| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <7cf3c13f-1cef-429b-a9f6-0404bbe721d3@linux.intel.com>
Date: Wed, 28 Jan 2026 08:50:46 -0800
From: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@...ux.intel.com>
To: Shuai Xue <xueshuai@...ux.alibaba.com>, linux-pci@...r.kernel.org,
linux-kernel@...r.kernel.org, linuxppc-dev@...ts.ozlabs.org,
bhelgaas@...gle.com, kbusch@...nel.org
Cc: mahesh@...ux.ibm.com, oohall@...il.com, Jonathan.Cameron@...wei.com,
terry.bowman@....com, tianruidong@...ux.alibaba.com, lukas@...ner.de
Subject: Re: [PATCH v7 3/5] PCI/AER: Report fatal errors of RCiEP and EP if
link recoverd
Hi,
On 1/23/2026 11:45 PM, Shuai Xue wrote:
> The AER driver has historically avoided reading the configuration space of
> an endpoint or RCiEP that reported a fatal error, considering the link to
> that device unreliable. Consequently, when a fatal error occurs, the AER
> and DPC drivers do not report specific error types, resulting in logs like:
>
> pcieport 0015:00:00.0: EDR: EDR event received
> pcieport 0015:00:00.0: EDR: Reported EDR dev: 0015:00:00.0
> pcieport 0015:00:00.0: DPC: containment event, status:0x200d, ERR_FATAL received from 0015:01:00.0
> pcieport 0015:00:00.0: AER: broadcast error_detected message
> pcieport 0015:00:00.0: AER: broadcast mmio_enabled message
> pcieport 0015:00:00.0: AER: broadcast resume message
> pcieport 0015:00:00.0: pciehp: Slot(21): Link Down/Up ignored
> pcieport 0015:00:00.0: AER: device recovery successful
> pcieport 0015:00:00.0: EDR: DPC port successfully recovered
> pcieport 0015:00:00.0: EDR: Status for 0015:00:00.0: 0x80
>
> AER status registers are sticky and Write-1-to-clear. If the link recovered
> after hot reset, we can still safely access AER status and TLP header of the
> error device. In such case, report fatal errors which helps to figure out the
> error root case.
>
> After this patch, the logs like:
>
> pcieport 0015:00:00.0: EDR: EDR event received
> pcieport 0015:00:00.0: EDR: Reported EDR dev: 0015:00:00.0
> pcieport 0015:00:00.0: DPC: containment event, status:0x200d, ERR_FATAL received from 0015:01:00.0
> pcieport 0015:00:00.0: AER: broadcast error_detected message
> + vfio-pci 0015:01:00.0: AER: Errors reported prior to reset
> + vfio-pci 0015:01:00.0: PCIe Bus Error: severity=Uncorrectable (Fatal), type=Transaction Layer, (Receiver ID)
> + vfio-pci 0015:01:00.0: device [144d:a80a] error status/mask=00001000/00400000
> + vfio-pci 0015:01:00.0: [12] TLP (First)
> + vfio-pci 0015:01:00.0: AER: TLP Header: 0x4a004010 0x00000040 0x01000000 0xffffffff
> pcieport 0015:00:00.0: AER: broadcast mmio_enabled message
> pcieport 0015:00:00.0: AER: broadcast resume message
> pcieport 0015:00:00.0: pciehp: Slot(21): Link Down/Up ignored
> pcieport 0015:00:00.0: AER: device recovery successful
> pcieport 0015:00:00.0: EDR: DPC port successfully recovered
> pcieport 0015:00:00.0: EDR: Status for 0015:00:00.0: 0x80
>
> Signed-off-by: Shuai Xue <xueshuai@...ux.alibaba.com>
> ---
LGTM. Few suggestions inline.
Reviewed-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@...ux.intel.com>
> drivers/pci/pci.h | 4 +++-
> drivers/pci/pcie/aer.c | 32 ++++++++++++++++++++++++++++----
> drivers/pci/pcie/dpc.c | 2 +-
> drivers/pci/pcie/err.c | 5 +++++
> 4 files changed, 37 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/pci/pci.h b/drivers/pci/pci.h
> index 58640e656897..bd020ba0cef0 100644
> --- a/drivers/pci/pci.h
> +++ b/drivers/pci/pci.h
> @@ -746,8 +746,10 @@ struct aer_err_info {
> struct pcie_tlp_log tlp; /* TLP Header */
> };
>
> -int aer_get_device_error_info(struct aer_err_info *info, int i);
> +int aer_get_device_error_info(struct aer_err_info *info, int i,
> + bool link_healthy);
> void aer_print_error(struct aer_err_info *info, int i);
> +void aer_report_frozen_error(struct pci_dev *dev);
>
> 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..4c0a2bbe9197 100644
> --- a/drivers/pci/pcie/aer.c
> +++ b/drivers/pci/pcie/aer.c
> @@ -1384,12 +1384,14 @@ EXPORT_SYMBOL_GPL(aer_recover_queue);
> * aer_get_device_error_info - read error status from dev and store it to info
> * @info: pointer to structure to store the error record
> * @i: index into info->dev[]
> + * @link_healthy: link is healthy or not
> *
> * Return: 1 on success, 0 on error.
> *
> * Note that @info is reused among all error devices. Clear fields properly.
> */
> -int aer_get_device_error_info(struct aer_err_info *info, int i)
> +int aer_get_device_error_info(struct aer_err_info *info, int i,
> + bool link_healthy)
> {
> struct pci_dev *dev;
> int type, aer;
> @@ -1420,7 +1422,8 @@ int aer_get_device_error_info(struct aer_err_info *info, int i)
> } else if (type == PCI_EXP_TYPE_ROOT_PORT ||
> type == PCI_EXP_TYPE_RC_EC ||
> type == PCI_EXP_TYPE_DOWNSTREAM ||
> - info->severity == AER_NONFATAL) {
> + info->severity == AER_NONFATAL ||
> + (info->severity == AER_FATAL && link_healthy)) {
>
> /* Link is still healthy for IO reads */
> pci_read_config_dword(dev, aer + PCI_ERR_UNCOR_STATUS,
> @@ -1447,17 +1450,38 @@ int aer_get_device_error_info(struct aer_err_info *info, int i)
> return 1;
> }
>
> +void aer_report_frozen_error(struct pci_dev *dev)
Since this function focuses specifically on printing fatal error details, would
aer_print_fatal_error() be a more descriptive name?
> +{
> + struct aer_err_info info;
> + int type = pci_pcie_type(dev);
> +
> + if (type != PCI_EXP_TYPE_ENDPOINT && type != PCI_EXP_TYPE_RC_END)
> + return;
> +
> + info.error_dev_num = 0;
> + info.severity = AER_FATAL;
> + info.level = KERN_ERR;
> + add_error_device(&info, dev);
> +
> + if (aer_get_device_error_info(&info, 0, true)) {
> + pci_err(dev, "Errors reported prior to reset\n");
The 'prior to reset' context depends on where this is called. I'd suggest moving
this log to the caller or removing it entirely to keep this helper generic.
> + aer_print_error(&info, 0);
> + }
> +
> + pci_dev_put(dev); /* pairs with pci_dev_get() in add_error_device() */
> +}
> +
> static inline void aer_process_err_devices(struct aer_err_info *e_info)
> {
> int i;
>
> /* Report all before handling them, to not lose records by reset etc. */
> for (i = 0; i < e_info->error_dev_num && e_info->dev[i]; i++) {
> - if (aer_get_device_error_info(e_info, i))
> + if (aer_get_device_error_info(e_info, i, false))
> aer_print_error(e_info, i);
> }
> for (i = 0; i < e_info->error_dev_num && e_info->dev[i]; i++) {
> - if (aer_get_device_error_info(e_info, i))
> + if (aer_get_device_error_info(e_info, i, false))
> handle_error_source(e_info->dev[i], e_info);
> }
> }
> diff --git a/drivers/pci/pcie/dpc.c b/drivers/pci/pcie/dpc.c
> index f6069f621683..21c4e8371279 100644
> --- a/drivers/pci/pcie/dpc.c
> +++ b/drivers/pci/pcie/dpc.c
> @@ -284,7 +284,7 @@ struct pci_dev *dpc_process_error(struct pci_dev *pdev)
> pci_warn(pdev, "containment event, status:%#06x: unmasked uncorrectable error detected\n",
> status);
> if (dpc_get_aer_uncorrect_severity(pdev, &info) &&
> - aer_get_device_error_info(&info, 0)) {
> + aer_get_device_error_info(&info, 0, false)) {
> aer_print_error(&info, 0);
> pci_aer_clear_nonfatal_status(pdev);
> pci_aer_clear_fatal_status(pdev);
> diff --git a/drivers/pci/pcie/err.c b/drivers/pci/pcie/err.c
> index bebe4bc111d7..0780ea09478b 100644
> --- a/drivers/pci/pcie/err.c
> +++ b/drivers/pci/pcie/err.c
> @@ -253,6 +253,11 @@ pci_ers_result_t pcie_do_recovery(struct pci_dev *dev,
> pci_warn(bridge, "subordinate device reset failed\n");
> goto failed;
> }
> +
> + /* Link recovered, report fatal errors of RCiEP or EP */
> + if (state == pci_channel_io_frozen)
To align with your comment regarding RCiEPs and EPs, should we explicitly
validate the device type here before calling the report function?
> + aer_report_frozen_error(dev);
> +
> }
>
> if (status == PCI_ERS_RESULT_NEED_RESET) {
--
Sathyanarayanan Kuppuswamy
Linux Kernel Developer
Powered by blists - more mailing lists