[PATCH] PCI/AER: Fix AER log missing in DPC case

[PATCH] PCI/AER: Fix AER log missing in DPC case

[Sizhe Liu]

In the current DPC error reporting case, some AER log information
is missing.

-- Error log abnormal (line breaks adjusted)
[  976.604003] pcieport 0000:20:00.0: DPC: containment event, status:
0x1f11: unmasked uncorrectable error detected
(------ AER error log supposed to be printed here, but missing ------)
[  976.604030] nvme nvme0: frozen state error detected, reset controller
[  977.812932] {4}[Hardware Error]: Hardware error from APEI
Generic Hardware Error Source: 0

Cause:
In aer_print_error(), PCIe AER errors is reported, and is rate-limited
by info->ratelimit_print[i]. There are two entry points for
aer_print_error().

1) Native AER
aer_isr_one_error_type() -> aer_process_err_devices() ->
aer_print_error()
2) DPC
dpc_process_error() -> aer_print_error()

The value of info->ratelimit_print[i] is initialized correctly in
the native AER case:
aer_isr_one_error_type() -> find_source_device() ->
find_device_iter() -> add_error_device()

In the DPC case, info->ratelimit_print[i] is not initialized and
alloc by 0 , so in aer_print_error(), it will directly return at line
if (!info->ratelimit_print[i])
This will result in lossing the AER log messages in the DPC case.

Solution:
1. Move the initialization of info->ratelimit_print[i] to
aer_ratelimit_print_init().
2. Add aer_ratelimit_print_init() in dpc_process_error().
3. Replace the initialization by aer_ratelimit_print_init()in
Native AER case.

Test with AER inject:
Set the DPC reporting priority in the BIOS and send
MalfTLP(AER FATAL ERROR) to device.

-- Error log normal (line breaks adjusted)
[ 5366.943807] pcieport 0000:20:00.0: DPC: containment event,
status:0x1f11: unmasked uncorrectable error detected
[ 5366.943826] pcieport 0000:20:00.0: PCIe Bus Error:
severity=Uncorrectable (Fatal), type=Transaction Layer, (Receiver ID)
[root@localhost ~]# [ 5366.943830] pcieport 0000:20:00.0:
device [19e5:a120] error status/mask=00040000/04580000
[ 5366.943833] pcieport 0000:20:00.0:    [18] MalfTLP   (First)
[ 5366.943836] pcieport 0000:20:00.0: AER:   TLP Header:
0x00000000 0x00000000 0x00000000 0x00000000
[ 5366.943843] nvme nvme0: frozen state error detected, reset controller
[ 5368.156778] {2}[Hardware Error]: Hardware error from APEI Generic
Hardware Error Source: 0

Fixes: a57f2bfb4a58 ("PCI/AER: Ratelimit correctable and non-fatal error logging")
Signed-off-by: Sizhe Liu <liusizhe5@huawei.com>
---
 drivers/pci/pci.h      |  1 +
 drivers/pci/pcie/aer.c | 35 +++++++++++++++++++++++------------
 drivers/pci/pcie/dpc.c |  1 +
 3 files changed, 25 insertions(+), 12 deletions(-)

diff --git a/drivers/pci/pci.h b/drivers/pci/pci.h
index 0e67014aa001..0cbcbcd52354 100644
--- a/drivers/pci/pci.h
+++ b/drivers/pci/pci.h
@@ -748,6 +748,7 @@ struct aer_err_info {
 
 int aer_get_device_error_info(struct aer_err_info *info, int i);
 void aer_print_error(struct aer_err_info *info, int i);
+void aer_ratelimit_print_init(struct pci_dev *dev, struct aer_err_info *e_info, int idx);
 
 int pcie_read_tlp_log(struct pci_dev *dev, int where, int where2,
 		      unsigned int tlp_len, bool flit,
diff --git a/drivers/pci/pcie/aer.c b/drivers/pci/pcie/aer.c
index e0bcaa896803..b73915b63327 100644
--- a/drivers/pci/pcie/aer.c
+++ b/drivers/pci/pcie/aer.c
@@ -925,6 +925,28 @@ int cper_severity_to_aer(int cper_severity)
 EXPORT_SYMBOL_GPL(cper_severity_to_aer);
 #endif
 
+/**
+ * aer_ratelimit_print_init - set flag whether error message is printed
+ * @dev: pointer to pci_dev to be rate-limited
+ * @e_info: pointer to error info
+ * @idx: index for ratelimit_print array
+ */
+void aer_ratelimit_print_init(struct pci_dev *dev, struct aer_err_info *e_info, int idx)
+{
+	/*
+	 * Ratelimit AER log messages.  "dev" is either the source
+	 * identified by the root's Error Source ID or it has an unmasked
+	 * error logged in its own AER Capability.  Messages are emitted
+	 * when "ratelimit_print[i]" is non-zero.  If we will print detail
+	 * for a downstream device, make sure we print the Error Source ID
+	 * from the root as well.
+	 */
+	if (aer_ratelimit(dev, e_info->severity)) {
+		e_info->ratelimit_print[idx] = 1;
+		e_info->root_ratelimit_print = 1;
+	}
+}
+
 void pci_print_aer(struct pci_dev *dev, int aer_severity,
 		   struct aer_capability_regs *aer)
 {
@@ -990,18 +1012,7 @@ static int add_error_device(struct aer_err_info *e_info, struct pci_dev *dev)
 	e_info->dev[i] = pci_dev_get(dev);
 	e_info->error_dev_num++;
 
-	/*
-	 * Ratelimit AER log messages.  "dev" is either the source
-	 * identified by the root's Error Source ID or it has an unmasked
-	 * error logged in its own AER Capability.  Messages are emitted
-	 * when "ratelimit_print[i]" is non-zero.  If we will print detail
-	 * for a downstream device, make sure we print the Error Source ID
-	 * from the root as well.
-	 */
-	if (aer_ratelimit(dev, e_info->severity)) {
-		e_info->ratelimit_print[i] = 1;
-		e_info->root_ratelimit_print = 1;
-	}
+	aer_ratelimit_print_init(dev, e_info, i);
 	return 0;
 }
 
diff --git a/drivers/pci/pcie/dpc.c b/drivers/pci/pcie/dpc.c
index fc18349614d7..d17adc642781 100644
--- a/drivers/pci/pcie/dpc.c
+++ b/drivers/pci/pcie/dpc.c
@@ -275,6 +275,7 @@ void dpc_process_error(struct pci_dev *pdev)
 			 status);
 		if (dpc_get_aer_uncorrect_severity(pdev, &info) &&
 		    aer_get_device_error_info(&info, 0)) {
+			aer_ratelimit_print_init(pdev, &info, 0);
 			aer_print_error(&info, 0);
 			pci_aer_clear_nonfatal_status(pdev);
 			pci_aer_clear_fatal_status(pdev);
-- 
2.33.0

Re: [PATCH] PCI/AER: Fix AER log missing in DPC case

[Jonathan Cameron]

On Tue, 27 Jan 2026 11:54:05 +0800
Sizhe Liu <liusizhe5@huawei.com> wrote:

> In the current DPC error reporting case, some AER log information
> is missing.

Wrap commit messages up to 75 chars, this is about 65.
Only do that for stuff where the formatting isn't fixed for other
reasons, such as the log below.

> 
> -- Error log abnormal (line breaks adjusted)
> [  976.604003] pcieport 0000:20:00.0: DPC: containment event, status:
> 0x1f11: unmasked uncorrectable error detected
> (------ AER error log supposed to be printed here, but missing ------)
> [  976.604030] nvme nvme0: frozen state error detected, reset controller
> [  977.812932] {4}[Hardware Error]: Hardware error from APEI
> Generic Hardware Error Source: 0
> 
This, but do remove timestamps as those don't matter to anyone.
Not wrapping helps if anyone is searching for this later.

Abnormal error log

  pcieport 0000:20:00.0: DPC: containment event, status: 0x1f11: unmasked uncorrectable error detected
(------ AER error log supposed to be printed here, but missing ------)
  nvme nvme0: frozen state error detected, reset controller
{4}[Hardware Error]: Hardware error from APEI Generic Hardware Error Source: 0


> Cause:
> In aer_print_error(), PCIe AER errors is reported, and is rate-limited
> by info->ratelimit_print[i]. There are two entry points for
> aer_print_error().
> 
> 1) Native AER
> aer_isr_one_error_type() -> aer_process_err_devices() ->
> aer_print_error()
> 2) DPC
> dpc_process_error() -> aer_print_error()
> 
> The value of info->ratelimit_print[i] is initialized correctly in
> the native AER case:

> aer_isr_one_error_type() -> find_source_device() ->
> find_device_iter() -> add_error_device()
> 
> In the DPC case, info->ratelimit_print[i] is not initialized and
> alloc by 0 , so in aer_print_error(), it will directly return at line
> if (!info->ratelimit_print[i])
> This will result in lossing the AER log messages in the DPC case.

losing

> 
> Solution:
> 1. Move the initialization of info->ratelimit_print[i] to
> aer_ratelimit_print_init().
> 2. Add aer_ratelimit_print_init() in dpc_process_error().
> 3. Replace the initialization by aer_ratelimit_print_init()in
> Native AER case.
> 
> Test with AER inject:
> Set the DPC reporting priority in the BIOS and send
> MalfTLP(AER FATAL ERROR) to device.
> 
> -- Error log normal (line breaks adjusted)
> [ 5366.943807] pcieport 0000:20:00.0: DPC: containment event,
> status:0x1f11: unmasked uncorrectable error detected
> [ 5366.943826] pcieport 0000:20:00.0: PCIe Bus Error:
> severity=Uncorrectable (Fatal), type=Transaction Layer, (Receiver ID)
> [root@localhost ~]# [ 5366.943830] pcieport 0000:20:00.0:
> device [19e5:a120] error status/mask=00040000/04580000
> [ 5366.943833] pcieport 0000:20:00.0:    [18] MalfTLP   (First)
> [ 5366.943836] pcieport 0000:20:00.0: AER:   TLP Header:
> 0x00000000 0x00000000 0x00000000 0x00000000
> [ 5366.943843] nvme nvme0: frozen state error detected, reset controller
> [ 5368.156778] {2}[Hardware Error]: Hardware error from APEI Generic
> Hardware Error Source: 0
Reformat this and drop timestamps as well:

Normal error log

  pcieport 0000:20:00.0: DPC: containment event, status:0x1f11: unmasked uncorrectable error detected
  pcieport 0000:20:00.0: PCIe Bus Error: severity=Uncorrectable (Fatal), type=Transaction Layer, (Receiver ID)
  pcieport 0000:20:00.0: device [19e5:a120] error status/mask=00040000/04580000
  pcieport 0000:20:00.0:    [18] MalfTLP   (First)
  pcieport 0000:20:00.0: AER:   TLP Header: 0x00000000 0x00000000 0x00000000 0x00000000
  nvme nvme0: frozen state error detected, reset controller
  {2}[Hardware Error]: Hardware error from APEI Generic Hardware Error Source: 0

> 
> Fixes: a57f2bfb4a58 ("PCI/AER: Ratelimit correctable and non-fatal error logging")
> Signed-off-by: Sizhe Liu <liusizhe5@huawei.com>

Otherwise looks good to me.

Thanks,

Jonathan

Re: [PATCH] PCI/AER: Fix AER log missing in DPC case

[Sizhe LIU]

On 2026/1/27 23:56, Jonathan Cameron wrote:
> On Tue, 27 Jan 2026 11:54:05 +0800
> Sizhe Liu <liusizhe5@huawei.com> wrote:
>
>> In the current DPC error reporting case, some AER log information
>> is missing.
> Wrap commit messages up to 75 chars, this is about 65.
> Only do that for stuff where the formatting isn't fixed for other
> reasons, such as the log below.
Got it, I will align the format.
>> -- Error log abnormal (line breaks adjusted)
>> [  976.604003] pcieport 0000:20:00.0: DPC: containment event, status:
>> 0x1f11: unmasked uncorrectable error detected
>> (------ AER error log supposed to be printed here, but missing ------)
>> [  976.604030] nvme nvme0: frozen state error detected, reset controller
>> [  977.812932] {4}[Hardware Error]: Hardware error from APEI
>> Generic Hardware Error Source: 0
>>
> This, but do remove timestamps as those don't matter to anyone.
> Not wrapping helps if anyone is searching for this later.
>
> Abnormal error log
>
>    pcieport 0000:20:00.0: DPC: containment event, status: 0x1f11: unmasked uncorrectable error detected
> (------ AER error log supposed to be printed here, but missing ------)
>    nvme nvme0: frozen state error detected, reset controller
> {4}[Hardware Error]: Hardware error from APEI Generic Hardware Error Source: 0
>
I will remove the timestamps and adjust the log accordingly.
>> Cause:
>> In aer_print_error(), PCIe AER errors is reported, and is rate-limited
>> by info->ratelimit_print[i]. There are two entry points for
>> aer_print_error().
>>
>> 1) Native AER
>> aer_isr_one_error_type() -> aer_process_err_devices() ->
>> aer_print_error()
>> 2) DPC
>> dpc_process_error() -> aer_print_error()
>>
>> The value of info->ratelimit_print[i] is initialized correctly in
>> the native AER case:
>> aer_isr_one_error_type() -> find_source_device() ->
>> find_device_iter() -> add_error_device()
>>
>> In the DPC case, info->ratelimit_print[i] is not initialized and
>> alloc by 0 , so in aer_print_error(), it will directly return at line
>> if (!info->ratelimit_print[i])
>> This will result in lossing the AER log messages in the DPC case.
> losing
Oops, good catch! Thanks :)
>> Solution:
>> 1. Move the initialization of info->ratelimit_print[i] to
>> aer_ratelimit_print_init().
>> 2. Add aer_ratelimit_print_init() in dpc_process_error().
>> 3. Replace the initialization by aer_ratelimit_print_init()in
>> Native AER case.
>>
>> Test with AER inject:
>> Set the DPC reporting priority in the BIOS and send
>> MalfTLP(AER FATAL ERROR) to device.
>>
>> -- Error log normal (line breaks adjusted)
>> [ 5366.943807] pcieport 0000:20:00.0: DPC: containment event,
>> status:0x1f11: unmasked uncorrectable error detected
>> [ 5366.943826] pcieport 0000:20:00.0: PCIe Bus Error:
>> severity=Uncorrectable (Fatal), type=Transaction Layer, (Receiver ID)
>> [root@localhost ~]# [ 5366.943830] pcieport 0000:20:00.0:
>> device [19e5:a120] error status/mask=00040000/04580000
>> [ 5366.943833] pcieport 0000:20:00.0:    [18] MalfTLP   (First)
>> [ 5366.943836] pcieport 0000:20:00.0: AER:   TLP Header:
>> 0x00000000 0x00000000 0x00000000 0x00000000
>> [ 5366.943843] nvme nvme0: frozen state error detected, reset controller
>> [ 5368.156778] {2}[Hardware Error]: Hardware error from APEI Generic
>> Hardware Error Source: 0
> Reformat this and drop timestamps as well:
>
> Normal error log
>
>    pcieport 0000:20:00.0: DPC: containment event, status:0x1f11: unmasked uncorrectable error detected
>    pcieport 0000:20:00.0: PCIe Bus Error: severity=Uncorrectable (Fatal), type=Transaction Layer, (Receiver ID)
>    pcieport 0000:20:00.0: device [19e5:a120] error status/mask=00040000/04580000
>    pcieport 0000:20:00.0:    [18] MalfTLP   (First)
>    pcieport 0000:20:00.0: AER:   TLP Header: 0x00000000 0x00000000 0x00000000 0x00000000
>    nvme nvme0: frozen state error detected, reset controller
>    {2}[Hardware Error]: Hardware error from APEI Generic Hardware Error Source: 0
I will reformat the AER log as above.
>> Fixes: a57f2bfb4a58 ("PCI/AER: Ratelimit correctable and non-fatal error logging")
>> Signed-off-by: Sizhe Liu <liusizhe5@huawei.com>
> Otherwise looks good to me.
>
> Thanks,
>
> Jonathan
>
A v2 patch will be sent for adjusting the commit log.


Kind regards,

Sizhe