kernel/0001-Add-support-for-deprecating-processors.patch

304 lines
8.8 KiB
Diff
Raw Normal View History

From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Laura Abbott <labbott@redhat.com>
Date: Wed, 13 Nov 2019 14:44:30 -0500
Subject: [PATCH] Add support for deprecating processors
Message-id: <20191113144431.9403-4-labbott@redhat.com>
Patchwork-id: 286081
O-Subject: [ARK INTERNAL PATCHv2 3/4] [redhat] Add support for deprecating processors
Bugzilla:
RH-Acked-by: Don Zickus <dzickus@redhat.com>
RH-Acked-by: Prarit Bhargava <prarit@redhat.com>
RH-Acked-by: Jiri Benc <jbenc@redhat.com>
This is the squashed version of the following patches
modifed to use the new CONFIG_RH_DISABLE_DEPRECATED option
add rh_check_supported
Message-id: <1525313397-75200-1-git-send-email-darcari@redhat.com>
Patchwork-id: 212189
O-Subject: [RHEL8.0 BZ 1565717 v2] x86: add rh_check_supported
Bugzilla: 1565717
RH-Acked-by: Steve Best <sbest@redhat.com>
RH-Acked-by: Mikulas Patocka <mpatocka@redhat.com>
Description:
Add code to verify that the booted x86 processor is supported by
Red Hat. In cases where the processor is not supported a critical
message is logged.
Initially the supported set of processors for RHEL8 remains the same
as RHEL7.
Bugzilla: http://bugzilla.redhat.com/1565717
Brew: https://brewweb.devel.redhat.com/taskinfo?taskID=15966242
Test: Booted brew build on supported (Haswell) and unsupported processor (CNL)
and achieved desired results.
Upstream: RHEL Only
Cc: David Arcari <darcari@redhat.com>
Cc: Mikulas Patocka <mpatocka@redhat.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Herton R. Krzesinski <herton@redhat.com>
update rh_check_supported processor list
Message-id: <1530618602-17477-1-git-send-email-darcari@redhat.com>
Patchwork-id: 223452
O-Subject: [RHEL8.0 BZ 1595918] x86: update rh_check_supported processor list
Bugzilla: 1595918
RH-Acked-by: Jarod Wilson <jarod@redhat.com>
RH-Acked-by: Prarit Bhargava <prarit@redhat.com>
RH-Acked-by: Steve Best <sbest@redhat.com>
Bugzilla: http://bugzilla.redhat.com/1595918
Upstream Status: RHEL_only
Build Info: https://brewweb.engineering.redhat.com/brew/taskinfo?taskID=16916208
Tested: Verified on a varity of supported and unsupported Intel and AMD systems.
Results as expected.
In RHEL8, we are marking some of the older processors as unsupported. No code
is being removed; however, the older processors are not supported and as such
are not candidates for certification.
Cc: David Arcari <darcari@redhat.com>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Steve Best <sbest@redhat.com>
Signed-off-by: Herton R. Krzesinski <herton@redhat.com>
Signed-off-by: Jakub Racek <jracek@redhat.com>
refresh: rename FAM6_ATOM again, see f2c4db1bd
mark whiskey-lake processor supported
Message-id: <1533125629-12870-1-git-send-email-darcari@redhat.com>
Patchwork-id: 225809
O-Subject: [ BZ 1609604] mark whiskey-lake processor supported
Bugzilla: 1609604
RH-Acked-by: Steve Best <sbest@redhat.com>
RH-Acked-by: Tony Camuso <tcamuso@redhat.com>
RH-Acked-by: Prarit Bhargava <prarit@redhat.com>
Bugzilla: http://bugzilla.redhat.com/1609604
Build Info: https://brewweb.devel.redhat.com/taskinfo?taskID=17474338
Upstream Status: RHEL Only
Tested: Successful excecution of platform-test suite.
Whiskey Lake is cpu model 142 with stepping 11. Stepping > 11 for cpu model
142 is not supported.
For model 158, the highest supported stepping is 10.
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: David Arcari <darcari@redhat.com>
Cc: Dave Young <dyoung@redhat.com>
Signed-off-by: Herton R. Krzesinski <herton@redhat.com>
mark intel knights landing and knights mill unsupported
Message-id: <1533580277-126323-1-git-send-email-darcari@redhat.com>
Patchwork-id: 226187
O-Subject: [RHEL8.0 BZ 1610493] mark intel knights landing and knights mill unsupported
Bugzilla: 1610493
RH-Acked-by: Tony Camuso <tcamuso@redhat.com>
RH-Acked-by: Prarit Bhargava <prarit@redhat.com>
RH-Acked-by: Steve Best <sbest@redhat.com>
Bugzilla: http://bugzilla.redhat.com/1610493
Build Info: https://brewweb.engineering.redhat.com/brew/taskinfo?taskID=17625348
Tested: verified on KNL/KNM system
Upstream Status: RHEL Only
Depends: http://bugzilla.redhat.com/1609604
We are marking KNL/KNM processors as unsupported. No code is being removed;
however, the older processors are not supported and as such are not
candidates for certification.
Upstream Status: RHEL only
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: David Arcari <darcari@redhat.com>
Cc: Dave Young <dyoung@redhat.com>
Signed-off-by: Herton R. Krzesinski <herton@redhat.com>
---
arch/x86/kernel/cpu/common.c | 1 +
arch/x86/kernel/setup.c | 132 ++++++++++++++++++++++++++++++++++-
2 files changed, 132 insertions(+), 1 deletion(-)
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
kernel-5.8.0-0.rc0.20200610git84fc461db99b.1 * Wed Jun 10 2020 Fedora Kernel Team <kernel-team@fedoraproject.org> [5.8.0-0.rc0.20200610git84fc461db99b.1] - 84fc461db99b rebase - Fix PATCHLEVEL for merge window ("Justin M. Forbes") - PCI: tegra: Revert raw_violation_fixup for tegra124 (Nicolas Chauvet) - Change ark CONFIG_COMMON_CLK to yes, it is selected already by other options ("Justin M. Forbes") - More module filtering for Fedora ("Justin M. Forbes") - Update filters for rnbd in Fedora ("Justin M. Forbes") - Fix up module filtering for 5.8 ("Justin M. Forbes") - More Fedora config work ("Justin M. Forbes") - RTW88BE and CE have been extracted to their own modules ("Justin M. Forbes") - Set CONFIG_BLK_INLINE_ENCRYPTION_FALLBACK for Fedora ("Justin M. Forbes") - Arm64 Use Branch Target Identification for kernel ("Justin M. Forbes") - Fedora config updates ("Justin M. Forbes") - Change value of CONFIG_SECURITY_SELINUX_CHECKREQPROT_VALUE ("Justin M. Forbes") - Fix configs for Fedora ("Justin M. Forbes") - Fix update_scripts.sh unselective pattern sub (David Howells) - Updated changelog for the release based on b0c3ba31be3e ("CKI@GitLab") - Drop the static path configuration for the Sphinx docs (Jeremy Cline) - Sign off generated configuration patches (Jeremy Cline) - Use __make macro instead of make (Tom Stellard) - redhat/configs: Enable CONFIG_SMC91X and disable CONFIG_SMC911X (Prarit Bhargava) [http://bugzilla.redhat.com/1722136] Resolves: rhbz# Signed-off-by: Justin M. Forbes <jforbes@fedoraproject.org>
2020-06-10 22:01:45 +00:00
index 8be042df12c3..a723808f4244 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
kernel-5.8.0-0.rc0.20200610git84fc461db99b.1 * Wed Jun 10 2020 Fedora Kernel Team <kernel-team@fedoraproject.org> [5.8.0-0.rc0.20200610git84fc461db99b.1] - 84fc461db99b rebase - Fix PATCHLEVEL for merge window ("Justin M. Forbes") - PCI: tegra: Revert raw_violation_fixup for tegra124 (Nicolas Chauvet) - Change ark CONFIG_COMMON_CLK to yes, it is selected already by other options ("Justin M. Forbes") - More module filtering for Fedora ("Justin M. Forbes") - Update filters for rnbd in Fedora ("Justin M. Forbes") - Fix up module filtering for 5.8 ("Justin M. Forbes") - More Fedora config work ("Justin M. Forbes") - RTW88BE and CE have been extracted to their own modules ("Justin M. Forbes") - Set CONFIG_BLK_INLINE_ENCRYPTION_FALLBACK for Fedora ("Justin M. Forbes") - Arm64 Use Branch Target Identification for kernel ("Justin M. Forbes") - Fedora config updates ("Justin M. Forbes") - Change value of CONFIG_SECURITY_SELINUX_CHECKREQPROT_VALUE ("Justin M. Forbes") - Fix configs for Fedora ("Justin M. Forbes") - Fix update_scripts.sh unselective pattern sub (David Howells) - Updated changelog for the release based on b0c3ba31be3e ("CKI@GitLab") - Drop the static path configuration for the Sphinx docs (Jeremy Cline) - Sign off generated configuration patches (Jeremy Cline) - Use __make macro instead of make (Tom Stellard) - redhat/configs: Enable CONFIG_SMC91X and disable CONFIG_SMC911X (Prarit Bhargava) [http://bugzilla.redhat.com/1722136] Resolves: rhbz# Signed-off-by: Justin M. Forbes <jforbes@fedoraproject.org>
2020-06-10 22:01:45 +00:00
@@ -1236,6 +1236,7 @@ static void __init early_identify_cpu(struct cpuinfo_x86 *c)
cpu_detect(c);
get_cpu_vendor(c);
get_cpu_cap(c);
+ get_model_name(c); /* RHEL: get model name for unsupported check */
get_cpu_address_sizes(c);
setup_force_cpu_cap(X86_FEATURE_CPUID);
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index a3767e74c758..a1b8cb9a2579 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -44,6 +44,7 @@
#include <asm/unwind.h>
#include <asm/vsyscall.h>
#include <linux/vmalloc.h>
+#include <asm/intel-family.h>
/*
* max_low_pfn_mapped: highest directly mapped pfn < 4 GB
@@ -753,7 +754,132 @@ static void __init trim_low_memory_range(void)
{
memblock_reserve(0, ALIGN(reserve_low, PAGE_SIZE));
}
-
+
+static bool valid_amd_processor(__u8 family, const char *model_id)
+{
+ bool valid;
+
+ switch(family) {
+ case 0x15:
+ valid = true;
+ break;
+
+ case 0x17:
+ valid = strstr(model_id, "AMD EPYC 7");
+ break;
+
+ default:
+ valid = false;
+ break;
+ }
+
+ return valid;
+}
+
+static bool valid_intel_processor(__u8 family, __u8 model, __u8 stepping)
+{
+ bool valid;
+
+ if (family != 6)
+ return false;
+
+ switch(model) {
+ case INTEL_FAM6_ATOM_GOLDMONT_D:
+ case INTEL_FAM6_ATOM_GOLDMONT_PLUS:
+
+ case INTEL_FAM6_BROADWELL:
+ case INTEL_FAM6_BROADWELL_G:
+ case INTEL_FAM6_BROADWELL_X:
+ case INTEL_FAM6_BROADWELL_D:
+
+ case INTEL_FAM6_HASWELL:
+ case INTEL_FAM6_HASWELL_G:
+ case INTEL_FAM6_HASWELL_L:
+ case INTEL_FAM6_HASWELL_X:
+ valid = true;
+ break;
+
+ case INTEL_FAM6_KABYLAKE:
+ valid = (stepping <= 10);
+ break;
+
+ case INTEL_FAM6_KABYLAKE_L:
+ valid = (stepping <= 11);
+ break;
+
+ case INTEL_FAM6_SKYLAKE_L:
+ case INTEL_FAM6_SKYLAKE:
+ case INTEL_FAM6_SKYLAKE_X:
+ /* stepping > 4 is Cascade Lake and is not supported */
+ valid = (stepping <= 4);
+ break;
+
+ default:
+ valid = false;
+ break;
+ }
+
+ return valid;
+}
+
+static void rh_check_supported(void)
+{
+ bool guest;
+
+ guest = (x86_hyper_type != X86_HYPER_NATIVE || boot_cpu_has(X86_FEATURE_HYPERVISOR));
+
+ /* RHEL supports single cpu on guests only */
+ if (((boot_cpu_data.x86_max_cores * smp_num_siblings) == 1) &&
+ !guest && is_kdump_kernel()) {
+ pr_crit("Detected single cpu native boot.\n");
+ pr_crit("Important: In this kernel, single threaded, single CPU 64-bit physical systems are unsupported.");
+ }
+
+ /*
+ * If the RHEL kernel does not support this hardware, the kernel will
+ * attempt to boot, but no support is provided for this hardware
+ */
+ switch (boot_cpu_data.x86_vendor) {
+ case X86_VENDOR_AMD:
+ if (!valid_amd_processor(boot_cpu_data.x86,
+ boot_cpu_data.x86_model_id)) {
+ pr_crit("Detected CPU family %xh model %d\n",
+ boot_cpu_data.x86,
+ boot_cpu_data.x86_model);
+ mark_hardware_unsupported("AMD Processor");
+ }
+ break;
+
+ case X86_VENDOR_INTEL:
+ if (!valid_intel_processor(boot_cpu_data.x86,
+ boot_cpu_data.x86_model,
+ boot_cpu_data.x86_stepping)) {
+ pr_crit("Detected CPU family %d model %d stepping %d\n",
+ boot_cpu_data.x86,
+ boot_cpu_data.x86_model,
+ boot_cpu_data.x86_stepping);
+ mark_hardware_unsupported("Intel Processor");
+ }
+ break;
+
+ default:
+ pr_crit("Detected processor %s %s\n",
+ boot_cpu_data.x86_vendor_id,
+ boot_cpu_data.x86_model_id);
+ mark_hardware_unsupported("Processor");
+ break;
+ }
+
+ /*
+ * Due to the complexity of x86 lapic & ioapic enumeration, and PCI IRQ
+ * routing, ACPI is required for x86. acpi=off is a valid debug kernel
+ * parameter, so just print out a loud warning in case something
+ * goes wrong (which is most of the time).
+ */
+ if (acpi_disabled && !guest)
+ pr_crit("ACPI has been disabled or is not available on this hardware. This may result in a single cpu boot, incorrect PCI IRQ routing, or boot failure.\n");
+}
+
/*
* Dump out kernel offset information on panic.
*/
@@ -1250,6 +1376,10 @@ void __init setup_arch(char **cmdline_p)
efi_apply_memmap_quirks();
#endif
+#ifdef CONFIG_RH_DISABLE_DEPRECATED
+ rh_check_supported();
+#endif
+
unwind_init();
}
--
2.26.2