signed-off-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
Co-developed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-11-yazen.ghannam@amd.com
The ECC symbol size was needed on legacy system to lookup the ECC syndrome.
This is not needed on modern systems because the ECC syndrome is explicitly
provided in the MCA information.
Remove the ECC symbol size discovery code for modern UMC-based systems.
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-10-yazen.ghannam@amd.com
The same function is used to calculate chip select size for all Zen-based
family/models. Therefore, a family/model function pointer is not necessary.
Drop the dbam_to_cs() function pointer for Family 17h and later systems.
Also, move the Family 17h function to avoid a forward declaration. Rename
it to indicate that the UMC Address Mask is used rather than the legacy
DBAM value.
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-9-yazen.ghannam@amd.com
Split get_csrow_nr_pages() into a legacy and modern versions in preparation
for further legacy/modern refactoring.
Also, rename f17_get_cs_mode() to match the new convention.
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-8-yazen.ghannam@amd.com
Use the "dct" and "umc" prefixes for legacy and modern versions
respectively.
Also, move the "dct" version to avoid a forward declaration, and fixup
some checkpatch warnings in the process.
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-7-yazen.ghannam@amd.com
Reportedly, clang cannot do interprocedural analysis:
https://lore.kernel.org/r/20230213-amd64_edac-wsometimes-uninitialized-v1-1-5bde32b89e02@kernel.org
and see that those arguments won't be used uninitialized.
So, yeah, the code's fine even without this. Normally, such a "fix"
won't be applied but that warning gets automatically enabled in -Wall
builds and when CONFIG_WERROR is set in allmodconfig builds, the build
fails.
So shut it up with a minimal fix as this code will see more
reorganization very soon.
[ bp: Write commit message. ]
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Link: https://lore.kernel.org/r/Y%2BqdVHidnrrKvxiD@dev-arch.thelio-3990X
The early_channel_count() function seems to have been useful in the past
for knowing how many EDAC mci structures to populate. However, this is no
longer needed as the maximum channel count for a system is used instead.
Remove the early_channel_count() helper functions and related code. Use the
size of the channel layer when iterating over channel structures.
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-6-yazen.ghannam@amd.com
PCI Function 0 is used on Family 17h and later only to read the "dhar"
value. This value is printed and provided through a module-specific
debug sysfs file. The value is not used for any Family 17h and later
code, and it does not have any apparent debug value on these systems.
Remove "dhar", Function 0 PCI IDs, and all related code.
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-5-yazen.ghannam@amd.com
PCI Function 6 is used on Family 17h and later to access scrub
registers. With scrub access removed, this function has no other use.
Remove all Function 6 PCI IDs and related code.
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-4-yazen.ghannam@amd.com
The scrub registers on AMD Family 17h and later may be inaccessible to
the OS. Furthermore, hardware designers recommend that the scrubbing
feature is managed by the firmware.
Remove support for the sdram_scrub_rate interface for AMD Family 17h
systems and later by not setting the scrub function pointers. The EDAC MC
core will then not expose the scrub files in sysfs.
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-3-yazen.ghannam@amd.com
EDAC PCI control is used to detect/report legacy PCI errors like
"Parity" and "SERROR". Modern AMD systems use PCIe Advanced Error
Reporting (AER), and legacy PCI errors should not be reported.
Remove EDAC PCI control setup on AMD Family 17h and later systems.
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230127170419.1824692-2-yazen.ghannam@amd.com
Call ghes_get_devices() to check whether ghes_edac should be used on the
platform where it is preferred over the corresponding chipset-specific
EDAC driver.
Unlike the existing edac_get_owner() check, the ghes_get_devices() check
works independent to the module_init ordering.
[ bp: Massage. ]
Suggested-by: Toshi Kani <toshi.kani@hpe.com>
Signed-off-by: Jia He <justin.he@arm.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lore.kernel.org/r/20221010023559.69655-6-justin.he@arm.com
amd_cache_northbridges() is exported by amd_nb.c and is called by
amd64-agp.c and amd64_edac.c modules at module_init() time so that NB
descriptors are properly cached before those drivers can use them.
However, the init_amd_nbs() initcall already does call
amd_cache_northbridges() unconditionally and thus makes sure the NB
descriptors are enumerated.
That initcall is a fs_initcall type which is on the 5th group (starting
from 0) of initcalls that gets run in increasing numerical order by the
init code.
The module_init() call is turned into an __initcall() in the MODULE=n
case and those are device-level initcalls, i.e., group 6.
Therefore, the northbridges caching is already finished by the time
module initialization starts and thus the correct initialization order
is retained.
Unexport amd_cache_northbridges(), update dependent modules to
call amd_nb_num() instead. While at it, simplify the checks in
amd_cache_northbridges().
[ bp: Heavily massage and *actually* explain why the change is ok. ]
Signed-off-by: Muralidhara M K <muralimk@amd.com>
Signed-off-by: Naveen Krishna Chatradhi <nchatrad@amd.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lore.kernel.org/r/20220324122729.221765-1-nchatrad@amd.com
Introduce a "family flags" bitmask that can be used to indicate any
special behavior needed on a per-family basis.
Add a flag to indicate a system uses the new register offsets introduced
with Family 19h Model 10h.
Use this flag to account for register offset changes, a new bitfield
indicating DDR5 use on a memory controller, and to set the proper number
of chip select masks.
Rework f17_addr_mask_to_cs_size() to properly handle the change in chip
select masks. And update code comments to reflect the updated Chip
Select, DIMM, and Mask relationships.
[uninitialized variable warning]
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: William Roche <william.roche@oracle.com>
Link: https://lore.kernel.org/r/20220202144307.2678405-3-yazen.ghannam@amd.com
Current AMD systems allow mixing of DIMM types within a system. However,
DIMMs within a channel, i.e. managed by a single Unified Memory
Controller (UMC), must be of the same type.
Handle this possible configuration by checking and setting the memory
type for each individual "UMC" structure.
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: William Roche <william.roche@oracle.com>
Link: https://lore.kernel.org/r/20220202144307.2678405-2-yazen.ghannam@amd.com
- Add support for DRR5 and new models 0x10-0x1f and 0x50-0x5f of AMD
family 0x19 CPUs to amd64_edac
- The usual set of fixes and cleanups
-----BEGIN PGP SIGNATURE-----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=85ZE
-----END PGP SIGNATURE-----
Merge tag 'edac_updates_for_v5.17_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras
Pull EDAC updates from Borislav Petkov:
- Add support for version 3 of the Synopsys DDR controller to
synopsys_edac
- Add support for DRR5 and new models 0x10-0x1f and 0x50-0x5f of AMD
family 0x19 CPUs to amd64_edac
- The usual set of fixes and cleanups
* tag 'edac_updates_for_v5.17_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras:
EDAC/amd64: Add support for family 19h, models 50h-5fh
EDAC/sb_edac: Remove redundant initialization of variable rc
RAS/CEC: Remove a repeated 'an' in a comment
EDAC/amd64: Add support for AMD Family 19h Models 10h-1Fh and A0h-AFh
EDAC: Add RDDR5 and LRDDR5 memory types
EDAC/sifive: Fix non-kernel-doc comment
dt-bindings: memory: Add entry for version 3.80a
EDAC/synopsys: Enable the driver on Intel's N5X platform
EDAC/synopsys: Add support for version 3 of the Synopsys EDAC DDR
EDAC/synopsys: Use the quirk for version instead of ddr version
Add the new family 19h models 50h-5fh PCI IDs (device 18h functions 0
and 6) to support Ryzen 5000 APUs ("Cezanne").
Signed-off-by: Marc Bevand <m@zorinaq.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Link: https://lore.kernel.org/r/20211221233112.556927-1-m@zorinaq.com
Add a new family type for AMD Family 19h Models 10h to 1Fh. Use this new
family type for Models A0h to AFh also.
Increase the maximum number of controllers from 8 to 12.
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lore.kernel.org/r/20211208174356.1997855-3-yazen.ghannam@amd.com
Define an address translation context struct. This will hold values that
will be passed between multiple functions.
Save return address, Node ID, and the Instance ID number to start.
Currently, the UMC number is used as the Instance ID, but future DF
versions may use another value.
Also include a "tmp" field to use when reading registers. This is to
avoid having to define a temporary variable in multiple functions.
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lkml.kernel.org/r/20211028175728.121452-5-yazen.ghannam@amd.com
The DF Indirect Access method allows for "Broadcast" accesses in which
case no specific instance is targeted. Add support using a reserved
instance ID of 0xFF to indicate a broadcast access. Set the FICAA
register appropriately.
Define helpers functions for instance and broadcast reads and use them
where appropriate.
Drop the "amd_" prefix since these functions are all static.
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lkml.kernel.org/r/20211028175728.121452-4-yazen.ghannam@amd.com
AMD Rome systems and later support interleaving between three identical
ranks within a channel.
Check for this mode by counting the number of enabled chip selects and
comparing their masks. If there are exactly three enabled chip selects
and their masks are identical, then three rank interleaving is enabled.
The size of a rank is determined from its mask value. However, three
rank interleaving doesn't follow the method of swapping an interleave
bit with the most significant bit. Rather, the interleave bit is flipped
and the most significant bit remains the same. There is only a single
interleave bit in this case.
Account for this when determining the chip select size by keeping the
most significant bit at its original value and ignoring any zero bits.
This will return a full bitmask in [MSB:1].
Fixes: e53a3b267f ("EDAC/amd64: Find Chip Select memory size using Address Mask")
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lkml.kernel.org/r/20211005154419.2060504-1-yazen.ghannam@amd.com
Instead of "open coding" DEVICE_ATTR, use the corresponding
helper macros DEVICE_ATTR_{RW,RO,WO} in amd64_edac.c
Some function names needed to be changed to match the device
conventions <foo>_show and <foo>_store, but the functionality
itself is unchanged.
The devices using EDAC_DCT_ATTR_SHOW() are left unchanged.
Reviewed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Dwaipayan Ray <dwaipayanray1@gmail.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Link: https://lore.kernel.org/r/20210713065130.2151-1-dwaipayanray1@gmail.com
The SYSCFG MSR continued being updated beyond the K8 family; drop the K8
name from it.
Suggested-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: Brijesh Singh <brijesh.singh@amd.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Joerg Roedel <jroedel@suse.de>
Link: https://lkml.kernel.org/r/20210427111636.1207-4-brijesh.singh@amd.com
amd64_edac was converted to CPU family autoprobing (from PCI device
IDs) to not have to add a new PCI device ID each time a new platform is
shipped but to support the whole family out-of-the-box.
However, this caused a lot of noise in dmesg even when the machine
doesn't have ECC DIMMs or ECC has been disabled in the BIOS:
EDAC MC: Ver: 3.0.0
EDAC amd64: F17h detected (node 0).
EDAC amd64: Node 0: DRAM ECC disabled.
EDAC amd64: F17h detected (node 1).
EDAC amd64: Node 1: DRAM ECC disabled.
EDAC amd64: F17h detected (node 2).
EDAC amd64: Node 2: DRAM ECC disabled.
EDAC amd64: F17h detected (node 3).
EDAC amd64: Node 3: DRAM ECC disabled.
EDAC amd64: F17h detected (node 4).
EDAC amd64: Node 4: DRAM ECC disabled.
EDAC amd64: F17h detected (node 5).
EDAC amd64: Node 5: DRAM ECC disabled.
EDAC amd64: F17h detected (node 6).
EDAC amd64: Node 6: DRAM ECC disabled.
EDAC amd64: F17h detected (node 7).
EDAC amd64: Node 7: DRAM ECC disabled.
or even
$ grep EDAC dmesg.log | sed 's/\[.*\] //' | sort | uniq -c
128 EDAC amd64: F17h detected (node 0).
128 EDAC amd64: Node 0: DRAM ECC disabled.
1 EDAC MC: Ver: 3.0.0
on a big machine. Yap, that's once per CPU for 128 of them.
So move the init messages after all probing has succeeded to avoid
unnecessary spew in dmesg.
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lkml.kernel.org/r/20210119164141.17417-1-bp@alien8.de
Families up to and including 0x16 allow access to the injection
hardware. Starting with family 0x17, access to those registers is
blocked by security policy.
Limit that only on the families which support it.
Suggested-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lkml.kernel.org/r/20201222180013.GD13463@zn.tnic
Merge them into the main driver and put them inside an EDAC_DEBUG
ifdeffery to simplify the driver and have all debugging/injection stuff
behind a debug build-time switch.
No functional changes.
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Link: https://lkml.kernel.org/r/20201215110517.5215-2-bp@alien8.de
There's no need for them to be in a separate file so merge them into the
main driver compilation unit like the other EDAC drivers do.
Drop now-unneeded function export, make the function static and shorten
static function names.
No functional changes.
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Link: https://lkml.kernel.org/r/20201215110517.5215-1-bp@alien8.de
Avadhut Naik