This is just a skeleton of how it might look. Several issues
arose while looking at this ... not all directly related to
the problem at hand.
Parts 1 & 2 are just cleanup. CEC should follow the same rules
as everyone else who wants to be on the mce notifier chain. No
real reason for it to have direct hooks into mce/core.c
Part 3 adds a field to struct mce, and defines the BIT fields
for each class of notifier. All EDAC drivers share the same BIT
since only one of them should be active.
Part 4 is where things are interesting and need a great deal more
thought. A bunch of things on the chain return NOTIFY_STOP which
prevents anything else on the chain from being run. For the moment
I ignored that semantic and added code everywhere to set the BIT
even though nobody else will see it. This is because I think at
least some of them should NOT be NOTIFY_STOP.
Part 5 is currently written to always call __print_mce() for
debugging. The "if (1 || ...)" obviously doesn't want the "1"
(though I'd like to add some /sys knob to flip a switch to force
printing for systems where something weird is happening and logs
are being lost).
Tony Luck (5):
x86/mce: Rename "first" function as "early"
x86/mce: Convert corrected error collector to use mce notifier
x86/mce: Add new "handled" field to "struct mce"
x86/mce: Fix all mce notifiers to update the mce->handled bitmask
x86/mce: Change default mce logger to check mce->handled
arch/x86/include/asm/mce.h | 15 ++++----
arch/x86/include/uapi/asm/mce.h | 9 +++++
arch/x86/kernel/cpu/mce/core.c | 53 +++++++---------------------
arch/x86/kernel/cpu/mce/dev-mcelog.c | 1 +
drivers/acpi/acpi_extlog.c | 1 +
drivers/acpi/nfit/mce.c | 1 +
drivers/edac/i7core_edac.c | 1 +
drivers/edac/mce_amd.c | 5 ++-
drivers/edac/pnd2_edac.c | 1 +
drivers/edac/sb_edac.c | 1 +
drivers/edac/skx_common.c | 1 +
drivers/ras/cec.c | 29 +++++++++++++++
12 files changed, 69 insertions(+), 49 deletions(-)
--
2.21.1
On Wed, Feb 12, 2020 at 12:46:47PM -0800, Tony Luck wrote:
> Part 4 is where things are interesting and need a great deal more
> thought. A bunch of things on the chain return NOTIFY_STOP which
> prevents anything else on the chain from being run. For the moment
> I ignored that semantic and added code everywhere to set the BIT
> even though nobody else will see it. This is because I think at
> least some of them should NOT be NOTIFY_STOP.
NOTIFY_STOP is just one mechanism for preventing every function
on the mce chain from reporting an error.
The other bit I'd like to reconsider is edac_get_report_status().
Back in the day we seemed to be paranoid about reporting the same
error more than once via all the different reporting mechanisms.
Since then I've had to track down numerous "Why didn't this error
get reported?" questions that frequently resolved to "It was reported,
but not in the place that you expected".
So now my attitude is "Let's just log it everywhere in so that
whatever log the user is checking, they'll find the error".
-Tony
On Wed, Feb 12, 2020 at 3:08 PM Luck, Tony <[email protected]> wrote:
>
> On Wed, Feb 12, 2020 at 12:46:47PM -0800, Tony Luck wrote:
> > Part 4 is where things are interesting and need a great deal more
> > thought. A bunch of things on the chain return NOTIFY_STOP which
> > prevents anything else on the chain from being run. For the moment
> > I ignored that semantic and added code everywhere to set the BIT
> > even though nobody else will see it. This is because I think at
> > least some of them should NOT be NOTIFY_STOP.
>
> NOTIFY_STOP is just one mechanism for preventing every function
> on the mce chain from reporting an error.
>
> The other bit I'd like to reconsider is edac_get_report_status().
> Back in the day we seemed to be paranoid about reporting the same
> error more than once via all the different reporting mechanisms.
>
> Since then I've had to track down numerous "Why didn't this error
> get reported?" questions that frequently resolved to "It was reported,
> but not in the place that you expected".
>
> So now my attitude is "Let's just log it everywhere in so that
> whatever log the user is checking, they'll find the error"
I HATE notifier chains for exceptions, and I REALLY HATE NOTIFY_STOP.
I don't suppose we could rig something up so that they are simply
notifiers (for MCE and, eventually, for everything) and just outright
prevent them from modifying the processing?
As an example that particularly bothers me, do_debug():
if (notify_die(DIE_DEBUG, "debug", regs, (long)&dr6, error_code,
SIGTRAP) == NOTIFY_STOP)
goto exit;
There is all kind of garbage hidden in there, and it's mostly
somewhere between slightly buggy and violently buggy. All this crap
should be open-coded.
On Wed, Feb 12, 2020 at 09:52:39PM -0800, Andy Lutomirski wrote:
> I HATE notifier chains for exceptions, and I REALLY HATE NOTIFY_STOP.
> I don't suppose we could rig something up so that they are simply
> notifiers (for MCE and, eventually, for everything) and just outright
> prevent them from modifying the processing?
As in: they all get executed unconditionally and there's no NOTIFY_STOP
and if they're not interested in the notification, they simply return
early?
Hohumm, sounds nicer.
--
Regards/Gruss,
Boris.
https://people.kernel.org/tglx/notes-about-netiquette
> On Feb 12, 2020, at 10:09 PM, Borislav Petkov <[email protected]> wrote:
>
> On Wed, Feb 12, 2020 at 09:52:39PM -0800, Andy Lutomirski wrote:
>> I HATE notifier chains for exceptions, and I REALLY HATE NOTIFY_STOP.
>> I don't suppose we could rig something up so that they are simply
>> notifiers (for MCE and, eventually, for everything) and just outright
>> prevent them from modifying the processing?
>
> As in: they all get executed unconditionally and there's no NOTIFY_STOP
> and if they're not interested in the notification, they simply return
> early?
>
> Hohumm, sounds nicer.
>
Exactly :)
> --
> Regards/Gruss,
> Boris.
>
> https://people.kernel.org/tglx/notes-about-netiquette
Parts 1 & 2 are just cleanup. CEC should follow the same rules
as everyone else who wants to be on the mce notifier chain. No
real reason for it to have direct hooks into mce/core.c
[No substantive change since RFC version 1, but note that
I have kept the change to make CEC a "normal" user of
the mce notifier chain. Result is a few checks for
if (mce->kflags & MCE_HANDLED_CEC) in EDAC etc. drivers.]
Part 3 adds a field to struct mce, and defines the BIT fields
for each class of notifier. All EDAC drivers share the same BIT
since only one of them should be active.
[Boris: Changed name of new field to "kflags" and made
it __u64, so plenty of space for possible future
other uses]
Part 4 Re-done since draft based on Luto and Tglx comments that
we should kill of all usage of NOTIFY_STOP. This patch
now gets rid of all but one. That's an AMD case where
it looks like they don't want to decode some particular
errors on a specific platform. The right fix for that
is to take Luto's advice and filter out before that item
gets to the notifier chain. We even already have a filter
function (filter_mce) to do that! But that change needs
to be handled by someone with the appropriate h/w.
Part 5 Now just checks for mce->kflags in the default handler at
the end of the chain to decide whether to print.
Part 6 NEW - add mce=print_all option to override default and
print everything to the console. Intended for debug, or
desperation scenarios where other logs are lost.
Part 7 NEW - Delete the code that tries to make sure only one
out of acpi_extlog and the current loaded EDAC driver
deals with an error.
Tony Luck (7):
x86/mce: Rename "first" function as "early"
x86/mce: Convert corrected error collector to use mce notifier
x86/mce: Add new "kflags" field to "struct mce"
x86/mce: Fix all mce notifiers to update the mce->kflags bitmask
x86/mce: Change default mce logger to check mce->kflags
x86/mce: Add mce=print_all option
x86/mce: Drop the EDAC report status checks
arch/x86/include/asm/mce.h | 15 +++----
arch/x86/include/uapi/asm/mce.h | 9 ++++
arch/x86/kernel/cpu/mce/core.c | 58 ++++++++------------------
arch/x86/kernel/cpu/mce/dev-mcelog.c | 5 +++
arch/x86/kernel/cpu/mce/internal.h | 1 +
drivers/acpi/acpi_extlog.c | 19 ++-------
drivers/acpi/nfit/mce.c | 1 +
drivers/edac/edac_mc.c | 61 ----------------------------
drivers/edac/i7core_edac.c | 5 ++-
drivers/edac/mce_amd.c | 9 +++-
drivers/edac/pnd2_edac.c | 8 ++--
drivers/edac/sb_edac.c | 7 ++--
drivers/edac/skx_common.c | 3 +-
drivers/ras/cec.c | 29 +++++++++++++
include/linux/edac.h | 8 ----
15 files changed, 91 insertions(+), 147 deletions(-)
base-commit: b19e8c68470385dd2c5440876591fddb02c8c402
--
2.21.1
The CEC code has its claws in a couple of routines in mce/core.c
Convert it to just register itself on the normal mce notifier
chain.
Signed-off-by: Tony Luck <[email protected]>
---
arch/x86/kernel/cpu/mce/core.c | 19 -------------------
drivers/ras/cec.c | 26 ++++++++++++++++++++++++++
2 files changed, 26 insertions(+), 19 deletions(-)
diff --git a/arch/x86/kernel/cpu/mce/core.c b/arch/x86/kernel/cpu/mce/core.c
index 3366807d8e58..06240cbe6f3e 100644
--- a/arch/x86/kernel/cpu/mce/core.c
+++ b/arch/x86/kernel/cpu/mce/core.c
@@ -542,21 +542,6 @@ bool mce_is_correctable(struct mce *m)
}
EXPORT_SYMBOL_GPL(mce_is_correctable);
-static bool cec_add_mce(struct mce *m)
-{
- if (!m)
- return false;
-
- /* We eat only correctable DRAM errors with usable addresses. */
- if (mce_is_memory_error(m) &&
- mce_is_correctable(m) &&
- mce_usable_address(m))
- if (!cec_add_elem(m->addr >> PAGE_SHIFT))
- return true;
-
- return false;
-}
-
static int mce_early_notifier(struct notifier_block *nb, unsigned long val,
void *data)
{
@@ -565,9 +550,6 @@ static int mce_early_notifier(struct notifier_block *nb, unsigned long val,
if (!m)
return NOTIFY_DONE;
- if (cec_add_mce(m))
- return NOTIFY_STOP;
-
/* Emit the trace record: */
trace_mce_record(m);
@@ -2588,7 +2570,6 @@ static int __init mcheck_late_init(void)
static_branch_inc(&mcsafe_key);
mcheck_debugfs_init();
- cec_init();
/*
* Flush out everything that has been logged during early boot, now that
diff --git a/drivers/ras/cec.c b/drivers/ras/cec.c
index c09cf55e2d20..d7f6718cbf8d 100644
--- a/drivers/ras/cec.c
+++ b/drivers/ras/cec.c
@@ -527,6 +527,29 @@ static int __init create_debugfs_nodes(void)
return 1;
}
+static int cec_notifier(struct notifier_block *nb, unsigned long val,
+ void *data)
+{
+ struct mce *m = (struct mce *)data;
+
+ if (!m)
+ return NOTIFY_DONE;
+
+ /* We eat only correctable DRAM errors with usable addresses. */
+ if (mce_is_memory_error(m) &&
+ mce_is_correctable(m) &&
+ mce_usable_address(m))
+ if (!cec_add_elem(m->addr >> PAGE_SHIFT))
+ return NOTIFY_STOP;
+
+ return NOTIFY_DONE;
+}
+
+static struct notifier_block cec_nb = {
+ .notifier_call = cec_notifier,
+ .priority = MCE_PRIO_CEC,
+};
+
void __init cec_init(void)
{
if (ce_arr.disabled)
@@ -546,8 +569,11 @@ void __init cec_init(void)
INIT_DELAYED_WORK(&cec_work, cec_work_fn);
schedule_delayed_work(&cec_work, CEC_DECAY_DEFAULT_INTERVAL);
+ mce_register_decode_chain(&cec_nb);
+
pr_info("Correctable Errors collector initialized.\n");
}
+late_initcall(cec_init);
int __init parse_cec_param(char *str)
{
--
2.21.1
From: Borislav Petkov <[email protected]>
Hi all,
here's what I have. I'd like to keep mce.kflags hidden for now.
The last patch is something tglx spotted yesterday and fixing that with
the MCE flags is pretty easy - was boxing with a wrapper struct around
struct mce and that gets really ugly.
Tony, I'm open to suggestions how to test it - I probably don't have an
access to such box which can trigger read errors on nvdimms or what was
the use case?
Thx.
Borislav Petkov (2):
x86/mce/amd, edac: Remove report_gart_errors
x86/mce: Fixup exception only for the correct MCEs
Tony Luck (7):
x86/mce: Rename "first" function as "early"
x86/mce: Convert the CEC to use the MCE notifier
x86/mce: Add a struct mce.kflags field
x86/mce: Fix all mce notifiers to update the mce->kflags bitmask
x86/mce: Change default MCE logger to check mce->kflags
x86/mce: Add mce=print_all option
EDAC: Drop the EDAC report status checks
arch/x86/include/asm/mce.h | 28 +++++++----
arch/x86/include/uapi/asm/mce.h | 1 +
arch/x86/kernel/cpu/mce/amd.c | 9 +++-
arch/x86/kernel/cpu/mce/core.c | 72 +++++++++++-----------------
arch/x86/kernel/cpu/mce/dev-mcelog.c | 5 ++
arch/x86/kernel/cpu/mce/internal.h | 1 +
arch/x86/kernel/cpu/mce/severity.c | 6 ++-
drivers/acpi/acpi_extlog.c | 19 ++------
drivers/acpi/nfit/mce.c | 1 +
drivers/edac/amd64_edac.c | 8 ----
drivers/edac/edac_mc.c | 61 -----------------------
drivers/edac/i7core_edac.c | 5 +-
drivers/edac/mce_amd.c | 28 ++---------
drivers/edac/mce_amd.h | 2 -
drivers/edac/pnd2_edac.c | 8 ++--
drivers/edac/sb_edac.c | 7 ++-
drivers/edac/skx_common.c | 3 +-
drivers/ras/cec.c | 33 ++++++++++++-
include/linux/edac.h | 8 ----
include/linux/ras.h | 5 --
20 files changed, 118 insertions(+), 192 deletions(-)
--
2.21.0
From: Tony Luck <[email protected]>
It isn't going to be first on the notifier chain when the CEC is moved
to be a normal user of the notifier chain.
Fix the enum for the MCE_PRIO symbols to list them in reverse order so
that the compiler can give them numbers from low to high priority. Add
an entry for MCE_PRIO_CEC as the highest priority.
[ bp: Use passive voice, add comments. ]
Signed-off-by: Tony Luck <[email protected]>
Signed-off-by: Borislav Petkov <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
---
arch/x86/include/asm/mce.h | 16 +++++++++-------
arch/x86/kernel/cpu/mce/core.c | 10 +++++-----
2 files changed, 14 insertions(+), 12 deletions(-)
diff --git a/arch/x86/include/asm/mce.h b/arch/x86/include/asm/mce.h
index 83b6ddafa032..689ac6e9c65f 100644
--- a/arch/x86/include/asm/mce.h
+++ b/arch/x86/include/asm/mce.h
@@ -144,14 +144,16 @@ struct mce_log_buffer {
struct mce entry[];
};
+/* Highest last */
enum mce_notifier_prios {
- MCE_PRIO_FIRST = INT_MAX,
- MCE_PRIO_UC = INT_MAX - 1,
- MCE_PRIO_EXTLOG = INT_MAX - 2,
- MCE_PRIO_NFIT = INT_MAX - 3,
- MCE_PRIO_EDAC = INT_MAX - 4,
- MCE_PRIO_MCELOG = 1,
- MCE_PRIO_LOWEST = 0,
+ MCE_PRIO_LOWEST,
+ MCE_PRIO_MCELOG,
+ MCE_PRIO_EDAC,
+ MCE_PRIO_NFIT,
+ MCE_PRIO_EXTLOG,
+ MCE_PRIO_UC,
+ MCE_PRIO_EARLY,
+ MCE_PRIO_CEC
};
struct notifier_block;
diff --git a/arch/x86/kernel/cpu/mce/core.c b/arch/x86/kernel/cpu/mce/core.c
index a6009efdfe2b..43b1519ad4e5 100644
--- a/arch/x86/kernel/cpu/mce/core.c
+++ b/arch/x86/kernel/cpu/mce/core.c
@@ -559,7 +559,7 @@ static bool cec_add_mce(struct mce *m)
return false;
}
-static int mce_first_notifier(struct notifier_block *nb, unsigned long val,
+static int mce_early_notifier(struct notifier_block *nb, unsigned long val,
void *data)
{
struct mce *m = (struct mce *)data;
@@ -580,9 +580,9 @@ static int mce_first_notifier(struct notifier_block *nb, unsigned long val,
return NOTIFY_DONE;
}
-static struct notifier_block first_nb = {
- .notifier_call = mce_first_notifier,
- .priority = MCE_PRIO_FIRST,
+static struct notifier_block early_nb = {
+ .notifier_call = mce_early_notifier,
+ .priority = MCE_PRIO_EARLY,
};
static int uc_decode_notifier(struct notifier_block *nb, unsigned long val,
@@ -2041,7 +2041,7 @@ __setup("mce", mcheck_enable);
int __init mcheck_init(void)
{
mcheck_intel_therm_init();
- mce_register_decode_chain(&first_nb);
+ mce_register_decode_chain(&early_nb);
mce_register_decode_chain(&mce_uc_nb);
mce_register_decode_chain(&mce_default_nb);
mcheck_vendor_init_severity();
--
2.21.0
From: Tony Luck <[email protected]>
Instead of keeping count of how many handlers are registered on the
MCE notifier chain and printing if below some magic value, look at
mce->kflags to see if anyone claims to have handled/logged this error.
[ bp: Do not print ->kflags in __print_mce(). ]
Signed-off-by: Tony Luck <[email protected]>
Signed-off-by: Borislav Petkov <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
---
arch/x86/kernel/cpu/mce/core.c | 19 +++----------------
1 file changed, 3 insertions(+), 16 deletions(-)
diff --git a/arch/x86/kernel/cpu/mce/core.c b/arch/x86/kernel/cpu/mce/core.c
index 5666a48a4bc9..fc879b6669d5 100644
--- a/arch/x86/kernel/cpu/mce/core.c
+++ b/arch/x86/kernel/cpu/mce/core.c
@@ -158,29 +158,17 @@ void mce_log(struct mce *m)
}
EXPORT_SYMBOL_GPL(mce_log);
-/*
- * We run the default notifier if we have only the UC, the first and the
- * default notifier registered. I.e., the mandatory NUM_DEFAULT_NOTIFIERS
- * notifiers registered on the chain.
- */
-#define NUM_DEFAULT_NOTIFIERS 3
-static atomic_t num_notifiers;
-
void mce_register_decode_chain(struct notifier_block *nb)
{
if (WARN_ON(nb->priority > MCE_PRIO_MCELOG && nb->priority < MCE_PRIO_EDAC))
return;
- atomic_inc(&num_notifiers);
-
blocking_notifier_chain_register(&x86_mce_decoder_chain, nb);
}
EXPORT_SYMBOL_GPL(mce_register_decode_chain);
void mce_unregister_decode_chain(struct notifier_block *nb)
{
- atomic_dec(&num_notifiers);
-
blocking_notifier_chain_unregister(&x86_mce_decoder_chain, nb);
}
EXPORT_SYMBOL_GPL(mce_unregister_decode_chain);
@@ -263,6 +251,7 @@ static void __print_mce(struct mce *m)
}
pr_cont("\n");
+
/*
* Note this output is parsed by external tools and old fields
* should not be changed.
@@ -602,10 +591,8 @@ static int mce_default_notifier(struct notifier_block *nb, unsigned long val,
if (!m)
return NOTIFY_DONE;
- if (atomic_read(&num_notifiers) > NUM_DEFAULT_NOTIFIERS)
- return NOTIFY_DONE;
-
- __print_mce(m);
+ if (!m->kflags)
+ __print_mce(m);
return NOTIFY_DONE;
}
--
2.21.0
From: Tony Luck <[email protected]>
If the handler took any action to log or deal with the error, set a bit
in mce->kflags so that the default handler on the end of the machine
check chain can see what has been done.
Get rid of NOTIFY_STOP returns. Make the EDAC and dev-mcelog handlers
skip over errors already processed by CEC.
Signed-off-by: Tony Luck <[email protected]>
Signed-off-by: Borislav Petkov <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
---
arch/x86/kernel/cpu/mce/core.c | 4 +++-
arch/x86/kernel/cpu/mce/dev-mcelog.c | 5 +++++
drivers/acpi/acpi_extlog.c | 5 +++--
drivers/acpi/nfit/mce.c | 1 +
drivers/edac/i7core_edac.c | 5 +++--
drivers/edac/mce_amd.c | 6 +++++-
drivers/edac/pnd2_edac.c | 5 +++--
drivers/edac/sb_edac.c | 5 ++++-
drivers/edac/skx_common.c | 4 ++++
drivers/ras/cec.c | 9 ++++++---
10 files changed, 37 insertions(+), 12 deletions(-)
diff --git a/arch/x86/kernel/cpu/mce/core.c b/arch/x86/kernel/cpu/mce/core.c
index b033b3589630..5666a48a4bc9 100644
--- a/arch/x86/kernel/cpu/mce/core.c
+++ b/arch/x86/kernel/cpu/mce/core.c
@@ -581,8 +581,10 @@ static int uc_decode_notifier(struct notifier_block *nb, unsigned long val,
return NOTIFY_DONE;
pfn = mce->addr >> PAGE_SHIFT;
- if (!memory_failure(pfn, 0))
+ if (!memory_failure(pfn, 0)) {
set_mce_nospec(pfn);
+ mce->kflags |= MCE_HANDLED_UC;
+ }
return NOTIFY_OK;
}
diff --git a/arch/x86/kernel/cpu/mce/dev-mcelog.c b/arch/x86/kernel/cpu/mce/dev-mcelog.c
index d089567a9ce8..c033e7ea9e3c 100644
--- a/arch/x86/kernel/cpu/mce/dev-mcelog.c
+++ b/arch/x86/kernel/cpu/mce/dev-mcelog.c
@@ -39,6 +39,9 @@ static int dev_mce_log(struct notifier_block *nb, unsigned long val,
struct mce *mce = (struct mce *)data;
unsigned int entry;
+ if (mce->kflags & MCE_HANDLED_CEC)
+ return NOTIFY_DONE;
+
mutex_lock(&mce_chrdev_read_mutex);
entry = mcelog->next;
@@ -56,6 +59,7 @@ static int dev_mce_log(struct notifier_block *nb, unsigned long val,
memcpy(mcelog->entry + entry, mce, sizeof(struct mce));
mcelog->entry[entry].finished = 1;
+ mcelog->entry[entry].kflags = 0;
/* wake processes polling /dev/mcelog */
wake_up_interruptible(&mce_chrdev_wait);
@@ -63,6 +67,7 @@ static int dev_mce_log(struct notifier_block *nb, unsigned long val,
unlock:
mutex_unlock(&mce_chrdev_read_mutex);
+ mce->kflags |= MCE_HANDLED_MCELOG;
return NOTIFY_OK;
}
diff --git a/drivers/acpi/acpi_extlog.c b/drivers/acpi/acpi_extlog.c
index 8596a106a933..9cc3c1f92db5 100644
--- a/drivers/acpi/acpi_extlog.c
+++ b/drivers/acpi/acpi_extlog.c
@@ -146,7 +146,7 @@ static int extlog_print(struct notifier_block *nb, unsigned long val,
static u32 err_seq;
estatus = extlog_elog_entry_check(cpu, bank);
- if (estatus == NULL)
+ if (estatus == NULL || (mce->kflags & MCE_HANDLED_CEC))
return NOTIFY_DONE;
memcpy(elog_buf, (void *)estatus, ELOG_ENTRY_LEN);
@@ -176,7 +176,8 @@ static int extlog_print(struct notifier_block *nb, unsigned long val,
}
out:
- return NOTIFY_STOP;
+ mce->kflags |= MCE_HANDLED_EXTLOG;
+ return NOTIFY_OK;
}
static bool __init extlog_get_l1addr(void)
diff --git a/drivers/acpi/nfit/mce.c b/drivers/acpi/nfit/mce.c
index f0ae48515b48..ee8d9973f60b 100644
--- a/drivers/acpi/nfit/mce.c
+++ b/drivers/acpi/nfit/mce.c
@@ -76,6 +76,7 @@ static int nfit_handle_mce(struct notifier_block *nb, unsigned long val,
*/
acpi_nfit_ars_rescan(acpi_desc, 0);
}
+ mce->kflags |= MCE_HANDLED_NFIT;
break;
}
diff --git a/drivers/edac/i7core_edac.c b/drivers/edac/i7core_edac.c
index b3135b208f9a..5860ca41185c 100644
--- a/drivers/edac/i7core_edac.c
+++ b/drivers/edac/i7core_edac.c
@@ -1815,7 +1815,7 @@ static int i7core_mce_check_error(struct notifier_block *nb, unsigned long val,
struct mem_ctl_info *mci;
i7_dev = get_i7core_dev(mce->socketid);
- if (!i7_dev)
+ if (!i7_dev || (mce->kflags & MCE_HANDLED_CEC))
return NOTIFY_DONE;
mci = i7_dev->mci;
@@ -1834,7 +1834,8 @@ static int i7core_mce_check_error(struct notifier_block *nb, unsigned long val,
i7core_check_error(mci, mce);
/* Advise mcelog that the errors were handled */
- return NOTIFY_STOP;
+ mce->kflags |= MCE_HANDLED_EDAC;
+ return NOTIFY_OK;
}
static struct notifier_block i7_mce_dec = {
diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index e58644d9c92b..2b5401db56ad 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -1046,6 +1046,9 @@ amd_decode_mce(struct notifier_block *nb, unsigned long val, void *data)
unsigned int fam = x86_family(m->cpuid);
int ecc;
+ if (m->kflags & MCE_HANDLED_CEC)
+ return NOTIFY_DONE;
+
pr_emerg(HW_ERR "%s\n", decode_error_status(m));
pr_emerg(HW_ERR "CPU:%d (%x:%x:%x) MC%d_STATUS[%s|%s|%s|%s|%s",
@@ -1146,7 +1149,8 @@ amd_decode_mce(struct notifier_block *nb, unsigned long val, void *data)
err_code:
amd_decode_err_code(m->status & 0xffff);
- return NOTIFY_STOP;
+ m->kflags |= MCE_HANDLED_EDAC;
+ return NOTIFY_OK;
}
static struct notifier_block amd_mce_dec_nb = {
diff --git a/drivers/edac/pnd2_edac.c b/drivers/edac/pnd2_edac.c
index bc47328eb485..1929a5dc8f94 100644
--- a/drivers/edac/pnd2_edac.c
+++ b/drivers/edac/pnd2_edac.c
@@ -1400,7 +1400,7 @@ static int pnd2_mce_check_error(struct notifier_block *nb, unsigned long val, vo
return NOTIFY_DONE;
mci = pnd2_mci;
- if (!mci)
+ if (!mci || (mce->kflags & MCE_HANDLED_CEC))
return NOTIFY_DONE;
/*
@@ -1429,7 +1429,8 @@ static int pnd2_mce_check_error(struct notifier_block *nb, unsigned long val, vo
pnd2_mce_output_error(mci, mce, &daddr);
/* Advice mcelog that the error were handled */
- return NOTIFY_STOP;
+ mce->kflags |= MCE_HANDLED_EDAC;
+ return NOTIFY_OK;
}
static struct notifier_block pnd2_mce_dec = {
diff --git a/drivers/edac/sb_edac.c b/drivers/edac/sb_edac.c
index 7d51c82be62b..f790f7d08688 100644
--- a/drivers/edac/sb_edac.c
+++ b/drivers/edac/sb_edac.c
@@ -3136,6 +3136,8 @@ static int sbridge_mce_check_error(struct notifier_block *nb, unsigned long val,
if (edac_get_report_status() == EDAC_REPORTING_DISABLED)
return NOTIFY_DONE;
+ if (mce->kflags & MCE_HANDLED_CEC)
+ return NOTIFY_DONE;
/*
* Just let mcelog handle it if the error is
@@ -3183,7 +3185,8 @@ static int sbridge_mce_check_error(struct notifier_block *nb, unsigned long val,
sbridge_mce_output_error(mci, mce);
/* Advice mcelog that the error were handled */
- return NOTIFY_STOP;
+ mce->kflags |= MCE_HANDLED_EDAC;
+ return NOTIFY_OK;
}
static struct notifier_block sbridge_mce_dec = {
diff --git a/drivers/edac/skx_common.c b/drivers/edac/skx_common.c
index 99bbaf629b8d..6f08a12f6b11 100644
--- a/drivers/edac/skx_common.c
+++ b/drivers/edac/skx_common.c
@@ -577,6 +577,9 @@ int skx_mce_check_error(struct notifier_block *nb, unsigned long val,
if (edac_get_report_status() == EDAC_REPORTING_DISABLED)
return NOTIFY_DONE;
+ if (mce->kflags & MCE_HANDLED_CEC)
+ return NOTIFY_DONE;
+
/* ignore unless this is memory related with an address */
if ((mce->status & 0xefff) >> 7 != 1 || !(mce->status & MCI_STATUS_ADDRV))
return NOTIFY_DONE;
@@ -616,6 +619,7 @@ int skx_mce_check_error(struct notifier_block *nb, unsigned long val,
skx_mce_output_error(mci, mce, &res);
+ mce->kflags |= MCE_HANDLED_EDAC;
return NOTIFY_DONE;
}
diff --git a/drivers/ras/cec.c b/drivers/ras/cec.c
index 6b42040bf956..569d9ad2c594 100644
--- a/drivers/ras/cec.c
+++ b/drivers/ras/cec.c
@@ -538,9 +538,12 @@ static int cec_notifier(struct notifier_block *nb, unsigned long val,
/* We eat only correctable DRAM errors with usable addresses. */
if (mce_is_memory_error(m) &&
mce_is_correctable(m) &&
- mce_usable_address(m))
- if (!cec_add_elem(m->addr >> PAGE_SHIFT))
- return NOTIFY_STOP;
+ mce_usable_address(m)) {
+ if (!cec_add_elem(m->addr >> PAGE_SHIFT)) {
+ m->kflags |= MCE_HANDLED_CEC;
+ return NOTIFY_OK;
+ }
+ }
return NOTIFY_DONE;
}
--
2.21.0
From: Borislav Petkov <[email protected]>
The severity grading code returns IN_KERNEL_RECOV error context for
errors which have happened in kernel space but from which the kernel can
recover. Whether the recovery can happen is determined by the exception
table entry having as handler ex_handler_fault() and which has been
declared at build time using _ASM_EXTABLE_FAULT().
IN_KERNEL_RECOV is used in mce_severity_intel() to lookup the
corresponding error severity in the severities table.
However, the mapping back from error severity to whether the error is
IN_KERNEL_RECOV is ambiguous and in the very paranoid case - which
might not be possible right now - but be better safe than sorry later,
an exception fixup could be attempted for another MCE whose address
is in the exception table and has the proper severity. Which would be
unfortunate, to say the least.
Therefore, mark such MCEs explicitly as MCE_IN_KERNEL_RECOV so that the
recovery attempt is done only for them.
Document the whole handling, while at it, as it is not trivial.
Reported-by: Thomas Gleixner <[email protected]>
Signed-off-by: Borislav Petkov <[email protected]>
---
arch/x86/include/asm/mce.h | 1 +
arch/x86/kernel/cpu/mce/core.c | 15 +++++++++++++--
arch/x86/kernel/cpu/mce/severity.c | 6 +++++-
3 files changed, 19 insertions(+), 3 deletions(-)
diff --git a/arch/x86/include/asm/mce.h b/arch/x86/include/asm/mce.h
index 5f04a24f30ea..c598aaab071b 100644
--- a/arch/x86/include/asm/mce.h
+++ b/arch/x86/include/asm/mce.h
@@ -136,6 +136,7 @@
#define MCE_HANDLED_NFIT BIT_ULL(3)
#define MCE_HANDLED_EDAC BIT_ULL(4)
#define MCE_HANDLED_MCELOG BIT_ULL(5)
+#define MCE_IN_KERNEL_RECOV BIT_ULL(6)
/*
* This structure contains all data related to the MCE log. Also
diff --git a/arch/x86/kernel/cpu/mce/core.c b/arch/x86/kernel/cpu/mce/core.c
index 4efe6c128887..02e1f165f148 100644
--- a/arch/x86/kernel/cpu/mce/core.c
+++ b/arch/x86/kernel/cpu/mce/core.c
@@ -1331,8 +1331,19 @@ void notrace do_machine_check(struct pt_regs *regs, long error_code)
local_irq_disable();
ist_end_non_atomic();
} else {
- if (!fixup_exception(regs, X86_TRAP_MC, error_code, 0))
- mce_panic("Failed kernel mode recovery", &m, msg);
+ /*
+ * Handle an MCE which has happened in kernel space but from
+ * which the kernel can recover: ex_has_fault_handler() has
+ * already verified that the rIP at which the error happened is
+ * a rIP from which the kernel can recover (by jumping to
+ * recovery code specified in _ASM_EXTABLE_FAULT()) and the
+ * corresponding exception handler which would do that is the
+ * proper one.
+ */
+ if (m.kflags & MCE_IN_KERNEL_RECOV) {
+ if (!fixup_exception(regs, X86_TRAP_MC, error_code, 0))
+ mce_panic("Failed kernel mode recovery", &m, msg);
+ }
}
out_ist:
diff --git a/arch/x86/kernel/cpu/mce/severity.c b/arch/x86/kernel/cpu/mce/severity.c
index 87bcdc6dc2f0..e1da619add19 100644
--- a/arch/x86/kernel/cpu/mce/severity.c
+++ b/arch/x86/kernel/cpu/mce/severity.c
@@ -213,8 +213,12 @@ static int error_context(struct mce *m)
{
if ((m->cs & 3) == 3)
return IN_USER;
- if (mc_recoverable(m->mcgstatus) && ex_has_fault_handler(m->ip))
+
+ if (mc_recoverable(m->mcgstatus) && ex_has_fault_handler(m->ip)) {
+ m->kflags |= MCE_IN_KERNEL_RECOV;
return IN_KERNEL_RECOV;
+ }
+
return IN_KERNEL;
}
--
2.21.0
From: Tony Luck <[email protected]>
Sometimes, when logs are getting lost, it's nice to just
have everything dumped to the serial console.
Signed-off-by: Tony Luck <[email protected]>
Signed-off-by: Borislav Petkov <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
---
arch/x86/kernel/cpu/mce/core.c | 7 ++++++-
arch/x86/kernel/cpu/mce/internal.h | 1 +
2 files changed, 7 insertions(+), 1 deletion(-)
diff --git a/arch/x86/kernel/cpu/mce/core.c b/arch/x86/kernel/cpu/mce/core.c
index fc879b6669d5..4efe6c128887 100644
--- a/arch/x86/kernel/cpu/mce/core.c
+++ b/arch/x86/kernel/cpu/mce/core.c
@@ -591,7 +591,7 @@ static int mce_default_notifier(struct notifier_block *nb, unsigned long val,
if (!m)
return NOTIFY_DONE;
- if (!m->kflags)
+ if (mca_cfg.print_all || !m->kflags)
__print_mce(m);
return NOTIFY_DONE;
@@ -1962,6 +1962,7 @@ void mce_disable_bank(int bank)
* mce=no_cmci Disables CMCI
* mce=no_lmce Disables LMCE
* mce=dont_log_ce Clears corrected events silently, no log created for CEs.
+ * mce=print_all Print all machine check logs to console
* mce=ignore_ce Disables polling and CMCI, corrected events are not cleared.
* mce=TOLERANCELEVEL[,monarchtimeout] (number, see above)
* monarchtimeout is how long to wait for other CPUs on machine
@@ -1990,6 +1991,8 @@ static int __init mcheck_enable(char *str)
cfg->lmce_disabled = 1;
else if (!strcmp(str, "dont_log_ce"))
cfg->dont_log_ce = true;
+ else if (!strcmp(str, "print_all"))
+ cfg->print_all = true;
else if (!strcmp(str, "ignore_ce"))
cfg->ignore_ce = true;
else if (!strcmp(str, "bootlog") || !strcmp(str, "nobootlog"))
@@ -2256,6 +2259,7 @@ static ssize_t store_int_with_restart(struct device *s,
static DEVICE_INT_ATTR(tolerant, 0644, mca_cfg.tolerant);
static DEVICE_INT_ATTR(monarch_timeout, 0644, mca_cfg.monarch_timeout);
static DEVICE_BOOL_ATTR(dont_log_ce, 0644, mca_cfg.dont_log_ce);
+static DEVICE_BOOL_ATTR(print_all, 0644, mca_cfg.print_all);
static struct dev_ext_attribute dev_attr_check_interval = {
__ATTR(check_interval, 0644, device_show_int, store_int_with_restart),
@@ -2280,6 +2284,7 @@ static struct device_attribute *mce_device_attrs[] = {
#endif
&dev_attr_monarch_timeout.attr,
&dev_attr_dont_log_ce.attr,
+ &dev_attr_print_all.attr,
&dev_attr_ignore_ce.attr,
&dev_attr_cmci_disabled.attr,
NULL
diff --git a/arch/x86/kernel/cpu/mce/internal.h b/arch/x86/kernel/cpu/mce/internal.h
index 74a01829c4f4..55f5c7b755f2 100644
--- a/arch/x86/kernel/cpu/mce/internal.h
+++ b/arch/x86/kernel/cpu/mce/internal.h
@@ -119,6 +119,7 @@ struct mca_config {
bool dont_log_ce;
bool cmci_disabled;
bool ignore_ce;
+ bool print_all;
__u64 lmce_disabled : 1,
disabled : 1,
--
2.21.0
From: Tony Luck <[email protected]>
The CEC code has its claws in a couple of routines in mce/core.c.
Convert it to just register itself on the normal MCE notifier chain.
[ bp: Make cec_add_elem() and cec_init() static. ]
Signed-off-by: Tony Luck <[email protected]>
Signed-off-by: Borislav Petkov <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
---
arch/x86/kernel/cpu/mce/core.c | 19 -------------------
drivers/ras/cec.c | 30 ++++++++++++++++++++++++++++--
include/linux/ras.h | 5 -----
3 files changed, 28 insertions(+), 26 deletions(-)
diff --git a/arch/x86/kernel/cpu/mce/core.c b/arch/x86/kernel/cpu/mce/core.c
index 43b1519ad4e5..b033b3589630 100644
--- a/arch/x86/kernel/cpu/mce/core.c
+++ b/arch/x86/kernel/cpu/mce/core.c
@@ -544,21 +544,6 @@ bool mce_is_correctable(struct mce *m)
}
EXPORT_SYMBOL_GPL(mce_is_correctable);
-static bool cec_add_mce(struct mce *m)
-{
- if (!m)
- return false;
-
- /* We eat only correctable DRAM errors with usable addresses. */
- if (mce_is_memory_error(m) &&
- mce_is_correctable(m) &&
- mce_usable_address(m))
- if (!cec_add_elem(m->addr >> PAGE_SHIFT))
- return true;
-
- return false;
-}
-
static int mce_early_notifier(struct notifier_block *nb, unsigned long val,
void *data)
{
@@ -567,9 +552,6 @@ static int mce_early_notifier(struct notifier_block *nb, unsigned long val,
if (!m)
return NOTIFY_DONE;
- if (cec_add_mce(m))
- return NOTIFY_STOP;
-
/* Emit the trace record: */
trace_mce_record(m);
@@ -2612,7 +2594,6 @@ static int __init mcheck_late_init(void)
static_branch_inc(&mcsafe_key);
mcheck_debugfs_init();
- cec_init();
/*
* Flush out everything that has been logged during early boot, now that
diff --git a/drivers/ras/cec.c b/drivers/ras/cec.c
index c09cf55e2d20..6b42040bf956 100644
--- a/drivers/ras/cec.c
+++ b/drivers/ras/cec.c
@@ -309,7 +309,7 @@ static bool sanity_check(struct ce_array *ca)
return ret;
}
-int cec_add_elem(u64 pfn)
+static int cec_add_elem(u64 pfn)
{
struct ce_array *ca = &ce_arr;
unsigned int to = 0;
@@ -527,7 +527,30 @@ static int __init create_debugfs_nodes(void)
return 1;
}
-void __init cec_init(void)
+static int cec_notifier(struct notifier_block *nb, unsigned long val,
+ void *data)
+{
+ struct mce *m = (struct mce *)data;
+
+ if (!m)
+ return NOTIFY_DONE;
+
+ /* We eat only correctable DRAM errors with usable addresses. */
+ if (mce_is_memory_error(m) &&
+ mce_is_correctable(m) &&
+ mce_usable_address(m))
+ if (!cec_add_elem(m->addr >> PAGE_SHIFT))
+ return NOTIFY_STOP;
+
+ return NOTIFY_DONE;
+}
+
+static struct notifier_block cec_nb = {
+ .notifier_call = cec_notifier,
+ .priority = MCE_PRIO_CEC,
+};
+
+static void __init cec_init(void)
{
if (ce_arr.disabled)
return;
@@ -546,8 +569,11 @@ void __init cec_init(void)
INIT_DELAYED_WORK(&cec_work, cec_work_fn);
schedule_delayed_work(&cec_work, CEC_DECAY_DEFAULT_INTERVAL);
+ mce_register_decode_chain(&cec_nb);
+
pr_info("Correctable Errors collector initialized.\n");
}
+late_initcall(cec_init);
int __init parse_cec_param(char *str)
{
diff --git a/include/linux/ras.h b/include/linux/ras.h
index 7c3debb47c87..1f4048bf2674 100644
--- a/include/linux/ras.h
+++ b/include/linux/ras.h
@@ -17,12 +17,7 @@ static inline int ras_add_daemon_trace(void) { return 0; }
#endif
#ifdef CONFIG_RAS_CEC
-void __init cec_init(void);
int __init parse_cec_param(char *str);
-int cec_add_elem(u64 pfn);
-#else
-static inline void __init cec_init(void) { }
-static inline int cec_add_elem(u64 pfn) { return -ENODEV; }
#endif
#ifdef CONFIG_RAS
--
2.21.0
From: Borislav Petkov <[email protected]>
... because no one should be interested in spurious MCEs anyway. Make
the filtering unconditional and move it to amd_filter_mce().
Signed-off-by: Borislav Petkov <[email protected]>
---
arch/x86/include/asm/mce.h | 3 ++-
arch/x86/kernel/cpu/mce/amd.c | 9 +++++++--
drivers/edac/amd64_edac.c | 8 --------
drivers/edac/mce_amd.c | 24 ------------------------
drivers/edac/mce_amd.h | 2 --
5 files changed, 9 insertions(+), 37 deletions(-)
diff --git a/arch/x86/include/asm/mce.h b/arch/x86/include/asm/mce.h
index f9cea081c05b..83b6ddafa032 100644
--- a/arch/x86/include/asm/mce.h
+++ b/arch/x86/include/asm/mce.h
@@ -127,6 +127,8 @@
#define MSR_AMD64_SMCA_MCx_DEADDR(x) (MSR_AMD64_SMCA_MC0_DEADDR + 0x10*(x))
#define MSR_AMD64_SMCA_MCx_MISCy(x, y) ((MSR_AMD64_SMCA_MC0_MISC1 + y) + (0x10*(x)))
+#define XEC(x, mask) (((x) >> 16) & mask)
+
/*
* This structure contains all data related to the MCE log. Also
* carries a signature to make it easier to find from external
@@ -347,5 +349,4 @@ umc_normaddr_to_sysaddr(u64 norm_addr, u16 nid, u8 umc, u64 *sys_addr) { return
#endif
static inline void mce_hygon_feature_init(struct cpuinfo_x86 *c) { return mce_amd_feature_init(c); }
-
#endif /* _ASM_X86_MCE_H */
diff --git a/arch/x86/kernel/cpu/mce/amd.c b/arch/x86/kernel/cpu/mce/amd.c
index 15c87b87b901..ea3cf714b7ad 100644
--- a/arch/x86/kernel/cpu/mce/amd.c
+++ b/arch/x86/kernel/cpu/mce/amd.c
@@ -577,14 +577,19 @@ bool amd_filter_mce(struct mce *m)
{
enum smca_bank_types bank_type = smca_get_bank_type(m->bank);
struct cpuinfo_x86 *c = &boot_cpu_data;
- u8 xec = (m->status >> 16) & 0x3F;
/* See Family 17h Models 10h-2Fh Erratum #1114. */
if (c->x86 == 0x17 &&
c->x86_model >= 0x10 && c->x86_model <= 0x2F &&
- bank_type == SMCA_IF && xec == 10)
+ bank_type == SMCA_IF && XEC(m->status, 0x3f) == 10)
return true;
+ /* NB GART TLB error reporting is disabled by default. */
+ if (c->x86 < 0x17) {
+ if (m->bank == 4 && XEC(m->status, 0x1f) == 0x5)
+ return true;
+ }
+
return false;
}
diff --git a/drivers/edac/amd64_edac.c b/drivers/edac/amd64_edac.c
index f91f3bc1e0b2..6bdc5bb8c8bc 100644
--- a/drivers/edac/amd64_edac.c
+++ b/drivers/edac/amd64_edac.c
@@ -4,9 +4,6 @@
static struct edac_pci_ctl_info *pci_ctl;
-static int report_gart_errors;
-module_param(report_gart_errors, int, 0644);
-
/*
* Set by command line parameter. If BIOS has enabled the ECC, this override is
* cleared to prevent re-enabling the hardware by this driver.
@@ -3681,9 +3678,6 @@ static int __init amd64_edac_init(void)
}
/* register stuff with EDAC MCE */
- if (report_gart_errors)
- amd_report_gart_errors(true);
-
if (boot_cpu_data.x86 >= 0x17)
amd_register_ecc_decoder(decode_umc_error);
else
@@ -3718,8 +3712,6 @@ static void __exit amd64_edac_exit(void)
edac_pci_release_generic_ctl(pci_ctl);
/* unregister from EDAC MCE */
- amd_report_gart_errors(false);
-
if (boot_cpu_data.x86 >= 0x17)
amd_unregister_ecc_decoder(decode_umc_error);
else
diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index 8874b7722b2f..e58644d9c92b 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -10,15 +10,8 @@ static struct amd_decoder_ops fam_ops;
static u8 xec_mask = 0xf;
-static bool report_gart_errors;
static void (*decode_dram_ecc)(int node_id, struct mce *m);
-void amd_report_gart_errors(bool v)
-{
- report_gart_errors = v;
-}
-EXPORT_SYMBOL_GPL(amd_report_gart_errors);
-
void amd_register_ecc_decoder(void (*f)(int, struct mce *))
{
decode_dram_ecc = f;
@@ -1030,20 +1023,6 @@ static inline void amd_decode_err_code(u16 ec)
pr_cont("\n");
}
-/*
- * Filter out unwanted MCE signatures here.
- */
-static bool ignore_mce(struct mce *m)
-{
- /*
- * NB GART TLB error reporting is disabled by default.
- */
- if (m->bank == 4 && XEC(m->status, 0x1f) == 0x5 && !report_gart_errors)
- return true;
-
- return false;
-}
-
static const char *decode_error_status(struct mce *m)
{
if (m->status & MCI_STATUS_UC) {
@@ -1067,9 +1046,6 @@ amd_decode_mce(struct notifier_block *nb, unsigned long val, void *data)
unsigned int fam = x86_family(m->cpuid);
int ecc;
- if (ignore_mce(m))
- return NOTIFY_STOP;
-
pr_emerg(HW_ERR "%s\n", decode_error_status(m));
pr_emerg(HW_ERR "CPU:%d (%x:%x:%x) MC%d_STATUS[%s|%s|%s|%s|%s",
diff --git a/drivers/edac/mce_amd.h b/drivers/edac/mce_amd.h
index 4e9c5e596c6c..4811b18d9606 100644
--- a/drivers/edac/mce_amd.h
+++ b/drivers/edac/mce_amd.h
@@ -7,7 +7,6 @@
#include <asm/mce.h>
#define EC(x) ((x) & 0xffff)
-#define XEC(x, mask) (((x) >> 16) & mask)
#define LOW_SYNDROME(x) (((x) >> 15) & 0xff)
#define HIGH_SYNDROME(x) (((x) >> 24) & 0xff)
@@ -77,7 +76,6 @@ struct amd_decoder_ops {
bool (*mc2_mce)(u16, u8);
};
-void amd_report_gart_errors(bool);
void amd_register_ecc_decoder(void (*f)(int, struct mce *));
void amd_unregister_ecc_decoder(void (*f)(int, struct mce *));
--
2.21.0
From: Tony Luck <[email protected]>
There can be many different subsystems register on the mce handler
chain. Add a new bitmask field and define values so that handlers can
indicate whether they took any action to log or otherwise handle an
error.
The default handler at the end of the chain can use this information to
decide whether to print to the console log.
Boris suggested a generic name and leaving plenty of spare bits for
possible future use.
[ bp: Move flag bits to the internal mce.h header and use BIT_ULL(). ]
Signed-off-by: Tony Luck <[email protected]>
Signed-off-by: Borislav Petkov <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
---
arch/x86/include/asm/mce.h | 8 ++++++++
arch/x86/include/uapi/asm/mce.h | 1 +
2 files changed, 9 insertions(+)
diff --git a/arch/x86/include/asm/mce.h b/arch/x86/include/asm/mce.h
index 689ac6e9c65f..5f04a24f30ea 100644
--- a/arch/x86/include/asm/mce.h
+++ b/arch/x86/include/asm/mce.h
@@ -129,6 +129,14 @@
#define XEC(x, mask) (((x) >> 16) & mask)
+/* mce.kflags flag bits for logging etc. */
+#define MCE_HANDLED_CEC BIT_ULL(0)
+#define MCE_HANDLED_UC BIT_ULL(1)
+#define MCE_HANDLED_EXTLOG BIT_ULL(2)
+#define MCE_HANDLED_NFIT BIT_ULL(3)
+#define MCE_HANDLED_EDAC BIT_ULL(4)
+#define MCE_HANDLED_MCELOG BIT_ULL(5)
+
/*
* This structure contains all data related to the MCE log. Also
* carries a signature to make it easier to find from external
diff --git a/arch/x86/include/uapi/asm/mce.h b/arch/x86/include/uapi/asm/mce.h
index 955c2a2e1cf9..5b59d80f1d4e 100644
--- a/arch/x86/include/uapi/asm/mce.h
+++ b/arch/x86/include/uapi/asm/mce.h
@@ -35,6 +35,7 @@ struct mce {
__u64 ipid; /* MCA_IPID MSR: only valid on SMCA systems */
__u64 ppin; /* Protected Processor Inventory Number */
__u32 microcode; /* Microcode revision */
+ __u64 kflags; /* Internal kernel use. See below */
};
#define MCE_GET_RECORD_LEN _IOR('M', 1, int)
--
2.21.0
From: Tony Luck <[email protected]>
When acpi_extlog was added, we were worried that the same error would
be reported more than once by different subsystems. But in the ensuing
years I've seen complaints that people could not find an error log
(because this mechanism suppressed the log they were looking for).
Rip it all out. People are smart enough to notice the same address from
different reporting mechanisms.
Signed-off-by: Tony Luck <[email protected]>
Signed-off-by: Borislav Petkov <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
---
drivers/acpi/acpi_extlog.c | 14 ---------
drivers/edac/edac_mc.c | 61 --------------------------------------
drivers/edac/pnd2_edac.c | 3 --
drivers/edac/sb_edac.c | 4 ---
drivers/edac/skx_common.c | 3 --
include/linux/edac.h | 8 -----
6 files changed, 93 deletions(-)
diff --git a/drivers/acpi/acpi_extlog.c b/drivers/acpi/acpi_extlog.c
index 9cc3c1f92db5..f138e12b7b82 100644
--- a/drivers/acpi/acpi_extlog.c
+++ b/drivers/acpi/acpi_extlog.c
@@ -42,8 +42,6 @@ struct extlog_l1_head {
u8 rev1[12];
};
-static int old_edac_report_status;
-
static u8 extlog_dsm_uuid[] __initdata = "663E35AF-CC10-41A4-88EA-5470AF055295";
/* L1 table related physical address */
@@ -229,11 +227,6 @@ static int __init extlog_init(void)
if (!(cap & MCG_ELOG_P) || !extlog_get_l1addr())
return -ENODEV;
- if (edac_get_report_status() == EDAC_REPORTING_FORCE) {
- pr_warn("Not loading eMCA, error reporting force-enabled through EDAC.\n");
- return -EPERM;
- }
-
rc = -EINVAL;
/* get L1 header to fetch necessary information */
l1_hdr_size = sizeof(struct extlog_l1_head);
@@ -281,12 +274,6 @@ static int __init extlog_init(void)
if (elog_buf == NULL)
goto err_release_elog;
- /*
- * eMCA event report method has higher priority than EDAC method,
- * unless EDAC event report method is mandatory.
- */
- old_edac_report_status = edac_get_report_status();
- edac_set_report_status(EDAC_REPORTING_DISABLED);
mce_register_decode_chain(&extlog_mce_dec);
/* enable OS to be involved to take over management from BIOS */
((struct extlog_l1_head *)extlog_l1_addr)->flags |= FLAG_OS_OPTIN;
@@ -308,7 +295,6 @@ static int __init extlog_init(void)
static void __exit extlog_exit(void)
{
- edac_set_report_status(old_edac_report_status);
mce_unregister_decode_chain(&extlog_mce_dec);
((struct extlog_l1_head *)extlog_l1_addr)->flags &= ~FLAG_OS_OPTIN;
if (extlog_l1_addr)
diff --git a/drivers/edac/edac_mc.c b/drivers/edac/edac_mc.c
index 75ede27bdf6a..5813e931f2f0 100644
--- a/drivers/edac/edac_mc.c
+++ b/drivers/edac/edac_mc.c
@@ -43,8 +43,6 @@
int edac_op_state = EDAC_OPSTATE_INVAL;
EXPORT_SYMBOL_GPL(edac_op_state);
-static int edac_report = EDAC_REPORTING_ENABLED;
-
/* lock to memory controller's control array */
static DEFINE_MUTEX(mem_ctls_mutex);
static LIST_HEAD(mc_devices);
@@ -60,65 +58,6 @@ static struct mem_ctl_info *error_desc_to_mci(struct edac_raw_error_desc *e)
return container_of(e, struct mem_ctl_info, error_desc);
}
-int edac_get_report_status(void)
-{
- return edac_report;
-}
-EXPORT_SYMBOL_GPL(edac_get_report_status);
-
-void edac_set_report_status(int new)
-{
- if (new == EDAC_REPORTING_ENABLED ||
- new == EDAC_REPORTING_DISABLED ||
- new == EDAC_REPORTING_FORCE)
- edac_report = new;
-}
-EXPORT_SYMBOL_GPL(edac_set_report_status);
-
-static int edac_report_set(const char *str, const struct kernel_param *kp)
-{
- if (!str)
- return -EINVAL;
-
- if (!strncmp(str, "on", 2))
- edac_report = EDAC_REPORTING_ENABLED;
- else if (!strncmp(str, "off", 3))
- edac_report = EDAC_REPORTING_DISABLED;
- else if (!strncmp(str, "force", 5))
- edac_report = EDAC_REPORTING_FORCE;
-
- return 0;
-}
-
-static int edac_report_get(char *buffer, const struct kernel_param *kp)
-{
- int ret = 0;
-
- switch (edac_report) {
- case EDAC_REPORTING_ENABLED:
- ret = sprintf(buffer, "on");
- break;
- case EDAC_REPORTING_DISABLED:
- ret = sprintf(buffer, "off");
- break;
- case EDAC_REPORTING_FORCE:
- ret = sprintf(buffer, "force");
- break;
- default:
- ret = -EINVAL;
- break;
- }
-
- return ret;
-}
-
-static const struct kernel_param_ops edac_report_ops = {
- .set = edac_report_set,
- .get = edac_report_get,
-};
-
-module_param_cb(edac_report, &edac_report_ops, &edac_report, 0644);
-
unsigned int edac_dimm_info_location(struct dimm_info *dimm, char *buf,
unsigned int len)
{
diff --git a/drivers/edac/pnd2_edac.c b/drivers/edac/pnd2_edac.c
index 1929a5dc8f94..c1f2e6deb021 100644
--- a/drivers/edac/pnd2_edac.c
+++ b/drivers/edac/pnd2_edac.c
@@ -1396,9 +1396,6 @@ static int pnd2_mce_check_error(struct notifier_block *nb, unsigned long val, vo
struct dram_addr daddr;
char *type;
- if (edac_get_report_status() == EDAC_REPORTING_DISABLED)
- return NOTIFY_DONE;
-
mci = pnd2_mci;
if (!mci || (mce->kflags & MCE_HANDLED_CEC))
return NOTIFY_DONE;
diff --git a/drivers/edac/sb_edac.c b/drivers/edac/sb_edac.c
index f790f7d08688..d414698ca324 100644
--- a/drivers/edac/sb_edac.c
+++ b/drivers/edac/sb_edac.c
@@ -3134,8 +3134,6 @@ static int sbridge_mce_check_error(struct notifier_block *nb, unsigned long val,
struct mem_ctl_info *mci;
char *type;
- if (edac_get_report_status() == EDAC_REPORTING_DISABLED)
- return NOTIFY_DONE;
if (mce->kflags & MCE_HANDLED_CEC)
return NOTIFY_DONE;
@@ -3526,8 +3524,6 @@ static int __init sbridge_init(void)
if (rc >= 0) {
mce_register_decode_chain(&sbridge_mce_dec);
- if (edac_get_report_status() == EDAC_REPORTING_DISABLED)
- sbridge_printk(KERN_WARNING, "Loading driver, error reporting disabled.\n");
return 0;
}
diff --git a/drivers/edac/skx_common.c b/drivers/edac/skx_common.c
index 6f08a12f6b11..423d33aef54f 100644
--- a/drivers/edac/skx_common.c
+++ b/drivers/edac/skx_common.c
@@ -574,9 +574,6 @@ int skx_mce_check_error(struct notifier_block *nb, unsigned long val,
struct mem_ctl_info *mci;
char *type;
- if (edac_get_report_status() == EDAC_REPORTING_DISABLED)
- return NOTIFY_DONE;
-
if (mce->kflags & MCE_HANDLED_CEC)
return NOTIFY_DONE;
diff --git a/include/linux/edac.h b/include/linux/edac.h
index 0f20b986b0ab..6eb7d55d7c3d 100644
--- a/include/linux/edac.h
+++ b/include/linux/edac.h
@@ -31,14 +31,6 @@ struct device;
extern int edac_op_state;
struct bus_type *edac_get_sysfs_subsys(void);
-int edac_get_report_status(void);
-void edac_set_report_status(int new);
-
-enum {
- EDAC_REPORTING_ENABLED,
- EDAC_REPORTING_DISABLED,
- EDAC_REPORTING_FORCE
-};
static inline void opstate_init(void)
{
--
2.21.0
> The last patch is something tglx spotted yesterday and fixing that with
> the MCE flags is pretty easy - was boxing with a wrapper struct around
> struct mce and that gets really ugly.
>
> Tony, I'm open to suggestions how to test it - I probably don't have an
> access to such box which can trigger read errors on nvdimms or what was
> the use case?
It passes my smoke tests (uncorrectable error consumed by application and
uncorrectable error consumed by mcsafe_memcpy()).
Tested-by: Tony Luck <[email protected]>
-Tony
On Tue, Apr 07, 2020 at 07:53:56PM +0000, Luck, Tony wrote:
> It passes my smoke tests (uncorrectable error consumed by application and
> uncorrectable error consumed by mcsafe_memcpy()).
That, of course, is even better, thanks for testing!
> Tested-by: Tony Luck <[email protected]>
Thx, I'll queue the whole pile next week.
--
Regards/Gruss,
Boris.
https://people.kernel.org/tglx/notes-about-netiquette
The following commit has been merged into the ras/core branch of tip:
Commit-ID: 9554bfe403bdfc084823df8695a01f28c680af61
Gitweb: https://git.kernel.org/tip/9554bfe403bdfc084823df8695a01f28c680af61
Author: Tony Luck <[email protected]>
AuthorDate: Fri, 14 Feb 2020 14:27:15 -08:00
Committer: Borislav Petkov <[email protected]>
CommitterDate: Tue, 14 Apr 2020 15:58:08 +02:00
x86/mce: Convert the CEC to use the MCE notifier
The CEC code has its claws in a couple of routines in mce/core.c.
Convert it to just register itself on the normal MCE notifier chain.
[ bp: Make cec_add_elem() and cec_init() static. ]
Signed-off-by: Tony Luck <[email protected]>
Signed-off-by: Borislav Petkov <[email protected]>
Tested-by: Tony Luck <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
---
arch/x86/kernel/cpu/mce/core.c | 19 -------------------
drivers/ras/cec.c | 30 ++++++++++++++++++++++++++++--
include/linux/ras.h | 5 -----
3 files changed, 28 insertions(+), 26 deletions(-)
diff --git a/arch/x86/kernel/cpu/mce/core.c b/arch/x86/kernel/cpu/mce/core.c
index 43b1519..b033b35 100644
--- a/arch/x86/kernel/cpu/mce/core.c
+++ b/arch/x86/kernel/cpu/mce/core.c
@@ -544,21 +544,6 @@ bool mce_is_correctable(struct mce *m)
}
EXPORT_SYMBOL_GPL(mce_is_correctable);
-static bool cec_add_mce(struct mce *m)
-{
- if (!m)
- return false;
-
- /* We eat only correctable DRAM errors with usable addresses. */
- if (mce_is_memory_error(m) &&
- mce_is_correctable(m) &&
- mce_usable_address(m))
- if (!cec_add_elem(m->addr >> PAGE_SHIFT))
- return true;
-
- return false;
-}
-
static int mce_early_notifier(struct notifier_block *nb, unsigned long val,
void *data)
{
@@ -567,9 +552,6 @@ static int mce_early_notifier(struct notifier_block *nb, unsigned long val,
if (!m)
return NOTIFY_DONE;
- if (cec_add_mce(m))
- return NOTIFY_STOP;
-
/* Emit the trace record: */
trace_mce_record(m);
@@ -2612,7 +2594,6 @@ static int __init mcheck_late_init(void)
static_branch_inc(&mcsafe_key);
mcheck_debugfs_init();
- cec_init();
/*
* Flush out everything that has been logged during early boot, now that
diff --git a/drivers/ras/cec.c b/drivers/ras/cec.c
index c09cf55..6b42040 100644
--- a/drivers/ras/cec.c
+++ b/drivers/ras/cec.c
@@ -309,7 +309,7 @@ static bool sanity_check(struct ce_array *ca)
return ret;
}
-int cec_add_elem(u64 pfn)
+static int cec_add_elem(u64 pfn)
{
struct ce_array *ca = &ce_arr;
unsigned int to = 0;
@@ -527,7 +527,30 @@ err:
return 1;
}
-void __init cec_init(void)
+static int cec_notifier(struct notifier_block *nb, unsigned long val,
+ void *data)
+{
+ struct mce *m = (struct mce *)data;
+
+ if (!m)
+ return NOTIFY_DONE;
+
+ /* We eat only correctable DRAM errors with usable addresses. */
+ if (mce_is_memory_error(m) &&
+ mce_is_correctable(m) &&
+ mce_usable_address(m))
+ if (!cec_add_elem(m->addr >> PAGE_SHIFT))
+ return NOTIFY_STOP;
+
+ return NOTIFY_DONE;
+}
+
+static struct notifier_block cec_nb = {
+ .notifier_call = cec_notifier,
+ .priority = MCE_PRIO_CEC,
+};
+
+static void __init cec_init(void)
{
if (ce_arr.disabled)
return;
@@ -546,8 +569,11 @@ void __init cec_init(void)
INIT_DELAYED_WORK(&cec_work, cec_work_fn);
schedule_delayed_work(&cec_work, CEC_DECAY_DEFAULT_INTERVAL);
+ mce_register_decode_chain(&cec_nb);
+
pr_info("Correctable Errors collector initialized.\n");
}
+late_initcall(cec_init);
int __init parse_cec_param(char *str)
{
diff --git a/include/linux/ras.h b/include/linux/ras.h
index 7c3debb..1f4048b 100644
--- a/include/linux/ras.h
+++ b/include/linux/ras.h
@@ -17,12 +17,7 @@ static inline int ras_add_daemon_trace(void) { return 0; }
#endif
#ifdef CONFIG_RAS_CEC
-void __init cec_init(void);
int __init parse_cec_param(char *str);
-int cec_add_elem(u64 pfn);
-#else
-static inline void __init cec_init(void) { }
-static inline int cec_add_elem(u64 pfn) { return -ENODEV; }
#endif
#ifdef CONFIG_RAS
The following commit has been merged into the ras/core branch of tip:
Commit-ID: 1df73b2131e3b33d518609769636b41ce00212de
Gitweb: https://git.kernel.org/tip/1df73b2131e3b33d518609769636b41ce00212de
Author: Borislav Petkov <[email protected]>
AuthorDate: Tue, 07 Apr 2020 13:49:58 +02:00
Committer: Borislav Petkov <[email protected]>
CommitterDate: Tue, 14 Apr 2020 16:01:49 +02:00
x86/mce: Fixup exception only for the correct MCEs
The severity grading code returns IN_KERNEL_RECOV error context for
errors which have happened in kernel space but from which the kernel can
recover. Whether the recovery can happen is determined by the exception
table entry having as handler ex_handler_fault() and which has been
declared at build time using _ASM_EXTABLE_FAULT().
IN_KERNEL_RECOV is used in mce_severity_intel() to lookup the
corresponding error severity in the severities table.
However, the mapping back from error severity to whether the error is
IN_KERNEL_RECOV is ambiguous and in the very paranoid case - which
might not be possible right now - but be better safe than sorry later,
an exception fixup could be attempted for another MCE whose address
is in the exception table and has the proper severity. Which would be
unfortunate, to say the least.
Therefore, mark such MCEs explicitly as MCE_IN_KERNEL_RECOV so that the
recovery attempt is done only for them.
Document the whole handling, while at it, as it is not trivial.
Reported-by: Thomas Gleixner <[email protected]>
Signed-off-by: Borislav Petkov <[email protected]>
Tested-by: Tony Luck <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
---
arch/x86/include/asm/mce.h | 1 +
arch/x86/kernel/cpu/mce/core.c | 15 +++++++++++++--
arch/x86/kernel/cpu/mce/severity.c | 6 +++++-
3 files changed, 19 insertions(+), 3 deletions(-)
diff --git a/arch/x86/include/asm/mce.h b/arch/x86/include/asm/mce.h
index 5f04a24..c598aaa 100644
--- a/arch/x86/include/asm/mce.h
+++ b/arch/x86/include/asm/mce.h
@@ -136,6 +136,7 @@
#define MCE_HANDLED_NFIT BIT_ULL(3)
#define MCE_HANDLED_EDAC BIT_ULL(4)
#define MCE_HANDLED_MCELOG BIT_ULL(5)
+#define MCE_IN_KERNEL_RECOV BIT_ULL(6)
/*
* This structure contains all data related to the MCE log. Also
diff --git a/arch/x86/kernel/cpu/mce/core.c b/arch/x86/kernel/cpu/mce/core.c
index 4efe6c1..02e1f16 100644
--- a/arch/x86/kernel/cpu/mce/core.c
+++ b/arch/x86/kernel/cpu/mce/core.c
@@ -1331,8 +1331,19 @@ void notrace do_machine_check(struct pt_regs *regs, long error_code)
local_irq_disable();
ist_end_non_atomic();
} else {
- if (!fixup_exception(regs, X86_TRAP_MC, error_code, 0))
- mce_panic("Failed kernel mode recovery", &m, msg);
+ /*
+ * Handle an MCE which has happened in kernel space but from
+ * which the kernel can recover: ex_has_fault_handler() has
+ * already verified that the rIP at which the error happened is
+ * a rIP from which the kernel can recover (by jumping to
+ * recovery code specified in _ASM_EXTABLE_FAULT()) and the
+ * corresponding exception handler which would do that is the
+ * proper one.
+ */
+ if (m.kflags & MCE_IN_KERNEL_RECOV) {
+ if (!fixup_exception(regs, X86_TRAP_MC, error_code, 0))
+ mce_panic("Failed kernel mode recovery", &m, msg);
+ }
}
out_ist:
diff --git a/arch/x86/kernel/cpu/mce/severity.c b/arch/x86/kernel/cpu/mce/severity.c
index 87bcdc6..e1da619 100644
--- a/arch/x86/kernel/cpu/mce/severity.c
+++ b/arch/x86/kernel/cpu/mce/severity.c
@@ -213,8 +213,12 @@ static int error_context(struct mce *m)
{
if ((m->cs & 3) == 3)
return IN_USER;
- if (mc_recoverable(m->mcgstatus) && ex_has_fault_handler(m->ip))
+
+ if (mc_recoverable(m->mcgstatus) && ex_has_fault_handler(m->ip)) {
+ m->kflags |= MCE_IN_KERNEL_RECOV;
return IN_KERNEL_RECOV;
+ }
+
return IN_KERNEL;
}
The following commit has been merged into the ras/core branch of tip:
Commit-ID: 3e0fdec858d82c829774f271e88b5ceb17051551
Gitweb: https://git.kernel.org/tip/3e0fdec858d82c829774f271e88b5ceb17051551
Author: Borislav Petkov <[email protected]>
AuthorDate: Tue, 07 Apr 2020 09:55:10 +02:00
Committer: Borislav Petkov <[email protected]>
CommitterDate: Tue, 14 Apr 2020 15:53:46 +02:00
x86/mce/amd, edac: Remove report_gart_errors
... because no one should be interested in spurious MCEs anyway. Make
the filtering unconditional and move it to amd_filter_mce().
Signed-off-by: Borislav Petkov <[email protected]>
Tested-by: Tony Luck <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
---
arch/x86/include/asm/mce.h | 3 ++-
arch/x86/kernel/cpu/mce/amd.c | 9 +++++++--
drivers/edac/amd64_edac.c | 8 --------
drivers/edac/mce_amd.c | 24 ------------------------
drivers/edac/mce_amd.h | 2 --
5 files changed, 9 insertions(+), 37 deletions(-)
diff --git a/arch/x86/include/asm/mce.h b/arch/x86/include/asm/mce.h
index f9cea08..83b6dda 100644
--- a/arch/x86/include/asm/mce.h
+++ b/arch/x86/include/asm/mce.h
@@ -127,6 +127,8 @@
#define MSR_AMD64_SMCA_MCx_DEADDR(x) (MSR_AMD64_SMCA_MC0_DEADDR + 0x10*(x))
#define MSR_AMD64_SMCA_MCx_MISCy(x, y) ((MSR_AMD64_SMCA_MC0_MISC1 + y) + (0x10*(x)))
+#define XEC(x, mask) (((x) >> 16) & mask)
+
/*
* This structure contains all data related to the MCE log. Also
* carries a signature to make it easier to find from external
@@ -347,5 +349,4 @@ umc_normaddr_to_sysaddr(u64 norm_addr, u16 nid, u8 umc, u64 *sys_addr) { return
#endif
static inline void mce_hygon_feature_init(struct cpuinfo_x86 *c) { return mce_amd_feature_init(c); }
-
#endif /* _ASM_X86_MCE_H */
diff --git a/arch/x86/kernel/cpu/mce/amd.c b/arch/x86/kernel/cpu/mce/amd.c
index 15c87b8..ea3cf71 100644
--- a/arch/x86/kernel/cpu/mce/amd.c
+++ b/arch/x86/kernel/cpu/mce/amd.c
@@ -577,14 +577,19 @@ bool amd_filter_mce(struct mce *m)
{
enum smca_bank_types bank_type = smca_get_bank_type(m->bank);
struct cpuinfo_x86 *c = &boot_cpu_data;
- u8 xec = (m->status >> 16) & 0x3F;
/* See Family 17h Models 10h-2Fh Erratum #1114. */
if (c->x86 == 0x17 &&
c->x86_model >= 0x10 && c->x86_model <= 0x2F &&
- bank_type == SMCA_IF && xec == 10)
+ bank_type == SMCA_IF && XEC(m->status, 0x3f) == 10)
return true;
+ /* NB GART TLB error reporting is disabled by default. */
+ if (c->x86 < 0x17) {
+ if (m->bank == 4 && XEC(m->status, 0x1f) == 0x5)
+ return true;
+ }
+
return false;
}
diff --git a/drivers/edac/amd64_edac.c b/drivers/edac/amd64_edac.c
index f91f3bc..6bdc5bb 100644
--- a/drivers/edac/amd64_edac.c
+++ b/drivers/edac/amd64_edac.c
@@ -4,9 +4,6 @@
static struct edac_pci_ctl_info *pci_ctl;
-static int report_gart_errors;
-module_param(report_gart_errors, int, 0644);
-
/*
* Set by command line parameter. If BIOS has enabled the ECC, this override is
* cleared to prevent re-enabling the hardware by this driver.
@@ -3681,9 +3678,6 @@ static int __init amd64_edac_init(void)
}
/* register stuff with EDAC MCE */
- if (report_gart_errors)
- amd_report_gart_errors(true);
-
if (boot_cpu_data.x86 >= 0x17)
amd_register_ecc_decoder(decode_umc_error);
else
@@ -3718,8 +3712,6 @@ static void __exit amd64_edac_exit(void)
edac_pci_release_generic_ctl(pci_ctl);
/* unregister from EDAC MCE */
- amd_report_gart_errors(false);
-
if (boot_cpu_data.x86 >= 0x17)
amd_unregister_ecc_decoder(decode_umc_error);
else
diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index 8874b77..e58644d 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -10,15 +10,8 @@ static struct amd_decoder_ops fam_ops;
static u8 xec_mask = 0xf;
-static bool report_gart_errors;
static void (*decode_dram_ecc)(int node_id, struct mce *m);
-void amd_report_gart_errors(bool v)
-{
- report_gart_errors = v;
-}
-EXPORT_SYMBOL_GPL(amd_report_gart_errors);
-
void amd_register_ecc_decoder(void (*f)(int, struct mce *))
{
decode_dram_ecc = f;
@@ -1030,20 +1023,6 @@ static inline void amd_decode_err_code(u16 ec)
pr_cont("\n");
}
-/*
- * Filter out unwanted MCE signatures here.
- */
-static bool ignore_mce(struct mce *m)
-{
- /*
- * NB GART TLB error reporting is disabled by default.
- */
- if (m->bank == 4 && XEC(m->status, 0x1f) == 0x5 && !report_gart_errors)
- return true;
-
- return false;
-}
-
static const char *decode_error_status(struct mce *m)
{
if (m->status & MCI_STATUS_UC) {
@@ -1067,9 +1046,6 @@ amd_decode_mce(struct notifier_block *nb, unsigned long val, void *data)
unsigned int fam = x86_family(m->cpuid);
int ecc;
- if (ignore_mce(m))
- return NOTIFY_STOP;
-
pr_emerg(HW_ERR "%s\n", decode_error_status(m));
pr_emerg(HW_ERR "CPU:%d (%x:%x:%x) MC%d_STATUS[%s|%s|%s|%s|%s",
diff --git a/drivers/edac/mce_amd.h b/drivers/edac/mce_amd.h
index 4e9c5e5..4811b18 100644
--- a/drivers/edac/mce_amd.h
+++ b/drivers/edac/mce_amd.h
@@ -7,7 +7,6 @@
#include <asm/mce.h>
#define EC(x) ((x) & 0xffff)
-#define XEC(x, mask) (((x) >> 16) & mask)
#define LOW_SYNDROME(x) (((x) >> 15) & 0xff)
#define HIGH_SYNDROME(x) (((x) >> 24) & 0xff)
@@ -77,7 +76,6 @@ struct amd_decoder_ops {
bool (*mc2_mce)(u16, u8);
};
-void amd_report_gart_errors(bool);
void amd_register_ecc_decoder(void (*f)(int, struct mce *));
void amd_unregister_ecc_decoder(void (*f)(int, struct mce *));