Re: [PATCH 1/2] PCI: Fix the wrong reading of register fields

[Ilpo Järvinen]

On Fri, 10 Jan 2025, Jiwei Sun wrote:

> From: Jiwei Sun <sunjw10@xxxxxxxxxx>
> 
> Since commit de9a6c8d5dbf ("PCI/bwctrl: Add pcie_set_target_speed() to set
> PCIe Link Speed"), there are two potential issues in the function
> pcie_failed_link_retrain().
> 
> (1) The macro PCIE_LNKCTL2_TLS2SPEED() and PCIE_LNKCAP_SLS2SPEED() just
> uses the link speed field of the registers. However, there are many other
> different function fields in the Link Control 2 Register or the Link
> Capabilities Register. If the register value is directly used by the two
> macros, it may cause getting an error link speed value (PCI_SPEED_UNKNOWN).
>
> (2) In the pcie_failed_link_retrain(), the local variable lnkctl2 is not
> changed after reading from PCI_EXP_LNKCTL2. It might cause that the
> removing 2.5GT/s downstream link speed restriction codes are not executed.

Thanks for finding these issues and coming up with a patch.

These cases seem two different problems to me so I'd put them into 
different patches, which would also make it easeir focus on 
describing the issue in case 2 which is currently a bit vague (when 
looking how de9a6c8d5dbf managed to break it by removing the lnkctl2 
change prior to writing into the reg).

> In order to avoid the above-mentioned potential issues, only keep link
> speed field of the two registers before using by pcie_set_target_speed()
> and reread the Link Control 2 Register before using.
> 
> Fixes: de9a6c8d5dbf ("PCI/bwctrl: Add pcie_set_target_speed() to set PCIe Link Speed")
> Signed-off-by: Jiwei Sun <sunjw10@xxxxxxxxxx>
> ---
>  drivers/pci/quirks.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
> index 76f4df75b08a..605628c810a5 100644
> --- a/drivers/pci/quirks.c
> +++ b/drivers/pci/quirks.c
> @@ -118,11 +118,13 @@ int pcie_failed_link_retrain(struct pci_dev *dev)
>  		ret = pcie_set_target_speed(dev, PCIE_SPEED_2_5GT, false);
>  		if (ret) {
>  			pci_info(dev, "retraining failed\n");
> +			oldlnkctl2 &= PCI_EXP_LNKCTL2_TLS;
>  			pcie_set_target_speed(dev, PCIE_LNKCTL2_TLS2SPEED(oldlnkctl2),
>  					      true);

I'd prefer these get fixed inside the macros so that the callers don't 
need to take handle what seem something that is always needed.

>  			return ret;
>  		}
>  
> +		pcie_capability_read_word(dev, PCI_EXP_LNKCTL2, &lnkctl2);
>  		pcie_capability_read_word(dev, PCI_EXP_LNKSTA, &lnksta);
>  	}
>  
> @@ -133,6 +135,7 @@ int pcie_failed_link_retrain(struct pci_dev *dev)
>  
>  		pci_info(dev, "removing 2.5GT/s downstream link speed restriction\n");
>  		pcie_capability_read_dword(dev, PCI_EXP_LNKCAP, &lnkcap);
> +		lnkcap &= PCI_EXP_LNKCAP_SLS;
>  		ret = pcie_set_target_speed(dev, PCIE_LNKCAP_SLS2SPEED(lnkcap), false);

-- 
 i.