On 8/27/2021 7:28 AM, Xueming(Steven) Li wrote: > > >> -----Original Message----- >> From: Kevin Traynor <ktray...@redhat.com> >> Sent: Thursday, August 26, 2021 6:46 PM >> To: Ferruh Yigit <ferruh.yi...@intel.com>; Ben Magistro >> <konce...@gmail.com>; dev@dpdk.org; Beilei Xing <beilei.x...@intel.com>; >> Luca Boccassi <bl...@debian.org>; Christian Ehrhardt >> <christian.ehrha...@canonical.com>; Xueming(Steven) Li >> <xuemi...@nvidia.com> >> Cc: ben.magis...@trinitycyber.com; stefan.baran...@trinitycyber.com; Qi >> Zhang <qi.z.zh...@intel.com> >> Subject: Re: [dpdk-dev] [PATCH] driver: i40evf device initialization >> >> + Christian and Xueming >> >> On 26/08/2021 11:25, Ferruh Yigit wrote: >>> On 8/25/2021 8:45 PM, Ben Magistro wrote: >>>> The i40evf driver is not initializing the eth_dev attribute which can >>>> result in a nullptr dereference. Changes were modeled after the >>>> iavf_dev_init() per suggestion from the mailing list[1]. >>>> >>>> [1] https://mails.dpdk.org/archives/dev/2021-August/217251.html >>>> >>>> Signed-off-by: Ben Magistro <konce...@gmail.com> >>> >>> i40evf will be removed in this release. But I guess it helps for >>> stable releases to first merge the fixes and later removed it, not sure. >>> >>> @Luca, @Kevin, do you prefer this patch directly to stable repos, or >>> through the main repo? >> >> I'll leave to Luca/Xueming and Christian to say if they have a preference, >> but I'd guess either way is fine from stable view once it has >> fixes/stable tags or LTS patch prefix (it doesn't have any of these at >> present). > > Yes, any option will make it being noticed by LTS maintainer: > 1. patches accepted by main with "fix" in subject > 2. patches accepted by main with "cc: sta...@dpdk.org" in commit message > 3. patches backported to LTS, sent to stable maillist with LTS prefix, for > example "[20.11]" >
Thanks Xueming, But is there a preferences for this case? The i40evf will be removed from main repo, is it better 1- first apply the fix and remove the component from main (I assume fix still will be bacported to LTS in this case) or 2- remove the i40evf from main (without fix), apply the fix directly to the LTS. Thanks, ferruh >> >>> i40evf won't be tested in the main anyway, since it would be removed >>> before -rc1 testing, so it looks like there won't be any difference from >>> testing point of view. >>> >>> >>>> --- >>>> drivers/net/i40e/i40e_ethdev_vf.c | 8 ++++++-- >>>> 1 file changed, 6 insertions(+), 2 deletions(-) >>>> >>>> diff --git a/drivers/net/i40e/i40e_ethdev_vf.c >>>> b/drivers/net/i40e/i40e_ethdev_vf.c >>>> index 0cfe13b7b2..ccdce9a16a 100644 >>>> --- a/drivers/net/i40e/i40e_ethdev_vf.c >>>> +++ b/drivers/net/i40e/i40e_ethdev_vf.c >>>> @@ -1564,8 +1564,9 @@ i40evf_dev_alarm_handler(void *param) static >>>> int i40evf_dev_init(struct rte_eth_dev *eth_dev) { >>>> - struct i40e_hw *hw >>>> - = I40E_DEV_PRIVATE_TO_HW(eth_dev->data->dev_private); >>>> + struct i40e_adapter *adapter = >>>> + I40E_DEV_PRIVATE_TO_ADAPTER(eth_dev->data->dev_private); >>>> + struct i40e_hw *hw = I40E_DEV_PRIVATE_TO_HW(adapter); >>>> struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(eth_dev); >>>> >>>> PMD_INIT_FUNC_TRACE(); >>>> @@ -1596,11 +1597,14 @@ i40evf_dev_init(struct rte_eth_dev *eth_dev) >>>> hw->device_id = pci_dev->id.device_id; >>>> hw->subsystem_vendor_id = pci_dev->id.subsystem_vendor_id; >>>> hw->subsystem_device_id = pci_dev->id.subsystem_device_id; >>>> + hw->bus.bus_id = pci_dev->addr.bus; >>>> hw->bus.device = pci_dev->addr.devid; >>>> hw->bus.func = pci_dev->addr.function; >>>> hw->hw_addr = (void *)pci_dev->mem_resource[0].addr; >>>> hw->adapter_stopped = 1; >>>> hw->adapter_closed = 0; >>>> + hw->back = I40E_DEV_PRIVATE_TO_ADAPTER(eth_dev->data->dev_private); >>>> + adapter->eth_dev = eth_dev; >>>> >>>> if(i40evf_init_vf(eth_dev) != 0) { >>>> PMD_INIT_LOG(ERR, "Init vf failed"); >>>> >>> >