[PATCH v2] PCI: qcom: Block accesses to downstream devices on link down

[PATCH v2] PCI: qcom: Block accesses to downstream devices on link down

[Qiang Yu]

After a PCIe link goes down, software may still access the BAR (MMIO)
space or configuration space of devices behind that link before recovery
has run. As the link is down, these accesses never complete, resulting in
a storm of Completion Timeout AERs.

Use the controller's ECAM blocker to drop these accesses to the PCIe
address space as soon as the link-down interrupt fires, so that Completion
Timeout AERs are reduced. The blocked range covers the entire address
space (base 0x0, all-ones limit), since the Root Port's own DBI/iATU
register space remains accessible regardless.

The range is programmed once in the host init path, since the range
registers are wiped by BCR reset. This leaves only the ECAM_BLOCKER_EN
bit to be flipped from the link-down IRQ handler, so the blocker can be
armed with a single fast register write for immediate effect.

The subsequent Root Port reset re-initialises the controller, which clears
the enable bit and re-programs the range for the fresh link.

Some IP revisions implement the ECAM blocker registers but do not wire
up the "global" interrupt used to deliver the link-down event that arms
the blocker, so there is no way to enable the blocker on those platforms.
Restrict blocker initialization to the post_init hooks of the IP
revisions that do have a global IRQ wired up
(qcom_pcie_post_init_2_3_3(), qcom_pcie_post_init_2_7_0() and
qcom_pcie_post_init_2_9_0()), instead of unconditionally programming it
from the common host_init path.

The link-down IRQ thread's blocker-enable write to PARF_SYS_CTRL can race
with a Root Port reset triggered independently through AER, which
reprograms PARF_SYS_CTRL as part of reinitializing the controller.
Serialize these with a per-controller mutex.

Signed-off-by: Qiang Yu <qiang.yu@oss.qualcomm.com>
---
Changes in v2:
- Only initialize the ECAM blocker on IP revisions that have the "global" IRQ wired up.
- Add a per-controller mutex to serialize the link-down IRQ thread's PARF_SYS_CTRL write against a concurrent Root Port reset from AER recovery.
- Link to v1: https://patch.msgid.link/20260810-ecam_blocker-v1-1-e588e07f68d4@oss.qualcomm.com

To: Manivannan Sadhasivam <mani@kernel.org>
To: Lorenzo Pieralisi <lpieralisi@kernel.org>
To: Krzysztof Wilczyński <kwilczynski@kernel.org>
To: Rob Herring <robh@kernel.org>
To: Bjorn Helgaas <bhelgaas@google.com>
Cc: linux-arm-msm@vger.kernel.org
Cc: linux-pci@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
---
 drivers/pci/controller/dwc/pcie-qcom.c | 83 +++++++++++++++++++++++++++++++---
 1 file changed, 77 insertions(+), 6 deletions(-)

diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c
index b58a607b713f..e8cf349579a8 100644
--- a/drivers/pci/controller/dwc/pcie-qcom.c
+++ b/drivers/pci/controller/dwc/pcie-qcom.c
@@ -20,6 +20,7 @@
 #include <linux/kernel.h>
 #include <linux/limits.h>
 #include <linux/init.h>
+#include <linux/mutex.h>
 #include <linux/of.h>
 #include <linux/of_pci.h>
 #include <linux/pci.h>
@@ -73,6 +74,23 @@
 #define PARF_BDF_TO_SID_TABLE_N			0x2000
 #define PARF_BDF_TO_SID_CFG			0x2c00
 
+/*
+ * ECAM blocker range registers. The blocked range has a write pair
+ * (WR_BASE/WR_LIMIT) and a read pair (RD_BASE/RD_LIMIT); each address is
+ * split into a low (32-bit) and a HI (upper 32-bit) register.
+ */
+#define PARF_BLOCK_SLV_AXI_WR_BASE		0x360
+#define PARF_BLOCK_SLV_AXI_WR_BASE_HI		0x364
+#define PARF_BLOCK_SLV_AXI_WR_LIMIT		0x368
+#define PARF_BLOCK_SLV_AXI_WR_LIMIT_HI		0x36c
+#define PARF_BLOCK_SLV_AXI_RD_BASE		0x370
+#define PARF_BLOCK_SLV_AXI_RD_BASE_HI		0x374
+#define PARF_BLOCK_SLV_AXI_RD_LIMIT		0x378
+#define PARF_BLOCK_SLV_AXI_RD_LIMIT_HI		0x37c
+
+#define PARF_ECAM_BASE				0x380
+#define PARF_ECAM_BASE_HI			0x384
+
 /* ELBI registers */
 #define ELBI_SYS_CTRL				0x04
 #define ELBI_SYS_STTS				0x08
@@ -90,6 +108,7 @@
 
 /* PARF_SYS_CTRL register fields */
 #define MAC_PHY_POWERDOWN_IN_P2_D_MUX_EN	BIT(29)
+#define ECAM_BLOCKER_EN				BIT(26)
 #define MST_WAKEUP_EN				BIT(13)
 #define SLV_WAKEUP_EN				BIT(12)
 #define MSTR_ACLK_CGC_DIS			BIT(10)
@@ -308,6 +327,7 @@ struct qcom_pcie {
 	struct gpio_desc *reset;
 	int global_irq;
 	bool use_pm_opp;
+	struct mutex hw_lock;
 };
 
 #define to_qcom_pcie(x)		dev_get_drvdata((x)->dev)
@@ -445,6 +465,25 @@ static void qcom_pcie_configure_dbi_atu_base(struct qcom_pcie *pcie)
 	}
 }
 
+static void qcom_pcie_init_ecam_blocker(struct qcom_pcie *pcie)
+{
+	struct dw_pcie *pci = pcie->pci;
+
+	/* ECAM base must match the DBI base address */
+	writel(lower_32_bits(pci->dbi_phys_addr), pcie->parf + PARF_ECAM_BASE);
+	writel(upper_32_bits(pci->dbi_phys_addr), pcie->parf + PARF_ECAM_BASE_HI);
+
+	writel(0, pcie->parf + PARF_BLOCK_SLV_AXI_WR_BASE);
+	writel(0, pcie->parf + PARF_BLOCK_SLV_AXI_WR_BASE_HI);
+	writel(U32_MAX, pcie->parf + PARF_BLOCK_SLV_AXI_WR_LIMIT);
+	writel(U32_MAX, pcie->parf + PARF_BLOCK_SLV_AXI_WR_LIMIT_HI);
+
+	writel(0, pcie->parf + PARF_BLOCK_SLV_AXI_RD_BASE);
+	writel(0, pcie->parf + PARF_BLOCK_SLV_AXI_RD_BASE_HI);
+	writel(U32_MAX, pcie->parf + PARF_BLOCK_SLV_AXI_RD_LIMIT);
+	writel(U32_MAX, pcie->parf + PARF_BLOCK_SLV_AXI_RD_LIMIT_HI);
+}
+
 static void qcom_pcie_2_1_0_ltssm_enable(struct qcom_pcie *pcie)
 {
 	struct dw_pcie *pci = pcie->pci;
@@ -990,6 +1029,8 @@ static int qcom_pcie_post_init_2_3_3(struct qcom_pcie *pcie)
 
 	dw_pcie_dbi_ro_wr_dis(pci);
 
+	qcom_pcie_init_ecam_blocker(pcie);
+
 	return 0;
 }
 
@@ -1104,6 +1145,8 @@ static int qcom_pcie_post_init_2_7_0(struct qcom_pcie *pcie)
 
 	qcom_pcie_set_slot_cap(pcie->pci);
 
+	qcom_pcie_init_ecam_blocker(pcie);
+
 	return 0;
 }
 
@@ -1322,6 +1365,8 @@ static int qcom_pcie_post_init_2_9_0(struct qcom_pcie *pcie)
 	for (i = 0; i < 256; i++)
 		writel(0, pcie->parf + PARF_BDF_TO_SID_TABLE_N + (4 * i));
 
+	qcom_pcie_init_ecam_blocker(pcie);
+
 	return 0;
 }
 
@@ -1382,6 +1427,18 @@ static void qcom_pcie_configure_ports(struct qcom_pcie *pcie)
 		dw_pcie_program_t_power_on(pcie->pci, port->l1ss_t_power_on);
 }
 
+static void qcom_pcie_enable_ecam_blocker(struct qcom_pcie *pcie)
+{
+	u32 sys_ctrl;
+
+	sys_ctrl = readl(pcie->parf + PARF_SYS_CTRL);
+	sys_ctrl |= ECAM_BLOCKER_EN;
+	writel(sys_ctrl, pcie->parf + PARF_SYS_CTRL);
+
+	/* Flush the write so the blocker is enabled before this function returns */
+	readl(pcie->parf + PARF_SYS_CTRL);
+}
+
 static int qcom_pcie_host_init(struct dw_pcie_rp *pp)
 {
 	struct dw_pcie *pci = to_dw_pcie_from_pp(pp);
@@ -1775,13 +1832,15 @@ static int qcom_pcie_reset_root_port(struct pci_host_bridge *bridge,
 	u32 val;
 	int ret;
 
+	mutex_lock(&pcie->hw_lock);
+
 	/* Wait for the pending transactions to be completed */
 	ret = readl_relaxed_poll_timeout(pcie->parf + PARF_STATUS, val,
 					 val & FLUSH_COMPLETED, 10,
 					 FLUSH_TIMEOUT_US);
 	if (ret) {
 		dev_err(dev, "Flush completion failed: %d\n", ret);
-		return ret;
+		goto out_unlock;
 	}
 
 	/* Clear the FLUSH_MODE to allow the core to be reset */
@@ -1795,7 +1854,7 @@ static int qcom_pcie_reset_root_port(struct pci_host_bridge *bridge,
 					 FLUSH_TIMEOUT_US);
 	if (ret) {
 		dev_err(dev, "Flush mode clear failed: %d\n", ret);
-		return ret;
+		goto out_unlock;
 	}
 
 	qcom_pcie_host_deinit(pp);
@@ -1803,12 +1862,12 @@ static int qcom_pcie_reset_root_port(struct pci_host_bridge *bridge,
 	ret = qcom_pcie_host_init(pp);
 	if (ret) {
 		dev_err(dev, "Host init failed\n");
-		return ret;
+		goto out_unlock;
 	}
 
 	ret = dw_pcie_setup_rc(pp);
 	if (ret)
-		return ret;
+		goto out_unlock;
 
 	/*
 	 * Re-enable global IRQ events as the PARF_INT_ALL_MASK register is
@@ -1822,11 +1881,14 @@ static int qcom_pcie_reset_root_port(struct pci_host_bridge *bridge,
 
 	ret = dw_pcie_wait_for_link(pci);
 	if (ret)
-		return ret;
+		goto out_unlock;
 
 	dev_dbg(dev, "Root Port reset completed\n");
 
-	return 0;
+out_unlock:
+	mutex_unlock(&pcie->hw_lock);
+
+	return ret;
 }
 
 static int qcom_pcie_link_transition_count(struct seq_file *s, void *data)
@@ -1878,6 +1940,11 @@ static irqreturn_t qcom_pcie_global_irq_thread(int irq, void *data)
 
 	if (test_and_clear_bit(INT_ALL_LINK_DOWN, &status)) {
 		dev_dbg(dev, "Received Link down event\n");
+
+		mutex_lock(&pcie->hw_lock);
+		qcom_pcie_enable_ecam_blocker(pcie);
+		mutex_unlock(&pcie->hw_lock);
+
 		for_each_pci_bridge(port, pp->bridge->bus) {
 			if (pci_pcie_type(port) == PCI_EXP_TYPE_ROOT_PORT)
 				pci_host_handle_link_down(port);
@@ -2204,6 +2271,10 @@ static int qcom_pcie_probe(struct platform_device *pdev)
 
 	INIT_LIST_HEAD(&pcie->ports);
 
+	ret = devm_mutex_init(dev, &pcie->hw_lock);
+	if (ret)
+		goto err_pm_runtime_put;
+
 	pci->dev = dev;
 	pci->ops = &dw_pcie_ops;
 	pp = &pci->pp;

---
base-commit: 3de6add8d15430cde96c615fd156216676762bf9
change-id: 20260810-ecam_blocker-f801b3a02bcd

Best regards,
--  
Qiang Yu <qiang.yu@oss.qualcomm.com>

Re: [PATCH v2] PCI: qcom: Block accesses to downstream devices on link down

[sashiko-bot]

> After a PCIe link goes down, software may still access the BAR (MMIO)
> space or configuration space of devices behind that link before recovery
> has run. As the link is down, these accesses never complete, resulting in
> a storm of Completion Timeout AERs.
> 
> Use the controller's ECAM blocker to drop these accesses to the PCIe
> address space as soon as the link-down interrupt fires, so that Completion
> Timeout AERs are reduced. The blocked range covers the entire address
> space (base 0x0, all-ones limit), since the Root Port's own DBI/iATU
> register space remains accessible regardless.
> 
> The range is programmed once in the host init path, since the range
> registers are wiped by BCR reset. This leaves only the ECAM_BLOCKER_EN
> bit to be flipped from the link-down IRQ handler, so the blocker can be
> armed with a single fast register write for immediate effect.
> [ ... ]
> with a Root Port reset triggered independently through AER, which
> reprograms PARF_SYS_CTRL as part of reinitializing the controller.
> Serialize these with a per-controller mutex.
> 
> Signed-off-by: Qiang Yu <qiang.yu@oss.qualcomm.com>

Sashiko has reviewed this patch and found no issues. It looks great!

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260819-ecam_blocker-v2-1-e7a8fdc1c5cb@oss.qualcomm.com?part=1

Re: [PATCH v2] PCI: qcom: Block accesses to downstream devices on link down

[Konrad Dybcio]

On 8/20/26 8:36 AM, Qiang Yu wrote:
> After a PCIe link goes down, software may still access the BAR (MMIO)
> space or configuration space of devices behind that link before recovery
> has run. As the link is down, these accesses never complete, resulting in
> a storm of Completion Timeout AERs.

[...]

>  static int qcom_pcie_host_init(struct dw_pcie_rp *pp)
>  {
>  	struct dw_pcie *pci = to_dw_pcie_from_pp(pp);
> @@ -1775,13 +1832,15 @@ static int qcom_pcie_reset_root_port(struct pci_host_bridge *bridge,
>  	u32 val;
>  	int ret;
>  
> +	mutex_lock(&pcie->hw_lock);

guard(mutex)(&pcie->hw_lock)

This decreases the number of AERs I see while de-authorizing the
PCIe link on a TBT3 connection to an ASUS PA27AC monitor from
"a whole lot" to just 1 on X1E80100 CRD

Tested-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com> # X1E80100 CRD + ASUS PA27AC

Konrad

Re: [PATCH v2] PCI: qcom: Block accesses to downstream devices on link down

[Manivannan Sadhasivam]

On Wed, 19 Aug 2026 23:36:54 -0700, Qiang Yu wrote:
> After a PCIe link goes down, software may still access the BAR (MMIO)
> space or configuration space of devices behind that link before recovery
> has run. As the link is down, these accesses never complete, resulting in
> a storm of Completion Timeout AERs.
> 
> Use the controller's ECAM blocker to drop these accesses to the PCIe
> address space as soon as the link-down interrupt fires, so that Completion
> Timeout AERs are reduced. The blocked range covers the entire address
> space (base 0x0, all-ones limit), since the Root Port's own DBI/iATU
> register space remains accessible regardless.
> 
> [...]

Applied, thanks!

[1/1] PCI: qcom: Block accesses to downstream devices on link down
      commit: 86624b718f94517b26f96c1d7beb7632e5562afd

Best regards,
-- 
மணிவண்ணன் சதாசிவம்

Re: [PATCH v2] PCI: qcom: Block accesses to downstream devices on link down

[Konrad Dybcio]

On 9/1/26 2:35 PM, Manivannan Sadhasivam wrote:
> 
> On Wed, 19 Aug 2026 23:36:54 -0700, Qiang Yu wrote:
>> After a PCIe link goes down, software may still access the BAR (MMIO)
>> space or configuration space of devices behind that link before recovery
>> has run. As the link is down, these accesses never complete, resulting in
>> a storm of Completion Timeout AERs.
>>
>> Use the controller's ECAM blocker to drop these accesses to the PCIe
>> address space as soon as the link-down interrupt fires, so that Completion
>> Timeout AERs are reduced. The blocked range covers the entire address
>> space (base 0x0, all-ones limit), since the Root Port's own DBI/iATU
>> register space remains accessible regardless.
>>
>> [...]
> 
> Applied, thanks!
> 
> [1/1] PCI: qcom: Block accesses to downstream devices on link down
>       commit: 86624b718f94517b26f96c1d7beb7632e5562afd

Could this be a fix for

https://lore.kernel.org/linux-arm-msm/b461c125-947b-4e06-b502-abb354e0cba6@oss.qualcomm.com/

?

(+Xilin)

Konrad

Re: [PATCH v2] PCI: qcom: Block accesses to downstream devices on link down

[Xilin Wu]

On 9/1/2026 8:39 PM, Konrad Dybcio wrote:
> On 9/1/26 2:35 PM, Manivannan Sadhasivam wrote:
>>
>> On Wed, 19 Aug 2026 23:36:54 -0700, Qiang Yu wrote:
>>> After a PCIe link goes down, software may still access the BAR (MMIO)
>>> space or configuration space of devices behind that link before recovery
>>> has run. As the link is down, these accesses never complete, resulting in
>>> a storm of Completion Timeout AERs.
>>>
>>> Use the controller's ECAM blocker to drop these accesses to the PCIe
>>> address space as soon as the link-down interrupt fires, so that Completion
>>> Timeout AERs are reduced. The blocked range covers the entire address
>>> space (base 0x0, all-ones limit), since the Root Port's own DBI/iATU
>>> register space remains accessible regardless.
>>>
>>> [...]
>>
>> Applied, thanks!
>>
>> [1/1] PCI: qcom: Block accesses to downstream devices on link down
>>        commit: 86624b718f94517b26f96c1d7beb7632e5562afd
> 
> Could this be a fix for
> 
> https://lore.kernel.org/linux-arm-msm/b461c125-947b-4e06-b502-abb354e0cba6@oss.qualcomm.com/
> 
> ?
> 
> (+Xilin)

I'm not sure. Actually, we still have PCIe-related issues during reboot 
after reverting the flattened USB patch, though at a much lower 
occurrence rate.

I hope this series fixes it, but I haven't tested it yet.

https://patchwork.kernel.org/project/linux-arm-msm/list/?series=1151871&state=%2A&archive=both

Case 1 log:

Reboot count: 97
Reboot in 10 seconds, press any key to abort... [   12.260389] qcom-apm 
gprsvc:service:2:1: CMD timeout for [1001021] opcode
[   12.615536] qcom-soundwire 3210000.soundwire: din-ports (0) mismatch 
with controller (1)
[   12.626249] qcom-soundwire 3210000.soundwire: dout-ports (5) mismatch 
with controller (6)
[   39.120833] systemd-shutdown[1]: Failed to set watchdog hardware 
timeout to 10min: Invalid argument
[   40.341886] ------------[ cut here ]------------
[   40.346737] phy_check_link_status+0x0/0x118: returned: -16
[   40.352528] WARNING: drivers/net/phy/phy.c:1352 at 
_phy_state_machine+0x190/0x330, CPU#1: kworker/u32:2/72
[   40.362549] Modules linked in: q6prm_clocks q6apm_lpass_dais 
snd_q6dsp_common q6apm_dai q6prm joydev mousedev algif_hash 
algif_skcipher af_alg bnep rtw89_8852be rtw89_8852b rtw89_8852b_common 
rtw89_pci rtw89_core mac80211 qcom_iris cfg80211 videobuf2_dma_contig 
btusb v4l2_mem2mem btmtk videobuf2_memops btrtl btbcm videobuf2_v4l2 
libarc4 btintel videodev snd_soc_wcd938x bluetooth snd_soc_wcd938x_sdw 
snd_soc_wcd_classh videobuf2_common snd_soc_wcd_common snd_soc_sc8280xp 
rfkill mc snd_soc_qcom_common snd_soc_wcd_mbhc qcom_spmi_adc_tm5 
qcom_spmi_adc5 snd_soc_qcom_sdw soundwire_qcom qcom_vadc_common 
regmap_sdw snd_soc_lpass_tx_macro soundwire_bus snd_soc_lpass_rx_macro 
qcom_spmi_temp_alarm snd_q6apm snd_soc_lpass_va_macro radxa_svc_glink 
qcomtee snd_soc_lpass_macro_common slimbus sg fastrpc qcom_scm_storage 
platform_profile qcom_rng libcomposite uinput pmic_pdcharger_ulog 
nfnetlink dwmac_tc956x stmmac gpio_tc956x pcs_xpcs phylink rtc_ds1307 
at24 rpmsg_ctrl tc956x_pci qrtr_smd rpmsg_char qcom_pd_mapper uas qcom_pon
[   40.362750]  i2c_qcom_geni videocc_sm8350 camcc_sc8280xp 
qcom_q6v5_pas qcom_pil_info qcom_common qcom_refgen_regulator 
pinctrl_sc8280xp_lpass_lpi ucsi_glink qcom_glink_smem pinctrl_lpass_lpi 
qcom_q6v5 lpasscc_sc8280xp typec_ucsi qcom_sysmon qcom_wdt [last 
unloaded: 842_decompress]
[   40.480794] CPU: 1 UID: 0 PID: 72 Comm: kworker/u32:2 Not tainted 
7.0.11+ #38 PREEMPT(lazy)  a69a8d0b840e8538e5b93598a4d4312fb8da4861
[   40.493213] Hardware name: Radxa Computer Co., Ltd. Radxa Dragon 
Q8B/RS782-D16S32W0, BIOS 6.0.260812.BOOT.MXF.1.1.c1-00167-MAKENA-1 
08/12/2026
[   40.506435] Workqueue: events_power_efficient phy_state_machine
[   40.512577] pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS 
BTYPE=--)
[   40.519788] pc : _phy_state_machine+0x190/0x330
[   40.524500] lr : _phy_state_machine+0x190/0x330
[   40.529209] sp : ffff800080a6bd20
[   40.532663] x29: ffff800080a6bd40 x28: 0000000000000000 x27: 
0000000000000000
[   40.540058] x26: ffff000080041028 x25: 0000000000000000 x24: 
ffff0000800c1605
[   40.547453] x23: ffff0000855b9cf8 x22: ffff000080041000 x21: 
0000000000000000
[   40.554845] x20: 0000000000000006 x19: ffff0000855b9800 x18: 
ffff3b3b60d68000
[   40.562236] x17: 000000040044ffff x16: ffffc4cc93b2a028 x15: 
ffffc4cc950c3c70
[   40.569632] x14: ffffc4cc94f8b480 x13: 0000000000000180 x12: 
ffff0007f5cf3480
[   40.577025] x11: 00000000000000c0 x10: 7ca63edf1413fc9b x9 : 
ffffc4cc92dc2df4
[   40.584427] x8 : ffff000080c9eba8 x7 : 0000000000000004 x6 : 
0000000000000000
[   40.591829] x5 : ffff3b3b60d68000 x4 : ffff000080c9da00 x3 : 
ffff00008e41da00
[   40.599223] x2 : 0000000000000000 x1 : 0000000000000000 x0 : 
ffff000080c9da00
[   40.606625] Call trace:
[   40.609174]  _phy_state_machine+0x190/0x330 (P)
[   40.613878]  phy_state_machine+0x34/0xc0
[   40.617953]  process_one_work+0x194/0x570
[   40.622129]  worker_thread+0x18c/0x320
[   40.626033]  kthread+0x134/0x148
[   40.629399]  ret_from_fork+0x10/0x20
[   40.633127] ---[ end trace 0000000000000000 ]---
[   40.946101] SError Interrupt on CPU5, code 0x00000000be000011 -- SError
[   40.946109] CPU: 5 UID: 0 PID: 335 Comm: irq/277-aerdrv Tainted: G 
M    W           7.0.11+ #38 PREEMPT(lazy) 
a69a8d0b840e8538e5b93598a4d4312fb8da4861
[   40.946115] Tainted: [M]=MACHINE_CHECK, [W]=WARN
[   40.946116] Hardware name: Radxa Computer Co., Ltd. Radxa Dragon 
Q8B/RS782-D16S32W0, BIOS 6.0.260812.BOOT.MXF.1.1.c1-00167-MAKENA-1 
08/12/2026
[   40.946119] pstate: 204000c5 (nzCv daIF +PAN -UAO -TCO -DIT -SSBS 
BTYPE=--)
[   40.946121] pc : pci_generic_config_read+0x48/0xd0
[   40.946128] lr : pci_generic_config_read+0x2c/0xd0
[   40.946130] sp : ffff80008002bbb0
[   40.946132] x29: ffff80008002bbb0 x28: ffff000088fd9200 x27: 
ffffc4cc9599029f
[   40.946136] x26: ffffc4cc959902a0 x25: ffff000085a6b000 x24: 
ffffc4cc950bfd68
[   40.946138] x23: 0000000000000115 x22: 0000000000000100 x21: 
0000000000000000
[   40.946140] x20: 0000000000000004 x19: ffff80008002bbf4 x18: 
ffff3b3b60df4000
[   40.946142] x17: ffff3b3b60df4000 x16: ffff800080028000 x15: 
0000000000001900
[   40.946143] x14: 00000f423fffe700 x13: 0000000000000000 x12: 
000000000000af3b
[   40.946145] x11: 000000000000b67e x10: 00000000010b218c x9 : 
ffffc4cc937b28c4
[   40.946147] x8 : ffffc4cc937b2898 x7 : ffff00008b4a8800 x6 : 
ffffc4cc95ed5000
[   40.946148] x5 : ffff80008002bc30 x4 : ffff80008002bbf4 x3 : 
0000000000000000
[   40.946150] x2 : 0000000000000130 x1 : 0000000000000000 x0 : 
0000000000000000
[   40.946153] Kernel panic - not syncing: Asynchronous SError Interrupt
[   40.946156] CPU: 5 UID: 0 PID: 335 Comm: irq/277-aerdrv Tainted: G 
M    W           7.0.11+ #38 PREEMPT(lazy) 
a69a8d0b840e8538e5b93598a4d4312fb8da4861
[   40.946158] Tainted: [M]=MACHINE_CHECK, [W]=WARN
[   40.946159] Hardware name: Radxa Computer Co., Ltd. Radxa Dragon 
Q8B/RS782-D16S32W0, BIOS 6.0.260812.BOOT.MXF.1.1.c1-00167-MAKENA-1 
08/12/2026
[   40.946161] Call trace:
[   40.946162]  show_stack+0x20/0x38 (C)
[   40.946171]  dump_stack_lvl+0x7c/0xa0
[   40.946176]  dump_stack+0x18/0x2c
[   40.946178]  vpanic+0x2a8/0x4b8
[   40.946181]  panic+0x68/0x70
[   40.946182]  nmi_panic+0x74/0x80
[   40.946185]  arm64_serror_panic+0x78/0x90
[   40.946187]  arm64_is_fatal_ras_serror+0x90/0x98
[   40.946189]  do_serror+0x38/0x60
[   40.946191]  el1h_64_error_handler+0x40/0x70
[   40.946196]  el1h_64_error+0x84/0x88
[   40.946197]  pci_generic_config_read+0x48/0xd0 (P)
[   40.946198]  pci_bus_read_config_dword+0x80/0xf0
[   40.946199]  pci_read_config_dword+0x34/0x60
[   40.946201]  aer_irq+0x50/0xf8
[   40.946204]  __handle_irq_event_percpu+0x74/0x3d0
[   40.946208]  handle_irq_event+0x4c/0x120
[   40.946210]  handle_fasteoi_irq+0xfc/0x1b0
[   40.946213]  handle_irq_desc+0x3c/0x68
[   40.946215]  generic_handle_domain_irq+0x20/0x40
[   40.946217]  gic_handle_irq+0x168/0x300
[   40.946218]  do_interrupt_handler+0x58/0xa0
[   40.946219]  el1_interrupt+0x48/0xb0
[   40.946222]  el1h_64_irq_handler+0x18/0x28
[   40.946223]  el1h_64_irq+0x84/0x88
[   40.946225]  handle_softirqs+0xb0/0x4b0 (P)
[   40.946228]  __do_softirq+0x1c/0x28
[   40.946229]  ____do_softirq+0x18/0x30
[   40.946231]  call_on_irq_stack+0x30/0x48
[   40.946234]  do_softirq_own_stack+0x24/0x50
[   40.946236]  __irq_exit_rcu+0x13c/0x168
[   40.946238]  irq_exit_rcu+0x18/0x30
[   40.946240]  el1_interrupt+0x4c/0xb0
[   40.946242]  el1h_64_irq_handler+0x18/0x28
[   40.946244]  el1h_64_irq+0x84/0x88
[   40.946245]  _raw_spin_unlock_irqrestore+0x14/0x70 (P)
[   40.946248]  pci_read_config_dword+0x34/0x60
[   40.946249]  find_device_iter+0xbc/0x1e8
[   40.946251]  __pci_walk_bus+0x68/0xa0
[   40.946253]  __pci_walk_bus+0x4c/0xa0
[   40.946254]  __pci_walk_bus+0x4c/0xa0
[   40.946255]  pci_walk_bus+0x38/0x60
[   40.946257]  aer_isr_one_error_type+0x54/0x338
[   40.946259]  aer_isr_one_error+0x104/0x180
[   40.946261]  aer_isr+0x78/0x120
[   40.946263]  irq_thread_fn+0x30/0xb8
[   40.946264]  irq_thread+0x1a8/0x400
[   40.946265]  kthread+0x134/0x148
[   40.946268]  ret_from_fork+0x10/0x20
[   41.146728] Kernel Offset: 0x44cc12c80000 from 0xffff800080000000
[   41.146730] PHYS_OFFSET: 0xfff1000080000000
[   41.146731] CPU features: 0x2000000,002e0005,00230501,540172ab
[   41.146732] Memory Limit: none
[   41.565732] ---[ end Kernel panic - not syncing: Asynchronous SError 
Interrupt ]---

Case 2 log:

Reboot count: 65
Reboot in 10 seconds, press any key to abort... [   13.281893] qcom-apm 
gprsvc:service:2:1: CMD timeout for [1001021] opcode
[   13.340861] qcom-soundwire 3210000.soundwire: din-ports (0) mismatch 
with controller (1)
[   13.351390] qcom-soundwire 3210000.soundwire: dout-ports (5) mismatch 
with controller (6)
[   41.120391] systemd-shutdown[1]: Failed to set watchdog hardware 
timeout to 10min: Invalid argument
[   42.128807] pcieport 0004:00:00.0: PCIe Bus Error: 
severity=Uncorrectable (Non-Fatal), type=Transaction Layer, (Requester ID)
[   42.140459] pcieport 0004:00:00.0:   device [17cb:010e] error 
status/mask=00004000/00400000
[   42.149053] pcieport 0004:00:00.0:    [14] CmpltTO                (First)
[   42.302328] ------------[ cut here ]------------
[   42.307148] phy_check_link_status+0x0/0x118: returned: -16
[   42.312945] WARNING: drivers/net/phy/phy.c:1352 at 
_phy_state_machine+0x190/0x330, CPU#3: kworker/u32:0/12
[   42.322951] Modules linked in: q6prm_clocks q6apm_lpass_dais 
snd_q6dsp_common q6prm q6apm_dai joydev mousedev algif_hash 
algif_skcipher af_alg bnep zram 842_decompress 842_compress 
lz4hc_compress lz4_compress rtw89_8852be rtw89_8852b rtw89_8852b_common 
rtw89_pci rtw89_core btusb mac80211 btmtk qcom_iris btrtl btbcm cfg80211 
videobuf2_dma_contig btintel v4l2_mem2mem videobuf2_memops libarc4 
videobuf2_v4l2 snd_soc_wcd938x bluetooth snd_soc_wcd938x_sdw videodev 
snd_soc_wcd_classh rfkill snd_soc_wcd_common snd_soc_wcd_mbhc regmap_sdw 
snd_soc_sc8280xp videobuf2_common qcom_spmi_adc_tm5 snd_soc_qcom_common 
qcom_spmi_adc5 snd_soc_lpass_tx_macro snd_soc_qcom_sdw 
snd_soc_lpass_va_macro snd_soc_lpass_rx_macro mc qcom_vadc_common 
snd_q6apm soundwire_qcom snd_soc_lpass_macro_common soundwire_bus 
qcom_spmi_temp_alarm radxa_svc_glink fastrpc qcom_scm_storage slimbus sg 
platform_profile qcomtee qcom_rng libcomposite uinput 
pmic_pdcharger_ulog nfnetlink rpmsg_ctrl rpmsg_char qrtr_smd 
dwmac_tc956x stmmac qcom_pd_mapper pcs_xpcs
[   42.323138]  phylink
[   42.415567]  gpio_tc956x qcom_q6v5_pas qcom_pil_info rtc_ds1307 at24 
qcom_common qcom_glink_smem tc956x_pci uas qcom_pon camcc_sc8280xp 
pinctrl_sc8280xp_lpass_lpi videocc_sm8350 i2c_qcom_geni 
qcom_refgen_regulator pinctrl_lpass_lpi ucsi_glink lpasscc_sc8280xp 
qcom_q6v5 qcom_wdt qcom_sysmon typec_ucsi
[   42.445394] CPU: 3 UID: 0 PID: 12 Comm: kworker/u32:0 Not tainted 
7.0.11+ #37 PREEMPT(lazy)  452f60429f6b383145c5e2c98b90ac68d2bd4e3a
[   42.457814] Hardware name: Radxa Computer Co., Ltd. Radxa Dragon 
Q8B/RS782-D16S32W0, BIOS 6.0.260812.BOOT.MXF.1.1.c1-00167-MAKENA-1 
08/12/2026
[   42.471031] Workqueue: events_power_efficient phy_state_machine
[   42.477171] pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS 
BTYPE=--)
[   42.484381] pc : _phy_state_machine+0x190/0x330
[   42.489080] lr : _phy_state_machine+0x190/0x330
[   42.493790] sp : ffff8000800e3d20
[   42.497233] x29: ffff8000800e3d40 x28: 0000000000000000 x27: 
0000000000000000
[   42.504629] x26: ffff000080041028 x25: 0000000000000000 x24: 
ffff0000800c2a05
[   42.512024] x23: ffff000085e524f8 x22: ffff000080041000 x21: 
0000000000000000
[   42.519419] x20: 0000000000000006 x19: ffff000085e52000 x18: 
ffff59480c02f000
[   42.526813] x17: ffff59480c02f000 x16: ffffa6bfe88a8cc8 x15: 
ffffa6bfe9e43c70
[   42.534206] x14: ffffa6bfe9d0b480 x13: 0000000000000180 x12: 
ffff0007f5d3a480
[   42.541598] x11: 00000000000000c0 x10: 71e4c81c85622a03 x9 : 
ffffa6bfe7b42df4
[   42.548992] x8 : ffff0000808d47a8 x7 : 0000000000000004 x6 : 
0000000000000000
[   42.556387] x5 : ffff59480c02f000 x4 : ffff0000808d3600 x3 : 
ffff000081c9b600
[   42.563777] x2 : 0000000000000000 x1 : 0000000000000000 x0 : 
ffff0000808d3600
[   42.571174] Call trace:
[   42.573724]  _phy_state_machine+0x190/0x330 (P)
[   42.578438]  phy_state_machine+0x34/0xc0
[   42.582513]  process_one_work+0x194/0x570
[   42.586688]  worker_thread+0x18c/0x320
[   42.590592]  kthread+0x134/0x148
[   42.593953]  ret_from_fork+0x10/0x20
[   42.597680] ---[ end trace 0000000000000000 ]---
[   43.035576] SError Interrupt on CPU5, code 0x00000000be000011 -- SError
[   43.035582] CPU: 5 UID: 0 PID: 0 Comm: swapper/5 Tainted: G   M    W 
          7.0.11+ #37 PREEMPT(lazy) 
452f60429f6b383145c5e2c98b90ac68d2bd4e3a
[   43.035586] Tainted: [M]=MACHINE_CHECK, [W]=WARN
[   43.035587] Hardware name: Radxa Computer Co., Ltd. Radxa Dragon 
Q8B/RS782-D16S32W0, BIOS 6.0.260812.BOOT.MXF.1.1.c1-00167-MAKENA-1 
08/12/2026
[   43.035588] pstate: 204000c5 (nzCv daIF +PAN -UAO -TCO -DIT -SSBS 
BTYPE=--)
[   43.035590] pc : pci_generic_config_read+0x48/0xd0
[   43.035595] lr : pci_generic_config_read+0x2c/0xd0
[   43.035596] sp : ffff80008002be00
[   43.035597] x29: ffff80008002be00 x28: ffff0000808f4800 x27: 
ffffa6bfea71029f
[   43.035600] x26: ffffa6bfea7102a0 x25: ffff000085926e00 x24: 
ffffa6bfe9e3fd68
[   43.035601] x23: 0000000000000105 x22: 0000000000000020 x21: 
0000000000000000
[   43.035603] x20: 0000000000000004 x19: ffff80008002be44 x18: 
ffff59480c075000
[   43.035604] x17: ffff59480c075000 x16: ffff800080028000 x15: 
0000000000000000
[   43.035606] x14: ffff0000808f4800 x13: ffff59480c075000 x12: 
00000000b474d91d
[   43.035607] x11: 0000000000000000 x10: 0000000000001000 x9 : 
ffffa6bfe85328c4
[   43.035609] x8 : ffffa6bfe8532898 x7 : ffff0007f7eb4800 x6 : 
ffffa6bfeac55000
[   43.035610] x5 : ffff80008002bea4 x4 : ffff80008002be44 x3 : 
0000000000000000
[   43.035612] x2 : 0000000000000090 x1 : 0000000000000000 x0 : 
0000000000000000
[   43.035614] Kernel panic - not syncing: Asynchronous SError Interrupt
[   43.035615] CPU: 5 UID: 0 PID: 0 Comm: swapper/5 Tainted: G   M    W 
          7.0.11+ #37 PREEMPT(lazy) 
452f60429f6b383145c5e2c98b90ac68d2bd4e3a
[   43.035617] Tainted: [M]=MACHINE_CHECK, [W]=WARN
[   43.035618] Hardware name: Radxa Computer Co., Ltd. Radxa Dragon 
Q8B/RS782-D16S32W0, BIOS 6.0.260812.BOOT.MXF.1.1.c1-00167-MAKENA-1 
08/12/2026
[   43.035619] Call trace:
[   43.035621]  show_stack+0x20/0x38 (C)
[   43.035629]  dump_stack_lvl+0x7c/0xa0
[   43.035633]  dump_stack+0x18/0x2c
[   43.035635]  vpanic+0x2a8/0x4b8
[   43.035637]  panic+0x68/0x70
[   43.035639]  nmi_panic+0x74/0x80
[   43.035641]  arm64_serror_panic+0x78/0x90
[   43.035644]  arm64_is_fatal_ras_serror+0x90/0x98
[   43.035646]  do_serror+0x38/0x60
[   43.035648]  el1h_64_error_handler+0x40/0x70
[   43.035651]  el1h_64_error+0x84/0x88
[   43.035652]  pci_generic_config_read+0x48/0xd0 (P)
[   43.035653]  pci_bus_read_config_dword+0x80/0xf0
[   43.035656]  pcie_capability_read_dword+0x90/0xc0
[   43.035657]  pcie_pme_irq+0x58/0xf0
[   43.035660]  __handle_irq_event_percpu+0x74/0x3d0
[   43.035664]  handle_irq_event+0x4c/0x120
[   43.035666]  handle_fasteoi_irq+0xfc/0x1b0
[   43.035668]  handle_irq_desc+0x3c/0x68
[   43.035670]  generic_handle_domain_irq+0x20/0x40
[   43.035672]  gic_handle_irq+0x168/0x300
[   43.035673]  call_on_irq_stack+0x30/0x48
[   43.035675]  do_interrupt_handler+0x88/0xa0
[   43.035676]  el1_interrupt+0x48/0xb0
[   43.035679]  el1h_64_irq_handler+0x18/0x28
[   43.035680]  el1h_64_irq+0x84/0x88
[   43.035681]  cpuidle_enter_state+0xc4/0x7c0 (P)
[   43.035682]  cpuidle_enter+0x3c/0x60
[   43.035686]  do_idle+0x168/0x320
[   43.035688]  cpu_startup_entry+0x3c/0x50
[   43.035689]  secondary_start_kernel+0x13c/0x180
[   43.035691]  __secondary_switched+0xc0/0xc8
[   43.036060] SError Interrupt on CPU4, code 0x00000000be000411 -- SError
[   43.237107] Kernel Offset: 0x26bf67a00000 from 0xffff800080000000
[   43.237109] PHYS_OFFSET: 0xfff1000080000000
[   43.237109] CPU features: 0x2000000,002e0005,00230501,540172ab
[   43.237111] Memory Limit: none
[   43.595524] ---[ end Kernel panic - not syncing: Asynchronous SError 
Interrupt ]---

> 
> Konrad
> 


-- 
Best regards,
Xilin Wu <sophon@radxa.com>

Re: [PATCH v2] PCI: qcom: Block accesses to downstream devices on link down

[Krishna Chaitanya Chundru]

On 9/1/2026 7:36 PM, Xilin Wu wrote:
> On 9/1/2026 8:39 PM, Konrad Dybcio wrote:
>> On 9/1/26 2:35 PM, Manivannan Sadhasivam wrote:
>>>
>>> On Wed, 19 Aug 2026 23:36:54 -0700, Qiang Yu wrote:
>>>> After a PCIe link goes down, software may still access the BAR (MMIO)
>>>> space or configuration space of devices behind that link before recovery
>>>> has run. As the link is down, these accesses never complete, resulting in
>>>> a storm of Completion Timeout AERs.
>>>>
>>>> Use the controller's ECAM blocker to drop these accesses to the PCIe
>>>> address space as soon as the link-down interrupt fires, so that Completion
>>>> Timeout AERs are reduced. The blocked range covers the entire address
>>>> space (base 0x0, all-ones limit), since the Root Port's own DBI/iATU
>>>> register space remains accessible regardless.
>>>>
>>>> [...]
>>>
>>> Applied, thanks!
>>>
>>> [1/1] PCI: qcom: Block accesses to downstream devices on link down
>>>        commit: 86624b718f94517b26f96c1d7beb7632e5562afd
>>
>> Could this be a fix for
>>
>> https://lore.kernel.org/linux-arm-msm/b461c125-947b-4e06-b502-abb354e0cba6@oss.qualcomm.com/
>>
>>
>> ?
>>
>> (+Xilin)
>
> I'm not sure. Actually, we still have PCIe-related issues during reboot
> after reverting the flattened USB patch, though at a much lower occurrence
> rate.
>
> I hope this series fixes it, but I haven't tested it yet.
>
> https://patchwork.kernel.org/project/linux-arm-msm/list/?series=1151871&state=%2A&archive=both
>
>
> Case 1 log:
>
> Reboot count: 97
> Reboot in 10 seconds, press any key to abort... [   12.260389] qcom-apm
> gprsvc:service:2:1: CMD timeout for [1001021] opcode
> [   12.615536] qcom-soundwire 3210000.soundwire: din-ports (0) mismatch with
> controller (1)
> [   12.626249] qcom-soundwire 3210000.soundwire: dout-ports (5) mismatch
> with controller (6)
> [   39.120833] systemd-shutdown[1]: Failed to set watchdog hardware timeout
> to 10min: Invalid argument
> [   40.341886] ------------[ cut here ]------------
> [   40.346737] phy_check_link_status+0x0/0x118: returned: -16
> [   40.352528] WARNING: drivers/net/phy/phy.c:1352 at
> _phy_state_machine+0x190/0x330, CPU#1: kworker/u32:2/72
> [   40.362549] Modules linked in: q6prm_clocks q6apm_lpass_dais
> snd_q6dsp_common q6apm_dai q6prm joydev mousedev algif_hash algif_skcipher
> af_alg bnep rtw89_8852be rtw89_8852b rtw89_8852b_common rtw89_pci rtw89_core
> mac80211 qcom_iris cfg80211 videobuf2_dma_contig btusb v4l2_mem2mem btmtk
> videobuf2_memops btrtl btbcm videobuf2_v4l2 libarc4 btintel videodev
> snd_soc_wcd938x bluetooth snd_soc_wcd938x_sdw snd_soc_wcd_classh
> videobuf2_common snd_soc_wcd_common snd_soc_sc8280xp rfkill mc
> snd_soc_qcom_common snd_soc_wcd_mbhc qcom_spmi_adc_tm5 qcom_spmi_adc5
> snd_soc_qcom_sdw soundwire_qcom qcom_vadc_common regmap_sdw
> snd_soc_lpass_tx_macro soundwire_bus snd_soc_lpass_rx_macro
> qcom_spmi_temp_alarm snd_q6apm snd_soc_lpass_va_macro radxa_svc_glink
> qcomtee snd_soc_lpass_macro_common slimbus sg fastrpc qcom_scm_storage
> platform_profile qcom_rng libcomposite uinput pmic_pdcharger_ulog nfnetlink
> dwmac_tc956x stmmac gpio_tc956x pcs_xpcs phylink rtc_ds1307 at24 rpmsg_ctrl
> tc956x_pci qrtr_smd rpmsg_char qcom_pd_mapper uas qcom_pon
> [   40.362750]  i2c_qcom_geni videocc_sm8350 camcc_sc8280xp qcom_q6v5_pas
> qcom_pil_info qcom_common qcom_refgen_regulator pinctrl_sc8280xp_lpass_lpi
> ucsi_glink qcom_glink_smem pinctrl_lpass_lpi qcom_q6v5 lpasscc_sc8280xp
> typec_ucsi qcom_sysmon qcom_wdt [last unloaded: 842_decompress]
> [   40.480794] CPU: 1 UID: 0 PID: 72 Comm: kworker/u32:2 Not tainted 7.0.11+
> #38 PREEMPT(lazy)  a69a8d0b840e8538e5b93598a4d4312fb8da4861
> [   40.493213] Hardware name: Radxa Computer Co., Ltd. Radxa Dragon
> Q8B/RS782-D16S32W0, BIOS 6.0.260812.BOOT.MXF.1.1.c1-00167-MAKENA-1 08/12/2026
> [   40.506435] Workqueue: events_power_efficient phy_state_machine
> [   40.512577] pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
> [   40.519788] pc : _phy_state_machine+0x190/0x330
> [   40.524500] lr : _phy_state_machine+0x190/0x330
> [   40.529209] sp : ffff800080a6bd20
> [   40.532663] x29: ffff800080a6bd40 x28: 0000000000000000 x27:
> 0000000000000000
> [   40.540058] x26: ffff000080041028 x25: 0000000000000000 x24:
> ffff0000800c1605
> [   40.547453] x23: ffff0000855b9cf8 x22: ffff000080041000 x21:
> 0000000000000000
> [   40.554845] x20: 0000000000000006 x19: ffff0000855b9800 x18:
> ffff3b3b60d68000
> [   40.562236] x17: 000000040044ffff x16: ffffc4cc93b2a028 x15:
> ffffc4cc950c3c70
> [   40.569632] x14: ffffc4cc94f8b480 x13: 0000000000000180 x12:
> ffff0007f5cf3480
> [   40.577025] x11: 00000000000000c0 x10: 7ca63edf1413fc9b x9 :
> ffffc4cc92dc2df4
> [   40.584427] x8 : ffff000080c9eba8 x7 : 0000000000000004 x6 :
> 0000000000000000
> [   40.591829] x5 : ffff3b3b60d68000 x4 : ffff000080c9da00 x3 :
> ffff00008e41da00
> [   40.599223] x2 : 0000000000000000 x1 : 0000000000000000 x0 :
> ffff000080c9da00
> [   40.606625] Call trace:
> [   40.609174]  _phy_state_machine+0x190/0x330 (P)
> [   40.613878]  phy_state_machine+0x34/0xc0
> [   40.617953]  process_one_work+0x194/0x570
> [   40.622129]  worker_thread+0x18c/0x320
> [   40.626033]  kthread+0x134/0x148
> [   40.629399]  ret_from_fork+0x10/0x20
> [   40.633127] ---[ end trace 0000000000000000 ]---
> [   40.946101] SError Interrupt on CPU5, code 0x00000000be000011 -- SError
> [   40.946109] CPU: 5 UID: 0 PID: 335 Comm: irq/277-aerdrv Tainted: G M   
> W           7.0.11+ #38 PREEMPT(lazy) a69a8d0b840e8538e5b93598a4d4312fb8da4861
> [   40.946115] Tainted: [M]=MACHINE_CHECK, [W]=WARN
> [   40.946116] Hardware name: Radxa Computer Co., Ltd. Radxa Dragon
> Q8B/RS782-D16S32W0, BIOS 6.0.260812.BOOT.MXF.1.1.c1-00167-MAKENA-1 08/12/2026
> [   40.946119] pstate: 204000c5 (nzCv daIF +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
> [   40.946121] pc : pci_generic_config_read+0x48/0xd0
> [   40.946128] lr : pci_generic_config_read+0x2c/0xd0
> [   40.946130] sp : ffff80008002bbb0
> [   40.946132] x29: ffff80008002bbb0 x28: ffff000088fd9200 x27:
> ffffc4cc9599029f
> [   40.946136] x26: ffffc4cc959902a0 x25: ffff000085a6b000 x24:
> ffffc4cc950bfd68
> [   40.946138] x23: 0000000000000115 x22: 0000000000000100 x21:
> 0000000000000000
> [   40.946140] x20: 0000000000000004 x19: ffff80008002bbf4 x18:
> ffff3b3b60df4000
> [   40.946142] x17: ffff3b3b60df4000 x16: ffff800080028000 x15:
> 0000000000001900
> [   40.946143] x14: 00000f423fffe700 x13: 0000000000000000 x12:
> 000000000000af3b
> [   40.946145] x11: 000000000000b67e x10: 00000000010b218c x9 :
> ffffc4cc937b28c4
> [   40.946147] x8 : ffffc4cc937b2898 x7 : ffff00008b4a8800 x6 :
> ffffc4cc95ed5000
> [   40.946148] x5 : ffff80008002bc30 x4 : ffff80008002bbf4 x3 :
> 0000000000000000
> [   40.946150] x2 : 0000000000000130 x1 : 0000000000000000 x0 :
> 0000000000000000
> [   40.946153] Kernel panic - not syncing: Asynchronous SError Interrupt
> [   40.946156] CPU: 5 UID: 0 PID: 335 Comm: irq/277-aerdrv Tainted: G M   
> W           7.0.11+ #38 PREEMPT(lazy) a69a8d0b840e8538e5b93598a4d4312fb8da4861
> [   40.946158] Tainted: [M]=MACHINE_CHECK, [W]=WARN
> [   40.946159] Hardware name: Radxa Computer Co., Ltd. Radxa Dragon
> Q8B/RS782-D16S32W0, BIOS 6.0.260812.BOOT.MXF.1.1.c1-00167-MAKENA-1 08/12/2026
> [   40.946161] Call trace:
> [   40.946162]  show_stack+0x20/0x38 (C)
> [   40.946171]  dump_stack_lvl+0x7c/0xa0
> [   40.946176]  dump_stack+0x18/0x2c
> [   40.946178]  vpanic+0x2a8/0x4b8
> [   40.946181]  panic+0x68/0x70
> [   40.946182]  nmi_panic+0x74/0x80
> [   40.946185]  arm64_serror_panic+0x78/0x90
> [   40.946187]  arm64_is_fatal_ras_serror+0x90/0x98
> [   40.946189]  do_serror+0x38/0x60
> [   40.946191]  el1h_64_error_handler+0x40/0x70
> [   40.946196]  el1h_64_error+0x84/0x88
> [   40.946197]  pci_generic_config_read+0x48/0xd0 (P)
> [   40.946198]  pci_bus_read_config_dword+0x80/0xf0
> [   40.946199]  pci_read_config_dword+0x34/0x60
> [   40.946201]  aer_irq+0x50/0xf8
> [   40.946204]  __handle_irq_event_percpu+0x74/0x3d0
> [   40.946208]  handle_irq_event+0x4c/0x120
> [   40.946210]  handle_fasteoi_irq+0xfc/0x1b0
> [   40.946213]  handle_irq_desc+0x3c/0x68
> [   40.946215]  generic_handle_domain_irq+0x20/0x40
> [   40.946217]  gic_handle_irq+0x168/0x300
> [   40.946218]  do_interrupt_handler+0x58/0xa0
> [   40.946219]  el1_interrupt+0x48/0xb0
> [   40.946222]  el1h_64_irq_handler+0x18/0x28
> [   40.946223]  el1h_64_irq+0x84/0x88
> [   40.946225]  handle_softirqs+0xb0/0x4b0 (P)
> [   40.946228]  __do_softirq+0x1c/0x28
> [   40.946229]  ____do_softirq+0x18/0x30
> [   40.946231]  call_on_irq_stack+0x30/0x48
> [   40.946234]  do_softirq_own_stack+0x24/0x50
> [   40.946236]  __irq_exit_rcu+0x13c/0x168
> [   40.946238]  irq_exit_rcu+0x18/0x30
> [   40.946240]  el1_interrupt+0x4c/0xb0
> [   40.946242]  el1h_64_irq_handler+0x18/0x28
> [   40.946244]  el1h_64_irq+0x84/0x88
> [   40.946245]  _raw_spin_unlock_irqrestore+0x14/0x70 (P)
> [   40.946248]  pci_read_config_dword+0x34/0x60
> [   40.946249]  find_device_iter+0xbc/0x1e8
> [   40.946251]  __pci_walk_bus+0x68/0xa0
> [   40.946253]  __pci_walk_bus+0x4c/0xa0
> [   40.946254]  __pci_walk_bus+0x4c/0xa0
> [   40.946255]  pci_walk_bus+0x38/0x60
> [   40.946257]  aer_isr_one_error_type+0x54/0x338
> [   40.946259]  aer_isr_one_error+0x104/0x180
> [   40.946261]  aer_isr+0x78/0x120
> [   40.946263]  irq_thread_fn+0x30/0xb8
> [   40.946264]  irq_thread+0x1a8/0x400
> [   40.946265]  kthread+0x134/0x148
> [   40.946268]  ret_from_fork+0x10/0x20
> [   41.146728] Kernel Offset: 0x44cc12c80000 from 0xffff800080000000
> [   41.146730] PHYS_OFFSET: 0xfff1000080000000
> [   41.146731] CPU features: 0x2000000,002e0005,00230501,540172ab
> [   41.146732] Memory Limit: none
> [   41.565732] ---[ end Kernel panic - not syncing: Asynchronous SError
> Interrupt ]---
>
> Case 2 log:
>
> Reboot count: 65
> Reboot in 10 seconds, press any key to abort... [   13.281893] qcom-apm
> gprsvc:service:2:1: CMD timeout for [1001021] opcode
> [   13.340861] qcom-soundwire 3210000.soundwire: din-ports (0) mismatch with
> controller (1)
> [   13.351390] qcom-soundwire 3210000.soundwire: dout-ports (5) mismatch
> with controller (6)
> [   41.120391] systemd-shutdown[1]: Failed to set watchdog hardware timeout
> to 10min: Invalid argument
> [   42.128807] pcieport 0004:00:00.0: PCIe Bus Error: severity=Uncorrectable
> (Non-Fatal), type=Transaction Layer, (Requester ID)
> [   42.140459] pcieport 0004:00:00.0:   device [17cb:010e] error
> status/mask=00004000/00400000
> [   42.149053] pcieport 0004:00:00.0:    [14] CmpltTO                (First)
> [   42.302328] ------------[ cut here ]------------
> [   42.307148] phy_check_link_status+0x0/0x118: returned: -16
> [   42.312945] WARNING: drivers/net/phy/phy.c:1352 at
> _phy_state_machine+0x190/0x330, CPU#3: kworker/u32:0/12
> [   42.322951] Modules linked in: q6prm_clocks q6apm_lpass_dais
> snd_q6dsp_common q6prm q6apm_dai joydev mousedev algif_hash algif_skcipher
> af_alg bnep zram 842_decompress 842_compress lz4hc_compress lz4_compress
> rtw89_8852be rtw89_8852b rtw89_8852b_common rtw89_pci rtw89_core btusb
> mac80211 btmtk qcom_iris btrtl btbcm cfg80211 videobuf2_dma_contig btintel
> v4l2_mem2mem videobuf2_memops libarc4 videobuf2_v4l2 snd_soc_wcd938x
> bluetooth snd_soc_wcd938x_sdw videodev snd_soc_wcd_classh rfkill
> snd_soc_wcd_common snd_soc_wcd_mbhc regmap_sdw snd_soc_sc8280xp
> videobuf2_common qcom_spmi_adc_tm5 snd_soc_qcom_common qcom_spmi_adc5
> snd_soc_lpass_tx_macro snd_soc_qcom_sdw snd_soc_lpass_va_macro
> snd_soc_lpass_rx_macro mc qcom_vadc_common snd_q6apm soundwire_qcom
> snd_soc_lpass_macro_common soundwire_bus qcom_spmi_temp_alarm
> radxa_svc_glink fastrpc qcom_scm_storage slimbus sg platform_profile qcomtee
> qcom_rng libcomposite uinput pmic_pdcharger_ulog nfnetlink rpmsg_ctrl
> rpmsg_char qrtr_smd dwmac_tc956x stmmac qcom_pd_mapper pcs_xpcs
> [   42.323138]  phylink
> [   42.415567]  gpio_tc956x qcom_q6v5_pas qcom_pil_info rtc_ds1307 at24
> qcom_common qcom_glink_smem tc956x_pci uas qcom_pon camcc_sc8280xp
> pinctrl_sc8280xp_lpass_lpi videocc_sm8350 i2c_qcom_geni
> qcom_refgen_regulator pinctrl_lpass_lpi ucsi_glink lpasscc_sc8280xp
> qcom_q6v5 qcom_wdt qcom_sysmon typec_ucsi
> [   42.445394] CPU: 3 UID: 0 PID: 12 Comm: kworker/u32:0 Not tainted 7.0.11+
> #37 PREEMPT(lazy)  452f60429f6b383145c5e2c98b90ac68d2bd4e3a
> [   42.457814] Hardware name: Radxa Computer Co., Ltd. Radxa Dragon
> Q8B/RS782-D16S32W0, BIOS 6.0.260812.BOOT.MXF.1.1.c1-00167-MAKENA-1 08/12/2026
> [   42.471031] Workqueue: events_power_efficient phy_state_machine
> [   42.477171] pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
> [   42.484381] pc : _phy_state_machine+0x190/0x330
> [   42.489080] lr : _phy_state_machine+0x190/0x330
> [   42.493790] sp : ffff8000800e3d20
> [   42.497233] x29: ffff8000800e3d40 x28: 0000000000000000 x27:
> 0000000000000000
> [   42.504629] x26: ffff000080041028 x25: 0000000000000000 x24:
> ffff0000800c2a05
> [   42.512024] x23: ffff000085e524f8 x22: ffff000080041000 x21:
> 0000000000000000
> [   42.519419] x20: 0000000000000006 x19: ffff000085e52000 x18:
> ffff59480c02f000
> [   42.526813] x17: ffff59480c02f000 x16: ffffa6bfe88a8cc8 x15:
> ffffa6bfe9e43c70
> [   42.534206] x14: ffffa6bfe9d0b480 x13: 0000000000000180 x12:
> ffff0007f5d3a480
> [   42.541598] x11: 00000000000000c0 x10: 71e4c81c85622a03 x9 :
> ffffa6bfe7b42df4
> [   42.548992] x8 : ffff0000808d47a8 x7 : 0000000000000004 x6 :
> 0000000000000000
> [   42.556387] x5 : ffff59480c02f000 x4 : ffff0000808d3600 x3 :
> ffff000081c9b600
> [   42.563777] x2 : 0000000000000000 x1 : 0000000000000000 x0 :
> ffff0000808d3600
> [   42.571174] Call trace:
> [   42.573724]  _phy_state_machine+0x190/0x330 (P)
> [   42.578438]  phy_state_machine+0x34/0xc0
> [   42.582513]  process_one_work+0x194/0x570
> [   42.586688]  worker_thread+0x18c/0x320
> [   42.590592]  kthread+0x134/0x148
> [   42.593953]  ret_from_fork+0x10/0x20
> [   42.597680] ---[ end trace 0000000000000000 ]---
> [   43.035576] SError Interrupt on CPU5, code 0x00000000be000011 -- SError
> [   43.035582] CPU: 5 UID: 0 PID: 0 Comm: swapper/5 Tainted: G   M    W
>          7.0.11+ #37 PREEMPT(lazy) 452f60429f6b383145c5e2c98b90ac68d2bd4e3a
> [   43.035586] Tainted: [M]=MACHINE_CHECK, [W]=WARN
> [   43.035587] Hardware name: Radxa Computer Co., Ltd. Radxa Dragon
> Q8B/RS782-D16S32W0, BIOS 6.0.260812.BOOT.MXF.1.1.c1-00167-MAKENA-1 08/12/2026
> [   43.035588] pstate: 204000c5 (nzCv daIF +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
> [   43.035590] pc : pci_generic_config_read+0x48/0xd0
> [   43.035595] lr : pci_generic_config_read+0x2c/0xd0
> [   43.035596] sp : ffff80008002be00
> [   43.035597] x29: ffff80008002be00 x28: ffff0000808f4800 x27:
> ffffa6bfea71029f
> [   43.035600] x26: ffffa6bfea7102a0 x25: ffff000085926e00 x24:
> ffffa6bfe9e3fd68
> [   43.035601] x23: 0000000000000105 x22: 0000000000000020 x21:
> 0000000000000000
> [   43.035603] x20: 0000000000000004 x19: ffff80008002be44 x18:
> ffff59480c075000
> [   43.035604] x17: ffff59480c075000 x16: ffff800080028000 x15:
> 0000000000000000
> [   43.035606] x14: ffff0000808f4800 x13: ffff59480c075000 x12:
> 00000000b474d91d
> [   43.035607] x11: 0000000000000000 x10: 0000000000001000 x9 :
> ffffa6bfe85328c4
> [   43.035609] x8 : ffffa6bfe8532898 x7 : ffff0007f7eb4800 x6 :
> ffffa6bfeac55000
> [   43.035610] x5 : ffff80008002bea4 x4 : ffff80008002be44 x3 :
> 0000000000000000
> [   43.035612] x2 : 0000000000000090 x1 : 0000000000000000 x0 :
> 0000000000000000
> [   43.035614] Kernel panic - not syncing: Asynchronous SError Interrupt
> [   43.035615] CPU: 5 UID: 0 PID: 0 Comm: swapper/5 Tainted: G   M    W
>          7.0.11+ #37 PREEMPT(lazy) 452f60429f6b383145c5e2c98b90ac68d2bd4e3a
> [   43.035617] Tainted: [M]=MACHINE_CHECK, [W]=WARN
> [   43.035618] Hardware name: Radxa Computer Co., Ltd. Radxa Dragon
> Q8B/RS782-D16S32W0, BIOS 6.0.260812.BOOT.MXF.1.1.c1-00167-MAKENA-1 08/12/2026
> [   43.035619] Call trace:
> [   43.035621]  show_stack+0x20/0x38 (C)
> [   43.035629]  dump_stack_lvl+0x7c/0xa0
> [   43.035633]  dump_stack+0x18/0x2c
> [   43.035635]  vpanic+0x2a8/0x4b8
> [   43.035637]  panic+0x68/0x70
> [   43.035639]  nmi_panic+0x74/0x80
> [   43.035641]  arm64_serror_panic+0x78/0x90
> [   43.035644]  arm64_is_fatal_ras_serror+0x90/0x98
> [   43.035646]  do_serror+0x38/0x60
> [   43.035648]  el1h_64_error_handler+0x40/0x70
> [   43.035651]  el1h_64_error+0x84/0x88
> [   43.035652]  pci_generic_config_read+0x48/0xd0 (P)
> [   43.035653]  pci_bus_read_config_dword+0x80/0xf0
> [   43.035656]  pcie_capability_read_dword+0x90/0xc0
> [   43.035657]  pcie_pme_irq+0x58/0xf0
> [   43.035660]  __handle_irq_event_percpu+0x74/0x3d0
> [   43.035664]  handle_irq_event+0x4c/0x120
> [   43.035666]  handle_fasteoi_irq+0xfc/0x1b0
> [   43.035668]  handle_irq_desc+0x3c/0x68
> [   43.035670]  generic_handle_domain_irq+0x20/0x40
> [   43.035672]  gic_handle_irq+0x168/0x300
> [   43.035673]  call_on_irq_stack+0x30/0x48
> [   43.035675]  do_interrupt_handler+0x88/0xa0
> [   43.035676]  el1_interrupt+0x48/0xb0
> [   43.035679]  el1h_64_irq_handler+0x18/0x28
> [   43.035680]  el1h_64_irq+0x84/0x88
> [   43.035681]  cpuidle_enter_state+0xc4/0x7c0 (P)
> [   43.035682]  cpuidle_enter+0x3c/0x60
> [   43.035686]  do_idle+0x168/0x320
> [   43.035688]  cpu_startup_entry+0x3c/0x50
> [   43.035689]  secondary_start_kernel+0x13c/0x180
> [   43.035691]  __secondary_switched+0xc0/0xc8
> [   43.036060] SError Interrupt on CPU4, code 0x00000000be000411 -- SError
> [   43.237107] Kernel Offset: 0x26bf67a00000 from 0xffff800080000000
> [   43.237109] PHYS_OFFSET: 0xfff1000080000000
> [   43.237109] CPU features: 0x2000000,002e0005,00230501,540172ab
> [   43.237111] Memory Limit: none
> [   43.595524] ---[ end Kernel panic - not syncing: Asynchronous SError
> Interrupt ]--- 
I am suspecting this issue is due absent of pcie shutdown callback to turn off
the pcie link as part of system shutdown.
suspecting few resources are turned off as shutdown, causing this issue.

can you try with shutdown patch once [1]

[1] [PATCH v4 0/2] PCI: qcom: Implement shutdown() to avoid SMMU/NoC errors on
reboot - Krishna Chaitanya Chundru
<https://lore.kernel.org/linux-pci/20260826-shutdown-v4-0-eb5fe9d454ae@oss.qualcomm.com/>

- Krishna Chaitanya.
>
>>
>> Konrad
>>
>
>