Re: [PATCH 4/5] PCI/FLIT: Add trace event for flit errors

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Infinite loop in threaded IRQ handler due to missing check for PCI device removal/error.
--

--- Patch [4]: [PATCH 4/5] PCI/FLIT: Add trace event for flit errors ---
commit e2a5453df78634ff77f29016af766d1994cda605
Author: Yazen Ghannam <yazen.ghannam@amd.com>

PCI/FLIT: Add trace event for flit errors

This commit emits a trace event for each erroneous flit logged by the Flit
Error Counter. It captures the counter control/status and both Error Log
registers to provide a structured record for userspace RAS tooling.

> diff --git a/drivers/pci/pcie/flit.c b/drivers/pci/pcie/flit.c
> index 12f2b6fbf1de0..83cf53f392913 100644
> --- a/drivers/pci/pcie/flit.c
> +++ b/drivers/pci/pcie/flit.c

[ ... ]

> @@ -84,6 +85,7 @@ static irqreturn_t flit_isr(int irq, void *context)
>  
>  		pci_read_config_dword(pdev, flit + PCI_FLIT_ERR_LOG2, &err_log2);
>  		pci_info(pdev, HW_ERR "  Error Log1: 0x%08x Error Log2: 0x%08x\n", err_log1, err_log2);
> +		trace_flit_event(pci_name(pdev), cntr_ctrl, cntr_sta, err_log1, err_log2);
>  
>  		pci_write_config_dword(pdev, flit + PCI_FLIT_ERR_LOG1, err_log1);
>  	} while (err_log1 & PCI_FLIT_ERR_LOG_MORE);

[Severity: High]
If the PCI device is hot-unplugged or the link goes down while flit_isr()
is executing, pci_read_config_dword() will fail and set err_log1 to all 1s
(0xFFFFFFFF).

Because 0xFFFFFFFF has all bits set, it will bypass any valid bit checks and
the loop condition (err_log1 & PCI_FLIT_ERR_LOG_MORE) will evaluate to true.

Could this trap the kernel thread in an infinite loop and flood the trace
buffer?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260714183950.1620016-1-yazen.ghannam@amd.com?part=4