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: <20200915203643.GA1426328@bjorn-Precision-5520>
Date:   Tue, 15 Sep 2020 15:36:43 -0500
From:   Bjorn Helgaas <helgaas@...nel.org>
To:     Ming Qiao <mqiao@...iper.net>
Cc:     bhelgaas@...gle.com, linux-pci@...r.kernel.org,
        linux-kernel@...r.kernel.org, Debjit Ghosh <dghosh@...iper.net>,
        Santhanakrishnan Balraj <sbalraj@...iper.net>,
        Rajat Jain <rajatja@...gle.com>
Subject: Re: [PATCH 1/3] PCI: Add quirks for Juniper FPGAs to set class code

On Tue, Sep 15, 2020 at 08:11:01AM -0700, Ming Qiao wrote:
> Some of the Juniper FPGAs do not report correct PCI class ID, which
> would confuse kernel APIs accessing the specific class of devices.
> Change them to PCI_CLASS_SYSTEM_OTHER << 8.

Please include a note about the consequence of the incorrect class ID,
i.e., what happens without this quirk?  Does the system panic?  Does
some device not work correctly?  If so, which, and what does the
problem look like to a user?

"Confusing kernel APIs" is pretty general and won't help a user who is
seeing a problem to find this patch.

> Also introduce Juniper vendor ID to be used in the quirks.
>     
> Signed-off-by: Debjit Ghosh <dghosh@...iper.net>
> Signed-off-by: Santhanakrishnan Balraj <sbalraj@...iper.net>
> Signed-off-by: Rajat Jain <rajatja@...gle.com>
> Signed-off-by: Ming Qiao <mqiao@...iper.net>
> ---
>  drivers/pci/quirks.c    | 25 +++++++++++++++++++++++++
>  include/linux/pci_ids.h |  2 ++
>  2 files changed, 27 insertions(+)
> 
> diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
> index 2a589b6..61344d2 100644
> --- a/drivers/pci/quirks.c
> +++ b/drivers/pci/quirks.c
> @@ -5632,3 +5632,28 @@ static void apex_pci_fixup_class(struct pci_dev *pdev)
>  }
>  DECLARE_PCI_FIXUP_CLASS_HEADER(0x1ac1, 0x089a,
>  			       PCI_CLASS_NOT_DEFINED, 8, apex_pci_fixup_class);
> +
> +/*
> + * PCI class reported by some Juniper FPGAs is not correct.
> + * Change it to SYSTEM.
> + */
> +static void quirk_jnx_fpga(struct pci_dev *dev)
> +{
> +	if (!dmi_match(DMI_BOARD_VENDOR, "Juniper Networks Inc."))
> +		return;

Why is the DMI_BOARD_VENDOR relevant to this quirk?  This check seems
to mean that the class code is programmable by the BIOS, and all
BIOSes program it correctly *except* the Juniper BIOS.

If this is just a silicon defect in the chips, you shouldn't need to
check the DMI_BOARD_VENDOR.

> +	dev->class = PCI_CLASS_SYSTEM_OTHER << 8;
> +}
> +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_JUNIPER, 0x0004, quirk_jnx_fpga);
> +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_JUNIPER, 0x006A, quirk_jnx_fpga);
> +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_JUNIPER, 0x006B, quirk_jnx_fpga);
> +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_JUNIPER, 0x006C, quirk_jnx_fpga);
> +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_JUNIPER, 0x006E, quirk_jnx_fpga);
> +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_JUNIPER, 0x0079, quirk_jnx_fpga);
> +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_JUNIPER, 0x0083, quirk_jnx_fpga);
> +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_JUNIPER, 0x0071, quirk_jnx_fpga);
> +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_JUNIPER, 0x00A7, quirk_jnx_fpga);
> +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_JUNIPER, 0x00A8, quirk_jnx_fpga);
> +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_JUNIPER, 0x00A9, quirk_jnx_fpga);
> +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_JUNIPER, 0x00AA, quirk_jnx_fpga);
> +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_XILINX, 0x0505, quirk_jnx_fpga);
> diff --git a/include/linux/pci_ids.h b/include/linux/pci_ids.h
> index 1ab1e24..bfbf8f1 100644
> --- a/include/linux/pci_ids.h
> +++ b/include/linux/pci_ids.h
> @@ -1859,6 +1859,8 @@
>  #define PCI_VENDOR_ID_ESDGMBH		0x12fe
>  #define PCI_DEVICE_ID_ESDGMBH_CPCIASIO4 0x0111
>  
> +#define PCI_VENDOR_ID_JUNIPER		0X1304

Please use "0x1304" (lower-case 'x') like the rest of the file.

>  #define PCI_VENDOR_ID_CB		0x1307	/* Measurement Computing */
>  
>  #define PCI_VENDOR_ID_SIIG		0x131f
> -- 
> 2.10.0
> 

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ