Skip to content

Commit

Permalink
x86/amd: Extend CPU erratum #1474 fix to more affected models
Browse files Browse the repository at this point in the history
Erratum #1474 has now been extended to cover models from family 17h ranges
00-2Fh, so the errata now covers all the models released under Family
17h (Zen, Zen+ and Zen2).

Additionally extend the workaround to Family 18h (Hygon), since it's based on
the Zen architecture and very likely affected.

Rename all the zen2 related symbols to fam17, since the errata doesn't
exclusively affect Zen2 anymore.

Reported-by: Andrew Cooper <[email protected]>
Signed-off-by: Roger Pau Monné <[email protected]>
Reviewed-by: Andrew Cooper <[email protected]>
  • Loading branch information
royger authored and andyhhp committed Dec 21, 2023
1 parent b74c735 commit 23db507
Showing 1 changed file with 14 additions and 13 deletions.
27 changes: 14 additions & 13 deletions xen/arch/x86/cpu/amd.c
Original file line number Diff line number Diff line change
Expand Up @@ -54,7 +54,7 @@ bool __read_mostly amd_acpi_c1e_quirk;
bool __ro_after_init amd_legacy_ssbd;
bool __initdata amd_virt_spec_ctrl;

static bool __read_mostly zen2_c6_disabled;
static bool __read_mostly fam17_c6_disabled;

static inline int rdmsr_amd_safe(unsigned int msr, unsigned int *lo,
unsigned int *hi)
Expand Down Expand Up @@ -978,24 +978,24 @@ void amd_check_zenbleed(void)
val & chickenbit ? "chickenbit" : "microcode");
}

static void cf_check zen2_disable_c6(void *arg)
static void cf_check fam17_disable_c6(void *arg)
{
/* Disable C6 by clearing the CCR{0,1,2}_CC6EN bits. */
const uint64_t mask = ~((1ul << 6) | (1ul << 14) | (1ul << 22));
uint64_t val;

if (!zen2_c6_disabled) {
if (!fam17_c6_disabled) {
printk(XENLOG_WARNING
"Disabling C6 after 1000 days apparent uptime due to AMD errata 1474\n");
zen2_c6_disabled = true;
fam17_c6_disabled = true;
/*
* Prevent CPU hotplug so that started CPUs will either see
* zen2_c6_disabled set, or will be handled by
* zen_c6_disabled set, or will be handled by
* smp_call_function().
*/
while (!get_cpu_maps())
process_pending_softirqs();
smp_call_function(zen2_disable_c6, NULL, 0);
smp_call_function(fam17_disable_c6, NULL, 0);
put_cpu_maps();
}

Expand Down Expand Up @@ -1294,8 +1294,8 @@ static void cf_check init_amd(struct cpuinfo_x86 *c)
amd_check_zenbleed();
amd_check_erratum_1485();

if (zen2_c6_disabled)
zen2_disable_c6(NULL);
if (fam17_c6_disabled)
fam17_disable_c6(NULL);

check_syscfg_dram_mod_en();

Expand All @@ -1307,15 +1307,16 @@ const struct cpu_dev amd_cpu_dev = {
.c_init = init_amd,
};

static int __init cf_check zen2_c6_errata_check(void)
static int __init cf_check amd_check_erratum_1474(void)
{
/*
* Errata #1474: A Core May Hang After About 1044 Days
* Set up a timer to disable C6 after 1000 days uptime.
*/
s_time_t delta;

if (cpu_has_hypervisor || boot_cpu_data.x86 != 0x17 || !is_zen2_uarch())
if (cpu_has_hypervisor ||
(boot_cpu_data.x86 != 0x17 && boot_cpu_data.x86 != 0x18))
return 0;

/*
Expand All @@ -1330,15 +1331,15 @@ static int __init cf_check zen2_c6_errata_check(void)
if (delta > 0) {
static struct timer errata_c6;

init_timer(&errata_c6, zen2_disable_c6, NULL, 0);
init_timer(&errata_c6, fam17_disable_c6, NULL, 0);
set_timer(&errata_c6, NOW() + delta);
} else
zen2_disable_c6(NULL);
fam17_disable_c6(NULL);

return 0;
}
/*
* Must be executed after early_time_init() for tsc_ticks2ns() to have been
* calibrated. That prevents us doing the check in init_amd().
*/
presmp_initcall(zen2_c6_errata_check);
presmp_initcall(amd_check_erratum_1474);

0 comments on commit 23db507

Please sign in to comment.