linux-kernel - Re: [PATCH v7 5/5] PCI/AER: Only clear error bits in pcie_clear_device

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

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <aYGpgSztxNMZO-oC@wunner.de>
Date: Tue, 3 Feb 2026 08:53:37 +0100
From: Lukas Wunner <lukas@...ner.de>
To: Shuai Xue <xueshuai@...ux.alibaba.com>
Cc: linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org,
	linuxppc-dev@...ts.ozlabs.org, bhelgaas@...gle.com,
	kbusch@...nel.org, sathyanarayanan.kuppuswamy@...ux.intel.com,
	mahesh@...ux.ibm.com, oohall@...il.com, Jonathan.Cameron@...wei.com,
	terry.bowman@....com, tianruidong@...ux.alibaba.com
Subject: Re: [PATCH v7 5/5] PCI/AER: Only clear error bits in
 pcie_clear_device_status()

On Sat, Jan 24, 2026 at 03:45:57PM +0800, Shuai Xue wrote:
> +++ b/drivers/pci/pci.c
> @@ -2246,7 +2246,7 @@ void pcie_clear_device_status(struct pci_dev *dev)
>  	u16 sta;
>  
>  	pcie_capability_read_word(dev, PCI_EXP_DEVSTA, &sta);
> -	pcie_capability_write_word(dev, PCI_EXP_DEVSTA, sta);
> +	pcie_capability_write_word(dev, PCI_EXP_DEVSTA, sta & PCI_EXP_DEVSTA_ERR);
>  }

I don't think there's any harm to write error bits which are currently 0,
so I'd just get rid of the pcie_capability_read_word() and directly write
the error bits.

> +++ b/include/uapi/linux/pci_regs.h
> @@ -534,6 +534,7 @@
>  #define  PCI_EXP_DEVSTA_NFED	0x0002	/* Non-Fatal Error Detected */
>  #define  PCI_EXP_DEVSTA_FED	0x0004	/* Fatal Error Detected */
>  #define  PCI_EXP_DEVSTA_URD	0x0008	/* Unsupported Request Detected */
> +#define  PCI_EXP_DEVSTA_ERR	0xf	/* Error bits */

There's only one user of PCI_EXP_DEVSTA_ERR and it feels a little
awkward to define a macro in a uapi header which does not correspond
to an "official" bit definition but is just there for convenience.

So maybe it's better to simply use the macros for the four bits in
pcie_clear_device_status()?  Might also be slightly clearer.

This patch could be submitted individually instead of being part
of this series.

Thanks,

Lukas