[v4] bus/pci/windows: support for PCI scan allowed and blocked lists
Checks
Commit Message
EAL -a and -b options are used to specify which PCI devices are
explicitly allowed or blocked during PCI bus scan. This evaluation
is missing in the Windows implementation of rte_pci_scan.
This patch provides this missing functionality, so that apps can specify
which NetUIO devices to ignore during PCI bus scan.
Signed-off-by: Khoa To <khot@linux.microsoft.com>
---
v4:
* Fix coding style warning with unaligned comments
v3:
* Move the check inside pci_scan_one
* Small change to pci_scan_one to malloc only after checks succeeded
v2:
* Truncate commit description lines to 75 charaters or less
drivers/bus/pci/windows/pci.c | 23 ++++++++++++++++-------
1 file changed, 16 insertions(+), 7 deletions(-)
Comments
On 2/28/2021 10:52 PM, Khoa To wrote:
> EAL -a and -b options are used to specify which PCI devices are
> explicitly allowed or blocked during PCI bus scan. This evaluation
> is missing in the Windows implementation of rte_pci_scan.
>
> This patch provides this missing functionality, so that apps can specify
> which NetUIO devices to ignore during PCI bus scan.
>
> Signed-off-by: Khoa To <khot@linux.microsoft.com>
> ---
> v4:
> * Fix coding style warning with unaligned comments
> v3:
> * Move the check inside pci_scan_one
> * Small change to pci_scan_one to malloc only after checks succeeded
> v2:
> * Truncate commit description lines to 75 charaters or less
>
> drivers/bus/pci/windows/pci.c | 23 ++++++++++++++++-------
> 1 file changed, 16 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/bus/pci/windows/pci.c b/drivers/bus/pci/windows/pci.c
> index f66258452..0d184f24d 100644
> --- a/drivers/bus/pci/windows/pci.c
> +++ b/drivers/bus/pci/windows/pci.c
> @@ -308,17 +308,24 @@ set_kernel_driver_type(PSP_DEVINFO_DATA device_info_data,
> static int
> pci_scan_one(HDEVINFO dev_info, PSP_DEVINFO_DATA device_info_data)
> {
> - struct rte_pci_device *dev;
> + struct rte_pci_device *dev = NULL;
> int ret = -1;
> char pci_device_info[REGSTR_VAL_MAX_HCID_LEN];
> struct rte_pci_addr addr;
> struct rte_pci_id pci_id;
>
> - dev = malloc(sizeof(*dev));
> - if (dev == NULL)
> + ret = get_device_pci_address(dev_info, device_info_data, &addr);
> + if (ret != 0)
> goto end;
>
> - memset(dev, 0, sizeof(*dev));
> + if (rte_pci_ignore_device(&addr)) {
> + /*
> + * We won't add this device, but we want to continue
> + * looking for supported devices
> + */
> + ret = ERROR_CONTINUE;
> + goto end;
> + }
>
> ret = get_pci_hardware_id(dev_info, device_info_data,
> pci_device_info, sizeof(pci_device_info));
> @@ -335,10 +342,12 @@ pci_scan_one(HDEVINFO dev_info, PSP_DEVINFO_DATA device_info_data)
> goto end;
> }
>
> - ret = get_device_pci_address(dev_info, device_info_data, &addr);
> - if (ret != 0)
> + dev = malloc(sizeof(*dev));
> + if (dev == NULL)
> goto end;
>
> + memset(dev, 0, sizeof(*dev));
> +
> dev->addr = addr;
> dev->id = pci_id;
> dev->max_vfs = 0; /* TODO: get max_vfs */
> @@ -417,7 +426,7 @@ rte_pci_scan(void)
> device_index++;
> /* we only want to enumerate net & netuio class devices */
> if (IsEqualGUID(&(device_info_data.ClassGuid),
> - &GUID_DEVCLASS_NET) ||
> + &GUID_DEVCLASS_NET) ||
> IsEqualGUID(&(device_info_data.ClassGuid),
> &GUID_DEVCLASS_NETUIO)) {
> ret = pci_scan_one(dev_info, &device_info_data);
This last change appears to be a formatting change. Was it required by
checkpatch?
Other than that...
Acked-by: Ranjit Menon <ranjit.menon@intel.com>
On Wed, 3 Mar 2021, Ranjit Menon wrote:
> On 2/28/2021 10:52 PM, Khoa To wrote:
>> EAL -a and -b options are used to specify which PCI devices are
>> explicitly allowed or blocked during PCI bus scan. This evaluation
>> is missing in the Windows implementation of rte_pci_scan.
>>
>> This patch provides this missing functionality, so that apps can specify
>> which NetUIO devices to ignore during PCI bus scan.
>>
>> Signed-off-by: Khoa To <khot@linux.microsoft.com>
>> ---
>> v4:
>> * Fix coding style warning with unaligned comments
>> v3:
>> * Move the check inside pci_scan_one
>> * Small change to pci_scan_one to malloc only after checks succeeded
>> v2:
>> * Truncate commit description lines to 75 charaters or less
>>
>> drivers/bus/pci/windows/pci.c | 23 ++++++++++++++++-------
>> 1 file changed, 16 insertions(+), 7 deletions(-)
>>
>
> This last change appears to be a formatting change. Was it required by
> checkpatch?
>
> Other than that...
>
> Acked-by: Ranjit Menon <ranjit.menon@intel.com>
>
Thanks, Ranjit. Yes, v4 was just some white-space fixes for a CI warning.
2021-02-28 22:52, Khoa To:
> EAL -a and -b options are used to specify which PCI devices are
> explicitly allowed or blocked during PCI bus scan. This evaluation
> is missing in the Windows implementation of rte_pci_scan.
>
> This patch provides this missing functionality, so that apps can specify
> which NetUIO devices to ignore during PCI bus scan.
>
> Signed-off-by: Khoa To <khot@linux.microsoft.com>
> ---
> v4:
> * Fix coding style warning with unaligned comments
> v3:
> * Move the check inside pci_scan_one
> * Small change to pci_scan_one to malloc only after checks succeeded
> v2:
> * Truncate commit description lines to 75 charaters or less
>
> drivers/bus/pci/windows/pci.c | 23 ++++++++++++++++-------
> 1 file changed, 16 insertions(+), 7 deletions(-)
Acked-by: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
04/03/2021 08:14, Dmitry Kozlyuk:
> 2021-02-28 22:52, Khoa To:
> > EAL -a and -b options are used to specify which PCI devices are
> > explicitly allowed or blocked during PCI bus scan. This evaluation
> > is missing in the Windows implementation of rte_pci_scan.
> >
> > This patch provides this missing functionality, so that apps can specify
> > which NetUIO devices to ignore during PCI bus scan.
not-only NetUIO
> >
> > Signed-off-by: Khoa To <khot@linux.microsoft.com>
>
> Acked-by: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
Applied with small rewording, thanks.
@@ -308,17 +308,24 @@ set_kernel_driver_type(PSP_DEVINFO_DATA device_info_data,
static int
pci_scan_one(HDEVINFO dev_info, PSP_DEVINFO_DATA device_info_data)
{
- struct rte_pci_device *dev;
+ struct rte_pci_device *dev = NULL;
int ret = -1;
char pci_device_info[REGSTR_VAL_MAX_HCID_LEN];
struct rte_pci_addr addr;
struct rte_pci_id pci_id;
- dev = malloc(sizeof(*dev));
- if (dev == NULL)
+ ret = get_device_pci_address(dev_info, device_info_data, &addr);
+ if (ret != 0)
goto end;
- memset(dev, 0, sizeof(*dev));
+ if (rte_pci_ignore_device(&addr)) {
+ /*
+ * We won't add this device, but we want to continue
+ * looking for supported devices
+ */
+ ret = ERROR_CONTINUE;
+ goto end;
+ }
ret = get_pci_hardware_id(dev_info, device_info_data,
pci_device_info, sizeof(pci_device_info));
@@ -335,10 +342,12 @@ pci_scan_one(HDEVINFO dev_info, PSP_DEVINFO_DATA device_info_data)
goto end;
}
- ret = get_device_pci_address(dev_info, device_info_data, &addr);
- if (ret != 0)
+ dev = malloc(sizeof(*dev));
+ if (dev == NULL)
goto end;
+ memset(dev, 0, sizeof(*dev));
+
dev->addr = addr;
dev->id = pci_id;
dev->max_vfs = 0; /* TODO: get max_vfs */
@@ -417,7 +426,7 @@ rte_pci_scan(void)
device_index++;
/* we only want to enumerate net & netuio class devices */
if (IsEqualGUID(&(device_info_data.ClassGuid),
- &GUID_DEVCLASS_NET) ||
+ &GUID_DEVCLASS_NET) ||
IsEqualGUID(&(device_info_data.ClassGuid),
&GUID_DEVCLASS_NETUIO)) {
ret = pci_scan_one(dev_info, &device_info_data);